[Crowbar]

Is there any sound card with lower than -100 dB noise/distortion of its input (preferably lower than -110 dB)? I want to avoid renting an Audio Precision to do distortion measurements since it's over two thousand bucks a month where I live. And I mean not just for 1 kHz, but at least 10 kHz. Most hardware has a high rise in distortion in the upper frequency range.

 

[borisov57]

Hello,

Various E-MU cards have good SNR and THD. You can search google for RMAA results for audio cards. Or you can made your own card based on good quality AD converter ( TI, Crystal, AKM ).

 

[jcx]

ESI Juli@ $130 low cost performance

EMU 1820M ~$400 a few less spurs, than Juli@

Lynx Studio ~$1K very good performance but still just using the best of Crystal/Cirrus monolithic audio adc/dac chips

I haven't heard of new competetion for these ~4yr old products unless you want to jump up to near AP level pricing

NI 5922 claims 114 dB SFDR @10 KHz input but cost is ~10K +/-

 

[Crowbar]

I wonder if this NI digitizer would be available for rent somewhere. Renting AP is about $2400 per month here, can't get any worse than that.

 

[amb]

E-MU 0404 USB is quite good. See my RMAA test results.

 

[pmillett]

I'm getting better than -130dBU with an M-Audio Audiophile 192.

See this thread on DIY Audio:

diyAudio Forums - Audiotester + M-audio - info + question

Pete

 

[jcx]

indirect Intermodulation measurements can get near the spot noise floor (~ -130 dB) for resolving specific orders of nonlinearity

add some narrow filters, low noise gain and detecting -160 dB distortion products is possible even with $130 Juli@

you just have to make the personal time vs debugged/traceable commercial solution trade-off decision

 

[heatmizer]

UsbPre:
Used in prosound really good unit.
USBPre Microphone Interface for Computer Audio | Sound Devices, LLC

 

[Crowbar]

Quote:

Originally Posted by jcx /img/forum/go_quote.gif indirect Intermodulation measurements can get near the spot noise floor (~ -130 dB) for resolving specific orders of nonlinearity add some narrow filters, low noise gain and detecting -160 dB distortion products is possible even with $130 Juli@ you just have to make the personal time vs debugged/traceable commercial solution trade-off decision

References to how to do this please? Basically I need these measurements for tweaking a discrete I/V design I'm working on.

BTW, anyone know how to measure jitter? I'm getting some custom clocks made with 0.3 ps jitter (from 10 Hz offset) and I want to see if the clock buffer/distribution parts I'm using add too much jitter on top of that.

 

[luvdunhill]

Quote:

Originally Posted by Crowbar /img/forum/go_quote.gif References to how to do this please? Basically I need these measurements for tweaking a discrete I/V design I'm working on. BTW, anyone know how to measure jitter? I'm getting some custom clocks made with 0.3 ps jitter (from 10 Hz offset) and I want to see if the clock buffer/distribution parts I'm using add too much jitter on top of that.

just curious, could I get a source for the custom clocks? Do you mean crystals, oscillators, or VCXO, or..?

 

[Crowbar]

XO or VCXO, I haven't decided yet (same jitter and price from what I understood of the quote). It's only cheap because I'm going to order a 100. I might organize a group order since I won't need that many (initially).

I need to measure if using a CDCV304 clock distributor will have a low enough additive jitter. That way I can avoid having to use the much more complex and pain-in-the-ass AD9510.

 

[jcx]

are you testing the DAC and I/V together? or using a lagrish R to turn a voltage wave source into a current input for the I/V stage alone?

One idea is to look for IMD products that can be well separated in frequency from the stimulus:
1:1 19KHz+20KHz gives 1KHz IMD product porportional to the 2nd order nonlinearity

10.5KHz + 20KHz gives 1KHz IMD portional to the third order distortion term ( 2*10.5 - 20)

you can arrange the difference frequency IMD product to fall in the passband of a filter/gain block that attenuates the source high frequenciy stimulus while amplifying the IMD product, easing the ADC requirements

There are some impressive speced ~1MHz hi res ADC chips from ADI, TI/BB and Linear in the past few years
TI/BB ADS8482 18 bit 1 MHz ADC claims -123 DB SFDR, -120 dB THD with 2KHz input, -110 dB THD @20KHz in - you'd need bandwidth restriction and averaging to push the S/N down to resolve these levels though

 

[Crowbar]

Since the DAC performance is not known, I'll need to isolate the I/V contribution to distortion, and thus need an ultra-low distortion sine generator. Then I can add the DAC to the loop and have an idea of its contribution.

The 19k+20k is probably my best bet, since sound card ADC analog filter would attenuate high harmonics and make testing near 20k not give useful results. I'm going to try to get I/V harmonics to be all under -120 dB, though probably -110 I'd be happy with (for 2nd and 3rd at least).

 

[Cauhtemoc]

Quote:

Originally Posted by Crowbar Quote: Originally Posted by jcx add some narrow filters, low noise gain and detecting -160 dB distortion products is possible even with $130 Juli@ References to how to do this please?

See here:

Distortion Analyser
Cordell Audio: Papers: Build a THD Analyzer

Quote:

Originally Posted by Crowbar Since the DAC performance is not known, I'll need to isolate the I/V contribution to distortion, and thus need an ultra-low distortion sine generator.

See here for a low distortion oscillator:

Ultra low distortion audio oszillator the never ending search for a pure sine

Quote:

Originally Posted by Crowbar BTW, anyone know how to measure jitter? I'm getting some custom clocks made with 0.3 ps jitter (from 10 Hz offset) and I want to see if the clock buffer/distribution parts I'm using add too much jitter on top of that.

Make friends with someone who has a R&S FSUP Signal Source Analyzer.

 

[Crowbar]

Unfortunately those THD analyzers don't show the individual harmonics, which is critical since different harmonics have a very different perceptual weighting.

The oscillator looks great though.