I just realized the one your looking at is a TI part.
Check the offerings from AD and LT.
Ill try to get one from TI. Maybe they will let me borrow one. None are audio compatible or easy to use. I think HPworks can be persuaded to work with some of these. Otherwise its demo software or roll your own.
The AD ADC EVM offings come with a LabView library but Labview is very expensive on it's own. It's tempting.
Hi RNM,
No but I've been considering it for ADC design as such. However, unless I misinterprete the datasheet the usable frequency range (- 3dB) is ~ 0.3*Fs meaning about 115 kHz for a 384 kHz sample frequency (see top figure on page 14).
I have also considered the ADS1675 EVM module as this would allow for investigating up to about 2 MHz. One caveat, though, is that TI's ADCPRO only samples up to 1024000 samples on a contiguous basis.
Best regards,
Jesper
No but I've been considering it for ADC design as such. However, unless I misinterprete the datasheet the usable frequency range (- 3dB) is ~ 0.3*Fs meaning about 115 kHz for a 384 kHz sample frequency (see top figure on page 14).
I have also considered the ADS1675 EVM module as this would allow for investigating up to about 2 MHz. One caveat, though, is that TI's ADCPRO only samples up to 1024000 samples on a contiguous basis.
Best regards,
Jesper
Last edited:
I would suggest another candidate:
AD7760
I originally wanted to build my own analyzer based on this chip (With some PLL auto-tuned Fliege notch filter topology in front of it), so I got me the AD7760/62/63 EVAL-system and really the basic distortion performance of this chip is very good (any residuals below 118dB, H2 and H3 heavily dominating at -3dBfs), if you drive the inputs firmly or even a little better if you bypass the internal buffers or use a better oscillator as I did (A heavily tweaked Hameg HM8037). This ADC could also be an excellent Audio ADC if you are in DXD or the like and don't mind slightly higher noise floors ...
Btw: The EVAL-kit is now for sale, since I gave up and got me a "Bruce Hofer hand tweaked" SYS-2722A recently. Not the bandwidth but still "enough" for me.
Thank you for the lead/info. A high sample rate means less/no filters and then we can see what is there when filters/reconstruction filters are not used and any artifacts that might be up there which can cause issues with the DUT or IM etc. Have you learned anything from these ADC ?
Thx-RNMarsh
Last edited:
What did you learn from it and why did you abandon it? And, how good did you get your oscillator to work?
Thx-RNMarsh
Last edited:
Hello again,
@RNM:
No, not yet. I've been considering the TI EVM but have been deterred a bit by the price (USD 500 + customs) and the fact that it "only" records up to 1024k samples contiguously. Otherwise I would really like to use it.
Greetings,
Jesper
@RNM:
No, not yet. I've been considering the TI EVM but have been deterred a bit by the price (USD 500 + customs) and the fact that it "only" records up to 1024k samples contiguously. Otherwise I would really like to use it.
Greetings,
Jesper
Besides the huge bandwidth and low distortion, the AD7760 has another nice feature: no rise of noisefloor above 20 kHz. This nasty effect makes ordinary Audio ADC chips not very feasible for measurments of higher order harmonics of e.g. a 20kHz fundamental.
The Fliege bandpass/Notch has a pretty nice property:
It can be tuned by change of a single resistor/capacitor, if you don't mind change in Q. This makes building "auto" tunable circuits much easier.
Here is the concept (Never realized, only partially simulated/evaluated):
Generator:
120 dB Audio DAC + 20dB auto tuned Fliege Bandpass = 140 dB residual distortion + fast settling + very low phase noise, not possible with a fast settling analog generator.
Analyzer:
118 dB ADC (AD7760) + 20dB generator synced modified Fliege Notch with 20dB attenuation of fundamental = 138 dB effective residual
Both connected to an XMOS USB Audio interface = compatible with standard software (VA, audioT etc.).
-> Way too much effort, so this stays a dream ...
Hameg HM8037, EVAL board
The optimization of the HM8037 sinewave oscillator (Which I used for testing the AD7760 EVAL) was three steps:
- Minimization of output of the voltage multiplier (AD633 or the like) by tweaking the voltage feedback loop gain/phase, to run it close to a nominal gain of 0
- Using better OPAs (LME instead of TL072)
- Chosing better components for the timing capacitors
Without these mods the generator is around 0.002% at 20kHz (If you find a good one).
With the modified HM8037 and on a good day (freshly adjusted) you get < 125 dB residuals at 1kHz and <120 dB at 20 kHz
Not bad for a 250 EUR used price instrument, but the mod comes with some drawbacks:
- Stable operation only up to about 40kHz
- Some sidebands (ALC loop instability) for frequencies above 10kHz (This can be improved, but I was simply too lazy)
- The output has some "dirt" on it, which is OK for a 0.002% (stock) instrument, but not really for a 120dB one ...
BTW: The AD7760 EVAL system is now sold .... !
The optimization of the HM8037 sinewave oscillator (Which I used for testing the AD7760 EVAL) was three steps:
- Minimization of output of the voltage multiplier (AD633 or the like) by tweaking the voltage feedback loop gain/phase, to run it close to a nominal gain of 0
- Using better OPAs (LME instead of TL072)
- Chosing better components for the timing capacitors
Without these mods the generator is around 0.002% at 20kHz (If you find a good one).
With the modified HM8037 and on a good day (freshly adjusted) you get < 125 dB residuals at 1kHz and <120 dB at 20 kHz
Not bad for a 250 EUR used price instrument, but the mod comes with some drawbacks:
- Stable operation only up to about 40kHz
- Some sidebands (ALC loop instability) for frequencies above 10kHz (This can be improved, but I was simply too lazy)
- The output has some "dirt" on it, which is OK for a 0.002% (stock) instrument, but not really for a 120dB one ...
BTW: The AD7760 EVAL system is now sold .... !
Last edited:
Your oscillator mods and parts upgrades are similar --- old opamp and parts upgrades made for a pretty decent source. Is the A-P analog source significantly better? Seems the modified osc's thd/+n is comparable with their analog source.
[ is there still an American military outpost in your town? There used to be an army communications site there which we passed traffic thru via my Frankfurt site]
-Thx-RNMarsh
[ is there still an American military outpost in your town? There used to be an army communications site there which we passed traffic thru via my Frankfurt site]
-Thx-RNMarsh
Last edited:
Good concept... also-
And if you had multiple notches sync'ed and tracking, you could remove any residual 2H and 3H for a ultra pure swept sine wave source.
-RNM
Last edited:
Victor has made me an ultra-low distortion 90Hz oscillator. It has a 90 sec settling time which is OK with me. With his new 90Hz, and the existing 1KHz and 10KHz all put into a single box and all powered from a 35vdc >100ma P.S., you will have the basis of an excellent test source. If you like, add an 11KHz osc and then do IM testing also.
Thanks to Victor !
-Richard Marsh
Last edited:
@Richard Marsh,
... sounds interesting ... might you have a link for this?
Best regards,
Jesper
P.S.: No, please wait answering. I'd rather ask a more focussed question about a suitable oscillator when I'm ready for it ;-)
Jesper
... sounds interesting ... might you have a link for this?
Best regards,
Jesper
P.S.: No, please wait answering. I'd rather ask a more focussed question about a suitable oscillator when I'm ready for it ;-)
Jesper
Last edited:
No link... been doing similar manually. Need to automate it only... maybe with this notch circuit as basis... was just an idea.
-THx RNMarsh
Would you mind to post screen shots of the two monitor outputs of the VP7722A ( via your HP spectrum analyzer )
and the readings of 2fo, 3fo etc. ?
BR
Gary
@JonLord: Might I ask you what is the reason you didn't connect the AD7760Evalboard to AD's converter-to-PC evalboard?
http://www.analog.com/static/imported-files/eval_boards/EVAL_CED1Z.pdf
Seems to me the software connects directly to the evalboard and thus allows for 1.25 MHz BW ...
Or did I miss something here?
Best regards,
Jesper