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

[Svart]

Unity is usually termed Gain=1(or -1 if it is inverting).

"Unity" is a gain of 0db where the input and output impedance are equal.  Zero gain can mean that the device has different input and output impedances among other things but the V and/or I gain can be 0.

 

[Svart]

As far as diode C is concerned, I think that lower is better.  You really need low impedance driving into an ADC and any parasitics can degrade your performance significantly as the ADC inputs are primarily capacitance.

Most diodes are much better than 10pf when you use better packages.  It's all in the lead and package for diodes so if you choose an SMD version in a small package you will likely get in the 3pf range.  However, at audio frequencies you aren't really concerned with 10pf of C in terms of frequency rolloff as much as you are concerned with the opamp's ability to slew the extra C.  So if you pick an opamp/driver that can drive a largely capacitive load then you won't have to worry about your choice in diodes.  The NE5532 can drive a moderate C load but I don't think it's that great for ADC work.

EDIT: I read all of the responses just now and see that JDB stated the same thing I just stated:

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).

But I suspect that the diodes were chosen for cost and effectiveness rather than performance.

 

[Moby]

Thanks Svart, besides capacitance is there chance that diode can generate some noise? I remember that zener's can do that, not sure for shotky's. Also , question about opa1632 if somebody is familiar with... I see from datasheet that it's able to drive low resistance what's great, but i can't figure the input impedance. In datasheet is written 34megohm (each input pin) but I read somewhere (can't remember where :-\) that input impedance is determined by series input resistors? If that's true it's pretty low .  ???

 

[jdbakker]

Thanks Svart, besides capacitance is there chance that diode can generate some noise? I remember that zener's can do that, not sure for shotky's.

That's not an issue when they're reverse biased. What is an issue is that the capacitance isn't constant, it varies with applied voltage. This can create distortion. So you want low capacitance, only low capacitance diodes tend to have a lower current rating and a higher forward voltage which makes them less useful in protecting the ADC. And, as Svart said, package parasitics start playing a role.

One way to ameliorate the effects of diode capacitance modulation is to lump a capacitor to ground in parallel with each of the ground-bound diodes. As long as its value is much larger than the capacitance of the diodes (and assuming the analog supply has sufficient decoupling) it will to a certain extent swamp the changes in capacitance. Then again, when you do that you'll need to take the filtering effect of these caps into account, and any mismatch in the capacitors will reduce the converter's CM rejection.

Also , question about opa1632 if somebody is familiar with... I see from datasheet that it's able to drive low resistance what's great, but i can't figure the input impedance. In datasheet is written 34megohm (each input pin) but I read somewhere (can't remember where :-\) that input impedance is determined by series input resistors? If that's true it's pretty low .  ???

TI has an appnote on fully differential opamps like the 1632, you may want to read that. In short, in the usual configuration you can compare the 1632 with a dual inverting amplifier. For normal mode signals this looks like a virtual ground, so the input impedance is indeed dominated by the resistors between source and input, and that may well be less than you find acceptable. Also, most converters need to be biased at mid-supply (2.5V for the AKM converter in the Delta 1010). If you use the 1632 (or any diff I/O opamp), this voltage is propagated to the input through the op-amps feedback network. This means you'll either need decoupling capacitors or a DC bias servo on your input.

JDB.

 

[dmills]

As to diode noise, the impedances are very low and the diodes operate a long way from breakdown, the thermal noise will be dominated by ADC self noise and noise present at the input (Even a cheap modern converter is better then most rooms!). Given the very low impedance at the ADC input, I cannot get excited about diode capacitance modulation either, particularly as many designs take the opamp low frequency feedback from that point so the opamp will tend to cancel it.

Regards, Dan.

 

[Moby]

Thanks guys. So, the opa1632 needs one more stage in front  I can't find any info about on TI site. Any recommended application?

 

[Svart]

Not to get off topic, welcome Dan.  Glad to see another knowledgeable individual here.

Moby: Check out the app notes for the TI PCM ADC line of parts.  They have example circuits that include the servos that JDB mentioned.

 

[Moby]

I searched TI site about recommended app note and can't find it. Just typical 1632 circuits with R-in 560ohm  :-\ I think that I understand what kind of circuit this servo amp is, but I would like to read it. Can you point me to the app-note pdf please 

 

[Svart]

http://focus.ti.com/lit/ds/symlink/pcm4202.pdf

Sorry it was in the datasheet, not the appnotes.

And I was also mistaken, it was a buffer for Vcom, not a servo for DC.

 

[Moby]

Thanks Svart, yes vcom buffer because vcom need to be driven from low z source. I learned something by reading all TI app notes and datasheets  I will continue this under thread I started before http://www.groupdiy.com/index.php?topic=27510.msg392874#msg392874  It's almost the same problem like in this thread and same topology can be used for "upgrading" M-audio analog stage. But to be honest for DAW like this I will go with input transformer option 

 

[Telmar]

Hi Folks,

thanks so far !!!

I was very busy last week and got no time to read or write something here, sorry for not replying.
Today I got one day off and I want to acquit oneself of a promise: The Output-Stage ... and the Power-Supply as a bonus ;o)

I´ll read and reply on this tomorrow!

 

[Moby]

I´ll read and reply on this tomorrow!

Go ahead 

 

[Telmar]

Thanks for your help so far !!!

At this point some things are much more clearer now, but I think I have to read some more app. notes, manuals, etc.

@dmills

You wrote: "C58 is kind of critical ... replace it with a similar dielectric part (probably COG?) ..."

Here is a picture of C58, I think it´s a film capacitor, but I am not sure.
What do you mean with the abbreviation "COG" ?
What kind of cap should I use instead of this one (picture) ?

 

[Telmar]

@ dmills

You mentioned that " ... even the AKM circuit seems to ignore the fact that the chip exports Vcom[L,R]
and instead prefer to generate a Va/2 signal for the bias."

Is there a good reason for doing so ?

Perhaps they did it to use the same voltage for all DACs (4x) in the unit,
so every bias is fed from the same source.
But I don´t know if that makes a difference ... 

 

[dmills]

A film cap will likely be fine, the ones to avoid are ceramics using things like X7R dielectrics.
C0G is a ceramic dielectric with good AC behaviour, unlike the common X7R parts that are really only good for decoupling.

Do the reference voltage in whatever way the data sheet indicates, so a divider for these parts or by buffering the output of the TI parts, it may be that there is some HF hash on the Vcom pin of the AKM part or something that makes it better to produce a reference externally.

Regards, Dan.