Distortion measurements, THD, IMD-what card?

[Eugene Dvoskin]

Hello,
I would like to ask you,what equipment are you using to measure distortions for your projects. I would expect that professional equipment is not affordable for most of DIY's. Are you using soundcards? What performance you can get from your card? Would like to figure out the best soundcard for the job. Looking for something affordable with good performance. Probably 24/192, to be able to measure up to 96 KHz
It also interesting, what software are you using together with your card.
Thanks.

 

[Pano]

A search of the forum will turn up a lot of good leads.

I've seen good results with the M-Audio Audiophile card and ARTA or AudioTester software.
At home I use M-Audio USB cards but am limited to 96Khz.

 

[jackinnj]

Problem with soundcards is that the iron-man test of an amplifier is the distortion of higher frequency signals -- this is where amplifier problems really manifest themselves. Soundcards aren't going to give you a lot of information on the 2nd and 3rd harmonics of a 19kHz signal.

If you're starting out, I would suggest looking at the Heathkit THD% and IM analyzers -- further -- take a look at Bob Cordell's "Distortion Magnifier" as well.

If you have some ability to program, Texas Instruments has some great USB digital signal processing development and mother boards. They come with a bit of canned software and the results I obtained were pretty darned good. You can get a lot of utility out of this stuff.

 

[jcx]

seeing higher harmonics directly can be helpful - but we really should be interested in audio frequency IMD - and we can measure that to very high resolution with prosumer soundcards and some helper circuitry

I have made indirect measurements of ~160 dB resolution with a US$130 [email protected] soundcard by looking for the difference frequency IMD products with a +30 dB gain/filter stage - part of the trick is to sum single tones from separate channels to avoid the IMD generation in the DAC and on card buffer circuitry

the [email protected] does 192K so its really as good as the "standard" audio 80KHz analyzers

 

[Eugene Dvoskin]

[email protected] looks like a good choice. I have PCM 2902 USB evaluation board from TI. It works very well for 16 bit and shows about 0.004% of THD (DAC to ADC), close to full range.
But I wanted to get like 10X better distortion reading and wider bandwidth. I think that 80-90KHz bandwidth should be good enough for most measurements. Main concern-linearity of converter and noise.

Any other than [email protected] good cards?

 

[Pano]

Well there is stuff like the PicoScope. I have the 3224 model here, it does 10Mhz, but is limited to 12 bits. Still pretty useful for FFT, tho.

Not in the same price range as the soundcards.

 

[Eugene Dvoskin]

Well there is stuff like the PicoScope. I have the 3224 model here, it does 10Mhz, but is limited to 12 bits. Still pretty useful for FFT, tho.

Not in the same price range as the soundcards.

12 bit resolution is limiting distortion measurements to 0.025% at the best. What are you getting with PicoScope?

 

[MiiB]

what software to use with the [email protected]

 

[Eugene Dvoskin]

E-MU 0202, 0204 and 0404 USB2 seems to be an interesting option. Portability and good performance. Unfortunately, they are using AK AD/DA. As I understand, some of Burr-Brown products are offering better THD performance.
But 0404 has SPDIF in/out, so it could be easily expanded with external AD/DA if required.
Has anybody tried one of these products?
Have you seen any other usb cards, supporting 24/192?
Or just USB to SPDIF in/out, supporting 24/192?

 

[Pano]

What are you getting with PicoScope?

I dunno. Here is a quick look at my computer soundcard playing a 1KHz tone.
Pico set to 156Khz limit, zoomed in. 1M termination (the scope).

It does allow you to see noise and other stuff pretty low.

 

[jcx]

RMAA has useful free version

Audacity open source sound editor lets you play/record

Ltspice can generate complex test .wav, and import the recorded response for analysis with its fft and .meas

SciLab is a free MatLab workalike for more complex signal processing of the .wav data

I've used all of the above freeware with the [email protected]

another interesting looking sw is Audio DiffMaker for null testing/residual difference meas with time alignment, FR correction

 

[Eugene Dvoskin]

I dunno. Here is a quick look at my computer soundcard playing a 1KHz tone.
Pico set to 156Khz limit, zoomed in. 1M termination (the scope).

It does allow you to see noise and other stuff pretty low.

Looks very good for 12bit. I'm just curious, why they name their DAC 12bit-16bit. THD also looks much better than I would expect from 12 bit.

 

[Pano]

Beats me.

It's not mine, I just borrow it from time to time. I do find it very handy.
Some of their stuff goes much higher, but is only 8 bit. 10Mhz is plenty for most audio use. For Class-D higher could be better, but even this one helps.

 

[BV]

I am using for THD an IMD measurements Lynx L22 card and ARTA software (free)
Some results with loopback

 

[Frex]

begemot61 ,

Hello,
Of my own experience, the AKM AD/DA converters are really excellent. If you find a sound card that use the AK5394A ADC you will get one of the better ADC system with very low THD specs.
TI or Cirrus IC aren't better, only maybe the ESS sabre ADC(ES9102), but not many infos are available on this IC and they are locked...
About your needs, i think too that sound card like [email protected] or others is very good starting point for low budget(less than 150$).
If you have more, Pico-scope is good too, but not exactly same instruments.
Wider bandwidth, lower resolution. It's depend on what you want to do with.
I personnaly use the digitial I/O of [email protected] sound-card (supporting 24/192l SR), with my DIY external ADCs. It's more easy to use than analog I/O of sound card (BNC inputs,+/-10V FS,DC coupled,..).
You can see more about this here.

I recommend you to absolutely avoid USB sound card for 192kHz operation.
This high sampling rate require high USB bandwidth that even USB2 don't work fine.
And of course, USB2 ask high processor resources.
For an external ADC/DAC, prefer Firewire you will avoid many problems.

About analysis software, two are very very good;
Spectraplus (not free, a 30 days free trial can be downloaded)
Oscillometerz (not free, shareware version fully working but stop analysis after 15s)
Many others FFT software exist, but i don't have find better than this.



Regards.


Frex

(Note; if i'm not wrong i think that ADC use in Lynx L22 of previous post is AK5394 from AKM !)

 

[Eugene Dvoskin]

begemot61 ,

Hello,
Of my own experience, the AKM AD/DA converters are really excellent. If you find a sound card that use the AK5394A ADC you will get one of the better ADC system with very low THD specs.
TI or Cirrus IC aren't better, only maybe the ESS sabre ADC(ES9102), but not many infos are available on this IC and they are locked...
About your needs, i think too that sound card like [email protected] or others is very good starting point for low budget(less than 150$).
If you have more, Pico-scope is good too, but not exactly same instruments.
Wider bandwidth, lower resolution. It's depend on what you want to do with.
I personnaly use the digitial I/O of [email protected] sound-card (supporting 24/192l SR), with my DIY external ADCs. It's more easy to use than analog I/O of sound card (BNC inputs,+/-10V FS,DC coupled,..).
You can see more about this here.

I recommend you to absolutely avoid USB sound card for 192kHz operation.
This high sampling rate require high USB bandwidth that even USB2 don't work fine.
And of course, USB2 ask high processor resources.
For an external ADC/DAC, prefer Firewire you will avoid many problems.

About analysis software, two are very very good;
Spectraplus (not free, a 30 days free trial can be downloaded)
Oscillometerz (not free, shareware version fully working but stop analysis after 15s)
Many others FFT software exist, but i don't have find better than this.

Regards.

Frex

(Note; if i'm not wrong i think that ADC use in Lynx L22 of previous post is AK5394 from AKM !)

Thank you.
Lynx card showed really excellent results. But it is not in my preferred price range. [email protected] sound-card is probably the best option for me with very good performance.
Speaking about usb, is it so bad for modern multi-core PC? This is not really mastering application. Usb2 works OK while transferring big data arrays for external HDD. Is latency so important for THD measurements? Or I'll get glitchy data that the software would not be able to properly process?
What if I will not use it for duplex mode and use external test signal generator.

Probably, the most convenient way for measuring system is to use SPDIF in/out and build external AD/DA with some buffers and attenuators. This is not very complicated and some sort of additional front end would be convenient for any sound card. Are there any options for just SPDIF board or USB-SPDIF, that support 24/192?
Regards,
Eugene

 

[mt490]

ARTA software (free)

As far as free packages go I like ARTA as well, it's a bit better featured than RMAA.

 

[Frex]

Hello begemot61,

Yes, a [email protected] sound card is the more cost effective solution, and these AD/DA converters are very good.
About USB link, if you use high sampling rate, even with powerful computer you risk to get discontinuous audio data that is problematic for FFT processing.
Of course, it is worse with full-duplex mode....
With external HDD you can achieve high data rate, and fall of data rate is not a problem. Contrariwise, with external USB sound card, if at any time the data rate fall below what it need, you will have loss of data!
These problem is not a concern with internal PCi card or with firewire interface.
But, of course you can try it at your own risk.

Note too that if you use external generator instead of sound card DAC out(full duplex mode), you will not easily do some measurements that allow full-duplex mode, like Bode plot (frequency response)...

And last, you're right. Using SPDIF I/O of sound card ([email protected]!) with external A/D unit is the best solution. It's that i do.
Did you have read about my AA5381external ADC project ?



Frex.

 

[Eugene Dvoskin]

Hi Frex,
Yes, I looked at you ADC project. I was not able to download schematic (for some reason it was forbidden ) but it is clear what it is. I'm thinking in the same direction. And you right, it is not completely clear what you will get, until you build it and test. I was thinking about more complicated front end, some built in attenuators, but other than this it is very similar to yours.
I also liked your evaluation of LT AN67 generator. You results look good, but I'm thinking of the structure that is more like old Bob Cordell's generator. It has some reserves left, while using modern components. And I'm thinking of different amplitude stabilization method as soon as I don't need fast amplitude setting (not going to do F sweep, just a couple of fixed frequencies). This may allow to redo amplitude diagram, and, may be, get better THD and noise. Not finalized my thoughts and have not done any simulations yet.

Happy New Year!

Thanks,
Eugene

 

[Ian Finch]

Picoscopes

Just a further word on Picoscope, their 4000 series scopes are up to 16-Bit and that gets me down another order or so and quite sufficient for DIY work. So as not to seem biased, there are similar products from The Netherlands called Handyscope, by Tie-Pie Engineering which offers this resolution and possibly higher now.

However, the software capabilities of the PCB card scopes are really tempting because they offer so many graphical variations which are commonly posted here. Agree about how good [email protected] cards are but there are some other well shielded cards of 24/192 capability now, even in the big-name-brand domestic market. To realise anything approaching 24 bit you need a really good interface/ attennuator with cards. This little extra can cost unless you DIY as per some excellent threads here. Pete Millet's recent creation comes to mind as the ultimate.