In the past few weeks I have been learning and testing REW for distortion measurements. I'm using the Linux version on a desktop PC with a Focusrite Solo USB interface. It's certainly great to have this available, especially at such a reasonable cost, so many thanks to John M. and all the contributors here for their efforts.
I do have one question though, and forgive me if this has already been covered, as I have not yet read all 65 pages of this thread, so here goes....
Has anyone had the opportunity to verify the accuracy of the REW distortion measurements against calibrated hardware distortion analyzers?
I don't mean this in a negative sense, I'm just wondering if it has ever been done. I have always worked in industries where every piece of equipment had to be calibrated or validated, so it's kind of a common question. TIA.
David
I do have one question though, and forgive me if this has already been covered, as I have not yet read all 65 pages of this thread, so here goes....
Has anyone had the opportunity to verify the accuracy of the REW distortion measurements against calibrated hardware distortion analyzers?
I don't mean this in a negative sense, I'm just wondering if it has ever been done. I have always worked in industries where every piece of equipment had to be calibrated or validated, so it's kind of a common question. TIA.
David
I would think the limit is the resolution of your soundcard. I have used ARTA and REW and the results were similar.
I have access to an AP2722 at work and Arta/Rew at home. If your interface is correctly calibrated and the software correctly configured it should be more or less the same measurements concerning distorsion. Cables are extremely important, good shielding, good ground, signal... I find Arta to be more accurate in general, at least to this AP, but it's negligible. I always use asio drivers, low samplerate 44-48, and no extra software correction/compensation whatsoever. I am confident about this AP measurements, always measured well vs datasheets. You obviously need a good interface and measure distorsions at least 10 times your loopback numbers. At home if it's less than 0.005% thd+n at 1k, then it's less than 0.005% at 1k. At work if it less than 0.001% thd+n 20-20k, then it's less than 0.001% thd+n 20-20k. The AP calibrates automatically on startup and as far as I can remember it has never been calibrated or validated by someone.
Forgot to add: with diy, find the input/output level sweetspot and stick to it.
Last edited:
Another way to look at accuracy assessment is to externally measure the rms voltage of a sinewave, and compare that to REW's RTA level (perhaps use a Y-axis of V or dBV for easier association). Then adjust the sinewave level to 10% of original and note the drop in RTA level - continue for 1% and 0.1% etc of original, although you may quickly run in to limitations from your signal source depending on what equipment you have. As long as harmonic levels are quite low, and your external meter is suitably accurate at the measurement levels being asked of it, then you likely have confirmation of the relative accuracy of the RTA Y-axis scale, and hence by definition the % distortion measurement accuracy as that is derived from harmonic level accuracy compared to fundamental.
Nowadays we just assume that a soundcard's ADC is linear from its LSB to MSB FS levels, and don't try and dig out the IC datasheet that may well indicate that level of non-linearity. Of course your measurement linearity is not just reliant on the ADC, but also any op-amp gain and passive attenuation within the soundcard as well as any probing circuitry.
Nowadays we just assume that a soundcard's ADC is linear from its LSB to MSB FS levels, and don't try and dig out the IC datasheet that may well indicate that level of non-linearity. Of course your measurement linearity is not just reliant on the ADC, but also any op-amp gain and passive attenuation within the soundcard as well as any probing circuitry.
You might be interested in this discussion on another forum:
https://www.audiosciencereview.com/...c-adc-chip-currently.13469/page-6#post-719529
Starting from this post but there are some posts on next pages, including posts from the author of REW .
Summing up, with right settings REW should be as accurate as AP software.
@trobbins :
Yes, this is also what i throught, and why i try to use a notch, and @YashN allready explained that i should do trace arithmetic.
But i still don't get it right.
Let's say i do the trace arithmetic + mic. cal file.; then i will feed (for testing) the Behringer with 1khz from Victor's through the notch-filter & with mic. cal file also ofcause.
But what about the 1khz spike, which normally shows up on the RTA window, how will i know how much i will have to turn up the input level on the Behringer
... Sry. guy's i simply cannot get this into my head.
@YashN
The schematic is this one :
I hope Victor don't mind, but i'am sure he doesen't.
The resistor values marked with * i was told is to compensate if the capacitor's are not excatly 22nF.
Jesper.
Yes, this is also what i throught, and why i try to use a notch, and @YashN allready explained that i should do trace arithmetic.
But i still don't get it right.
Let's say i do the trace arithmetic + mic. cal file.; then i will feed (for testing) the Behringer with 1khz from Victor's through the notch-filter & with mic. cal file also ofcause.
But what about the 1khz spike, which normally shows up on the RTA window, how will i know how much i will have to turn up the input level on the Behringer
... Sry. guy's i simply cannot get this into my head.@YashN
The schematic is this one :
I hope Victor don't mind, but i'am sure he doesen't.
The resistor values marked with * i was told is to compensate if the capacitor's are not excatly 22nF.
Jesper.
Thank you all:
@bucks bunny
@beyah
@trobbins
@BamboszeK
plus all that may follow for sharing your experiences and insights. It's a big help.
@bucks bunny
@beyah
@trobbins
@BamboszeK
plus all that may follow for sharing your experiences and insights. It's a big help.
Hey here.
It's an amazing piece of tool such a setup with soundcard, loadresistor's etc... should have build one years ago, but well...
I'am tuning my "shop"-amp. right now (LM4780 cloneamp.), changing the grounding and so, to have much better RTA/FFT cause of that.
Eventhrough i used a lot of hour's reading (sorry if i missed some of it in this thread) i have some quistion's i would kindly ask here.
If i look at my typical "loopback" with a 1khz sine from Victors osc. it will look like this one :
I see that the soundcard is "producing" some THD's, not much, but some.
If i then measure an poweramp, it might be that this poweramp's noisefloor are very low, so that the soundcards own THD cannot hide there.
EDIT :: Soundcard is modded Behringer UMC202hd
Jesper.
It's an amazing piece of tool such a setup with soundcard, loadresistor's etc... should have build one years ago, but well...
I'am tuning my "shop"-amp. right now (LM4780 cloneamp.), changing the grounding and so, to have much better RTA/FFT cause of that.
Eventhrough i used a lot of hour's reading (sorry if i missed some of it in this thread) i have some quistion's i would kindly ask here.
If i look at my typical "loopback" with a 1khz sine from Victors osc. it will look like this one :
I see that the soundcard is "producing" some THD's, not much, but some.
If i then measure an poweramp, it might be that this poweramp's noisefloor are very low, so that the soundcards own THD cannot hide there.
- Will i then see the sum of both soundcard and poweramp then ?
- I belive that if measuring an amp with a much higher noisefloor, this does not matter? - is this correct ?
EDIT :: Soundcard is modded Behringer UMC202hd
Jesper.
Nice numbers, the harmonics you are seeing are certainly from your soundcard input given the specs of Viktor's gen. A poweramp with such specs would be a very, very quiet and distorsion free amp! In reality you're probably going to have much more THD+N at an amplifier output, so these numbers you get in this direct setup will be negligible.Hey here.
It's an amazing piece of tool such a setup with soundcard, loadresistor's etc... should have build one years ago, but well...
I'am tuning my "shop"-amp. right now (LM4780 cloneamp.), changing the grounding and so, to have much better RTA/FFT cause of that.
Eventhrough i used a lot of hour's reading (sorry if i missed some of it in this thread) i have some quistion's i would kindly ask here.
If i look at my typical "loopback" with a 1khz sine from Victors osc. it will look like this one :
View attachment 1051632
I see that the soundcard is "producing" some THD's, not much, but some.
If i then measure an poweramp, it might be that this poweramp's noisefloor are very low, so that the soundcards own THD cannot hide there.
Thank's all
- Will i then see the sum of both soundcard and poweramp then ?
- I belive that if measuring an amp with a much higher noisefloor, this does not matter? - is this correct ?
EDIT :: Soundcard is modded Behringer UMC202hd
Jesper.
Nice FFT overall - very clean baseline devoid of mains 50/60Hz, but I think everything you see there is an artifact of your Behringer if you are using Victors osc. The 5th harmonic is especially bad - even though quite low, is an odd higher order “mechanical” sound and may be slightly audible to some sensitive ears if this were used as the DAC.
Hey. Thank's for the feedback guy's!
Changing the GND connection's, for improved 50Hz hum in this old CloneAmp.
I was never able to do this before!
Eventhrough THD numbers didn't change a lot, the 50Hz peaking-through the amp. is lower now.
There must be some more modding to do with my other amp's now
EDIT:: Sometime REW trick's me, and having the 1KhZ off, i therefore replaced the picture, where the mod was shown.
Jesper.
Changing the GND connection's, for improved 50Hz hum in this old CloneAmp.
I was never able to do this before!
Eventhrough THD numbers didn't change a lot, the 50Hz peaking-through the amp. is lower now.
There must be some more modding to do with my other amp's now
EDIT:: Sometime REW trick's me, and having the 1KhZ off, i therefore replaced the picture, where the mod was shown.
Jesper.
Last edited:
Ryssen, you may find it worthwhile to do some reading of previous thread posts to better understand how the soundcard is the principle generator of distortion when doing distortion testing using REW. For a loopback setup, the soundcard's DAC and related circuitry will include its own distortion, and similarly the ADC and its related circuitry will add to that inherent distortion. Even if you generate a signal externally, REW will show the soundcard's ADC side circuitry contribution to distortion, which may dominate for a very low distortion signal from the likes of Victor's oscillator, or be dominated by the external source like your phone derived signal.
Using REW relies on you providing adequate signal conditions to be confident of what you are trying to measure. Your first plot shows that even the fundamental signal appears to be buried in the noise floor of your measurement. If you read the thread then you can come across ways to suppress the noise floor, especially when trying to identify harmonic levels in a loopback measurement, such as using the coherent averaging feature of REW.
Using REW relies on you providing adequate signal conditions to be confident of what you are trying to measure. Your first plot shows that even the fundamental signal appears to be buried in the noise floor of your measurement. If you read the thread then you can come across ways to suppress the noise floor, especially when trying to identify harmonic levels in a loopback measurement, such as using the coherent averaging feature of REW.
Hi all...
I'am playing with my notch filter, and i think i got it right.
Here is what i did (i know it's also stated at page42) :
Calibrated the voltage at the generator-window (705mV)
Took a sweep with notchfilter
Took a sweep without notchfilter
Trace arethmetic (With-notch / Without-notch) - This file saved and loaded as mic. calibration.
I then adjusted the oscilator for 1008mV and connected it like this : Oscilator ---> Notch ---> Behringer
Then i enter 1.008V in the manual fundamental
But i was told, and also see that the graph looks a little strange
I'am not sure if i missed something?
Another thing. How do i handle the FS sine Vrms dialog on top of RTA window ?
It's because, that when i test a poweramp, i typically load it with 2.8v for having 1w with my 8ohm power resistors.
So i suppose that i should enter 2.8v in the fundamental then ?
But if FS sine Vrms is set to say 1v, the dB will go positive (E.g. 5.35dB) - This is in the window where all the distortions numbers are. (Top left)
Jesper.
I'am playing with my notch filter, and i think i got it right.
Here is what i did (i know it's also stated at page42) :
Calibrated the voltage at the generator-window (705mV)
Took a sweep with notchfilter
Took a sweep without notchfilter
Trace arethmetic (With-notch / Without-notch) - This file saved and loaded as mic. calibration.
I then adjusted the oscilator for 1008mV and connected it like this : Oscilator ---> Notch ---> Behringer
Then i enter 1.008V in the manual fundamental
But i was told, and also see that the graph looks a little strange
I'am not sure if i missed something?
Another thing. How do i handle the FS sine Vrms dialog on top of RTA window ?
It's because, that when i test a poweramp, i typically load it with 2.8v for having 1w with my 8ohm power resistors.
So i suppose that i should enter 2.8v in the fundamental then ?
But if FS sine Vrms is set to say 1v, the dB will go positive (E.g. 5.35dB) - This is in the window where all the distortions numbers are. (Top left)
Jesper.
Attachments
