Post by: Adam Schaible on April 27, 2010, 01:28:09 PM
I posted in the SMAART forum and didn't get much help.
My goal is to phase align my tops and subs, but before that I need to time align them. I'm fairly green with SMAART - I've read the quick start guide and know my way around the software, but I'm not able to configure my measurements correctly, and/or I'm having trouble configuring my hardware.
I am aligning KF650z's to SRX 728's. I'm going to make a preset stacked, but I'd like to be able to repeat this process on site for my larger gigs when I change my sub placement.
My signal chain is as follows (ignoring SMAART):
iPod L&R into A&H iLive, L&R outs into a UX8800 (80hz crossover), right now I'm doing mono - UX8800 is taking both A&B for the outputs, into an I-Tech 5000HD, then out to the speaker. The iLive is setup as L-R-Sub and the sub output is going straight into another i-tech HD and I'm using the built in processing to cross the subs.
I have an M-Audio profire 2626 interface, with a DBX reference mic. I'm sending both channels of the iPod input into aux 1 and sending aux 1 out into the reference channel of the interface. Smaart see's both of them.
I'm not sure, but I don't believe there is any more or less delay in the aux bus than there is in the master, so I think this should work - however should as I understand reference should PROBABLY come out of the L out, and I should just use the L channel for this?
Advice would be greatly appreciated!
Post by: Rob Spence on April 27, 2010, 01:41:55 PM
Actually, once you have done one side, you are really done if the 2 stacks are the same.
Post by: Mac Kerr on April 27, 2010, 02:03:38 PM
| Adam Schaible wrote on Tue, 27 April 2010 13:28 |
My signal chain is as follows (ignoring SMAART): iPod L&R into A&H iLive, L&R outs into a UX8800 (80hz crossover), right now I'm doing mono - UX8800 is taking both A&B for the outputs, into an I-Tech 5000HD, then out to the speaker. The iLive is setup as L-R-Sub and the sub output is going straight into another i-tech HD and I'm using the built in processing to cross the subs. I have an M-Audio profire 2626 interface, with a DBX reference mic. I'm sending both channels of the iPod input into aux 1 and sending aux 1 out into the reference channel of the interface. Smaart see's both of them. |
Use the iPod only for listening to music to pass judgement on the changes you make. Use pink noise from the console for measuring. If Aux 1 is your reference send, send pink noise to that, and to the subs. When you have a stable phase trace take a snapshot by hitting the space bar. Now send noise to the 650s and ref, but not the subs. Look at the phase trace at the crossover frequency between the 650s and the subs. You want the phase trace from the mains and the subs to lay on top of one another at the crossover frequency. Without changing anything else, still with a live view of the mains, add delay to the mains. As you add delay the phase trace should move toward the saved trace from the subs. If it moves away, start over with the mains as the saved trace and delay the subs.
Mac
Post by: Adam Schaible on April 27, 2010, 03:07:30 PM
I was using the iPod as the pink source but that makes sense to just use the console. I'm using SMAART 7 - is there a particular combination of settings I should be using to collect this data? It seems to wrap a LOT - like 7+ times, and although the slope is constant, it seems like my measurement configuration is suspect.
Post by: Mac Kerr on April 27, 2010, 03:15:55 PM
| Adam Schaible wrote on Tue, 27 April 2010 15:07 |
Ok thanks Mac, I was using the iPod as the pink source but that makes sense to just use the console. I'm using SMAART 7 - is there a particular combination of settings I should be using to collect this data? It seems to wrap a LOT - like 7+ times, and although the slope is constant, it seems like my measurement configuration is suspect. |
Since you are only interested in the area around the crossover frequency between the sub and the mains, zoom in so you only see the range from say, 50Hz to 250Hz. This will get rid of a lot of the distracting wraps.
Mac
Post by: Adam Schaible on April 27, 2010, 03:27:17 PM
Post by: Art Welter on April 27, 2010, 03:36:44 PM
| Adam Schaible wrote on Tue, 27 April 2010 13:07 |
Ok thanks Mac, I was using the iPod as the pink source but that makes sense to just use the console. I'm using SMAART 7 - is there a particular combination of settings I should be using to collect this data? It seems to wrap a LOT - like 7+ times, and although the slope is constant, it seems like my measurement configuration is suspect. |
You must use Auto sm to measure the delay, then insert the delay time, otherwise the phase trace will not be correct, which can easily cause multiple wraps in the HF.
A large phase discontinuity around the crossover point may indicate a polarity reversal. Polarity should be checked prior to adding delay to the LF or MF.
Post by: Adam Schaible on April 27, 2010, 03:41:38 PM
When running pink noise through the subs, the results of the find delay feature vary extremely (like 300ms). When I do it with the tops it is pretty accurate, but comes up with over 16ms which seems way higher than I expect. I anticipate 1-2 for the desk, and maybe up to 5ms for the ux8800.
Post by: Adam Schaible on April 27, 2010, 03:50:54 PM
http://www.rationalacoustics.com/files/v7GS_Ch4_Transfer_Fun ction.pdf and read the v6 manual.. That's what i was doing, but perhaps the issue was that the pink noise was coming from my iPod and not the console. I don't really think that's the problem but it's something I will try?
Post by: Mac Kerr on April 27, 2010, 03:55:42 PM
| Adam Schaible wrote on Tue, 27 April 2010 15:27 |
My assumption was that the phase trace on my subs should be flat after the HPF and before the LPF - is that assumption incorrect? |
Yes, that is incorrect.
Mac
Post by: Mac Kerr on April 27, 2010, 04:03:43 PM
| Adam Schaible wrote on Tue, 27 April 2010 15:41 |
Is this equivelent to the "delay finder" in smaart 7? When running pink noise through the subs, the results of the find delay feature vary extremely (like 300ms). When I do it with the tops it is pretty accurate, but comes up with over 16ms which seems way higher than I expect. I anticipate 1-2 for the desk, and maybe up to 5ms for the ux8800. |
The delay finder, and inserting that delay into Smaart (or any FFT) is to compensate primarily for the time of flight from the speaker to the microphone. With digital gear that also adds latency you will compensate for that as well.
It is important in this measurement to not move the mic between measurements. It must be in the same place for the subs and the mains. As I said before, do not reset the delay after you have captured the phase trace of the subs.
Often when Smaart returns an obviously wrong delay time it is because the ref and measurement signals are reversed.
Mac
Post by: Art Welter on April 27, 2010, 04:12:18 PM
| Adam Schaible wrote on Tue, 27 April 2010 13:50 |
I just checked this PDF: http://www.rationalacoustics.com/files/v7GS_Ch4_Transfer_Fun ction.pdf and read the v6 manual.. That's what i was doing, but perhaps the issue was that the pink noise was coming from my iPod and not the console. I don't really think that's the problem but it's something I will try? |
To eliminate as many variables as possible, I'd suggest using the noise generated from Smaart.
Test outside if possible, you only need enough level to get good coherency.
As you progress, you will find corrective EQ adjustments will smooth the phase response, but you won't eliminate the "drooping" phase response with conventional filters.
More than 720 degrees of phase rotation from 40 Hz to 18kHz would be common.
Post by: Adam Schaible on April 27, 2010, 04:14:10 PM
Post by: Adam Schaible on April 27, 2010, 04:18:10 PM
| Art Welter wrote on Tue, 27 April 2010 15:12 | ||
To eliminate as many variables as possible, I'd suggest using the noise generated from Smaart. Test outside if possible, you only need enough level to get good coherency. As you progress, you will find corrective EQ adjustments will smooth the phase response, but you won't eliminate the "drooping" phase response with conventional filters. More than 720 degrees of phase rotation from 40 Hz to 18kHz would be common. |
I will try this .. I tried before but for some reason it wasn't making any noise!
Thanks again Art.
Post by: Mac Kerr on April 27, 2010, 04:19:10 PM
| Adam Schaible wrote on Tue, 27 April 2010 16:14 |
I was attempting to do this in my basement, and then once I had learned the technique well enough, get the gear outside and do a better job. Could/would the boundaries create that much of an issue? I don't mind lugging it all outside, but I'd rather learn how to measure inside, and maximize my time outside. |
The close reflections will lower the coherence, but you should still be able to get measurements that work. As Art said, you don't need a lot of level. When you are doing more precise work with system eq you will probably want to be outdoors, or in a large room, like a real venue.
How far away from your system is the measurement mic?
Mac
Post by: Mac Kerr on April 27, 2010, 04:22:01 PM
| Adam Schaible wrote on Tue, 27 April 2010 16:18 | ||||
I will try this .. I tried before but for some reason it wasn't making any noise! Thanks again Art. |
Either noise source should be fine, but I disagree here with Art on which eliminates the most variables. I find running the noise from the console and being able to route it to both the ref in of Smaart and the system under test from one point easier. YMMV.
Mac
Post by: Adam Schaible on April 27, 2010, 04:24:36 PM
| Mac Kerr wrote on Tue, 27 April 2010 15:19 | ||
The close reflections will lower the coherence, but you should still be able to get measurements that work. As Art said, you don't need a lot of level. When you are doing more precise work with system eq you will probably want to be outdoors, or in a large room, like a real venue. How far away from your system is the measurement mic? Mac |
I had it about 5 ft from the cabs, about 2 feet off of the ground. I didn't have it cranked, but probably like 95dbish at that distance. Would it be better to turn down some?
Post by: Mac Kerr on April 27, 2010, 04:29:25 PM
| Adam Schaible wrote on Tue, 27 April 2010 16:24 | ||||
I had it about 5 ft from the cabs, about 2 feet off of the ground. I didn't have it cranked, but probably like 95dbish at that distance. Would it be better to turn down some? |
The easy answer is try it. Turning down and shooting another delay time shouldn't take long. I rarely touch the number of averages, maybe someone with more experience with different averages will chime in with help getting a good delay time on a sub in your conditions.
Mac
Post by: Art Welter on April 27, 2010, 04:35:17 PM
I would really suggest going outside, all the LF room reflections in a small room make it harder to understand what is going on.
Post by: Evan Kirkendall on April 27, 2010, 05:20:14 PM
Measurements were made 10' away from the cab in the ground plane. The pink trace is the combined sub and low of the system once they've been aligned.
Evan
Post by: Phillip_Graham on April 28, 2010, 09:29:14 AM
| Evan Kirkendall wrote on Tue, 27 April 2010 17:20 |
You want your phase trace to look similar to the pink trace in this shot: Measurements were made 10' away from the cab in the ground plane. The pink trace is the combined sub and low of the system once they've been aligned. Evan |
Adam,
Not to confuse you further, but Evan's posted way of doing this and what Mac is describing are not using the phase information in the same way, though the end result is the same. The difference is why Mac's phase traces will not be flat near the XO point, but Evan's phase traces will be.
Mac is instructing you to pick a fixed reference delay time in Smaart (by using AutoSM/Delay Finder) and then adjust the top delay till the slope of the sub phase and the top phase are the same in the crossover region.
Evan's way is manually adjusting the reference delay till you have zero apparent group delay at the desired XO point. You do this for each of the boxes, and then subtract the adjusted reference delay times from each other to find out the relative delay difference between the boxes.
In the specific example Evan posted, Evan appears to have done:
1. Adjust the SMAART reference delay time for the sub (i.e. Blue) trace until it looks like the blue trace on the screen. Typically starting with an AutoSM and then adding additional delay.
2. Shoot the top trace (i.e. Orange) using the current (i.e. Blue) reference delay time.
The next step would then be:
3. Adjust the Smaart reference delay until the Orange curve lays on top of the Blue curve.
4. Compare the reference delay of the Blue curve to the Orange, and apply the necessary differential delay to the appropriate box. Usually that is the top box, but not always.
The final result of this, using the original Blue reference delay time, is the pink curve displayed.
Both this method and Mac's method achieve the same final result (i.e. phase alignment at the XO). Mac's method is easier to explain, but can be more difficult to perform.
I personally find the method Evan shows easier to do in difficult acoustic environments, and that is the method I teach new users first. Also, in a class setting, the method Evan is doing allows me to expound a deeper conceptual understanding of what is really going on.
Since I taught Evan this method, he can expound on whether he felt that it helped give him a deeper understanding of what the phase trace is really saying.
Post by: Adam Schaible on April 28, 2010, 10:33:59 AM
Post by: Tim McCulloch on April 28, 2010, 11:16:58 AM
Re-read Phill's post. Using delay within SMAART to flatten the phase trace in the frequency range of interest is manipulating the *display* to make it easier to line up. It's not how flat the trace is but how well the traces from 2 adjacent pass-bands overlay.
Harry teaches both methods in his class, and folks usually grok 1 method first and then slowly figure out the other. Don't freak out if you don't "get it" the first time. Keep practicing, making bad measurements, etc until you get it. It will happen.
Have fun, good luck.
Tim Mc
Post by: Adam Schaible on April 28, 2010, 11:25:43 AM
And thanks again Tim et all.
Post by: Mac Kerr on April 28, 2010, 11:33:29 AM
| Adam Schaible wrote on Wed, 28 April 2010 11:25 |
Well, I might be more confused now than prior to re-reading Phillip's post. I really don't see the orange laying on the blue at all? I understand that everything above 200-250hz is not relevant in this situation and I do understand the method -- you're just adding the delay to SMAART rather than your DSP until it lines up, and then transferring that delay to the DSP -- or am I off base? My problem is I seem to be interpreting the measurement graph incorrectly. When I look between 63hz and 125hz the orange line (tops) has a negative slope and the blue line has a positive slop (subs). I thought that would indicate a polarity issue - but clearly the magnitude trace shows pretty good summation. And thanks again Tim et all. |
The orange trace doesn't line up because it is the pre adjustment trace. The pink trace is after adjustment. I still think my way is simpler (I think I learned it from 606, but I could be wrong). Remember that you are looking for the difference between the 2 delay times. The time you end up with in Smaart is not the delay time for the tops.
Mac
Post by: Adam Schaible on April 28, 2010, 11:37:37 AM
I agree, that method is simpler -- I understand you're looking for the difference between the initial Auto SM delay and the new delay to get them to lay on top of each other.
However, mine looked similar - but my reaction was to invert the polarity of my subs because that made the lines roughly 2.25ms apart. I would have had to delay quite a bit more without the polarity reversed.
So, how do I know if I should reverse polarity, or add more delay?
Post by: Bennett Prescott on April 28, 2010, 11:53:57 AM
| Adam Schaible wrote on Wed, 28 April 2010 11:37 |
So, how do I know if I should reverse polarity, or add more delay? |
Heh... first, do this about a thousand times with different systems set up different ways in different locations.
Seriously, it's not just one or the other. If inverting polarity allows you to get your desired phase alignment with less delay, then do that. Just don't get into any arguments about absolute polarity with anyone, I wouldn't win one of those either...
Post by: Rich Grisier on April 28, 2010, 12:01:16 PM
| Adam Schaible wrote on Tue, 27 April 2010 13:28 |
My goal is to phase align my tops and subs, but before that I need to time align them. |
Here's a simple method I've used to time align a top:
1. System is on, but volume is off.
2. Send a sine wave into the FOH board with the frequency set to the top/sub crossover frequency. The signal generator in Room EQ Wizard can do this.
3. Reverse the phase on the signal going to the sub. I use a speaker cable with a banana connector flipped over at the amp end.
4. Set up a flat mic (dbx rta mic) in front of the system.
5. With the pfl ON and fader down on the rta mic, bring the sine wave volume up so you can hear it.
6. While monitoring the rta mic level, adjust the delay time of the tops. The level seen by the mic should start to drop as you increase the delay time. The level will reach a low point and then start to come back up. At that point there is too much delay. The delay is properly set when the level is at its lowest point.
Of course this doesn't help you with SMAART at all, but it does achieve your goal.
Post by: Adam Schaible on April 28, 2010, 12:03:08 PM
One more question -- is there any difference between time and phase alignment? It may be a fairly obvious question, as I really don't see how one could adjust time alignment without affecting phase and vice versa.
Also - would the measurements change if the source volume changes? When doing my subs, if I got them too loud my furnace/etc would resonate so I turned the signal generator down some. For the tops, I bumped it up a bit.
I think I have it close now.
Post by: Adam Schaible on April 28, 2010, 12:05:12 PM
| Rich Grisier wrote on Wed, 28 April 2010 11:01 | ||
Here's a simple method I've used to time align a top: 1. System is on, but volume is off. 2. Send a sine wave into the FOH board with the frequency set to the top/sub crossover frequency. The signal generator in Room EQ Wizard can do this. 3. Reverse the phase on the signal going to the sub. I use a speaker cable with a banana connector flipped over at the amp end. 4. Set up a flat mic (dbx rta mic) in front of the system. 5. With the pfl ON and fader down on the rta mic, bring the sine wave volume up so you can hear it. 6. While monitoring the rta mic level, adjust the delay time of the tops. The level seen by the mic should start to drop as you increase the delay time. The level will reach a low point and then start to come back up. At that point there is too much delay. The delay is properly set when the level is at its lowest point. Of course this doesn't help you with SMAART at all, but it does achieve your goal. |
That's how I did it initially but I think with a sine wave you can be 360 degree's out (or 180 + polarity reversal of course) and still have perfect summation, but not have achieved your goal. Am I off base here?
Post by: Evan Kirkendall on April 28, 2010, 12:34:39 PM
| Adam Schaible wrote on Wed, 28 April 2010 11:37 |
So, how do I know if I should reverse polarity, or add more delay? |
Adam,
A polarity reversal is about 7ms of delay. I figured this out when I went to align one of my PA's. I could flip the top polarity and get things to sum up nicely. 7ms of delay also worked as well. I like to keep the polarity consistent in my rig, so I went with the delay.
I found the way Phill taught me a lot easier then trying to get slopes to line up. Both require a little finesse with either SMARRT's or the DSP's delay time to get things in line. I noticed a huge difference in the low mid section of my PA once things were nicely aligned. It took me a few hundred measurements before I could finally start getting good, reliable results. There's a lot of things that are easy to mess up, and a lot of things that can interfere with your measurements.
Mic position is VERY important when doing the measurements. For anything below 400hz, I like to use ground plane measurements. There's no floor bounce to deal with, and it's a lot easier to do. For proper ground plane measurements, you'll want the mic pointed at the ground with the capsule touching the ground. You'll want the capsule about 45 degrees to the ground as well. Being outdoors away from anything is very important. Buildings can skew results.
Evan
Post by: Adam Schaible on April 28, 2010, 01:16:00 PM
My results are interesting because I expected the FIR latency to be higher in the UX8800 - it has to be atleast 1 cycle right? I suppose since I am operating in bi-amp mode that FIR latency is probably less than 1ms around 1000hz (not sure what the MF/HF crossover is in the KF650z).
Why is it that the tops are typically delayed to the subs?? And I can't figure out why my tops would have less latency than the subs - I suppose this would be a theory question but perhaps it applies here?
Post by: Phillip_Graham on April 28, 2010, 01:32:21 PM
| Adam Schaible wrote on Wed, 28 April 2010 11:37 |
Ok, that makes more sense now. I agree, that method is simpler -- I understand you're looking for the difference between the initial Auto SM delay and the new delay to get them to lay on top of each other. |
Actually, you are not doing this. You are using AutoSM to get you in the ballpark of a starting delay time, then you are adding additional delay till the group delay at the XO point appears to be zero.
You then repeat this for the other box, and you look at the differential between the two manually adjusted delay times. The initial AutoSM times only matter to get you started.
| Quote: |
However, mine looked similar - but my reaction was to invert the polarity of my subs because that made the lines roughly 2.25ms apart. I would have had to delay quite a bit more without the polarity reversed. |
Adam, what exactly do you mean by "because that made the lines roughly 2.25ms apart"? What lines are you perceiving to be 2.25ms apart? I don't understand what you are trying to say here.
Post by: Phillip_Graham on April 28, 2010, 01:37:44 PM
| Adam Schaible wrote on Wed, 28 April 2010 12:03 |
Ok, thanks Bennett.. One more question -- is there any difference between time and phase alignment? It may be a fairly obvious question, as I really don't see how one could adjust time alignment without affecting phase and vice versa. |
If you adjust the "time alignment" that causes a specific change in phase for every frequency, one that is different for every frequency. The two effects are intimately tied.
We would really like to have direct control over the pure phase at every frequency, but for reasons far too long to elaborate on here we can generally only adjust the delay. We use this delay adjustment to try to get as close to the desired phase behavior in the frequencies that matter around the crossover point. Its not exact, but it can get the desired phase pretty close over a small area of bandwidth.
| Quote: |
Also - would the measurements change if the source volume changes? When doing my subs, if I got them too loud my furnace/etc would resonate so I turned the signal generator down some. For the tops, I bumped it up a bit. I think I have it close now. |
Any nonlinearity, or spurious noise (like the furnace) will pollute the measurement, though averaging increases noise immunity. The 95dB level you stated is much louder than generally necessary to get a valid measurement.
Post by: Adam Schaible on April 28, 2010, 01:42:00 PM
| Phillip Graham wrote on Wed, 28 April 2010 12:32 | ||||
Actually, you are not doing this. You are using AutoSM to get you in the ballpark of a starting delay time, then you are adding additional delay till the group delay at the XO point appears to be zero. You then repeat this for the other box, and you look at the differential between the two manually adjusted delay times. The initial AutoSM times only matter to get you started.
Adam, what exactly do you mean by "because that made the lines roughly 2.25ms apart"? What lines are you perceiving to be 2.25ms apart? I don't understand what you are trying to say here. |
My technical writing skills are far from adequate - I'll try to do better.
Prior to flipping the polarity fo the subs, my phase slops were similar but opposite - I concluded from that that my subs were out of polarity. When I flipped the polarity of the subs, the tops only needed 2.25ms for the lines to lay more-or-less on top of each other.
What is a little surprising to me is that I was using a LR48db LP on the subs and a LR24db HP on the tops, but the slope of the phase trace was very similar.
Post by: Adam Schaible on April 28, 2010, 01:43:08 PM
| Phillip Graham wrote on Wed, 28 April 2010 12:32 | ||||
Actually, you are not doing this. You are using AutoSM to get you in the ballpark of a starting delay time, then you are adding additional delay till the group delay at the XO point appears to be zero. You then repeat this for the other box, and you look at the differential between the two manually adjusted delay times. The initial AutoSM times only matter to get you started.
Adam, what exactly do you mean by "because that made the lines roughly 2.25ms apart"? What lines are you perceiving to be 2.25ms apart? I don't understand what you are trying to say here. |
When you use the term group delay -- does this mean you add delay until the slope around the crossover frequency is 0?
Post by: Phillip_Graham on April 28, 2010, 01:43:30 PM
| Evan Kirkendall wrote on Wed, 28 April 2010 12:34 |
Adam, A polarity reversal is about 7ms of delay. |
Evan,
This generality is not true. A polarity swap is equivalent to 180deg of phase shift at the frequency in question.
If that frequency is 71Hz, the delay time is 7ms. If the frequency was 142Hz, the equivalent delay is only 3.5ms.
The polarity reversal you applied is about 7ms at 71Hz, which presumably was the center of the crossover band for the measurement you were making.
| Quote: |
I figured this out when I went to align one of my PA's. I could flip the top polarity and get things to sum up nicely. 7ms of delay also worked as well. I like to keep the polarity consistent in my rig, so I went with the delay. |
And this was only true for this case because of the nature of this particular sub/tops' acoustic crossover point.
I am glad you found the way that I present doing the alignment more helpful than the align the slopes method.
Post by: Phillip_Graham on April 28, 2010, 01:52:01 PM
| Adam Schaible wrote on Wed, 28 April 2010 13:42 |
My technical writing skills are far from adequate - I'll try to do better. Prior to flipping the polarity fo the subs, my phase slops were similar but opposite |
I still don't know what you mean by "similar but opposite," this statement is no clearer to me than the previous, sorry.
| Quote: |
What is a little surprising to me is that I was using a LR48db LP on the subs and a LR24db HP on the tops, but the slope of the phase trace was very similar. |
I don't know that a 48dB filter would be the one I choose, but welcome to the world of measurements, where the resultant acoustic response is what matters, regardless of what you think about the electrical filter behavior.
FWIW, the combination of a vented box's natural acoustic response near the low corner frequency and an electrical LR24 will produce an aggregate acoustic response that is not that unlike a 48dB filter. Vented box is 4th order (approximately) and the electrical filter is also 4th order: 4+4=8
Post by: Evan Kirkendall on April 28, 2010, 01:54:28 PM
| Phillip Graham wrote on Wed, 28 April 2010 13:43 | ||||
Evan, This generality is not true. A polarity swap is equivalent to 180deg of phase shift at the frequency in question. If that frequency is 71Hz, the delay time is 7ms. If the frequency was 142Hz, the equivalent delay is only 3.5ms. The polarity reversal you applied is about 7ms at 71Hz, which presumably was the center of the crossover band for the measurement you were making.
And this was only true for this case because of the nature of this particular sub/tops' acoustic crossover point. I am glad you found the way that I present doing the alignment more helpful than the align the slopes method. |
And that's why I never stop learning Phill.
It did work in my case, and I should have included the frequency range it worked in for me. I'm slowly getting SMAART figured out thanks to you.
Evan
Post by: Phillip_Graham on April 28, 2010, 01:55:01 PM
| Adam Schaible wrote on Wed, 28 April 2010 13:43 |
When you use the term group delay -- does this mean you add delay until the slope around the crossover frequency is 0? |
Yes Adam, that is correct.
The group delay is the negative derivative of the slope of the phase trace. IOW the instantaneous slope of the phase curve at any frequency.
If the phase curve is flat (i.e slope of zero), or nearly so, then the group delay is zero or nearly so.
Post by: Adam Schaible on April 28, 2010, 01:59:37 PM
| Phillip Graham wrote on Wed, 28 April 2010 12:52 | ||||
I still don't know what you mean by "similar but opposite," this statement is no clearer to me than the previous, sorry.
I don't know that a 48dB filter would be the one I choose, but welcome to the world of measurements, where the resultant acoustic response is what matters, regardless of what you think about the electrical filter behavior. FWIW, the combination of a vented box's natural acoustic response near the low corner frequency and an electrical LR24 will produce an aggregate acoustic response that is not that unlike a 48dB filter. Vented box is 4th order (approximately) and the electrical filter is also 4th order: 4+4=8 |
Ok, sorry - let me re-phrase.
Prior to flipping my subwoofer polarity, the slope around the crossover region was rising with frequency, but the tops were falling as frequency rose at roughly the same rate. After flipping the polarity of my subs, the graph produced showed the subwoofer polarity falling as frequency rises.
I think my difficulty explaining this is due to the relatively infamiliarity with reference points and the terminology around the science -- I hope this is better, and I appreciate the help.
Post by: Phillip_Graham on April 28, 2010, 02:02:42 PM
| Evan Kirkendall wrote on Wed, 28 April 2010 13:54 |
And that's why I never stop learning Phill. It did work in my case, and I should have included the frequency range it worked in for me. |
Yes, including the frequency would have made all things right
Polarity reversal is equivalent to 180deg, which in turn is half a wavelength.
So in your case 7ms*2=14ms (that is a full wavelength is 14ms).
Then to determine the frequency of the wave with this period, divide into time:
(1000ms/s)/(14ms)=71.4(1/s)
Inverse seconds, or (1/s), is the same as Hz, so an equivalent statement in new units is 71.4Hz.
Never underestimate the power of dimensional analysis for analyzing problems
Post by: Phillip_Graham on April 28, 2010, 02:11:47 PM
| Adam Schaible wrote on Wed, 28 April 2010 13:59 |
Prior to flipping my subwoofer polarity, the slope around the crossover region was rising with frequency, but the tops were falling as frequency rose at roughly the same rate. After flipping the polarity of my subs, the graph produced showed the subwoofer polarity falling as frequency rises. |
This is much more clear, but unfortunately now casts some doubts for me on whether you are really understanding the concepts. Though this was your previous measurement, and you seem to be tracking well with the thread, so perhaps you have figured it out since that first measurement.
Don't feel bad, this is not a simple concept for most people to grasp, especially if you are not as least passingly familiar with trig, calculus, and wave representation with sines and cosines.
I am sure that Harry, Paul, Arthur, and Jamie could all tell stories about SMAART class attenders suffering in silence through the class, not really grasping, but too embarrassed to ask a question.
Kudos for being willing to figure the whole thing out in public. Once you get the basics, I highly recommend one of the classes, as it will open your eyes to many other measurement details. It is really a topic that demands hands on learning.
| Quote: |
I think my difficulty explaining this is due to the relatively infamiliarity with reference points and the terminology around the science -- I hope this is better, and I appreciate the help. |
Much better, keep plugging at it!
Post by: Art Welter on April 28, 2010, 02:15:00 PM
| Adam Schaible wrote on Wed, 28 April 2010 11:16 |
Thanks Evan, My results are interesting because I expected the FIR latency to be higher in the UX8800 - it has to be atleast 1 cycle right? I suppose since I am operating in bi-amp mode that FIR latency is probably less than 1ms around 1000hz (not sure what the MF/HF crossover is in the KF650z). Why is it that the tops are typically delayed to the subs?? And I can't figure out why my tops would have less latency than the subs - I suppose this would be a theory question but perhaps it applies here? |
A ported sub is a phase inversion device, the tops generally will be in front of that phase inversion, and need to be delayed.
I agree with Bennett:
“If inverting polarity allows you to get your desired phase alignment with less delay, then do that.”
Although you can get a smooth phase alignment by adding delay to compensate for a polarity inversion, the system will not be transient aligned.
To push the point, if the time difference is 7ms (71.4 Hz as Phillip explained) between LF and MF, you could “phase slope align” the transition at 14ms, 21ms, 28ms, 35ms.
The LF would be obviously leading the MF at 35ms, the MF would sound like a slap echo of the LF. The lead time difference is there at the lesser leading times, just less obvious.
Art Welter
Post by: Adam Schaible on April 28, 2010, 02:26:06 PM
I will stop making excuses for not going to a class at some point, but as you stated I think I would benefit from knowing the basics.
Just to go way back to the basics to make sure I haven't made any incorrect assumptions --
Using method #1, my objective is to get the phase graph of my subs and tops to lay on top of each other. Am I correct so far?
Carrying that a step further...
(The only reason I restate is that I may be misunderstanding some terms that were used above and thus acting incorrectly based upon that misunderstanding).
Step 1: Generate some pink noise through just the subs and allow SMAART to find the delay -- Auto SM in 6 or before, in 7 I think it's just called delay finder
Step 2: Set the phase averaging - I used 4 seconds. Take a snapshot
Step 3: Turn off the subs, and turn on the tops.
Step 4: Without adjusting the internal SMAART delay, take another phase measurement. Add delay to the tops to move the phase measurement across the X axis of the SMAART graph until the slope of the tops lays on the slope of the subs.
Am I off track?
Could you elaborate a bit more on "This is much more clear, but unfortunately now casts some doubts for me on whether you are really understanding the concepts."
I definitely don't disagree, I'm just trying to identify the part of the concept I'm not understanding.
Thanks again!
I think some of my measurements might be somewhat suspect because I was pretty close to the stack and not on ground plane. I tried to get the relative mic-to-driver distances as close to each other as possible for this measurement, which meant it was about 1m away and 6" lower than the Hf driver, almost on axis with the LF driver. I would imagine at this distance delay figures might be a little weird because the distance difference between the subs and MF driver is far in comparison to doing the measurement at 10ft away.
Post by: Phillip_Graham on April 28, 2010, 03:00:00 PM
| Adam Schaible wrote on Wed, 28 April 2010 14:26 |
Using method #1, my objective is to get the phase graph of my subs and tops to lay on top of each other. Am I correct so far? |
Yes, this is correct. IN THE RANGE OF THE CROSSOVER you want the slopes of the phase of both sub and top to be the same (same steepness and direction). The slope direction should be down in 99% of all cases.
If the two traces have the same slope, but do not lay directly on top of each other, that is ok. You want the absolute difference between to the two traces with the same slope to be less than 120deg, and ideally less than 60 deg.
| Quote: |
Step 1: Generate some pink noise through just the subs and allow SMAART to find the delay -- Auto SM in 6 or before, in 7 I think it's just called delay finder Step 2: Set the phase averaging - I used 4 seconds. Take a snapshot Step 3: Turn off the subs, and turn on the tops. Step 4: Without adjusting the internal SMAART delay, take another phase measurement. Add delay to the tops to move the phase measurement across the X axis of the SMAART graph until the slope of the tops lays on the slope of the subs. |
I might of written this differently, but the essence seems correct as I perceive what you wrote.
| Quote: |
Could you elaborate a bit more on "This is much more clear, but unfortunately now casts some doubts for me on whether you are really understanding the concepts." |
If you were clear (before this thread) on what the slope of the phase trace was telling you relative to the delay time of a given frequency, you would have not though that having one slope "up" and the other "down" was correct or appropriate.
Six inches one way or another is not a big deal when the frequencies involved here have 10+ feet wavelength. The difference in level depending on placement close to the boxes, however, will affect the measured acoustic XO point.
Post by: Rob Spence on April 28, 2010, 03:11:34 PM
| Adam Schaible wrote on Wed, 28 April 2010 14:26 |
Step 1: Generate some pink noise through just the subs and allow SMAART to find the delay -- Auto SM in 6 or before, in 7 I think it's just called delay finder |
I had thought that finding the delay using the subs was not the preferred way as the frequencies tended to confound the software delay finder?
Post by: Evan Kirkendall on April 28, 2010, 03:13:31 PM
| Rob Spence wrote on Wed, 28 April 2010 15:11 | ||
I had thought that finding the delay using the subs was not the preferred way as the frequencies tended to confound the software delay finder? |
I find that auto SM will get you close, and then you just have to play with the delay a little until the phase trace smooths out.
Evan
Post by: Phil Lewandowski on April 28, 2010, 03:13:32 PM
| Phillip Graham wrote on Wed, 28 April 2010 15:00 |
Yes, this is correct. IN THE RANGE OF THE CROSSOVER you want the slopes of the phase of both sub and top to be the same (same steepness and direction). The slope direction should be down in 99% of all cases. If the two traces have the same slope, but do not lay directly on top of each other, that is ok. You want the absolute difference between to the two traces with the same slope to be less than 120deg, and ideally less than 60 deg. |
Hey Phillip,
Just a quick reality check here for me as well, in the above where you talk about getting the "phase traces to lay on top of each other" (Or within 60-120 degrees with similar group delay?) would that be in reference to the way that Mac Kerr was talking about and uses to phases align?
Because it seemed that the way you taught Evan was the other way that you said many people grasp first, by wanting to get zero group delay at the XO point for both the tops and subs separately and then using the difference in the delay times in Smaart to get your delay time to phase align?
What exactly are you looking for to signify "zero group delay" at the XO point? From Evan's example it looks like he would have adjusted the Smaart delay individually for the tops and subs in different measurement until the phase traces are "around" the same spot, meaning a zero group delay, and then using the difference in each delay it took to get them do that at XO?
Does this sound on the right track?
The first way, phases traces on top of each other is slightly familiar to me, and makes sense because I believe that was the way that Bob McCarthy went through and explained in his book and it did/does make sense.
Thanks also Adam for starting this topic, as this was something I am also now getting into, and this will help to clarify!
Thanks all,
Phil
Post by: Phil Lewandowski on April 28, 2010, 03:18:11 PM
| Phillip Graham wrote on Wed, 28 April 2010 13:55 | ||
Yes Adam, that is correct. The group delay is the negative derivative of the slope of the phase trace. IOW the instantaneous slope of the phase curve at any frequency. If the phase curve is flat (i.e slope of zero), or nearly so, then the group delay is zero or nearly so. |
Cool, so that partially answers my question from above. But now when I look back at Evan's picture and look at the individual Phase traces (Blue and Orange), would those be an example of zero group delay around 125hz, or Evan acoustic XO point?
Since it looks like they are opposite slops although they do start to both get towards a slop of zero between 100 and 150hz?
Thanks Much,
Phil
Post by: Phillip_Graham on April 28, 2010, 03:18:49 PM
| Rob Spence wrote on Wed, 28 April 2010 15:11 | ||
I had thought that finding the delay using the subs was not the preferred way as the frequencies tended to confound the software delay finder? |
As long as successive AutoSM tests produce stable values for the reference delay, it is a suitable starting point. I will typically shoot 3 successive AutoSM/LG to make sure that it settles on the same point, within a few samples (0.02ms/sample)
Usually it will latch to stable point. It doesn't really matter exactly where in the sub arrival, as long as it finds a stable ballpark inside the sub energy arrival, you can start there. The zero point for the phase in these type of aligments is relative, as it is the differential between the sub and top that matters.
You can always raise the LP on the subs to 1kHz and shoot that as your starting AutoSM, and then manually tweak the delay time (by adding delay) as you lower the LP corner freq.
Aligning by method 2 (the one I espoused) will get you comfortable manipulating the reference delay time to suit your ends.
Post by: Evan Kirkendall on April 28, 2010, 03:21:01 PM
"Zero group delay" is where your phase is flat, or close to it. A falling phase slope indicates frequencies arriving later then the ones at "0."
In my traces, I start with the subs, and play with the delay time until I can get the phase as close to 0 as possible in it's bandpass. I write down that delay time, say 10ms and redo it with just the low band of the tops. Say I get 14ms on the low band of the tops. After subtracting the difference between the 2 delay times, you end up with 4ms of delay on the tops. If you add that 4ms of delay and DO NOT touch the delay on SMAART, the phase should smooth out considerably when you have both the tops and the subs playing. If so, as seen in my picture, you will have aligned the subs and tops.
Phill G can correct me if I'm wrong.
Evan
Post by: Phil Lewandowski on April 28, 2010, 03:26:19 PM
| Evan Kirkendall wrote on Wed, 28 April 2010 15:21 |
Phil, "Zero group delay" is where your phase is flat, or close to it. A rising phase slope indicates frequencies arriving later then the ones at "0." In my traces, I start with the subs, and play with the delay time until I can get the phase as close to 0 as possible in it's bandpass. I write down that delay time, say 10ms and redo it with just the low band of the tops. Say I get 14ms on the low band of the tops. After subtracting the difference between the 2 delay times, you end up with 4ms of delay on the tops. If you add that 4ms of delay and DO NOT touch the delay on SMAART, the phase should smooth out considerably when you have both the tops and the subs playing. If so, as seen in my picture, you will have aligned the subs and tops. Phill G can correct me if I'm wrong. Evan |
Thanks Evan,
So in theory when using, what we are now calling "Method 2" (Yours and Phill's); We are looking to get as close to a zero group delay, or slope of zero, around the crossover region with both the subs and tops LF?
Is that the correct goal with this method?
Thanks,
Phil
Post by: Phillip_Graham on April 28, 2010, 03:33:48 PM
| Phil Lewandowski wrote on Wed, 28 April 2010 15:13 |
Hey Phillip, Just a quick reality check here for me as well, in the above where you talk about getting the "phase traces to lay on top of each other" (Or within 60-120 degrees with similar group delay?) would that be in reference to the way that Mac Kerr was talking about and uses to phases align? |
No, the goal of having them lay on top of each other is the SAME in both cases. The difference with the second type of alignment is that you will by trying to align two flattish "U" shaped traces on top of each other.
In Mac's example the both phases will be sloping downward from left to right across the phase trace. Your still trying to get them to line on top of each other, but now you are trying to match two sloped lines, rather than two "U" shaped regions.
The only difference between Mac's method and the one I discuss, really, is that in "mine" you are artificially forcing the apparent group delay at the XO point to be zero ms, or close to that. In Mac's case, you accept that the lows are lagging the highs (i.e. the downward phase slopes) and try to align those slopes.
In Mac's case AutoSM will typically still produce a "U" in the phase trace, but it will typically be at 2-8kHz, rather in the sub frequencies we are actually trying to align. This is because the low end energy lags the high energy in real loudspeakers, for reasons that would be another large thread.
| Quote: |
Because it seemed that the way you taught Evan was the other way that you said many people grasp first, by wanting to get zero group delay at the XO point for both the tops and subs separately and then using the difference in the delay times in Smaart to get your delay time to phase align? |
I, personally, teach that method. I teach that method because I first explain the phase trace, and group delay, and what the slopes mean, and how real loudspeakers behave, and why. By the time we are in front of SMAART, you have a solid conceptual basis for why the "U" exists, and what it means. It then becomes an "AHA" moment when you can play with the reference delay time to move the U, and therefore realize when, how, and why it moves.
That is how I teach it for people who hire me to train them. It is different than the SMAART curriculum, which covers both methods. I don't think their covering both methods is wrong, I would just rather go into more conceptual detail on one method.
| Quote: |
What exactly are you looking for to signify "zero group delay" at the XO point? |
Phase trace slope near zero (i.e. flat horizontal line).
| Quote: |
From Evan's example it looks like he would have adjusted the Smaart delay individually for the tops and subs in different measurement until the phase traces are "around" the same spot, meaning a zero group delay, and then using the difference in each delay it took to get them do that at XO? |
Exactly right. Evan's screen shot shows the beginning of the process, ONE of the middle steps, and the final result after the processor delay is changed.
| Quote: |
The first way, phases traces on top of each other is slightly familiar to me, and makes sense because I believe that was the way that Bob McCarthy went through and explained in his book and it did/does make sense. |
As I said above, it really is two sides of the same coin.
Post by: Phillip_Graham on April 28, 2010, 03:38:49 PM
| Phil Lewandowski wrote on Wed, 28 April 2010 15:18 |
...when I look back at Evan's picture and look at the individual Phase traces (Blue and Orange), would those be an example of zero group delay around 125hz, or Evan acoustic XO point? |
Pink trace, the final response, has near zero group delay from about 50Hz to 90Hz. The blue trace for the sub is similar.
The Orange trace, which is evidently the tops before any additional added processor delay, has near zero group delay about an octave higher. Adding delay to the tops in the processor brought the apparent group delay of the tops into line with the Blue trace, and produced the Pink trace.
| Quote: |
Since it looks like they are opposite slops although they do start to both get towards a slop of zero between 100 and 150hz? |
I am not clear what you are trying to say with this. Hopefully the paragraph above answered your question.
Post by: Phil Lewandowski on April 28, 2010, 03:38:58 PM
| Phillip Graham wrote on Wed, 28 April 2010 15:33 | ||||||||||
No, the goal of having them lay on top of each other is the SAME in both cases. The difference with the second type of alignment is that you will by trying to align two flattish "U" shaped traces on top of each other. In Mac's example the both phases will be sloping downward from left to right across the phase trace. Your still trying to get them to line on top of each other, but now you are trying to match two sloped lines, rather than two "U" shaped regions. The only difference between Mac's method and the one I discuss, really, is that in "mine" you are artificially forcing the apparent group delay at the XO point to be zero ms, or close to that. In Mac's case, you accept that the lows are lagging the highs (i.e. the downward phase slopes) and try to align those slopes. In Mac's case AutoSM will typically still produce a "U" in the phase trace, but it will typically be at 2-8kHz, rather in the sub frequencies we are actually trying to align. This is because the low end energy lags the high energy in real loudspeakers, for reasons that would be another large thread.
I, personally, teach that method. I teach that method because I first explain the phase trace, and group delay, and what the slopes mean, and how real loudspeakers behave, and why. By the time we are in front of SMAART, you have a solid conceptual basis for why the "U" exists, and what it means. It then becomes an "AHA" moment when you can play with the reference delay time to move the U, and therefore realize when, how, and why it moves. That is how I teach it for people who hire me to train them. It is different than the SMAART curriculum, which covers both methods. I don't think their covering both methods is wrong, I would just rather go into more conceptual detail on one method.
Phase trace slope near zero (i.e. flat horizontal line).
Exactly right. Evan's screen shot shows the beginning of the process, ONE of the middle steps, and the final result after the processor delay is changed.
As I said above, it really is two sides of the same coin. |
Ok, Thank you Phillip, that makes sense. I know see the forced "U's" or zero group delay in Evan's pic.
Phil
Post by: Phillip_Graham on April 28, 2010, 03:41:00 PM
| Phil Lewandowski wrote on Wed, 28 April 2010 15:26 |
Thanks Evan, So in theory when using, what we are now calling "Method 2" (Yours and Phill's); We are looking to get as close to a zero group delay, or slope of zero, around the crossover region with both the subs and tops LF? Is that the correct goal with this method? Thanks, Phil |
Phil,
Yes, that is the goal. Once you determine the differential in delay, you apply it in the processor, to bring the real boxes into alignment. Think of your changing the reference delay as sort of a virtual alignment that is then made physical by the processor delay.
Post by: Phil Lewandowski on April 28, 2010, 03:43:12 PM
| Phillip Graham wrote on Wed, 28 April 2010 15:41 | ||
Phil, Yes, that is the goal. Once you determine the differential in delay, you apply it in the processor, to bring the real boxes into alignment. Think of your changing the reference delay as sort of a virtual alignment that is then made physical by the processor delay. |
Gotcha, Thanks Again, this all *starting* to become become clearer, slowly.
Now if the darn wind would stop here and I could get outside to do some practicing with a couple hundred measurements...
Thanks,
Phil
Post by: Phillip_Graham on April 28, 2010, 03:44:50 PM
| Evan Kirkendall wrote on Wed, 28 April 2010 15:21 |
Phil, "Zero group delay" is where your phase is flat, or close to it. A falling phase slope indicates frequencies arriving later then the ones at "0." |
As Evan says,
If the slope is down (falling) from left to right, those frequencies LAG in time. If the slope is up (from left to right) then those frequencies are LEADING relative to the chosen reference delay time.
| Quote: |
In my traces, I start with the subs, and play with the delay time until I can get the phase as close to 0 as possible in it's bandpass. I write down that delay time, say 10ms and redo it with just the low band of the tops. Say I get 14ms on the low band of the tops. After subtracting the difference between the 2 delay times, you end up with 4ms of delay on the tops. If you add that 4ms of delay and DO NOT touch the delay on SMAART, the phase should smooth out considerably when you have both the tops and the subs playing. If so, as seen in my picture, you will have aligned the subs and tops. Evan |
Evan describes the process and result well.
Post by: Adam Schaible on April 28, 2010, 04:57:49 PM
| Phillip Graham wrote on Wed, 28 April 2010 14:00 | ||||||
Yes, this is correct. IN THE RANGE OF THE CROSSOVER you want the slopes of the phase of both sub and top to be the same (same steepness and direction). The slope direction should be down in 99% of all cases. If the two traces have the same slope, but do not lay directly on top of each other, that is ok. You want the absolute difference between to the two traces with the same slope to be less than 120deg, and ideally less than 60 deg.
I might of written this differently, but the essence seems correct as I perceive what you wrote.
If you were clear (before this thread) on what the slope of the phase trace was telling you relative to the delay time of a given frequency, you would have not though that having one slope "up" and the other "down" was correct or appropriate. Six inches one way or another is not a big deal when the frequencies involved here have 10+ feet wavelength. The difference in level depending on placement close to the boxes, however, will affect the measured acoustic XO point. |
Gotcha, I think my writing was not clear (common theme). Wouldn't reversing the polarity of the subwoofer flip the graph on the X axis? One slope up and one down seemed innapropriate, thus the polarity reversal... but again, I'm grasping.
Post by: Art Welter on April 28, 2010, 05:18:51 PM
| Adam Schaible wrote on Wed, 28 April 2010 14:57 |
Wouldn't reversing the polarity of the subwoofer flip the graph on the X axis? One slope up and one down seemed innapropriate, thus the polarity reversal... but again, I'm grasping. |
A polarity flip will move the phase trace 180 degrees "up" or "down", the left to right curve will still look the same.
Post by: Mac Kerr on April 28, 2010, 05:26:36 PM
| Adam Schaible wrote on Wed, 28 April 2010 14:26 |
Could you elaborate a bit more on "This is much more clear, but unfortunately now casts some doubts for me on whether you are really understanding the concepts." I definitely don't disagree, I'm just trying to identify the part of the concept I'm not understanding. Thanks again! I think some of my measurements might be somewhat suspect because I was pretty close to the stack and not on ground plane. I tried to get the relative mic-to-driver distances as close to each other as possible for this measurement, which meant it was about 1m away and 6" lower than the Hf driver, almost on axis with the LF driver. I would imagine at this distance delay figures might be a little weird because the distance difference between the subs and MF driver is far in comparison to doing the measurement at 10ft away. |
Just to restate the maybe not obvious, both methods are showing you the same information. In the Phil Method™ you are using the offset delay in Smaart to change the display to make the curve easier to read. In the Mac Method™ you make the curve easier to read by adjusting the scale that Smaart displays. In either case you are looking for the difference in offset delays after you have brought the traces together.
While ground plane measurements are always good, they are not as critical for measurements of such long wavelengths as the sub to lows crossover. Unless the mic is set quite high, with waves that big the first reflection will be very close in time and phase to the direct sound. With high frequency measurements the reflection will be many phase rotations off when compared to the direct signal.
Mac
Post by: Mac Kerr on April 28, 2010, 05:28:58 PM
| Art Welter wrote on Wed, 28 April 2010 17:18 | ||
A polarity flip will move the phase trace 180 degrees "up" or "down", the left to right curve will still look the same. |
You make the left to right correction by adding delay to the tops.
Mac
ps some screenshots would make it easier to know what you are seeing.
Post by: Adam Schaible on April 28, 2010, 06:11:54 PM
I was having difficulty with the delay finder, so I did the delay finder with the tops, then turned those off and measured the subs -- saved the phase trace (and I experimented with diff. slopes) and then "aligned" the tops. Luckily, I came up with the same measurement, polarity reversed and 2.25ms of delay on the subs.
Post by: Art Welter on April 28, 2010, 06:38:21 PM
Try reversing the top polarity and delaying it instead.
Edit: What is the physical offset distance between the bass cone and the mid cone?
Post by: Evan Kirkendall on April 28, 2010, 07:18:21 PM
Your subs also drop off fast below 70hz. Are you using a high HPF in this trace?
Evan
Post by: Adam Schaible on April 28, 2010, 07:25:09 PM
| Art Welter wrote on Wed, 28 April 2010 17:38 |
I would guess you have now put the subs one wavelength behind at the crossover point. Try reversing the top polarity and delaying it instead. Edit: What is the physical offset distance between the bass cone and the mid cone? |
The kf650 is right on top of the 728 and its a 3 way cab - the sub cone is maybe 12-18 inches away from the LF cone but the MF cone is probably 30 inches away.
I'm not yet able to figure out the parameters you used to make you're recommendation - could you elaborate?
Post by: Adam Schaible on April 28, 2010, 07:29:25 PM
| Evan Kirkendall wrote on Wed, 28 April 2010 18:18 |
You've got something going on at 90hz that's not right. The dip in coherence makes me think it's a reflection, or you're not in phase or you're using too steep of filters... The summed response of the system puts you down like 24dB at 90hz! Your subs also drop off fast below 70hz. Are you using a high HPF in this trace? Evan |
Yes I have an 18db butterworth at 31ish per jbl recommendations. I need to check the sub channel eq as it may be weird - I think I pulled 63hz down 15db yesterday on accident. That brings up another question - should phase alignment be done again after eq tweaks?
Also, in your plot how were you able to sum both traces in the software?
Thanks!
Post by: Evan Kirkendall on April 28, 2010, 08:07:14 PM
| Adam Schaible wrote on Wed, 28 April 2010 19:29 |
Also, in your plot how were you able to sum both traces in the software? Thanks! |
It's not done in the software, it's the actual measurement from the summation of the speakers.
Evan
Post by: Adam Schaible on April 28, 2010, 08:10:03 PM
Post by: Rob Spence on April 28, 2010, 11:06:46 PM
| Art Welter wrote on Wed, 28 April 2010 18:38 |
I would guess you have now put the subs one wavelength behind at the crossover point. Try reversing the top polarity and delaying it instead. Edit: What is the physical offset distance between the bass cone and the mid cone? |
One thing to remember is that for some subs such as horn loaded ones, it takes a while for the sound to get out. Mine are about 6.5 ms so the tops have to wait for it.
Post by: Phillip_Graham on April 29, 2010, 09:10:40 AM
| Adam Schaible wrote on Wed, 28 April 2010 18:11 |
I did just that! I was having difficulty with the delay finder, so I did the delay finder with the tops, then turned those off and measured the subs -- saved the phase trace (and I experimented with diff. slopes) and then "aligned" the tops. Luckily, I came up with the same measurement, polarity reversed and 2.25ms of delay on the subs. |
Adam,
It would appear that there are serious issues with this particular measurement. You need to start from scratch.
Post the following:
1. Sub only, LP applied, and reference delay time
2. Top only, no delay, no HP all frequencies
3. Top with HP, no delay all frequencies
4. #3 zoomed in from 50-250Hz
As for when you perform eq relative to phase alignment, generally you equalized each bandpass first, and then do the phase alignment. Its more complicated that that, but that is an ok generality.
I am not sure where Art is trying to lead you with all his questions, but I don't immediately see the relevance to what appears to be a much more fundamental problem here.
Your crazy parametric is probably a non equalizable effect of measuring in a room node. It may also be a visual trick the measurement is playing on you because of the narrow zoom range of your SMAART plot, and the overly steep 48dB LR LP
Post by: Phillip_Graham on April 29, 2010, 10:10:39 AM
| Rob Spence wrote on Wed, 28 April 2010 23:06 |
One thing to remember is that for some subs such as horn loaded ones, it takes a while for the sound to get out. Mine are about 6.5 ms so the tops have to wait for it. |
I am pretty certain Art is well aware of all these effects. He is stabbing at something else, but I cannot discern what it is. Perhaps he will elaborate.
It should be mentioned that even non-horn loaded LF boxes have substantial group delay because of the combination of crossover filters and box acoustic response.
Your horn boxes are going to have about 6ms of additional acoustic propagation delay, but they also have all of the filter effects.
I have found, after doing enough analysis and optimization, that I no longer even think about the physical propagation delay. By this I mean that I don't think about the physical distance to the speaker, as it usually has no correlation to the true arrival time. The box's position is space is where my measured phase curve tells me it is at the frequencies I care about. This is irregardless of where the box seems to be physically.
The arrival time, of course, changes (sometimes dramatically) as a consequence of the phase response of the box. So a speaker has no one specific arrival time.
Even a box with excellent phase response, like the Fulcrum Acoustic DX1265, cannot escape the pull of the fact that all speakers are high pass filters. The DX1265 has about 5ms of group delay at 90Hz, relative to the essentially dead flat phase phase response from 500Hz to 15kHz.
Post by: Evan Kirkendall on April 29, 2010, 05:04:25 PM
First, my set up:
Dual 18, SLS960 on top.
Audix TR40A measurement mic in ground plane.
About 10' of distance between cabs and mic.
IT4k's for subs/tops
SMAART V5.4
M-Audio preamp
So, the first measurement I made was with the top muted, and just the sub on. It should be noted everything you see is post EQ, and post HPF/LPF.
What you are seeing is what it looks like with no delayed added in SMAART. Yes, the magnitude response looks good, but the coherence and phase are shit. The phase trace just keeps wrapping and wrapping.
This trace is when you add the auto small function in. Delay has been added and the coherence/phase traces look a lot nicer now.
Now, I've fined tuned the delay by hand and things are looking good. So, we write down the delay time of 33ms.
Now, it's time for the tops. Sub gets muted, and I apply a LPF at 300hz to keep the mids and highs from interfering with my measurements. Remember, we're only looking for sub/low alignments right now, so the rest of the box doesn't matter.
Again, this is the top with no delay added.
This is once the auto small function has been used. I do not normally fine tune the lows/mids/highs once auto sm locks in on a time. I simply repeat the auto small test a few times to make sure it locks in on a time. Normally once it locks in, the phase looks pretty good. For simplicity's sake, I rounded the delay off to an even number, which ended up being 19ms.
So, we take our 2 delay times(33, 19) and we subtract them, giving us a total delay of 14ms on the tops. I applied the 14ms to my tops and this is what it looked like:
The blue trace is the total phase response of the system. Here's a cleaner shot with just the finished results:
Pretty good all around, and it sounded good too. So, my day was finished. Hopefully this is helpful for Adam, and anyone trying to align their PA's.
Evan
Post by: Mac Kerr on April 29, 2010, 05:20:19 PM
Mac
Post by: Evan Kirkendall on April 29, 2010, 05:30:26 PM
I actually cross my subs over at 75hz, and my tops at 110hz, leaving an electronic gap from 75-110hz. I have my subs at +6, and my tops at -3, so it bumps the acoustic crossover point up to around 90hz. I should have mentioned that in my first post. My points of zero group delay are close to the crossover point, and could have been a little closer, but I was dealing with wind and limited time, so I settled for that.
Evan
Post by: Adam Schaible on April 29, 2010, 06:09:28 PM
On that note, I do run my subs a little harder as well - so using active LR slopes at the same frequency probably means my acoustic crossover is higher than my electronic one, thus perhaps phase should be aligned there.
It seems like you just use the delay finder (auto sm) on the subs, then on the tops, and subtracted?
I might be following this wrong though...
Post by: Mac Kerr on April 29, 2010, 06:25:00 PM
| Evan Kirkendall wrote on Thu, 29 April 2010 17:30 |
Mac, I actually cross my subs over at 75hz, and my tops at 110hz, leaving an electronic gap from 75-110hz. I have my subs at +6, and my tops at -3, so it bumps the acoustic crossover point up to around 90hz. I should have mentioned that in my first post. My points of zero group delay are close to the crossover point, and could have been a little closer, but I was dealing with wind and limited time, so I settled for that. Evan |
Your points of zero group delay (matching phase) are more than an octave apart. The whole point of phase alignment between mains and subs is that the phase matches at the crossover point. If you have 2 electrical crossover points you need to get the phase to match at the acoustical crossover, not at the 2 different electrical crossover points.
Mac
Post by: Mac Kerr on April 29, 2010, 06:44:12 PM
| Adam Schaible wrote on Thu, 29 April 2010 18:09 |
It seems like you just use the delay finder (auto sm) on the subs, then on the tops, and subtracted? I might be following this wrong though... |
That would be an impulse alignment and may be a good starting point, but the impulse point will be derived from the upper part of whichever frequency band you are looking at, and as you can see from the phase traces the phase is not the same through the whole band. To get good summation, which is what you are trying to do, you need the bands on both sides of the crossover to be in phase.
Mac
Post by: Mike Christy on April 29, 2010, 07:27:07 PM
Post by: Adam Schaible on April 29, 2010, 07:36:25 PM
Post by: Mac Kerr on April 29, 2010, 07:51:36 PM
| Adam Schaible wrote on Thu, 29 April 2010 19:36 |
Aside from the 90hz weirdness, what's so different about those measurements than mine? |
Which measurement are you referring to? A link would be nice.
IIRC I have not seen one of your measurements where the 2 phase traces lay on top of one another.
In Mike's screenshot the acoustic crossover, which you can see from where the sub and low magnitude response curves cross is graphically about half way between 63Hz and 125Hz on the log scale frequency axis. If you follow that point up to the phase traces you will see that the 2 phase traces for subs and lows lay right on top of one another at that frequency. That is what you are shooting for.
Mac
Post by: Adam Schaible on April 29, 2010, 08:40:34 PM
http://srforums.prosoundweb.com/index.php/fa/29790/23677/
Light green is the tops delayed, light pink is the subs, dark red is the tops pre delay.
Post by: Mac Kerr on April 29, 2010, 09:32:56 PM
| Adam Schaible wrote on Thu, 29 April 2010 20:40 |
Here's the link to the image: http://srforums.prosoundweb.com/index.php/fa/29790/23677/ Light green is the tops delayed, light pink is the subs, dark red is the tops pre delay. |
If the crossover is just above 80Hz that seems pretty good. The anomaly just above 90Hz seems to be effecting everything, it might help to figure out what that is and fix it, then check your phase response again. Where the 2 phase traces cross they line up with the same amount of phase difference from zero. That means they are moving together at that frequency.
Mac
Post by: Adam Schaible on April 29, 2010, 09:39:11 PM
I did this in the opposite manner -- could you tell me if this is an issue?
I found reference delay from my tops, then did the phase trace from my subs, and then aligned my tops to that. The reason being I couldn't get a reliable reference delay from my subs, but now I think that may be due to reversing polarity on the subs?
Post by: Mac Kerr on April 29, 2010, 09:49:39 PM
| Adam Schaible wrote on Thu, 29 April 2010 21:39 |
Moving the reference mic 10 feet back and about 3 feet off of the floor dramatically reduced the dip at 90hz to about 3db. I did this in the opposite manner -- could you tell me if this is an issue? I found reference delay from my tops, then did the phase trace from my subs, and then aligned my tops to that. The reason being I couldn't get a reliable reference delay from my subs, but now I think that may be due to reversing polarity on the subs? |
It can be hard to get an impulse response from a sub because they don't pass an impulse. Using the Smaart offset delay from your mains is probably fine. With the Phil Method™ you fudge the delay time anyway to get the display to a point where the phase trace has zero slope, and then work from there. The important thing either way is to get the 2 traces to lay on top of each other at the crossover frequency.
What does it look like if you do not reverse the polarity? Changing polarity should move the curve vertically by 180 degrees.
Mac
Post by: Phillip_Graham on April 30, 2010, 09:36:08 AM
| Mike Christy wrote on Thu, 29 April 2010 19:27 |
Maybe this will help... I think what Mac is saying is that you want something like this. Below is a Triple-8 and a Growler, with cross over points of 95Hz and 110Hz, I tried to align the phase traces with the top's DSP delay: |
For those still playing along at home, Mike's plot is a sample of what "Mac's method" looks like when zoomed in on the low frequencies.
Looks good Mike
Post by: Mike Christy on April 30, 2010, 09:41:38 AM
Post by: Phillip_Graham on April 30, 2010, 09:44:32 AM
| Mac Kerr wrote on Thu, 29 April 2010 18:25 | ||
Your points of zero group delay (matching phase) are more than an octave apart. The whole point of phase alignment between mains and subs is that the phase matches at the crossover point. If you have 2 electrical crossover points you need to get the phase to match at the acoustical crossover, not at the 2 different electrical crossover points. Mac |
I am confident that Evan knows the method well enough to defend himself, but lets use Mac's valid point to have a teachable moment for those following the thread. SMAART old timers keep your mouths shut
"If one wanted to shift the point of apparent zero group delay on Evan's sub trace upwards in frequency, what would should you do to SMAART's reference delay, and why?
There's no grade, so feel free to take a stab!
Post by: Adam Schaible on April 30, 2010, 09:52:34 AM
Ok, F!
Post by: Evan Kirkendall on April 30, 2010, 10:16:03 AM
Evan
Post by: Phillip_Graham on May 03, 2010, 06:46:04 PM
| Adam Schaible wrote on Fri, 30 April 2010 09:52 |
I think you would add reference delay -- and my jibberish reasoning is that since the subs and filters are essentially adding "delay" themselves in the form of phase shift, you must "wait" longer until the 0 degree point of the wave form matches up with the 0 degree point of the reference signal. Ok, F! |
Adam,
While I was hoping that others would answer, I guess not that many are following the thread, or want to take a stab.
In the interest of not leaving you hanging, I will clarify. Evan posted the correct answer, but I asked him to remove it to see if others would contribute.
You actually want to subtract, not add reference delay, though the essence of your "jibberish" why is essentially correct.
If you can parse out what doesn't make sense to you in the "jibberish" I will take a stab at clarifying it for you.
For starters, think about this. If you start by accepting that a certain XO filter causes a fixed amount of phase shift, then we can build on that as a basis why the lows are behind in time.
Lets say our imaginary filter causes 90deg, or 1/4 wavelength, of phase shift at 1kHz. The full period of 1kHz is 1ms.
Therefore the 90 deg phase shift causes a delay equal to 1ms/4, or 0.25ms.
If we do the same thought experiment at 100Hz, the period of the wave is 10ms, and 1/4 wavelength (i.e. 90 degrees) is 2.5ms. The XO filter causes the same phase shift, regardless of frequency, but at low frequencies this corresponds into much greater delay.
Hopefully that is clear to you.
Post by: Rob Spence on May 06, 2010, 10:21:41 AM
A second question
In Evan's setup, he has the top box right on top of the sub. Would it not change things when the top box is placed where it would be when in use? Or not?
thanks
Post by: David Stiles on May 06, 2010, 11:34:52 PM
| Rob Spence wrote on Thu, 06 May 2010 21:21 |
Re: Evan's graphs, I don't understand why the delay is 32ms when I would expect it to be closer to 10ms if the measurement is 10ft from the speakers. A second question In Evan's setup, he has the top box right on top of the sub. Would it not change things when the top box is placed where it would be when in use? Or not? thanks |
i also wondered about that delay time.
Post by: Adam Schaible on May 07, 2010, 08:16:48 AM
1: Lots of other items contribute to the delay. With the method Evan is using, he is adding reference delay in SMAART to get the "simulated" group delay to be 0 at the accoustic crossover point. Sources of delay are the actual signal processor, console, the electronic crossover, and the physical distance between the source and the ref mic.
2: Yes, if physical positioning changes, then these results may be skewed. One problem is that once the sources move away from each other, then phase alignment is then limited to only a physical point in space. In other words, if in point A you are 45 feet from the subs and 60 feet from the tops, and point B is 30 feet from the subs, 30 feet from the tops, you can't phase align both of those areas. This is one of the advantages of flying subwoofer arrays that are physically near the tops.
Post by: Phillip_Graham on May 07, 2010, 11:44:14 AM
| Rob Spence wrote on Thu, 06 May 2010 10:21 |
Re: Evan's graphs, I don't understand why the delay is 32ms when I would expect it to be closer to 10ms if the measurement is 10ft from the speakers. A second question In Evan's setup, he has the top box right on top of the sub. Would it not change things when the top box is placed where it would be when in use? Or not? thanks |
Rob,
The delay time is much greater than the physical distance due to the phase altering effects of the sub's high and low pass filters, as a well as its acoustic phase response.
The physical distance to the speakers, especially when you are near them as in Evan's, is rarely the dominant delay-causing entity. Being able to represent the true arrival accounting for all of the effects, is one of measurement software's "big wins."
Ultimately the propagation delay involved is the delay time that gives you the group delay response you are looking for, and the physical time of flight is only one component of that.
As for your second question, yes things would change in the real-use situation. But the boxes can be remeasured and aligned quickly in the field once the final sub/top configuration is set.
Post by: Doug Fowler on May 07, 2010, 03:32:10 PM
| Quote: |
I would expect it to be closer to 10ms if the measurement is 10ft from the speakers. |
If the microphone is 10 feet from the sub, I would expect to see 20-24 msec delay time.
Experiment -
Place the microphone on the sub grille (direct radiator preferred for this experiment).
Put your measurement software in 'auto delay' mode.
Run the sub LPF up over 1KHz.
Use LR24 for the LPF.
Now, lower the LPF frequency and observe the delay time get loooooonger. I'll wager it ends up around 11msec, give or take. The actual peak time will vary depending on the filter type and slope.
Why?
As Phill mentioned, the LPF at bass frequencies is pushing the time back.
And yes there are still people out there that wish to get their 'sub delay time' by running the LPF up over 1KHz. Whatever.
The fact is since because the desired offset time is relative between the sub and LF section of what you are aligning it to, you may not need the 'initial' delay time at all, other than to 'true up' the phase display in the area of interest.
Edit: I cover this at length in class with demonstrations. Until one understands what the LPF is doing, and why other frequencies below the LPF are later still, one can't expect to be confident about an alignment.
For those who already understand this, that part of class is a bit of a 'yawner', but invariably there are several (or most, depending) who never knew this and are somewhat amazed.
Post by: David Stiles on May 09, 2010, 12:25:19 AM
Post by: Desmond Lee on May 09, 2010, 02:21:07 PM
I would like to ask you a question. How would you do it when you have both flown and ground stack subs?
I have an experience that I have both flown and ground stacked subs which are not the same kind. What I got is VerTec subs flown and Turbosound 21" horn subs ground stacked.
I would like to know how will you align them. Thanks.
Des
Post by: Phillip_Graham on May 09, 2010, 02:38:10 PM
| David Stiles wrote on Sun, 09 May 2010 00:25 |
can someone using the group delay method show the results of the alignment in normal phase display also?. showing the resultant phase response at the acoustic xover points pls.i would like to see that to help me understand the possible differences in results. |
David,
There is no difference in results, only paths to getting there, and relative difficulty of those paths. Either method ends up with the same results.
If you do it my/evan/zero group delay way, you get graphs that look like the ones Evan posted.
If you do it Macs/phase slope method, it will look like what Mike Christy posted earlier (and I repeat below):
There is little point in showing both methods for the same system, as the end plots would look like evans', and the one immediately above.
Post by: Phillip_Graham on May 09, 2010, 02:45:12 PM
| Desmond Lee wrote on Sun, 09 May 2010 14:21 |
Philip, I would like to ask you a question. How would you do it when you have both flown and ground stack subs? I have an experience that I have both flown and ground stacked subs which are not the same kind. What I got is VerTec subs flown and Turbosound 21" horn subs ground stacked. I would like to know how will you align them. Thanks. Des |
Des,
There are plenty of subtleties that could be discussed about this kind of configuration, and where/how one aligns the various boxes together.
In principle, though, the flown sub array is treated as part of the main PA, and phase aligned like it was a small speaker stacked on top of sub. Whether one tunes this combination flat, or with a bass "hastack" is one but one of the variables to consider.
Once the flown PA/sub combo is acting as one big speaker, then this completed whole will be aligned with the ground based subs, taking into consideration the numerous compromises that are made when integrating these two subsystems.
All of the details and compromises in regards to such an array are complicated, and beyond the scope of this thread. But the physical process simply iterates at each subsystem of the PA.
Post by: Desmond Lee on May 09, 2010, 06:02:37 PM
align the flown subs to the main PA first, then align the ground subs to the flown subs?
When align ground subs to the flown subs, the lows in the main PA need to be on as well or only the flown subs?
In this situation, still putting the mic close to the ground will get best results? Should the mic be place at one side of the system or at the center, like FOH position?
Des
Post by: Langston Holland on May 09, 2010, 06:49:52 PM
| Phill wrote: |
There are plenty of SUBtleties that could be discussed about this kind of configuration,... |
Nice! :)
Post by: Phillip_Graham on May 10, 2010, 07:22:59 AM
| Desmond Lee wrote on Sun, 09 May 2010 18:02 |
So Phillip, align the flown subs to the main PA first, then align the ground subs to the flown subs? When align ground subs to the flown subs, the lows in the main PA need to be on as well or only the flown subs? In this situation, still putting the mic close to the ground will get best results? Should the mic be place at one side of the system or at the center, like FOH position? Des |
I suggest starting with the main pa hangs, and then moving on to the subs. The flown subs behavior will be predicable, ie they will lag the main pa like one would expect from physics.
The ground stacked subs may end up ahead or behind the main pa after it is aligned.
The rest of your questions don't have definitive, clear, short answers, and would best be answered in one of the advanced measurement classes available from Meyer, doug fowler, or the upcoming new SMAART advanced seminars.
Post by: Desmond Lee on May 10, 2010, 10:22:53 AM
The rest of your questions don't have definitive, clear, short answers, and would best be answered in one of the advanced measurement classes available from Meyer, doug fowler, or the upcoming new SMAART advanced seminars.[/quote]
I understand, thanks for your reply.
Des
Post by: Matt Harris on May 11, 2010, 11:50:28 PM
I flatten the group delay for the subs. Then, instead of moving the reference delay for the tops and then using the difference for the delay in my DSP I just add in the delay at my DSP while measuring until the phase slopes match. It saves a step, and theres no math involved
Post by: Tim McCulloch on May 11, 2010, 11:52:53 PM
| Matt Harris wrote on Tue, 11 May 2010 22:50 |
Nice thread guys. I would like to mention I use the Phil Method with one modification. I flatten the group delay for the subs. Then, instead of moving the reference delay for the tops and then using the difference for the delay in my DSP I just add in the delay at my DSP while measuring until the phase slopes match. It saves a step, and theres no math involved |
That is the way Jamie Anderson demonstrated the method when I took his class @ EAW in 2004.
Have fun, good luck.
Tim Mc
Post by: Leo Melkote on May 13, 2010, 12:31:23 PM
Post by: Adam Schaible on May 13, 2010, 12:58:46 PM
I know the thought is to measure at ground plane to reduce the effects of reflections, however - if our ears are 6' high, shouldn't we be measuring there?
Post by: Mac Kerr on May 13, 2010, 01:02:10 PM
| Adam Schaible wrote on Thu, 13 May 2010 12:58 |
To take another spin on this .... I know the thought is to measure at ground plane to reduce the effects of reflections, however - if our ears are 6' high, shouldn't we be measuring there? |
No, because your ears are connected to a brain that does a huge amount of post processing on the sound. When measuring speakers we want to eliminate as many variables as possible to find out what the real free field response is, so that we can compare them to some known standard (like flat response). Using a ground plane measurement eliminates reflections, and the comb filtering they produce.
When setting up a sound system in a performance venue, your ears are the final arbiter. The measurement however should be as unbiased as possible.
Mac
Post by: Art Welter on May 13, 2010, 01:23:19 PM
| Leo Melkote wrote on Thu, 13 May 2010 10:31 |
What are the criteria for ground plane measurements? If the stack is two dual 18 subs tall, can I still get accurate measurements in the ground plane? |
What the microphone hears is a combination of the direct and reflected sounds, as well as on and off axis response if measuring an array. A measurement can be “accurate”, yet not be indicative of the speaker’s actual response.
The most accurate response curve for a full range speaker would be made on axis at a large distance relative to the speaker size in a large anechoic chamber, or the speaker lifted far enough off the ground to be considered in “free space”.
However, a sub is usually used on the ground, so measurement on the ground will be indicative of it’s actual behavior.
Measuring on the ground plane eliminates the reflected path length difference a microphone has at mid stack or head height, but when measuring close to two stacked dual 18 subs, the microphone will be hearing a large time (path length) difference between the top cone and the bottom. That stack at one or two meters would show a very different response than if the same were measured at around 10 meters. Raising the microphone to the middle of the stack height would again show different response curves at the two distances.
A 30 Hz wavelength is 37.5 feet long, a 100 Hz wavelength is 11.25 feet long. For the 100 Hz wave to be within 1/4 wavelength (where they combine constructively) requires a path length difference of 2.81 feet or less.
With an 8 foot high stack, to get the path length difference down to 2.81 feet or less would require the ground plane microphone to be at around 10 feet or more from the stack. A further distance would be required to avoid a bumpy graph above 100 Hz.
Art Welter
Post by: Leo Melkote on May 13, 2010, 02:32:59 PM
| Art Welter wrote on Thu, 13 May 2010 12:23 |
A 30 Hz wavelength is 37.5 feet long, a 100 Hz wavelength is 11.25 feet long. For the 100 Hz wave to be within 1/4 wavelength (where they combine constructively) requires a path length difference of 2.81 feet or less. With an 8 foot high stack, to get the path length difference down to 2.81 feet or less would require the ground plane microphone to be at around 10 feet or more from the stack. A further distance would be required to avoid a bumpy graph above 100 Hz. Art Welter |
Thanks, that was very useful.
Post by: Desmond Lee on May 14, 2010, 04:26:26 PM
both your post has gave me answers for my previous questions to Phillip. Thank you guys.
Des
Post by: Art Welter on May 15, 2010, 12:12:26 AM
| Desmond Lee wrote on Fri, 14 May 2010 14:26 |
Mac & Art, both your post has gave me answers for my previous questions to Phillip. Thank you guys. Des |
You're welcome.
Measurement is addictive.
Art Welter