Hi Demian,
I raised the question of software dependency on the internal generator of the QA400 with Matt and he said all the software does on the FFT side is measure.
Unless explicitly set to use gen1 for THD calculation then the SW just uses the highest
measured signal to calculate THD, THD+N.
I asked Matt if they could add in the option of specifying a frequency to use for THD calc
but it's not working the way I envisioned it. I also asked for an option to have markers track what they are placed on. If they can get this working then THD can be calculated on the markers used.
Well, this is becoming a pattern. I had an A-P and I used the FFT in it and same thing! The FFT showed higher distortion than swept sine measurement from same instrument. I thought something was wrong with A-P and took it to Thailand for repair in October. Its still there. Maybe nothing is broken with it?!
So I started on the 339A and then the ShibaSoku and the QA400. Similar results. I think the A-P (earier version) didnt have the ADC/DAC precision available today. So, it read higher harmonics than the QA does. [But a lot of people used it and published the DUT results with the FFT.]
I can see the harmonic structure change (2nd relative to 3rd) as I lower/raise the input level to the QA400. So, I know I Must use a notch before it. That is my next step. I'd like to see QA make a variable notch filter... I hate making this stuff... so time consuming for the users. Note: QA spec's the -400 as a 105dB range. I need a lot better and the notch into the QA400 might do it for me.
Thx-RNMarsh
Last edited:
Sorry, killed my post because I realized that I didn't read the one I was answering correctly.
I was saying that there are much better windows than the ordinary flat top one finds in standard FFT software. I'm using some of those disclosed here for critical work:
http://www.rssd.esa.int/SP/LISAPATHFINDER/docs/Data_Analysis/GH_FFT.pdf
I'll need to install the QA400 on my new computer to get it up and running.
Samuel
Last edited:
I mean like the way 339A measures or other non FFT instrument... all analog.
Direct analog input is used... without notch on fundemental.
Really low and near max level. Above 0.5 volts and below -90.
No complaint about windowing or FFT software at this time.... adc issue, i'm sure --- well, maybe sure.
Thx-RNMarsh
Direct analog input is used... without notch on fundemental.
Really low and near max level. Above 0.5 volts and below -90.
No complaint about windowing or FFT software at this time.... adc issue, i'm sure --- well, maybe sure.
Thx-RNMarsh
Last edited:
Hi Rick,
Do you mean swept tuned? That would be an analog spectrum analyzer.
Neither the 339a or the ShibaSoku are swept tuned. The 339a is direct and the ShibaSoku, after reading the patent paper, is a very clever approach to resolving harmonic distortion, typical of Japanese engineers. Sample the fundamental and reconstruct it at a fixed lower frequency so that fixed frequency filters can be used. It's not heterodyne as I thought but it's has similarities.
I was thinking the A-P swept sine. But all analog is the meaning and not FFT adc. I Thought a 24/192 would be better than they are for this purpose..... I mean with -144 noise floor you would think it is better than it is with thd, too. Is the Analog Devices much better... looks great on paper... esp. thd at -60db. Its my current favorite for audio use. Thx-RNMarsh
The AKM 5394a is the best I have measured so far. Its possible the new one from ESS, 9108, could be better. Since the demo board is $450 I won't be finding out. I have an app note for bonding two AKM ADC's together to get a lower noise floor. The linearity on any of these is exceptional, especially considering the limited voltage range they have to work with.
Here is the app note: http://www.akm.com/AppsNotes/AN120516.pdf
If you want to go down this road I'll follow along. I have no time to do it myself but I can get help from AKM if you need it.
If you want to go down this road I'll follow along. I have no time to do it myself but I can get help from AKM if you need it.
Well not immediately. Down the road. I have to finish current projects before tackling this.
I also have to figure out how to get the data into the computer.
I might be able to funnel it through an EMU 0204 with a bit of mods. It would be nice to have a driver for the maximum sample rate of the 5394A.
The demo board for the 5394 is pretty expensive. The PCB art work is shown in the manual.
Does AKM make PCB file available?
Hello,
I confirm that AK5394 is the lowest THD ADC available. I've made some time ago an ADC box using this IC and no more recent IC allow better THD and lower out of band noise (especially at 192k sampling rate).
I've build also recently another ADC based on the Cirrus CS5381 ADC and i was a little bit dissapointed because results are significantly lower than AK5394.
The fact is that both ADC and DAC IC will no longer interest IC manufacturers. All these IC were made from a long time. I don't know why, and it's a pity...
Actually, i you don't want spend money and time to build your own, probably the better solution is the TI ADC EVM (PCM4222) for 149USD. It's results performances are very good.
I had plan to build an ADC using the AK5388 ADC that have 4 channels, and use a CPLD to increase performance by SQRT(N channels)dB used in parrallel mode.
Unfortunatly, AKM parts are not easy to get (at least in EU) and i don't have so much time right now.
The ESS ES9102 ADC seem also to be interesting, but it exist from about two years and i don't have read any real world specs that it can really reach. I think also that signed a NDA to use an IC is not very DIY friendly !
About AKM and Cirrus ADC, if somebody are interested,i could post my FFT results.
Regards.
Frex
I confirm that AK5394 is the lowest THD ADC available. I've made some time ago an ADC box using this IC and no more recent IC allow better THD and lower out of band noise (especially at 192k sampling rate).
I've build also recently another ADC based on the Cirrus CS5381 ADC and i was a little bit dissapointed because results are significantly lower than AK5394.
The fact is that both ADC and DAC IC will no longer interest IC manufacturers. All these IC were made from a long time. I don't know why, and it's a pity...
Actually, i you don't want spend money and time to build your own, probably the better solution is the TI ADC EVM (PCM4222) for 149USD. It's results performances are very good.
I had plan to build an ADC using the AK5388 ADC that have 4 channels, and use a CPLD to increase performance by SQRT(N channels)dB used in parrallel mode.
Unfortunatly, AKM parts are not easy to get (at least in EU) and i don't have so much time right now.
The ESS ES9102 ADC seem also to be interesting, but it exist from about two years and i don't have read any real world specs that it can really reach. I think also that signed a NDA to use an IC is not very DIY friendly !
About AKM and Cirrus ADC, if somebody are interested,i could post my FFT results.
Regards.
Frex
I am late to the party by about 2-3 years!
Yes, I/we are VERY interested. We need the best there is for measurement uses.
And, I'll look for the PCM4222 as well. Do you or anyone you know have this in a completed package -- maybe from china/internet source?
Thank you... -Richard
Yes, I/we are VERY interested. We need the best there is for measurement uses.
And, I'll look for the PCM4222 as well. Do you or anyone you know have this in a completed package -- maybe from china/internet source?
Thank you... -Richard
Last edited:
Richard:
I can lend you my demo board for the TI PCM4222.
I am using an ESI Juli@ card to bring the data into my PC. I use Praxis as my primary go to software solution, mostly because I have it and know how to get a lot from it. However HPW-works or Arta or others are similar. I don't think any of them are nuts enough to include 16M point fft's.
I could lean on ESS to get a demo of the chip but I don't think its worth the effort. The AKM is hard to get but I think its now available at NuHorizons. Its steep as well at $350. It could be simplified a lot into a smaller PCB or use the 4 ch version and make the ultra low noise version. I am game (but not free now) to work on it. I need to finish the QA400 interface first.
The QA400 uses the Cirrus CS4272 CS4272 : 114 dB, 24-Bit, 192 kHz Stereo Codec which seems to be pretty good and a good value but not in the same league.
The Audio Widget has been designed and has support for the AKM 5394A but the current PCB's don't have ADC capability. This would be another way to get the data into a PC. The current version from Q N K T C USB-I2S Module and Analog Board has a pretty decent AKM DAC (higher second harmonic but not enormously than the premium parts) and could be kludged to run the ADC I think. Too many projects, too little time.
I can lend you my demo board for the TI PCM4222.
I am using an ESI Juli@ card to bring the data into my PC. I use Praxis as my primary go to software solution, mostly because I have it and know how to get a lot from it. However HPW-works or Arta or others are similar. I don't think any of them are nuts enough to include 16M point fft's.
I could lean on ESS to get a demo of the chip but I don't think its worth the effort. The AKM is hard to get but I think its now available at NuHorizons. Its steep as well at $350. It could be simplified a lot into a smaller PCB or use the 4 ch version and make the ultra low noise version. I am game (but not free now) to work on it. I need to finish the QA400 interface first.
The QA400 uses the Cirrus CS4272 CS4272 : 114 dB, 24-Bit, 192 kHz Stereo Codec which seems to be pretty good and a good value but not in the same league.
The Audio Widget has been designed and has support for the AKM 5394A but the current PCB's don't have ADC capability. This would be another way to get the data into a PC. The current version from Q N K T C USB-I2S Module and Analog Board has a pretty decent AKM DAC (higher second harmonic but not enormously than the premium parts) and could be kludged to run the ADC I think. Too many projects, too little time.
I am too far along my Life Line to spend even as much as I have on trying to measure circuit distortion. Running out of time to do everything I want. I am in good health BUT time is getting shorter on this end of it.
I havent been able to figure out the ShibaSoku fft output. it doesnt seem to be there.... maybe why I was able to buy it cheap. Or its just me. I'll play some more with it.
I'd like to try the PCM4222 demo board to measure with but it needs a USB as I dont want yet more kludges to deal with.... that is what is appealing about the QA400... bundled. Ideas? [$350 is ok for AKM if it works as needed] Thx-Richard
Last edited:
for ADC http://www.analog.com/static/imported-files/data_sheets/AD7690.pdf looks pretty good with -125 THD/spurs - though only "typical" spec and not a "audio" ADC @400 ks/s 18-bit
Last edited:
It doesnt change much --- from NL it is .000015% and at 600 ohms it is .00002%. (1KHz).
-RNM
Last edited: