M-Audio Delta 1010 Output-Stage

Hi folks,

for those who are interested in the M-Audio Delta 1010,
here is a "layout-pic" of the output-stage.

Some thoughts ...

The strangest thing about the whole Delta 1010 unit is the output-stage (but that´s just my personal point of view).


The signal is provided by the AK4393 DAC as a balanced signal at pin 21 (AOUTR+) and pin 20 (AOUTR-) for output-channel 8.

Then it´s passed by C167 over R163 and R 165 into the inverting-input of U22A (pin 2) for the +sinal (red trace)
and via C159, R152 and R154 into the inverting-input of U22B (pin 6) for the -signal (green trace).

U22 adds a gain of approximately 4,6 and does the filtering, so far so good.

But then both signals are mixed together by U25B at unity-gain and the output becomes unbalanced.
The Delta 1010 has balanced-ins but unbalanced-outs only, there is no chance to get a balanced-signal out of this unit!

When I bought it I thought it´s balanced when switched to +4dBU and unbalanced when switched to -10dBV, but that´s not the case.
You can see what´s happening when the unit is switched to -10dBV, simply follow the yellow trace in the lower right corner.
Also look what C169 and R161 are doing (green trace of output-jack)!

... and here comes the schematic ... mission done !

Thanks Telmar,
  I too am disappointed about the unbalanced output.
It looks like C163 and C169 are electrolytics. Did you change those at all? Perhaps some sonic improvement could be had there...

Yeah, but that´s easy to change. Simply throw everything out that´s not needed, like this ...

And it´s maybe a good idea to make R153 and R164 trim able, to change the 18dBFS to +4dBU setting.

By the way the Delta 1010 is set to +4dBU @ approximately -16dBFS.

Newer researches have shown that this calibration is too hot.
As a reaction to this a lot of units are switchable from +4dBU @ -18dBFS to +4dBU @ -9dBFS now.

By the way ... do I really need R157 and R161 (1K) ?

Any other suggestions ?

Here is how it could be done in the real world, red shows the +signal, blue the -signal.

Of cause R157 and R161 (1K) have to be soldered back in if necessary.

But don´t forget to cut the R157-ground-connection (yellow circle)!

Telmar said:

[...] there is no chance to get a balanced-signal out of this unit! [...]

Click to expand...

The output is balanced in +4dBu-mode. The impedance for both positive and negative outputs is the same: 1k in series with 10uF (assuming 0R opamp output impedance, which certainly isn't true for higher frequencies). The output is not differential, though, but having a differential output is much less important than having balanced impedances. Have a look at this Jensen application note for more details.

Telmar said:

Yeah, but that´s easy to change. Simply throw everything out that´s not needed, like this ...

Click to expand...

But then your output includes all the common-mode noise that the DAC produces, and there's a good chance that imperfections in your cabling and the receiver will transform some of the CM noise to normal mode.

If you must have differential out (why?), I would recommend keeping U19A; have it drive a THAT1646 or some other line driver.

JDB.
[I do agree that that -10dBV mode is a bit fishy]

looks like C163 and C169 are electrolytics. Did you change those at all? Perhaps some sonic improvement could be had there...

That´s right, C163 and C169 are electrolytics and both are resposible for the audio quality.
But don´t forget C159 and C167 those are critical items, too !

By the way I did a blind-test with four different brands.

Panasonic FC-series, ELNA Silmic II, United Chemi-Con LXZ-series and stock caps.
The difference was audible and I was able to repeat the same result several times in a blind-test !

stock > weak, lifeless, small

LXZ > neutral, balanced, open

FC > punchy, vital, crisp

Silmic II > round, soft, fat

But don´t get carried away with this. The difference is audible (as I mentioned before) but NOT HUGE !!!
I tried to find some words to describe the different sound-characters of those caps.

Hi JDB and thanks for your help !

I´ll read the Jensen application Note and reply tomorrow (it´s too late by now).

That´s right, C163 and C169 are electrolytics and both are resposible for the audio quality.
But don´t forget C159 and C167 those are critical items, too !


Oh my, yes! I didn't notice those on my first look.
I sense a Digi-key order in my future...
Maybe it would be a good idea to double-up and make those non-polar.

Actually, I looked, and there are bi-polar Panasonics with those values at Digi-key!

Differential outputs can gain you 6db of noise performance if (and only if) the noise is dominated by the line receiver and the pickup in the wiring that is not cancelled sufficiently.
I do however consider that 1K as a build out resistor is pushing things a bit (100r would be better, I normally try for 50r), those resistors are there to ensure opamp stability when driving a long capacitive cable.

Regards, Dan.

dmills said:

Differential outputs can gain you 6db of noise performance if (and only if) the noise is dominated by the line receiver and the pickup in the wiring that is not cancelled sufficiently.

Click to expand...

How? I can't figure a scenario where this would be the case, assuming of course that the two lines (and the receiver) have balanced impedance and the total signal power is the same.

JDB.
[on the other hand: most common differential output schemes have about 3dB more noise than single-ended balanced drive, given the same circuitry/resistor values]

The differential signal can be 6db higher for a given supply rail then the single ended case.

As ever with interfacing you do have to consider both ends and how they interact, so it is not always clear cut.
In most cases of course the dominant noise source will be elsewhere and then it really does not matter which topology you use.

Regards, Dan.

jdbakker said:

Telmar said:

[...] there is no chance to get a balanced-signal out of this unit! [...]

Click to expand...

The output is balanced in +4dBu-mode. The impedance for both positive and negative outputs is the same: 1k in series with 10uF (assuming 0R opamp output impedance, which certainly isn't true for higher frequencies). The output is not differential, though, but having a differential output is much less important than having balanced impedances. Have a look at this Jensen application note for more details.

Click to expand...

Yes JDB you are right, I mixed something up. I was talking about differential balanced outputs that not exist.
That was the reason I was disappointed about the 1010, differential Ins but "only" impedance balanced Outs.

It seems obvious that impedance balancing is the big deal in other words a must have and differential is a "nice bonus".

The cool thing in the "differential-case" is you got two OpAmps driving half the output-voltage,
for example 2 x 0,644 Volt instead of 1 x 1,228 Volt @ 4dBU. And the current draw will drop also.
In other words it´s less work for two opamps instead of one and the NJM5532 is not the most powerful device anyway.

jdbakker said:

Telmar said:

Yeah, but that´s easy to change. Simply throw everything out that´s not needed, like this ...

Click to expand...

But then your output includes all the common-mode noise that the DAC produces, and there's a good chance that imperfections in your cabling and the receiver will transform some of the CM noise to normal mode.

Click to expand...

My cables are at maximum 2 meters (6,5 feet) in and 2 meters (6,5 feet) out of the converter.
It´s a good quality cable with 8 x 2 cores + shield and new Neutrik TRS-plugs on both sides.
Is it really that dangerous to transform those DAC common-mode noise to normal mode with such short cables ?

To be honest I got no clue and no real life experience with impedance-balanced vs. differential-balanced.
I recognised that the most pro audio equipment has differential ins and outs so I thought it might be better to adapt my converters to differential-balanced, too.
But perhaps I was wrong and that´s not necessary, or let´s say I have to change a lot of things inside the units and will get a very small enhancement.

tchgtr said:

Oh my, yes! I didn't notice those on my first look.
I sense a Digi-key order in my future...
Maybe it would be a good idea to double-up and make those non-polar.

Actually, I looked, and there are bi-polar Panasonics with those values at Digi-key!

Click to expand...

I´ve seen a configuration with doubled caps as you describe it in an old Studiomaster Console some weeks ago.
Does that make an audible difference ?

I´ve seen a configuration with doubled caps as you describe it in an old Studiomaster Console some weeks ago.
Does that make an audible difference ?

I just recapped and chip-swapped an old Studiomaster Mixdown for use in my home studio, and love it...
If electros are in the signal path, and can't be replaced by film, this is supposed to be the way to go. I don't know enough to explain the theory behind it, but I think I might try some of those bipolar Panasonic caps in my Delta 1010.
Should be an easy job, and the price is quite reasonable.
I'm enjoying the Delta threads, and learning a lot...

The caps at the input (C159,C167) do not need to be non polar as they have ~2.5V bias across them, and I would (If I did anything) just replace with a classier electrolytic (Good modern electros with a little bias across them are actually seldom the villain of the piece).

Bipolar might be appropriate at the output however (and reduce that build out resistor value IMHO).

Regards, Dan.

dmills said:

The caps at the input (C159,C167) do not need to be non polar as they have ~2.5V bias across them, and I would (If I did anything) just replace with a classier electrolytic (Good modern electros with a little bias across them are actually seldom the villain of the piece).

Bipolar might be appropriate at the output however (and reduce that build out resistor value IMHO).

Click to expand...

Gotcha Dan,
    By build-out resistors, I assume you mean R161, R157.
This is all fantastic info. I run my Delta through the Studiomaster and mix old-school using the computer like a glorified 2-inch machine, so every bit of fidelity I can squeeze out of my rig is vital. I've just come to dislike plug-ins as a rule, and prefer the EQs on the board. I work faster that way and depend on the extra analog grit for character.
One day I'll be able to afford better gear, but right now I'm enjoying the DIY experience, and learning so much.
Can't wait to build some clones...