TI PCM4222EVM

[Dyonight]

Hi!

I got one of those evaluation board a while ago and decided to hook it up yesterday.

The EVM is connected via AES/EBU to a RME MadiFx. The EVM is the master clock (I discovered that "Master Clock" and "Word Clock" are not the same thing...)

When it initialse, before any signal have passed through it, the mixer in Totalmix shows an input noise of -107db on both channels.

Then when I feed signal through it all is good, but when I stop the input signal (I even disconnect the xlr cable) the "idle" input noise of the EVM hang around -89db.

If I reset both switch on the EVM the noise floor drop back to the initial -107db.

What is happening? Is there any thing I can do to avoid this behaviour?

Thanks!

 

[gyraf]

I really think this is a question for the TI customer service. Those guys know their stuff, and are great at tech support...

/Jakob E.

 

[Dyonight]

Oh haven't tought they would care, I'll definitely do that immediately!

 

[57sputnik]

Not sure if this affects your situation but there is a floating input (without a pull-up) on one of the configuration DIP switches. When open the state is indeterminate.

I've seen a similar thing with the noise floor and I forget what I did other than resetting it.

Take a look at the thread I authored here: Using The TI PCM4222 A/D Evaluation Module - Pro Audio Design Forum

 

[Dyonight]

Thank you! I'll take a look right now!

 

[Dyonight]

I think I found what you're refering to:

"With RN13 out-of-circuit, the PCM4222EVM DITFMT and DITMONO lines may float to an indeterminate state when switched "Hi."

Mine is currently set with both of those switch set to "Low", so I guess that does not apply, I'll read the whole thread again though!

57sputnik, do you remember if you managed to correct this behavior? The very last post of this thread explain that exact problem but I'm not sure if there's a solution explained.

I contacted TI and they should get back to me within 2 days via email, I'll update the thread when I know more.

 

[Dyonight]

There's a lot "DC/AC Offset" discussions in that thread, could it be related to what I'm experiencing?

 

[Dyonight]

Well... I was using the "Test Tone" of a Behringer cable tester (running on batteries) but today I tried with a 1073 Audio Maintenance preamp and the noise floor followed that of the preamp (around -90db).

So I guess there was something left floating with the Behringer cable tester and as soon as the signal reached a threashold, the 4222 "woke up" and could not find a valid "rest position" when the signal dropped below that treashold.

As you can tell, I know roughly nothing and can't even tell where I should start to learn those things.

Anyone can help me find the terms to research what I'm describbing here? Does "DC Offest" and stuff of that nature have anything to do with anything?

No need for a full course... only the right key words would help a lot.

Thanks!

 

[Khron]

If you look closely, on the schematic, the two analog input pairs are "unterminated", or at least completely "floating". So with nothing connected, the opamps will pick up whatever random interference might reach their inputs, and the ADC will convert those into the higher noise floor you've observed.

What signal source have you used for your testing?

 

[Dyonight]

Hi! Thanks a lot for your input!

I was using the "Test Tone" of a Behringer cable tester (running on batteries) but I tried with a microphone into a 1073 Audio Maintenance preamp and the noise floor followed that of the preamp (around -90db).

When you say "unterminated" or "floating", what is missing in front of the opamp to make it feel "terminated"? It means there's nothing tying the input lines to the ground when nothing is plugged in? Yeah... that's my current level, thanks for your patience if you choose to answer

There are some unused pads on the board to add a some components to customize the board, I'll add some pictures to see if someone could find something interesting.

 

[Khron]

Not necessarily resistors to ground, but a resistor (say, 10k, or even lower) between the "hot" and "cold" terminal of each analog input (ie. between XLR pin 2 and 3). The opamps used in the front end there amplify the difference between the "hot" and "cold" input pin.

 

[Dyonight]

ok I see! thanks!! I'll give that a try!

 

[57sputnik]

FWIW the PCM4222EVM "leaks" common mode DC out of its input connectors. As supplied from the factory there are no input coupling capacitors.

 

[Dyonight]

Are the unpopulated pads near the inputs there to add those capacitors?

Which value would you suggest?

 

[Rochey]

That's because it was designed for the evaluation of a chip, along with an audio precision, not to be used in an end equipment.
DC blocking caps would have caused a High Pass Filter that would not have shown the ADC's true potential.

 

[Dyonight]

Yes that make absoulte sense and I get that this board was not desinged for someone like me but now that I own one I'd love to have some help to optimise it for my use case. I've put it for sale on reverb so if it sell before I can get some help I'll simply get something else and stop thinking about it.

I understand that I lack the knowledge to make it work the way I'd like and even to be part of this forum, I thought I could get some advices but I feel that I'm bothering everyone by me lack of knowledge (understandably so, that's not a complaint). I'll probably just give up and get the products that already do what I need and move farward.

Thanks everyone for your time!

 

[Matt Syson]

You must real ALL the data sheets for devices like those on Eval boards. The full technical descriptions of ALL the facilities the chips themselves can offer as some, even 'great' designers overlook things on occasion. 20 years ago I was looking at Crystal Semiconductor'ADCs and in the 'small print' several pages into the description it mentions auto zeroing and dC calibration that sHOULD be done after chips have reached normal operating temperature, typically say 5 - 20 minutes after power up.
my thought was 'how many commercial convertors mention the need or implement a 'calibration' phase that SHOULD be performed say 10 minutes or so after power up (or alternative provisions be made to render autocal to be unnecessary.)? ADCs can and do have DC drift compensation or may have a VERY LOW filter (below a couple of Hz) implemented in software.
I would suspect this is an area where the difference between 'cheap and cheerful and 'high end' gear starts to emerge. See also: maximum capacitive loading of op amp outputs, and maximum input current limitation into OP amp inputs (spikes outside the power rails or differentialy).
Matt S

 

[Dyonight]

Thanks a lot Matt! Now I have a lot to read about!