Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[Ivan Beaver]

PS: Merry Christmas Ivan, how are the lights this year?

They are up and running fine.

I have added 2 new songs and working on 2 more.

I think a nice old Spike Jones (nuttin for Christmas) would be a nice addition.

[Ivan Beaver]

Now you're scaring me a little... Performing these tests will be harder than static power measurements that customers rarely get correct. The higher the power, the harder to control sundry variables.

By "testing" I don't mean engineering type tests-but rather simple things like: walking the pattern of a loudspeaker and listening for where the sound cuts off at what freq, maybe a simple low freq sweep for the low corner, doing some side by side tests regarding sensitivity-freq response-max outputs-arraybility etc, off axis response of mics and so forth.

You know, things that are easy to do with the tools on hand, at least to give an idea of how things compare.

But of course it is also easy to screw up the tests and come up with "results"/conclusions that are wrong based on incorrect tests.

I know I have caught myself making wrong decisions based on incorrect or faulty tests.

I try to be a bit more careful these days-------

[Keith Broughton]

Normalizing to test the amp for all three loads using the same crest factor will only reveal optimal performance for one load and understate the other two.. 

I have been reading this thread and get most of the idea. As usual, no simple answer and more so now with these ultra high  power amps that seem to have nuclear capabilities.
Anyway, let's see if I have this right.
For a more accurate performance specification, an amp should be tested at 3 different load impedances and with 3 different crest factor test sources to reveal how the amp will perform at those loads.
Higher crest factor for lower impedance load and lower crest factor for higher impedance loads.
These tests should be taken over a minimum time period of up to 10 seconds or more.
In any of those cases, it will not clearly indicate performance variables as impedance changes within the load range tested as these variables are speaker specific.
Comments or corrections?

[radulescu_paul_mircea]

For a more accurate performance specification, an amp should be tested at 3 different load impedances and with 3 different crest factor test sources to reveal how the amp will perform at those loads.
Higher crest factor for higher impedance load and lower crest factor for lower loads.
These tests should be taken over a minimum time period of up to 10 seconds or more.
In any of those cases, it will not clearly indicate performance variables as impedance changes within the load range tested as these variables are speaker specific.
Comments or corrections?

IMO they should be tested this way, but also state the thermal cappacity of the amp for continous use, for us to know better what kind of signals could be used with them at lower loads. From what i see, I could use this amp with all chanels in 2 Ohm
But only for high dynamic music. If I use dubstep or more compressed music, it is better to use them in 4-8 ohms, because i will have high power for a small moment and after that it will limit. For me it's clear. One could do a rock concert with half the amplifiers as an EDM concert, at the same peak SPL.

[Ivan Beaver]

IMO they should be tested this way, but also state the thermal cappacity of the amp for continous use, for us to know better what kind of signals could be used with them at lower loads. From what i see, I could use this amp with all chanels in 2 Ohm
But only for high dynamic music. If I use dubstep or more compressed music, it is better to use them in 4-8 ohms, because i will have high power for a small moment and after that it will limit. For me it's clear. One could do a rock concert with half the amplifiers as an EDM concert, at the same peak SPL.

The problem is that real music varies much more than a simple test signal.

It can get as high as a sine wave and be as little as next to no signal.

So the amp in reality goes from a really heavy load to no load in a matter of seconds.

And just because a test signal is "standard" does not mean that it applies to the real world.

Take the various "shaped" signals used in loudspeaker power handling.  THey all roll off the top and the low end.

I get the top end, but would "argue" that rolling off the low end does not represent much of "modern" music, that is really bass heavy.

It would be fine for older music, but the freq content has changed over the years.

So sadly, as much as I like specs and measurements, it does come down to how well something works for somebody in a particular situation/setup and having the experience to figure it out-as compared to something else in the same situation.

Which is why I like side by side tests and HIGHLY encourage them

[Mac Kerr]

it does come down to how well something works for somebody in a particular situation/setup and having the experience to figure it out-as compared to something else in the same situation.

Which is why I like side by side tests and HIGHLY encourage them

I would think this statement would argue toward testing with the same crest factor signal at the 3 different rated load impedances. Changing 2 variables (crest factor and load impedance) is not a good way to make comparisons.

The fact that real music is not well represented by a sine wave does not negate the need to be able to make comparisons with the test results that are available.

Mac

[John Roberts {JR}]

I have been reading this thread and get most of the idea. As usual, no simple answer and more so now with these ultra high  power amps that seem to have nuclear capabilities.
Anyway, let's see if I have this right.
For a more accurate performance specification, an amp should be tested at 3 different load impedances and with 3 different crest factor test sources to reveal how the amp will perform at those loads.

To characterize if for those three different loads, yes it will return three different results.


Using the k10 power vs time and impedance plot because it's handy, we see a few things I consider relative.
#1 for 100 seconds and longer, all three load impedances receive the same power.   
#2 for very brief (100mSec or less) they deliver significantly higher power into lower load impedance.
#3 the difference (crest factor) between short peaks and long term average is different for each impedance because the peak power is different.

To measure a higher impedance load using the same crest factor (ratio) as we could realize from the lower impedance we must drop the average power a bunch lower to maintain the ratio. Similarly trying to use the higher impedance crest factor with a lower impedance means ignoring available peak power.   

Higher crest factor for higher impedance load and lower crest factor for lower loads.

Actually opposite... higher crest factor for lower impedance, due to higher short term peak power etc.

These tests should be taken over a minimum time period of up to 10 seconds or more.

?? When I designed my own tone-burst test set (decades ago), I included multiple variables. Burst on-time (how many cycles of a sine wave play each burst), off-time or repetition rate for that burst to repeat, and the steady state continuous signal level to play between bursts (like real music).

So step one is determine how long of a burst do you want to cleanly reproduce (before PS collapses or limits), then how often do you want this burst to repeat, to be musically meaningful. Too short and it's not like a transient but a complex waveform, to much rest between bursts and it will overstate performance. Then lastly increasing the continuous level also parallels how real music behaves, with the higher the continuous level the lower the burst peaks possible so lower crest factor ratio. The amplifier under test, should be able to maintain this burst test indefinitely, while the adjustments will interact and affect results. Higher continuous means less peak or burst power, etc.

I can imagine a range of crest factor ratio measurements depending on the rep rate of the burst. (Didn't I say this gets complicated.)  Back last century there was an IHF dynamic headroom spec that offered a standardized burst on/off time, and I used that for my one funny amp design back in the 80s, (100W peak, 35W continuous).  But all these moving parts are kind of arbitrary and will vary between user applications.

   

In any of those cases, it will not clearly indicate performance variables as impedance changes within the load range tested as these variables are speaker specific.
Comments or corrections?

Yup speaker impedance with temperature is another variable, while my steady state burst test will allow speakers and amps to stabilize at a working temperature.

While my burst test might return a single figure crest factor we need to ASSume some nominal servings for burst on/off time etc.

JR

PS: If the K10 was designed to meet the old IHF dynamic headroom spec, the short term power would only be specified for 20 mSec, not 100 mSec, so that is generous and 5x longer than what the old spec would require from design engineers. A spec for this would drive design engineers to look good using this spec... so those nominal variables need to be carefully selected or it could drive amp designer to add unneeded cost. I kind of prefer the way it is now, vs that alternative, and like powered speakers even better. 

[radulescu_paul_mircea]

The problem is that real music varies much more than a simple test signal.

It can get as high as a sine wave and be as little as next to no signal.

So the amp in reality goes from a really heavy load to no load in a matter of seconds.

And just because a test signal is "standard" does not mean that it applies to the real world.

Take the various "shaped" signals used in loudspeaker power handling.  THey all roll off the top and the low end.

I get the top end, but would "argue" that rolling off the low end does not represent much of "modern" music, that is really bass heavy.

It would be fine for older music, but the freq content has changed over the years.

So sadly, as much as I like specs and measurements, it does come down to how well something works for somebody in a particular situation/setup and having the experience to figure it out-as compared to something else in the same situation.

Which is why I like side by side tests and HIGHLY encourage them

I also like side-by-side comparisons and I try to make them whenever I can. I like the method Ricci used, to connect many series-paralel speakers (that one knows well and have previously measured) to give different impedances, measure the system sensitivity at 2,83 V and then measure the Avg and peak SPL at 10 meters. If one have enough speakers to not have more than 1dB power compression, one can calculate what amount of power his amp has.
For example I tested using 6 Qube q218 wih drivers conected to get 8, 4 and 2 ohm (nominal) and compared a Crest CC5500 vs a Powersoft K6.  Here in my country I do these tests, but IDK anyone else who does it and I can tell I am not installing nowhere near as many sound systems as the others around here.
Anyway. Until now I never needed a power amp to give me 5kW continously at 2 ohms. I like to have around 6 dB headroom so I always use only RMS limiters on subs and choose the power amps acordingly.

[John Roberts {JR}]

I would think this statement would argue toward testing with the same crest factor signal at the 3 different rated load impedances. Changing 2 variables (crest factor and load impedance) is not a good way to make comparisons.

The fact that real music is not well represented by a sine wave does not negate the need to be able to make comparisons with the test results that are available.

Mac

This would make sense if the customer knew the actual crest factor that his application really needs. Since physics dictates that the same amp can deliver higher crest factor into lower Z loads, accurately specifying the amp, and showing it in the best light is to reveal how much it "could" deliver into each load impedance.

Counter intuitively if I knew my genre demanded a higher crest factor than this amp delivers into 8 ohms I might be tempted to load it at 4 ohms or lower, but i wouldn't pursue crest factor in a more must be better, since the amp is working harder the rest of the time.

I suspect the number of customers who know the crest factor required for their application, is a much smaller subset of people who actually understand what crest factor is.

So yes perhaps this is TMI for most, but i see no reason to expect the amp maker to not show their amp in the best light for the consumers who "think" they understand what the numbers mean.

I really think trying to educate all consumers about this is a massive waste of time, because for the reasons I have enumerated, there will not be an ah hah moment at the end of the class where they can apply some single number CF specs that don't represent all situations.

JR

PS: Regarding listening tests, the amp designers have already done that in the course of designing the amp, and if this isn't their first rodeo they have learned difficult applications from customer feedback to benchmark their designs against. So trust but verify, but trust the respected amp makers to have thought about this stuff before we did.

[Ivan Beaver]

I suspect the number of customers who know the crest factor required for their application, is a much smaller subset of people who actually understand what crest factor is.

And just to "further complicate" the issue, how many customers even know what the ACTUAL impedance of their speakers?

They may "think" the load is 2 or 4 ohms, but it could be much higher (so the amp is not working hard at all) or a bit lower (at some freq) and the amp is actually working harder than expected.

A simple one number rating is not even beginning to tell the story of the real impedance load.

So in a typical EDM "sweep", the amp will go through periods of having a very light load to a very heavy load-depending on freq.

I doubt many every even look at the impedance CURVE of their loudspeakers vs the freq content of the material they are using.  Add to that looking at the crest factor and the number of people becomes probably non existent. 

But when talking about how much "real" power the amp can deliver in a real world setting, and all of a sudden those things that are totally ignored BECOME REAL IMPORTANT-if you are looking for REAL answers, and not some simple easy number "spec".

[ProSoundWeb Community]