I need a new oscilloscope - 300€?

[Tubetec]

I wish between us we could come up with a dedicated test set up, never mind waiting for the price to come down.
Clearly in terms of distortion/noise measurement the newer audio A/D chipsets do what we want , for bandwidth measurement maybe not  , seems like all audio based convertors tail off bandwidth way early in 768khz mode.

I tried to get more info on Nihtila's A/D but no luck .
Maybe with our own analog front end we could extend the usable response with a switchable  filter that compensates for the chips low pass out to a few hundred khz , it still doesnt take the place of a proper scope for mhz range but for audio bandwidth it would be plenty good enough .

 

[jacomart]

I wish between us we could come up with a dedicated test set up, never mind waiting for the price to come down.
Clearly in terms of distortion/noise measurement the newer audio A/D chipsets do what we want , for bandwidth measurement maybe not  , seems like all audio based convertors tail off bandwidth way early in 768khz mode.

I tried to get more info on Nihtila's A/D but no luck .
Maybe with our own analog front end we could extend the usable response with a switchable  filter that compensates for the chips low pass out to a few hundred khz , it still doesnt take the place of a proper scope for mhz range but for audio bandwidth it would be plenty good enough .

Nihtila ADC AK5572 uses AK5572 model manufactured by AKM, it is a 32 bit delta sigma two channel ADC. From nihtila.com the stereo card costs £ 87 while the mono card costs £ 76. It could be an interesting product to try as, from the AKM datasheet, it has Fs (max) = 768KHz, THD + N = -112dB and S / N = 121dB (124dB mono).
The component (SMD) is priced at around £ 4. Attached the (very detailed) manual of the evaluation board by AKM (no idea about the price).

Cheers
JM

 

[Tubetec]

Thanks Jaco ,
£4 for the AK5572 seems good value , they do tend put a huge mark up on eval modules though . At 192khz it clearly shows bandwidth above to around 80khz. No info on 384 or 768 khz though . 

 

[jacomart]

Thanks Jaco ,
£4 for the AK5572 seems good value , they do tend put a huge mark up on eval modules though . At 192khz it clearly shows bandwidth above to around 80khz. No info on 384 or 768 khz though .

fs=192KHZ BW=83.7KHz (+0.001/-0.037dB) BW=100.2KHz (-6dB)
fs=384KHz BW=81.75KHz (-0.1dB) BW=114KHz (-1dB)
fs=768KHz BW=26.25KHz (-0.1dB) BW=83.75KHz (-1dB)
Unfortunately there aren't many info about noise, S/(N+D) is stated at about 110dB with fs=192KHZ and BW=40KHz, no info for fs=384KHz and 786KHz but I guess that noise and distortion become relevant.
The combination fs = 384KHz BW = 114KHz at -1dB might be interesting but noise and distortion remain unknown.
Maybe someone interested (and not retired like me  ;D) could fill in the form and ask AKM directly for the missing information.
https://www.akm.com/eu/en/support/inquiries/contact2/

Cheers
JM

 

[abbey road d enfer]

£4 for the AK5572 seems good value , they do tend put a huge mark up on eval modules though .

I think it's the price to pay for not having to deal with the QFN package.

 

[jacomart]

at this point I believe that using a SAR ADC is preferable to Delta-Sigma technology.  there are Texas ADCs with 20 bit resolution (ADS8900B) that could give amazing results with fs = 1 MHz.

https://www.ti.com/lit/ds/symlink/ads8900b.pdf

https://www.ti.com/tool/ADS8900BEVM-PDK

https://www.ti.com/tool/PSIEVM


Cheers
JM

 

[Tubetec]

I had been wondering about instrumentation type ADCs, what type of SPI interface board would be required here, seems to be a few cheap options on ebay using usb ethernet or aduino .What are the chances  this module would be seen by REW or using waveforms software ?

 

[jacomart]

I had been wondering about instrumentation type ADCs, what type of SPI interface board would be required here, seems to be a few cheap options on ebay using usb ethernet or aduino .What are the chances  this module would be seen by REW or using waveforms software ?

The PSI (precision signal injector) and the SPI interface are part of the kit
https://www.ti.com/diagrams/ads8900bevm-pdk_ads8900bevm-pdk-box.jpg

the second question is great, you will probably find the answer on the TI forum at :
https://e2e.ti.com/search?category=forum&q=ADS8900BEVM-PDK#serp=3

I think a standard interface like SPI shouldn't have any problems but better be sure.

Cheers
JM

 

[jacomart]

What are the chances  this module would be seen by REW or using waveforms software ?

What I learned today from John Mulcahy is that "If a device appears as an audio interface that your OS recognises then REW should be able to get signals from it. REW does not directly access hardware, the Java runtime takes care of that. REW just asks the runtime what audio interfaces there are."
I know that Java supports SPI and Raspberry has an SPI interface, I suppose that using one RB and a simple wired breadboard with 1xTLC1549 (10 bit SAR ADC priced around 6 €) and 1x10kohm trimmer the test can be done.

Cheers
JM

Edited

TLC1549 schematic inserted (I'm sorry for the "spaghetti" drawing I should modify the symbol, but I got pretty lazy)

Finally, after this long series of reflections on an ADC that can really work as an audio measurement tool even for THD and S / N on a BW of some hundred kHz, I believe that the commercial sound card option, and in general, the use of the delta-sigma ADC is limiting in terms of BW, while the SAR option, which could extend BW up to about 500kHz and have a very small form factor, thanks to the possible use of Raspberry pi, requires heavy development both hardware and (even more) software because applications such as REW, and most likely Waveforms as well, requires the operating system to recognize the device as a sound card. There may be an open source application that can be modified for that purpose but I honestly don't know how much it is worth for a DIYer.

 

[Tubetec]

I found this SARS eval module that looks usefull ,

https://www.digikey.com/en/products/detail/analog-devices-inc./EVAL-AD4020FMCZ/7649207

92 dollars ,

Data sheet here ,

https://www.analog.com/en/products/AD4020.html#product-overview

Maybe its able to output a 768khz 20 bit stream which could more readily be converted to audio .

Unfortunately the FPGA controller board that goes along with it is 200 odd dollars , it connects up via an odd ball multipin connector , but tapping off the four wire spi shouldnt be an issue .

It looks like the XMOS XU200 series controller has an SPI capabillity ,
https://www.xmos.ai/download/lib_spi-[userguide](3.0.2rc1).pdf
Coding would be needed but it might not be any trouble to someone whos familiar with xmos audio.

 

[jacomart]

there would remain the problem of the software that should be developed from scratch, isn’t it?

Cheers
JM

Addendum:
perhaps it would be better to open a new thread, probably there are people who have not read the oscilloscope thread but might be interested in making a measurement tool for audio.  what do you think?

 

[Bo Deadly]

Addendum:
perhaps it would be better to open a new thread, probably there are people who have not read the oscilloscope thread but might be interested in making a measurement tool for audio.  what do you think?

Yup. I'm probably not going to contribute but I have been reading the posts and looking into the various parts. If I ever have an infinite amount of resources and time I would love to take a fancy ADC, a GPU micro and a small high contrast display to create an analyzer that is targeted at the < 1MHz range. So if you create a new thread I'll at least follow along.

 

[Tubetec]

Well it might not be that complicated Jacco ,
It might involve a we-write of the xmos firmware  and a flash programmer .
Ive posted a topic on Xcore.com to try and find out more .
Of course to implement additional functions like noise shaping etc would take more effort but I'll bet theres lots of code around that with small changes could utilise the power of the microcontroler to process the input signal. 

 

[jacomart]

Well it might not be that complicated Jacco ,
It might involve a we-write of the xmos firmware  and a flash programmer .
Ive posted a topic on Xcore.com to try and find out more .
Of course to implement additional functions like noise shaping etc would take more effort but I'll bet theres lots of code around that with small changes could utilise the power of the microcontroler to process the input signal.

I think the hardware is not a big problem, I imagine a good SAR ADC (TI or AD) with SPI interface connected with a flat cable to the SPI port of a Raspberry pi (or even a Beaglebone Black) with the HDMI port connected to something like this, all packaged in a nice instrumentation box, and a bunch of lines of Python or C++/C# or Java code to manage the data and the user interface. Honestly, despite being a retiree I don't have much time and (I hope you will understand) not even the desire to sit down and write code like a madman (in the past I often did, I confess) I prefer to do small manual jobs like what I'm doing now restoring an old tube guitar amplifier (1964) in very bad shape. I will have to replace the speakers, repair some scratches in the wooden cabinet and put the new tolex in addition to repairing the electrical part, not a small job in short.
But if anyone needs an opinion or advice I will always be here to grumble without problems, you can bet on it!

Cheers
JM

this too is very nice!

 

[jacomart]

In case someone was still interested in the "project" this is what someone from TI replied when I asked how I could make an OS recognize an ADC system built with an AD8900B as if it were an audio interface:

"You'll need to design a board with a host processor that will enumerate as a USB UAC1.0 or UAC2.0 Audio class device and then it will show up as a sound card in your OS.  You'll then need to set up device communications with the ADS8900B and configure the processor peripherals and DMA to capture, packetize and stream the data out of the USB Audio Class format.  The ADS8900BEVM-PDK evaluation system not be able to be modified to fit this need, but the daughter card with the ADS8900B on it could be reused.  I'm not sure we have anything in terms of starter boards or systems for the UAC 1.0/2.0 processor and you'll need to look at popular vendors in this space such as C-Media or XMOS."

Honestly some of these things are beyond my abilities mainly as an "analog man". Once I would have asked some of my colleagues to think about it ... and in the meantime I would have dedicated myself to more funny things.  8)

Here the XMOS USB Audio Design Guide

Cheers
JM

by chance I found this which looks very interesting! Because "if it exists and works it's useless to design it again" -  the funny thing is that it's produced 20 km from my house!

 

[abbey road d enfer]

Thanks for this info Jaco. I know I can't use it for any practical purpose, but one of our younger members may be willing to carry the torch...

 

[Tubetec]

Thanks Jaco ,
Ive looked at the Amanero board alright , I think it like most xmos boards is only configured for usage with a DAC or in other words as an audio output via USB .

https://www.diyinhk.com/shop/audio-kits/142-xmos-multichannel-high-quality-usb-tofrom-i2sdsd-spdif-pcb.html 
Heres another xmos board that has I/O capabillity , it says with custom firmware 768khz is possible , in this case as we only need an input we wouldnt need to allocate cores or tiles to process an output signal . 

I tried posting a question on Xmos exchange forum , but it seems fairly deserted now.
Ive never been much into coding either but would be nice if we could find someone who knows about xmos audio. Ill try posting on other xmos forums that are more popular .

 

[jacomart]

Ive looked at the Amanero board alright , I think it like most xmos boards is only configured for usage with a DAC or in other words as an audio output via USB .

In fact, my intent was to use a SAR ADC connected via i2c bus to the Amanero card which, connected to a PC, would allow to use a software such as REW.

Cheers
JM

 

[Tubetec]

There does seem to be a few sar ADC's that do I2C but it appears to be limited to 400khz ,
SPI goes to 35 Mhz and LVDS upto 200Mhz
some more info below

https://miscircuitos.com/delta-sigma-sar-adc-converter/

I did check back on the Amanero board and its only configured for output , so even though the xmos chipset itself supports inputs there not configured for use on that particular board .

 

[jacomart]

There does seem to be a few sar ADC's that do I2C but it appears to be limited to 400khz ,
SPI goes to 35 Mhz and LVDS upto 200Mhz
some more info below

https://miscircuitos.com/delta-sigma-sar-adc-converter/

I did check back on the Amanero board and its only configured for output , so even though the xmos chipset itself supports inputs there not configured for use on that particular board .

I agree that the SPI + dedicated firmware solution can produce absolutely more professional results, my latest proposal is instead a low cost solution for a measurement system that would still have guaranteed good performances up to almost 200kHz, a BW that I believe is adequate to achieve accurate  measurements.  As engineer, logically, my aspiration would be to always achieve even better performances and I can only support your idea.

Cheers
JM