Audio test interface

As I mentioned in the other thread http://www.groupdiy.com/index.php?topic=51698.0
please find attached the block dgm of the audio test interface I'm designing.
Obviously, in terms of performance, it cannot challenge an AP, but I think it is up to the task for many of us. It comes from my own experience, where I have an old 1970's oscillator, with those useful and ergonomic things called knobs, but it's not super-low distortion, about 0.1%; I also have a semi-automatic THD meter, not bad but has only unbalanced ins. A souncard is capable of generating super-low THD signals and making super-complex measurements, but lacks in the I/O department.
The unit should be a 1RU piece of about 6" depth, with all the connections on the front panel (except mains).

Please comment, suggest, embarass me, insult me, praise me...

Caveat#1: I know the CMR test is not fully compliant to the standard, but in practice, I don't think it makes much difference.

this looks pretty much exactly to what I am up too.... I will be shooting for a 'AP grade' sine gen though. Just ordered a PCM4222 eval board as a high quality AD converter, §i might roll my own DAC and use an AES-EBU soundcard. the generator is only spot frequencies, but lowest distortion (supposed at least). I will post some results when I get the PCM4222 AD converter, should be with DHL right now ;-)

cherio,

Michael

What are the filters for?

Samuel

Samuel Groner said:

What are the filters for?

Samuel

Click to expand...

I suspect your question means: why analog filters when the souncard + software can provide all the existing and imaginary filters?
This unit is meant to operate also with traditional instruments, such as my THD analyser/voltmeter, that lack this kind of filters.
The second one may be a psophometric type (noise filter).

I see!

Samuel

It seems there is a thread on much the same topic at diyaudio::

http://www.diyaudio.com/forums/equipment-tools/231401-quantasylum-qa400.html

Cheers

Ian

I'm super interested in this.

My EMU-1212M's DSP tone generator can output up to 90khz at low distortion, so something like this would be super useful for testing.

Very interesting Abbey.

I wonder if it would be possible to put in a tunable notch filter for removal of the excitation signal from the return of the DUT.

Many of us do not have this kind of functionality with our basic test sets. Like me

Be great to see the distortion spectrum of a DUT without the excitation signal.
Nothing too flash - maybe a switched 100Hz, 1KHz and 10KHz kind of thing.

Thanks
Alex

alexc said:

Very interesting Abbey.

I wonder if it would be possible to put in a tunable notch filter for removal of the excitation signal from the return of the DUT.

Click to expand...

I've made provision for a connector that allows inserting a custom filter in the chain.

Be great to see the distortion spectrum of a DUT without the excitation signal.
Nothing too flash - maybe a switched 100Hz, 1KHz and 10KHz kind of thing.

Click to expand...

Beware that as soon as you tread in this territory, you're probably going to need tuning of the filters. It can become rapidly overly complex.

I wonder if it would be possible to put in a tunable notch filter for removal of the excitation signal from the return of the DUT.

Click to expand...

You'd want to have the passive notch filter ahead of any active electronics, or the line receiver etc. will limit distortion performance.

Samuel

Here's the schemo.
PCB size is 16x4.3
As usual, comments welcome.

I got my PCM422 eval board today - finally. I will do some tests to check the specifications and whether i can make it work as a precision audio analyzer. 150$ is just a great price I think. me totally excited ;-)

There are a few input conditioning schematics going on, specially on the QA400 thread, but I need a symmetric output into the PCM4222 .... so i might pick up the best features and roll my own in the end. I want some more gain options as well, may be -40dB / -20dB / 0dB / 20dB / 40dB

I bought an usb to I2S interface, but the bad spirits had a go at me again - no full duplex at 96kHz/24bit - doh!

will keep you guys updated.

cheers,

michael

audiomixer said:

I bought an usb to I2S interface, but the bad spirits had a go at me again - no full duplex at 96kHz/24bit - doh!

Click to expand...

Which one? I was looking at the new Silicon Labs USB-to-I2S bridge. It doesn't support 24-bit words so it's not really all that exciting. I traded messages back and forth with one of their support guys; it seems like the product was made basically for connecting to an iPad or other portable thing, so it was deemed that "the users really don't need better than 16 bits."  I suggested to him a couple of applications where you do want better than 16 bits and after a couple of messages, I think he agreed. Maybe in the next flavor of the part (CP2114).

Anyways, the reason you can't do full duplex 96 kHz/24-bit stereo is because that configuration exceeds the available bandwidth of a Full Speed device.

Here's the problem.  First, you have to realize that there are different versions of the overall USB spec. Version 1.1 was the standard which had two speeds, Low (1.2 Mbps) and Full (12 Mbps). The main improvement with Version 2.0 was the addition of the High Speed (480 Mbps) option. (Version 3.0 added the Super Speed and higher bus power capability.)

In addition to the overall spec, each Device Class has its own spec with its own revision. USB Audio Class version 1.0 was written before the the USB Version 2.0 spec and its High Speed configuration option was released. This means that it is limited to Full Speed (12 Mbps) operation. You can't do High Speed with USB Audio Class v1.0. (The reasons for this are somewhat technical; one might think that just enumerating with High Speed endpoints instead of Full Speed would work, but apparently there are issues with High Speed isochronous transfers that broke USB Audio.)

And with all of the bandwidth limits and such, you cannot do full duplex stereo 96 kHz/24-bit audio on Full Speed USB. You can do a 96 kHz/24-bit stereo DAC or ADC but not both.

On 31 May 2006, the USB Audio Class 2.0 spec was released. This version added support for the High Speed endpoints and worked with the High Speed isochronous transfers. Native drivers for Audio Class 2.0 were added with Mac OS X 10.5 (Leopard) which was released in October 2007 (but USB AC 2.0 support wasn't added immediately).

And here's the point: as of this writing, Microsoft still has not added USB Audio Class 2.0 support to any version of Windows, including Windows 8. That's right, almost seven years after the spec was released, Microsoft has not supported it with a native driver.

So the options for a developer who wishes to design a USB audio gizmo that supports USB Audio Class 2.0 on Windows are to write their own driver, pay one of the handful of driver writing companies (Thesycon, Jungo) a lot of money for their driver, or basically punt.

There's no obvious reason why Microsoft hasn't released this class driver, but the lack of it and the resulting costs of licensing a driver from a third party limit the number of USB audio devices which can do better than 48 kHz 24 bit stereo full duplex.

-=a

I got an ministreamer from minidsp.com

not that there is something fundamentally wrong with the device, just not fully suited for the job, so more or less my mistake. a friend ordered there anyway so I jumped on the occasion.

I have been looking at alternative solutions like getting a networked I/O - like the AVB solution from dsp4you.com - they are from the same group, but targeting a more professional market. the price is a little steep, so I am still reluctant to shell out the cash. might still wait a little longer for a proper DANTE solution. there you have a software ASIO driver.

thanks for your precise check-up on USB audio - actually I should know but I was not aware of the full details.

- Michael

audiomixer said:

I got an ministreamer from minidsp.com
not that there is something fundamentally wrong with the device, just not fully suited for the job,

Click to expand...

These do not integrate a codec, so you still have to provide I2S to analog and analog to I2S to get it working in the analog realm.

I have been looking at alternative solutions like getting a networked I/O - like the AVB solution from dsp4you.com - they are from the same group, but targeting a more professional market. the price is a little steep, so I am still reluctant to shell out the cash. might still wait a little longer for a proper DANTE solution. there you have a software ASIO driver.

Click to expand...

I believe it's somewhat overkill if your aim is two-channel measurement. But it's true that, just in order to have 24/96 full duplex, you need a better protocol than USB2. I would say at the moment, FW is your best bet (ASIO drivers are tried and tested), but I think USB 3.0 is coming.

I have a firewire interface (focusrite sapphire 56), but the 4 pin connector on the laptops is so much of a pain so that I am in sweat every time I record something. just look at it and it breaks the connection. I opened up my laptop and checked whether I can put in a 6 pin connector, just hadn't the nerve for that mod till now. will test that today.

I got the PCM4222 to get best in class performance [may be, still needs to be verified]. Codecs are far behind on the THD+N side as far as I could see. still have to get my -100dB vs  FS THD+N performance though. So this has driven the decision towards a AES-EBU / spdif / I2S solution with an converter and an digital I/O.

real USB2.0 has enough power for full duplex 96kHz / 24bit, but you do need a dedicated driver / chip solution. I might just wait for the DANTE interfaces to pop up and go via ethernet.

cheers,

Michael

abbey road d enfer said:

But it's true that, just in order to have 24/96 full duplex, you need a better protocol than USB2.

Click to expand...

Not quite. You need to use USB 2.0 High Speed which requires USB Audio Class 2.0 (see my post upthread).
USB 2.0 Full Speed cannot do 24/96 full duplex.

I have an XMOS multi-channel eval kit that does like 10 in 10 out 24/96 using USB 2.0 High Speed and AC 2.0 with nary a glitch.

FW of course works well. I don't know if there is any difference in Audio Class support for the USB 3.0 SuperSpeed.
-a

XMOS is one of the obvious choices, also for AVB (thru ethernet) but I am still a little reluctant to go that way. one interface I will give a try is the Yellowtec PUC2 lite with AES-EBU. I wonder what chip they use.

- Michael

OK, here it is, well, at least the board is there, waiting for a front panel and case.
Everything works according to plan (only two mistakes).
Input noise is below -97 dBu, THD immeasurable with my current set-up.
I'll test it with the AP next week.

Nice !

Would you eventually make the pcb available ?

This interface + the Q400 is more than enough for my needs.

Waiting for your report with the AP testing.