M-Audio Delta 1010 Input Stage - how does it work ?

Hi folks!

I am currently doing some improvements to my M-Audio Delta 1010.
The Output was easy, but the input looks a little bit confusing to me.

So I thought it´s maybe a good idea to write the schems down to discuss it,
perhaps there are some more people interested in modding their units.

What you see in the drawing is the input-channel #8 without the +4/-10-switch.
The schem starts at the point where your signal hits the +4dBU-circuit.
If you´re working with professional equipment you don´t need -10dBV anyway,
so I decided to pull all switches out and shorten those contacts.

Okay, I understand what this circuit does basically,
but what does the third OpAmp (U19) with those schottkey-diodes ?
Can anybody explain that to me please ?

Is anybody interested in the output-schematics
and what I did to my unit ?
Should I start a seperate post ?

If somebody is interested to see the circuit in the real world,
here is a picture with the signal-flow drawen at the pcb.
Watch for the green dot with a "1" in it, that´s a connection!
I thought it´s not a good idea to draw that green trace all over those parts ...

Something is missing, that would show what establishes the d.c. bias for things to the right of the coupling caps (C186, C196).

Otherwise the extra amp is accomplishing some common-mode feedback and some sort of overlaod protection in conjunction with two of the schottkies; the other two just hard-clamp for excursions heading below ground.

What is the value and type of C151?  The symbol looks like an attempt to indicate a bipolar electrolytic.

See pg 16 of the datasheet for the ADC: http://www.asahi-kasei.co.jp/akm/en/product/ak5383/ak5383_f03e.pdf

Note that the part wants to see about +2.5V d.c. on the inputs.

I would like to see what you have done on the outputs.
Thanks!

@ bcarso

Thanks a lot for the infos so far. I also thought U19A is there to protect the ADC from overload,
but I was not sure how it`s working exactly.

Something is missing, that would show what establishes the d.c. bias for things to the right of the coupling caps (C186, C196).

I think that´s it, I double checked everything. There´s no dc except for +/- 15 V for the OpAmps.


What is the value and type of C151?  The symbol looks like an attempt to indicate a bipolar electrolytic.

Sorry, it´s the first time I used EAGLE, so I am not used to it. C151 should be a standard cap (ceramic I think).
It´s a very small smd-cap, very difficult to measure. I tried it and the meter showed something around 1nF.
I got a Fluke 26III Multimeter. It´s no problem to measure 100nF or 22nF, but around 1nF it gets a little bit tricky. 

@ tchgtr

I´ll start a Delta 1010 output-mod-post next week.
I´ve to take some pictures and convert my hand-drawings in a viewable version 

It would make sense if the cathodes of D29 and D31 plus the right side of R23 were connected to the +5Vanalog rail of the converter chip. Your picture shows a trace going north out of view coming from R23, are you sure it doesn't meander back to the converter? In that case, D29/D31 are overvoltage protection like bcarso said, and U19A gets a filtered +2.5V on its positive input.

If so, U19A works to keep the converter input centered at +2.5V, and it suppresses common-mode signals to a certain extent. (This is not how you normally want to do that, BTW. If I read the schematic right the CM voltage is cancelled by presenting a low CM impedance; this tends to hurt CM rejection).

JDB.
[love the annotated board photograph, BTW]

Agree with JD---the diode cathodes-R23 node has to also be going to 5V.

!!! YOU WERE RIGHT !!!

I corrected my failure in the schematic,
now it looks like this:

It would make sense if the cathodes of D29 and D31 plus the right side of R23 were connected to the +5Vanalog rail of the converter chip. Your picture shows a trace going north out of view coming from R23, are you sure it doesn't meander back to the converter? In that case, D29/D31 are overvoltage protection like bcarso said, and U19A gets a filtered +2.5V on its positive input.

If so, U19A works to keep the converter input centered at +2.5V, and it suppresses common-mode signals to a certain extent. (This is not how you normally want to do that, BTW. If I read the schematic right the CM voltage is cancelled by presenting a low CM impedance; this tends to hurt CM rejection).

Thanks for the explanation so far !!!
Is there a better way to protect the converter-input from overloading ?
Perhaps as AKM suggested it in their AK5383-datasheet (Fig. 8 / page 16) ?

JDB.
[love the annotated board photograph, BTW]

Thanks, I did that to see more clearly. It really helps, especially when things are so tiny 

The diodes are a very standard part of protecting the input of an ADC chip from over voltage (you typically find AD parts specified under abs. max ratings for -0.3V - Va + 0.3V or there about at these pins).

As far as input common mode goes. the circuit is pretty horrible, as with real world sources over long lines you would really want a much higher CM impedance to minimise the effect of source imbalance and cable characteristic mismatch, but for the intended use (cheapish prosumer stuff where the cables will probably not exceed 10M or so), it will do the job.

Personally, I have never been a big fan of the design approach that runs two parallel single ended signal chains and then relies on the ADC chip to act as a difference amp, it just feels nasty at best, still it has the considerable virtue of being simple and cheap.

C58 is kind of critical, and I would ensure that you replace it with a similar dielectric part (probably COG?) and keep the lead length short (SMT would be good). This capacitor is one of the decidedly non trivial details when getting delta sigma converters to really work well.

I am slightly surprised to see that even the AKM circuits seem to ignore the fact that the chip exports Vcom{L,R}, and instead prefer to generate a Va/2 signal for the bias. This iis in contrast to the BB/Texas approach that very heavily favours buffering Vcom and using that as the DC bias.

The schematics given in the data sheet ignore the input protection diodes, but if you look at the specified abs max input levels you can see why they are required. 

Regards, Dan.

I have a similar  "problem" to solve with AK5393 (behringrr DCX2496 dsp crossover) and wonder how to add ADC input over protection  ???

Personally, I have never been a big fan of the design approach that runs two parallel single ended signal chains and then relies on the ADC chip to act as a difference amp, it just feels nasty at best, still it has the considerable virtue of being simple and cheap.

Click to expand...

I agree, but what you suggest ?

I am slightly surprised to see that even the AKM circuits seem to ignore the fact that the chip exports Vcom{L,R}, and instead prefer to generate a Va/2 signal for the bias. This iis in contrast to the BB/Texas approach that very heavily favours buffering Vcom and using that as the DC bias.

The schematics given in the data sheet ignore the input protection diodes, but if you look at the specified abs max input levels you can see why they are required.

Click to expand...

Again, can you give some suggestion about? Also, I'm not sure will the ADC be burned over specified max input voltage or just the signal will be distorted ???

The last time I did an ADC input stage (but it was MUCH higher end), I went for a THAT corporation balanced line receiver (with a common mode choke on the input and input protection diodes), followed by a single ended multiple feedback filter going over at about 70Khz (To ensure that nothing got through anywhere it could alias) which then drove a TI differential driver chip  to produce the differential signal for the ADC, the adc has the usual input protection diodes and 2N7 COG ceramic as per the data sheet. That design was using TI/BB parts rather then the AKM stuff but the principle is the same.

For hacking that card around, I would go with a simple opamp line receiver (the traditional 4 * 10K resistors), followed by a passive single pole low pass into a pair of opamps to provide the differential drive, the clamp diodes can stay as they are.

The ADC will clip when the differential input voltage exceeds Va, and according to the absolute maximum ratings, either pin  MUST NOT exceed Va + 0.3V and must NEVER go below Gnd - 0.3V. In addition the current into these pins MUST not exceed 10mA at any time (SCR Latchup I assume). It is likely that the only time these pins draw significant current is when the input falls outside {Gnd, Va}.

Regards, Dan.

For hacking that card around, I would go with a simple opamp line receiver (the traditional 4 * 10K resistors), followed by a passive single pole low pass into a pair of opamps to provide the differential drive, the clamp diodes can stay as they are.

Click to expand...

Single opamp receiver? I presume that you are talking about unbalanced input? Correct if I'm wrong (it will be much easier to understand if you draw some circuit). At the end , what you think about BB OPA1632? Is it recommended to keep protection diodes in that case too? Does the input of ADC (AK5383-typ 14kohm, AK5393-typ 4kohm) need to be terminated or it can be driven directly?

Dan: welcome.

Telmar: THAT Corporation design note 133 is worth a read; it's similar to what Dan described doing (without the MFB LP filter). If you want to tinker, I suggest leaving D29-32 and R137/R152 in place and driving your signal into the left side of R137/R152. That way it's harder to kill your converters while trying to design a good input stage.

Moby said:

For hacking that card around, I would go with a simple opamp line receiver (the traditional 4 * 10K resistors), followed by a passive single pole low pass into a pair of opamps to provide the differential drive, the clamp diodes can stay as they are.

Click to expand...

Single opamp receiver? I presume that you are talking about unbalanced input?

Click to expand...

I suspect he is talking about the standard differential amplifier configuration.

[quote author=Moby]At the end , what you think about BB OPA1632? Is it recommended to keep protection diodes in that case too? Does the input of ADC (AK5383-typ 14kohm, AK5393-typ 4kohm) need to be terminated or it can be driven directly?[/quote]

Anything that can drive the ADC input beyond its maximum ratings needs the protection diodes. This includes the 1632, if you're running it at a high enough voltage to not have its THD dominate the entire converter. As for the input impedance, it's not an issue as long as your driver output Z is low enough (and you've implemented the mandatory lowpass filter).

JDB.

IIRC there are some amps with externally-adjustable internal clamping out there.  Unfortunately I think the intersection of that set with the set of good audio amps is empty  :'(

Personally , I think that opa1632 is probably best way to go for driving ADC. When I say best I think that it's simple enough and it suppose to be good sound vice... I never listened the 1632 but it seems worth of trying. Regarding clamping diodes, how important is diode capacitance? I see that BAT85 is stated as 10pf. Maybe some other can do the job better  ??? BAT41? 2pf... ???

Since people are asking about op-amps in other threads...I'm confused why  in this circuit is the output of the two op-amps there fed back into the negative side without a resistor reducing the voltage?  That would make it a zero-gain device wouldn't it?  My apologies if this is a very stupid question.

> a zero-gain device wouldn't it?

"zero gain" and "unity gain" are different.