Wayne's 1570 summing amp

[juniorhifikit]

I'd love to see how the mix amp and how the overall project turned out. Glad you were able to get it going. I was afraid I had orphaned you in that effort while I got construction of the house finished. Please do start that thread and share what you can.

Not at All Wayne! I'm forever in your debt for the information you provided. Here's the circuit that we ended up with. We never got the design with the servo on the output to be flat - there was always a low frequency rise. Servo on the input spec'd better/flatter, but we were left with a bunch of DC offset at the output. I ended up putting a cap at the output of the 1240 (not the most elegant solution).

I seem to have misplaced the RMAA test results. If I find them I'll post them. If anyone has any thoughts, I'm all ears.

[JR.]

Damn...I got kicked me out again and lost my first answer.

OK short version... consider adding a small cap at junction of 2x100k to ground and another resistor in series from there to the servo input. A pole well above the dominant servo pole shouldn't alter servo stability (I hope). The op07 is one of those slow guys, in a low voltage place.

caveat,,, this is more of my over-engineering making this better than it needs to be, but its only a few cents worth of parts.

JR

[juniorhifikit]

I did some quick RMAA tests to see if swapping the OP07 for a 5534 would make any difference (in an effort to consolidate parts) - not much dif. Here's the plot:


Noise figure was -92dB, which is as low as the sound card goes (straight wire in-to-out). Might be better than that, but I wont know without better test equipment. I should also mention that this is measured through a channel with an automated VCA at unity gain - not just the summing amp.

[mediatechnology]

Scott - Thank you for sending that. I'm curious about the differences that we're seeing around 20 kHz. What was the value of C31?

I would encourage anyone reading this thread to right-click Scott's schematic and view it offline so that it's not cropped by the board. It's a complete solution.

I like the way you drew the 1646 stage showing the critically short ground path and the connection of the bypass caps to pin 3. The 1646's sensitivity to inductance from pin 3 to ground causes us to layout and bypass that stage a little differently. The input coupling cap is also large, which for the 1646, is very good idea.

[juniorhifikit]

Scott - Thank you for sending that. I'm curious about the differences that we're seeing around 20 kHz. What was the value of C31?

Ah, that whopping 0.25dB drop at 20KHz? C31 is currently 100nF, but I seem to remember trying 10uF and it made no difference (could be wrong though...) If you remember Wayne, we tried the servo both on the input and the output, with all kinds of variations. This was the one that behaved the best. I wish I could find all of those RMAA tests I did...

[mediatechnology]

Do you suppose that we're seeing (very minor) soundcard differences? After I posted that I began to wonder if you did the RMAA overlay by having a 553X in one channel and an OP07 in the other. The error is so small and I don't see how it could be the servo amps used. It's so small I have to wonder if it's measurement variation.

I didn't want the OP-07 to get blamed for anything.

[juniorhifikit]

They were two separate passes through the same signal chain, with just the servo opamp changed. I need to get in front of an Audio Precision, as the sound card I'm using for ball park tests doesn't do this circuit justice.

[mediatechnology]

If you get the chance you might do a comparison run of the OP07 with 10K/100nF at the output and 10K/10uF.

With 100K as a servo Rin, the OP07 might be running out of BW enough to subtract that 0.25 dB in the top octave. You might also increase the servo Rin to 1M and also see if it changes.

JR might well be right in the OPA188 post when the servo Rin is low.

I'm curious.

[JR.]

I would be surprised (if I was right ) this way... I suspect that servo path will be dominated by the 100nF integrator cap across the opamp.

The 1/4dB may be a measurement artifact, or something too subtle for me to easily see... There might be some odd input capacitance interaction with an electrolytic cap to ground on servo output, but this seems unexpected at only 20kHz.

JR

[mediatechnology]

I would be surprised (if I was right ) this way.

Well, I wouldn't be surprised if you were right.

I suspect that servo path will be dominated by the 100nF integrator cap across the opamp.

That was my original thought when you made the comment in the OPA188 thread. But I'm not so sure.

When I ran the simple integrator servo tests with low value Rin, I did see a slight HF rise at the OP07 output come up out of the DC component. It could have been crosstalk.

With the servo inverting the input and feeding a very slight amount of HF back, attenuation would result.

With a 100K/10uF filter on the output of the OP07 it ain't gonna happen. But with 10K/100nF would it?

I remain curious...