Print

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

[Rob Spence]

To expand a little, Rob, Russ is saying that without applying a delay to the reference signal, it will be "ahead" of the measurement signal by virtue of AD/DA conversion and latency created by the processing of the DR.

Use the delay finder.

Ok, will do. Note that both inputs are going through DSP. Reference through the old DSP and measurement through new one.

[Tim McCulloch]

Ok, will do. Note that both inputs are going through DSP. Reference through the old DSP and measurement through new one.

I suggest you do a loop back test for each DSP, whichever has the greatest delay should be your measurement side when you compare.

You need to save the traces on screen, then manipulate the DUT until the magnitude and phase traces overlay the original DSP traces.  You don't have to go back and forth to BOTH processors in real time.

[Rob Spence]

So, today I set up to measure again. I stuck a Y cable  in the Pink line and sent it to a dsp and to the reference input to SMAART. For today I only looked at the HF section and so took the HF output of the dsp to the measurement input.
I did this for each dsp, setting the delay for each. I was surprised that my old one had 8.88ms of delay while the new one was only 2.4ms.

Following is a screen capture of the two measurements.

The purple is the NAV480 and the pink is the Venue360 loaded with the same (as close as allowed) values as the Sabine.

I am not sure I need to tweak anything at this point before I set up the rig (after the heat wave) and do an acoustic measurement. I will look at the LF sections first but I wanted to get some feedback from y'all first.

[David Sturzenbecher]

So, today I set up to measure again. I stuck a Y cable  in the Pink line and sent it to a dsp and to the reference input to SMAART. For today I only looked at the HF section and so took the HF output of the dsp to the measurement input.
I did this for each dsp, setting the delay for each. I was surprised that my old one had 8.88ms of delay while the new one was only 2.4ms.

Following is a screen capture of the two measurements.

The purple is the NAV480 and the pink is the Venue360 loaded with the same (as close as allowed) values as the Sabine.

I am not sure I need to tweak anything at this point before I set up the rig (after the heat wave) and do an acoustic measurement. I will look at the LF sections first but I wanted to get some feedback from y'all first.

Rob,
I have measured the transfer function of a lot of DSPs, and i have never seen one produce responses so jagged (that didn’t have FIR).  These sure look like acoustical measurements.

[Russell Ault]

Rob,
I have measured the transfer function of a lot of DSPs, and i have never seen one produce responses so jagged (that didn’t have FIR).  These sure look like acoustical measurements.

I was going to say.

Rob, the approach you're describing (measuring the old DSP on its own, capturing that trace, and then measuring the second on its own while adjusting it to match) is the right approach, but what DSP settings are you using that are producing those traces? It looks like you're using a lot of filters.

Also, did you turn off the coherence trace?

-Russ

[Rob Spence]

I was going to say.

Rob, the approach you're describing (measuring the old DSP on its own, capturing that trace, and then measuring the second on its own while adjusting it to match) is the right approach, but what DSP settings are you using that are producing those traces? It looks like you're using a lot of filters.

Also, did you turn off the coherence trace?

-Russ

There are 4 filters for each band plus the HP & LP of the crossover.
I realized that I didn’t apply any smoothing. Been a while since I did this.

I turned off the coherence trace for the screen shot. I would not expect much coherence given the output is expected to be much different than the input?

[David Sturzenbecher]

There are 4 filters for each band plus the HP & LP of the crossover.
I realized that I didn’t apply any smoothing. Been a while since I did this.

I turned off the coherence trace for the screen shot. I would not expect much coherence given the output is expected to be much different than the input?

If your measurement is purely in the electrical domain, you don't need any smoothing.  Something is wrong with your setup.   It shouldn't be jagged like that at all.  It should look nearly identical to the EQ curve pictures on the screen.

[Russell Ault]

I turned off the coherence trace for the screen shot. I would not expect much coherence given the output is expected to be much different than the input?

Coherence is a measure of how likey the two signals are to be related, not just how similar they are. When measuring something like an EQ coherence should be 100% for every frequency in the passband (if your delay is set properly), dropping down only when a pass filter has dropped the signal close to the noise level.

There are 4 filters for each band plus the HP & LP of the crossover.

Just so we're clear, is your measurement signal just a cable running directly from the output of the DSP to the input of your USB interface, or is there a speaker and a measurement microphone in your test setup as well?

-Russ

[Rob Spence]

Coherence is a measure of how likey the two signals are to be related, not just how similar they are. When measuring something like an EQ coherence should be 100% for every frequency in the passband (if your delay is set properly), dropping down only when a pass filter has dropped the signal close to the noise level.

Just so we're clear, is your measurement signal just a cable running directly from the output of the DSP to the input of your USB interface, or is there a speaker and a measurement microphone in your test setup as well?

-Russ

Thanks by the way to everyone who is helping me with this. I really do appreciate it.

The setup is as follows:

MacBook connected to USB-Pre.
Pink noise from SMAART V7 to the output.
One output of the USB-pre through a Y cable to the dsp & and an input.
The output of the dsp to the other input.

My system sounds amazing with current settings on old DSP. After 10 years I had not needed to make any changes other than to back line delay depending on the venue. I want to make new dsp behave the same.

My first attempt at setting delay got me strange negative numbers (like 500ms+) so I swapped inputs and then got what seemed to be realistic numbers - 8.88ms for the old unit and 2.4ms for the dbx Venue360.

Please ask anything. I took SMAART classes a long time ago and only used it a few times in the meantime. I find the spectrograph useful for determining splay when using multiple speakers.

Anyway, thanks again for helping.

[Russell Ault]

MacBook connected to USB-Pre.
Pink noise from SMAART V7 to the output.
One output of the USB-pre through a Y cable to the dsp & and an input.
The output of the dsp to the other input.

Okay, that's definitely the correct setup, but the traces don't look right at all for the filters you're describing. What does the trace look like if you bypass everything in the DSP (including the HPF and LPF)?

-Russ

[ProSoundWeb Community]