Console input - match resistors?

http://www.douglas-self.com/ampins/books/book3.htm is written by the designer (with Gareth Connor Reply #18) of your Soundcraft 200B and describes most of the circuitry and philosophy behind it.

Worth getting even if you don't have one as it has much of use and interest to anyone involved with audio design.  I don't agree with everything he says but he's always worth close study.  And you can always believe his careful measurements even if you don't agree with his interpretation.

You don't have to actually CUT traces to test-drive the circuit changes.

If you can pump/wick the pads cleanly enough to kill any contact, you can try mods just with soldering the jumpers directly to component leads - IF you remove the solder from the pads efficiently.

Works for me, ymmv.

p.s. TLE2072's are good. I have recommended them to various forum members here, haha. Certainly the price (as free samples) was unbeatable - and they sound like a TL072 "sound" that was a little "mastered".

So I was finally able to try the mod. I used a channel where the offset from the transistor pair was worse. Bottom line: The preamp could be cranked up some more before the servo amp distorted badly.

living sounds said:

So I was finally able to try the mod. I used a channel where the offset from the transistor pair was worse. Bottom line: The preamp could be cranked up some more before the servo amp distorted badly.

Click to expand...

Can you measure & post the voltages at the outputs of the 3 OPAs at min & max gain for this new circuit? 
Is this with a formerly bad channel?

Please be precise with your measurements and your description of these measurements.

ricardo said:

Can you measure & post the voltages at the outputs of the 3 OPAs at min & max gain for this new circuit? 
Is this with a formerly bad channel?

Please be precise with your measurements and your description of these measurements.

Click to expand...

I'll do it next week, there's too much stuff going on right now unfortunately. All of the channels are sort of bad, since I modded all of them (though some show more DC offset than others). I still think John's solution with the different resistors at the servo input would be great, since I could also lower the noise by driving the input transistors higher.

I already forgot what I suggested so when we revisit this be clear about what suggestion you are following.

JR

ricardo said:

Can you measure & post the voltages at the outputs of the 3 OPAs at min & max gain for this new circuit? 
Is this with a formerly bad channel?

Please be precise with your measurements and your description of these measurements.

Click to expand...

Have finally been able to make the measurements. There is no measureable offset on the outputs of op amp 1 and two (pin 1 + 7 of IC1), offset on op amp 3 (IC2 pin 1) stays at +3.1V DC regardless of how the pot is set. Currently there is a 1.5uf cap in the C12 position.

Thanks!

If "audio" opamps both stay at 0 Volts, then it "works as advertised" (if you performed the mods according to "my" schematics).

However, if the "servo" opamp stays at a constant voltage at all positions of the gain-pot, there may be something weird going on after all (or not).

Hard to tell over the internet.

I second TV's comment.

The offset on IC2 p1 is just the servo doing its job and getting rid of the offsets introduced by shorting out C9/10

Thanks! I've now tried and replaced R11 and R15 with lower values (25k) as John suggested, and this (also settting the console to -10db input operation via the jumpter) seems - for all practical purposes - to lessen the problem sufficiently. It also lowers the noise on the input, which supposedly comes from the pot (as I've read somewhere).

However, if there was a way to modify the servo in a way that does not render it useless and makes it sonically transparent I would still like to try that out. The servo with a cap at 2uf does have the same frequency response as with 1000uf (I could match noise curves perfectly down to single digit frequencies), but it still does not sound the same (taking away low end punch), and this shows up in the difference signal between channels, too. Am I hearing phase changes?

living sounds said:

However, if there was a way to modify the servo in a way that does not render it useless and makes it sonically transparent I would still like to try that out. The servo with a cap at 2uf does have the same frequency response as with 1000uf (I could match noise curves perfectly down to single digit frequencies), but it still does not sound the same (taking away low end punch), and this shows up in the difference signal between channels, too. Am I hearing phase changes?

Click to expand...

Now that Guru TV has solved the objective problem (huge offset), we can start work on the subjective stuff (the sound).

You need to do 2 things.

1  Measure the frequency & phase response using something like Right Mark, for both "good" and "bad", and post the results.  We are interested in the region below 1kHz to as low a frequency as convenient, eg 1Hz.  Also your matched noise curves.

2  Post sound / music samples of what you are describing; both on "good sound as you want it" channels and original, unmodified (bad sound) channels recorded simultaneously.  eg on L & R tracks on a stereo channel.  192kHz MP3 or preferably WMA or MP4 will do.  You can split an attenuated line output to feed the 2 mike channels.

If the audible difference is most obvious in the "difference" signal, please post samples of that as above.  Explain how you get this "difference" signal too.

living sounds said:

Am I hearing phase changes?

Click to expand...

Yes, I'm 99'99% positive about that.

This is a tricky thing to solve even with uber-complicated servos.

The one of the simplest (but efficient) thing that will lessen the "sound" of servo is to simply attenuate it's "output" - similar to the schemo I posted here.

But to do that, you must first get both transistors to match as good as it gets, because this way the servo "output" (or headroom) wil be significantly lowered.

When you understand the relationship between these factors, you can "fix" these things intuitively.

But you also know that you are "running in circles".


Btw. is this the same desk you were "fixing" the EQ's? IIRC I once also calculated some eq mods - I thik it was your desk?

Here's a sim. The top graph shows phase/freq response of your (from your description) current config.

If you want to fiddle with phase response, you can also try to add a RC after the servo, like in the below schem. The phase response change is visible in the second graph. Of course, this is only the starting point for "euphonic" component value changes.

If you're concerned with input impedance symmetry, you can also try this servo mod (insert a RC before the servo input). Optimal values would be 75k-82k for R and 2u2 for C - but you can "get by" by reusing the 100k that you pulled from preamps. See attached graphs - phase curve is "almost ideal" when using 75k (or 82k), and ever so slightly bumpy with 100k - both at max. gain, which doesn't look so bad to me at all ...

I've replaced resistors 11 and 15 as well as changed the input sensitivity on all channels now, and have concluded that for all practical purposes this is actually good enough. It's not technically ideal, but it does the job flawlessly. So I'll leave it at that and move on to the large pile of other things that need to be done.

Thanks again to everybody for your help, this has improved my console AND provided an education!

living sounds said:

I've replaced resistors 11 and 15 as well as changed the input sensitivity on all channels now, and have concluded that for all practical purposes this is actually good enough. It's not technically ideal, but it does the job flawlessly. So I'll leave it at that and move on to the large pile of other things that need to be done.

Click to expand...

LS, could you post a schematic of the final circuit you ended up with?

Loadsa gurus have posted suggestions.  But even more valuable than their suggestions, is exact details of a properly working circuit.  I'm sure the gurus will second that.

The problem you have solved, getting rid of C9/10 is of great interest to many.  The final solution is probably still to be found but concrete evidence, tested working circuits like yours, are really thin on the ground.

And of course, thanks to TV as head of the guru team ...  ;D

ricardo said:

LS, could you post a schematic of the final circuit you ended up with?

Loadsa gurus have posted suggestions.  But even more valuable than their suggestions, is exact details of a properly working circuit.  I'm sure the gurus will second that.

The problem you have solved, getting rid of C9/10 is of great interest to many.  The final solution is probably still to be found but concrete evidence, tested working circuits like yours, are really thin on the ground.

And of course, thanks to TV as head of the guru team ...  ;D

Click to expand...

I can't post a schematic right now, but all I've really done is replace R11 and R15 with 28k resistors, and then removed the jumper J1 so the input is no longer attenuated (originally for -10db operation). C12 is still at 1000uf, so the servo isn't doing much. The input is hotter than before, but for what I'm doing this is fine, and I also don't need the whole gain range.

So this is probably not what an engineer would call a properly working circuit, but it works fine for me. I haven't solved the problem, but following John's suggestion the problem was sufficiently reduced.

I've looked around to find a plugin/program that would allow me to measure phase response properly, but haven't found it. As for being able to hear the problem, the converters unfortunately destroy more than the console, so recording it won't show much.

But again, lot's of thanks to "The Gurus", I've learned a ton and can use my console the way I like now!

living sounds said:

I can't post a schematic right now, but all I've really done is replace R11 and R15 with 28k resistors, and then removed the jumper J1 so the input is no longer attenuated (originally for -10db operation). C12 is still at 1000uf, so the servo isn't doing much. The input is hotter than before, but for what I'm doing this is fine, and I also don't need the whole gain range.

So this is probably not what an engineer would call a properly working circuit, but it works fine for me. I haven't solved the problem, but following John's suggestion the problem was sufficiently reduced.

Click to expand...

Did you change the servo so it acts on the input as TV suggested?  His Reply #47 & #51

ricardo said:

Did you change the servo so it acts on the input as TV suggested?  His Reply #47 & #51

Click to expand...

I tried it, and it made a little difference, but I opted not to cut traces and rewire connections on all the channels.

I made more tests and put the TL072 back in as the servo op amp. Turns out the problem with the servo's sound was related to the new op amp (TLE2072), with the TL072 it sounds actually better with a quality 250nf cap in place. So now I can use the full range of the gain pot, there's no DC and the noise is lowered.  8)