ProSoundWeb Community

Sound Reinforcement - Forums for Live Sound Professionals - Your Displayed Name Must Be Your Real Full Name To Post In The Live Sound Forums => Audio Measurement and Testing => Topic started by: Johnny Diaz on November 12, 2011, 11:55:53 am

Title: Using Smaart to determine delays on a processor
Post by: Johnny Diaz on November 12, 2011, 11:55:53 am
I would like to use Smaart to determine how much delay there is from the input stage to the output stage of my dsp. How would I accomplish this?

Title: Re: Using Smaart to determine delays on a processor
Post by: Timo Beckman on November 12, 2011, 12:14:29 pm
I would like to use Smaart to determine how much delay there is from the input stage to the output stage of my dsp. How would I accomplish this?

Hook up smaart to the input of the processor and the output of the processor to your soundcard and put pink noise trough it (turn of phantom power if you do this) . Hit auto small/large . The value you get is the processor latency .
Title: Re: Using Smaart to determine delays on a processor
Post by: Johnny Diaz on November 12, 2011, 12:49:30 pm
Hook up smaart to the input of the processor and the output of the processor to your soundcard and put pink noise trough it (turn of phantom power if you do this) . Hit auto small/large . The value you get is the processor latency .

So I would connect my reference signal to the input of the dsp and the measurement signal to the output of the dsp?  Would the pink noise be through smaart? 
Title: Re: Using Smaart to determine delays on a processor
Post by: Doug Fowler on November 12, 2011, 12:55:04 pm
So I would connect my reference signal to the input of the dsp and the measurement signal to the output of the dsp?  Would the pink noise be through smaart?

Split the reference signal between the DSP input and reference channel for your analyzer.  DSP output is the measurement channel.  You can use the noise generator if you wish.

Title: Re: Using Smaart to determine delays on a processor
Post by: Ivan Beaver on November 12, 2011, 01:30:03 pm
Split the reference signal between the DSP input and reference channel for your analyzer.  DSP output is the measurement channel.  You can use the noise generator if you wish.
Or any other input signal-be it a sine wave-music-whatever.

All that most of the various systems are doing is comparing a known input to an unknown output-and showing the difference-be it in time-phase-amplitude etc.

With of course a bunch of various measurement parameters associated with the resultant measurement-windowing-number of samples-averaging-weighting etc.

The same difference can look very different-depending on the parameters of the above factors.
Title: Re: Using Smaart to determine delays on a processor
Post by: Paul Tucci on November 12, 2011, 11:38:50 pm
I would like to use Smaart to determine how much delay there is from the input stage to the output stage of my dsp. How would I accomplish this?

Johhny,

I prefer to answer in general terms rather than with specifics. Yours is a relationship question. What you're asking to do is find the relative timing difference between what you have in this hand compared to what you have in that hand.  This timing measurement has two points of interest. That's where you put the probes. One is "before" and one is "after" the device under test. In your case, the device under test is the DSP. The REFERENCE side of the measurement is the "before" the DSP so grab that point of interest as Doug said. That's a simple way to do it. Simple is good. The "after" is obviously the output of the DSP. This measurement (Auto Small/Large in earlier versions or Find in version 7) is in the Time domain. The answer we'll get back is the difference in arrival time between the two points of interest where we put the probes. If yours were a Frequency Response question, we would still have two points of interest and need two probes. ie, if you were to ask " Does + 6dB at 80HZ line on my equalizer actually produce a 6dB electronic boost?"  Grab two points of interest, one "before" the EQ for your Reference and one  "after" the EQ for your measurement side. 

I found the two probe concept helpful in my understanding of separating a single channel from a dual channel comparison so I offer that up to you.

Your experiment is exactly what sparked my interest  years back. Do this for us if you would please. After you get your "What's the arrival time difference between the input and output of my DSP?" measurement done, and you feel confident it's valid, try adding a 100Hz Low pass filter to your DSP and measure again. Report back.

Happy measuring

Tucc

(added after the fact to further the two probe / dual channel measurement concept)...

These dual channel comparisons happen often in the course of a gig day. Firstly, after finding Sparky, you measure the voltage between two electrical points of interest. Ground and a hot leg, then ground and another hot leg, then yet another hot leg, ground to neutral, hot to hot, hot to the other hot. These are all two channel measurements that answer voltage difference questions. So is getting change from the gas station after fueling the truck. The Reference channel is the price, the Measurement channel is the amount paid, the difference is the change.

Title: Re: Using Smaart to determine delays on a processor
Post by: Doug Fowler on November 13, 2011, 02:22:13 pm

Your experiment is exactly what sparked my interest  years back. Do this for us if you would please. After you get your "What's the arrival time difference between the input and output of my DSP?" measurement done, and you feel confident it's valid, try adding a 100Hz Low pass filter to your DSP and measure again. Report back.


Heh heh, here we go ....
Title: Re: Using Smaart to determine delays on a processor
Post by: Art Welter on November 17, 2011, 05:29:58 pm
Heh heh, here we go ....
Yup, after hearing that the Driverack PA + had the mid output (or was it the HF) reverse polarity, decided to check my older DRPA, it fortunately has LF,MF, and HF all in polarity.
However, the LF,MF, and HF are all reversed from the input polarity :^(.
Title: Re: Using Smaart to determine delays on a processor
Post by: Preston Soper on November 30, 2011, 10:01:38 pm



Your experiment is exactly what sparked my interest  years back. Do this for us if you would please. After you get your "What's the arrival time difference between the input and output of my DSP?" measurement done, and you feel confident it's valid, try adding a 100Hz Low pass filter to your DSP and measure again. Report back.
Heh heh, here we go ....

XTA224
SoundDevices USB Pre
MBP(Boot in 32bit)
V7.3.0.2
Averaging:16
Phase Smoothing: None
Mag Smoothing: None
Weighting: None
FFT: 16k


Here are some screenshots of the experiment Paul suggested.
Title: Re: Using Smaart to determine delays on a processor
Post by: Paul Tucci on November 30, 2011, 10:07:31 pm




XTA224
MBP
V7.3.0.2
Averaging:16
Phase Smoothing: None
Mag Smoothing: None
Weighting: None
FFT: 16k


Here are some screenshots of the experiment Paul suggested.

Preston,

Glad you're playing along at home.  Try a 100 Hz low pass instead. ;)  Measure the time arrival of that and a full range signal. (Impluse Response) That's where it started. 

PT
Title: Re: Using Smaart to determine delays on a processor
Post by: Preston Soper on November 30, 2011, 10:10:58 pm
Preston,

Glad you're playing along at home.  Try a 100 Hz low pass instead. ;)  Measure the time arrival of that and a full range signal. (Impluse Response) That's where it started. 

PT

Wow,
Missed that one I thought your previous post was HPF not LPF.
Silly me :)

XTA224
SoundDevices USB Pre
MBP(Boot in 32bit)
V7.3.0.2
Averaging:16
Phase Smoothing: None
Mag Smoothing: None
Weighting: None
FFT: 16k

Here is the example in Transfer mode:
Delay times ranged from 6.60ms to 7.65ms
The trace below is @7.21ms
Title: Re: Using Smaart to determine delays on a processor
Post by: Preston Soper on November 30, 2011, 10:41:06 pm
Preston,

Glad you're playing along at home.  Try a 100 Hz low pass instead. ;)  Measure the time arrival of that and a full range signal. (Impluse Response) That's where it started. 

PT

XTA224
SoundDevices USB Pre
MBP(Boot in 32bit)
V7.3.0.2
Averaging:16
Phase Smoothing: None
Mag Smoothing: None
Weighting: None
FFT: 16k
Time Constant: 341ms

Delay Time  7.1 to 7.6 ms
Peak varied with each measurement.

Title: Re: Using Smaart to determine delays on a processor
Post by: Paul Tucci on December 01, 2011, 08:37:41 am
and the full range measurement??
Title: Re: Using Smaart to determine delays on a processor
Post by: Preston Soper on December 01, 2011, 09:22:42 am
and the full range measurement??

XTA224
SoundDevices USB Pre
MBP(Boot in 32bit)
V7.3.0.2
Averaging:16
Phase Smoothing: None
Mag Smoothing: None
Weighting: None
FFT: 16k
Time Constant: 341ms

10Hz-LR-24db - 22K-LR-24db
Delay @ Peak 2.2ms

Title: Re: Using Smaart to determine delays on a processor
Post by: Paul Tucci on December 01, 2011, 11:05:31 am
The point being that the same DSP exhibited different latency as measured by the Impulse Response of Smaart.
The full range signal showed a latency of 2.2 mSec, the low passed signal showed a latency of 7+ mSec.

How can that be?


...and so it all began...

PT
Title: Re: Using Smaart to determine delays on a processor
Post by: Martyn ferrit Rowe on December 01, 2011, 12:18:57 pm
And when you shared this with me a long time ago a small fire started............. 8)

ferrit
Title: Re: Using Smaart to determine delays on a processor
Post by: Doug Fowler on December 01, 2011, 12:29:51 pm
The point being that the same DSP exhibited different latency as measured by the Impulse Response of Smaart.
The full range signal showed a latency of 2.2 mSec, the low passed signal showed a latency of 7+ mSec.

How can that be?


...and so it all began...

PT

I use "auto delay" in ST class and sweep a low pass filter to demonstrate the effect.  It tends to get the uninitiated thinking in a different direction.
Title: Re: Using Smaart to determine delays on a processor
Post by: Jay Barracato on December 02, 2011, 05:46:38 pm
The point being that the same DSP exhibited different latency as measured by the Impulse Response of Smaart.
The full range signal showed a latency of 2.2 mSec, the low passed signal showed a latency of 7+ mSec.

How can that be?


...and so it all began...

PT

So are you going to leave us guessing?

At 100hz my sabine navigator goes to 3.05 ms from 1.50 ms for the full range signal using the delay finder. It is a difference of 1.1 ms using the IR.

My best guess is that it is a function of the wavelength and the fact that the longer wavelengths vibrate slower.
Title: Re: Using Smaart to determine delays on a processor
Post by: Mac Kerr on December 02, 2011, 06:34:28 pm
So are you going to leave us guessing?

It's probably just a phase he's going through.

Mac