An Improved Servo for the THAT1510 and THAT1512

Where we discuss new analog design ideas for Pro Audio and modern spins on vintage ones.
User avatar
mediatechnology
Posts: 5444
Joined: Sat Aug 11, 2007 2:34 pm
Location: Oak Cliff, Texas
Contact:

Re: An Improved Servo for the THAT1510 and THAT1512

Post by mediatechnology »

JR. wrote:Interesting to revisit this old thread... I've got at least one new question... Do they still make chopper stabilized opamps? IIRC they were slow as dirt, but very low offset voltage. Probably too noisy for a mic preamp.

JR
I think they do. However, the OP07 is probably good enough when the servo is taken from the output as shown in the OP.

If the servo input is taken from emitters, as in "Phantom Menace I," I don't think the LT1012 is even good enough.

IIRC, to minimize clicking at high gains, the Vos across the gain resistor had to be held <<30 uV. Below 10 uV I considered it to be good enough. I used the LT1012 in the DC preamp this way and it did work. But, using the new topo it could work better. If you add up all the LT1012 errors in the topology where the servo wraps from the emitters back to the input they quickly grow beyond 30 uV. And it's not just DC Vos errors from Vbe mis-match. The servo has to correct DC rectification from differential RF imbalance. That may just be a protoboard issue but I found it to be very real. I could "play" the servo like a Theramin.

I easily got <<30 uV at high gains with the "improved" topology. Also advantageous is the decrease in servo Rin. Fortunately, the servo makes it's greatest errors as the gain is stepped downward so the decrease in gain as the dynamic servo error increases tend to cancel. You'll note that the 1570/5171 uses this same topology. It's not new, one of our members here have used it in a commercial design. I just didn't know it when I came up with it. Thanks JP for letting us use it. :)
ricardo
Posts: 262
Joined: Sun Apr 10, 2011 7:24 am

Re: An Improved Servo for the THAT1510 and THAT1512

Post by ricardo »

The more I stare at these ICs, the more I'm convinced the real answer is a 4700uF Unobtainium CGO cap the size of a DIP8. ;) Other caps are too big or don't act like proper caps. In it's absence, Wayne's servo is probably the best all round.
You'll note that the 1570/5171 uses this same topology. It's not new, one of our members here have used it in a commercial design.
THAT 5171 datasheet quotes 1.5nV/rtHz when used with 1570. Anyone know what's possible with 1510 etc. with full RFI & P48V protection? Never having used these ICs for real, I'm beginning to appreciate the Millenia Media (1nV/rtHz) and Earthworks.

Wonder if you can get rid of the OP07 driving p5 by using offset trim on the servo OP07. ie trim for no output on the servo at Lo gain. This would give +/- 5mV output offset but saves one package. If it works, it might mean you could use any OPA instead of OP07
User avatar
mediatechnology
Posts: 5444
Joined: Sat Aug 11, 2007 2:34 pm
Location: Oak Cliff, Texas
Contact:

Re: An Improved Servo for the THAT1510 and THAT1512

Post by mediatechnology »

The low gain offset trim corrects only the output diff amp offset. Though you might be able to offset a single OP07, adjustment would be problematic since the servo would have to have its loop closed. The adjustment would require a lot of trial and error. I tried this and didn't like it.

As shown, the "high gain" servo is jumpered off in FT to adjust the Vos of the output diff amp so they don't interact.

Another problem is that if you offset the servo for low gain to correct the output amp Vos, it's going to "see" error at high gain that isn't really error resulting from input Vos. That's going to re-introduce offset voltage across Rgain and cause clicks.

What you want is a 1510 that "perfectly" amplifies it's own input Vos. (And the output Vos from different cap leakage, different input Ios, RF etc.) At high preamp gains, the output stage Vos doesn't matter much because the gain within the servo feedback loop is high. As a percentage of the total potential Vos error at high preamp gain, the 1-5 mV is pretty small. As the preamp gain is lowered, the output stage Vos becomes a larger percentage of the total error.

OP07s run about 30 cents. If having a second one bugs you from a real-estate perspective just eliminate it and ground pin 5. If the minimum gain of the preamp is >15 dB or so it isn't needed. The commercial design doesn't have the second OPA driving P5. There are dual OP07-class op amps. I think Roger found a TI part but I tend to shy away from TI because their delivery times are often quite long. OP07s are made by everyone and their dogs.

See bruno2000's preamp here: viewtopic.php?f=7&t=339
ricardo
Posts: 262
Joined: Sun Apr 10, 2011 7:24 am

Re: An Improved Servo for the THAT1510 and THAT1512

Post by ricardo »

mediatechnology wrote:... adjustment would be problematic since the servo would have to have its loop closed. The adjustment would require a lot of trial and error.
As shown, the "high gain" servo is jumpered off in FT to adjust the Vos of the output diff amp so they don't interact.
I was thinking of moving the 100k from the 2M to the -ve input of the servo instead of jumpering to earth for FT. This disconnects the servo and gives something sensible to measure the output of the servo. Could increase the 100k to help too.
Another problem is that if you offset the servo for low gain to correct the output amp Vos, it's going to "see" error at high gain that isn't really error resulting from input Vos. That's going to re-introduce offset voltage across Rgain and cause clicks.
Please correct me if I'm wrong but I think most OPA (including OP07) offset trims are actually i/p offset trims as they vary the current in the i/p devices. Doing this would "match" the OP07 to the 1510 so the servo gives no correction at the 1510's original "output offset".

Have we gone as far as we can without soldering irons?
User avatar
mediatechnology
Posts: 5444
Joined: Sat Aug 11, 2007 2:34 pm
Location: Oak Cliff, Texas
Contact:

Re: An Improved Servo for the THAT1510 and THAT1512

Post by mediatechnology »

I was thinking of moving the 100k from the 2M to the -ve input of the servo instead of jumpering to earth for FT. This disconnects the servo and gives something sensible to measure the output of the servo. Could increase the 100k to help too.
Not sure I follow you. What closes the loop around the servo for measurement to keep the top OP07 from being open loop?

As an alternative, couldn't the servo integration cap be paralleled (via a FT link) with a value equal to the servo Rin for FT adjustment? The link going from the output could then go to either the 2M or a 220K Rfb. (A link looking like a SPDT switch.) With the output jumpered to the 10K, the servo loop is closed with inverting unity gain for "1510 low gain" adjustment. (No correction is fed to the 1510 input.) The Vos trim - however it's done - could then be adjusted to null out the 1510's internal unity/low gain offset. Moving the link to the operating position opens the 10K, leaving only Cfb, and connects the OP07 output to the 2M injection resistor.

About that 220K servo Rin: It could be made 10K and the Cfb scaled accordingly to 1 or 2.2 uF. That would reduce the Ibias error developed across the resistor. I used 100 nF to avoid a larger NP cap.
Please correct me if I'm wrong but I think most OPA (including OP07) offset trims are actually i/p offset trims as they vary the current in the i/p devices. Doing this would "match" the OP07 to the 1510 so the servo gives no correction at the 1510's original "output offset".
Yes, they usually are. You could use the trim connections on the OP07 but it might not have enough adjustment range.

If you wanted to use a single OP07 and introduce deliberate input offset, wouldn't it be just as easy to offset the non-inverting input by +/- 1-5 mV rather than ground it? Because the OP07 is a "precision" part, the internal OP07 trim might not have enough range to give you up to +/- 5 mV.
Have we gone as far as we can without soldering irons?
We're long past the academic stage. bruno200 has been using these in his remote truck for over a year. The schematic is in the post. viewtopic.php?f=7&t=339

If you want to try a single OP07 and offset the servo itself go for it! There's more than one way to skin a catfish. I hadn't thought about making the servo a unity gain stage for FT. The only reason why that second OP07 is there is to buffer the trim pot so that the added R in the ref pin leg (Pin 5) didn't blow the 1510s common mode rejection.
ricardo
Posts: 262
Joined: Sun Apr 10, 2011 7:24 am

Re: An Improved Servo for the THAT1510 and THAT1512

Post by ricardo »

mediatechnology wrote:As an alternative, couldn't the servo integration cap be paralleled (via a FT link) with a value equal to the servo Rin for FT adjustment? The link going from the output could then go to either the 2M or a 220K Rfb. (A link looking like a SPDT switch.) With the output jumpered to the 10K, the servo loop is closed with inverting unity gain for "1510 low gain" adjustment. (No correction is fed to the 1510 input.) The Vos trim - however it's done - could then be adjusted to null out the 1510's internal unity/low gain offset. Moving the link to the operating position opens the 10K, leaving only Cfb, and connects the OP07 output to the 2M injection resistor.
That's exactly my suggestion but I'd use 220k or more from servo output and switch that either to the 2M (servo closed) or across the servo cap (inverting buffer for Final test).
If you wanted to use a single OP07 and introduce deliberate input offset, wouldn't it be just as easy to offset the non-inverting input by +/- 1-5 mV rather than ground it? Because the OP07 is a "precision" part, the internal OP07 trim might not have enough range to give you up to +/- 5 mV.
Can't find any info on range of OP07 trim but I was worrying about real estate. OK. I'm being mean over 2 resistors. ;)
Last edited by ricardo on Wed Apr 20, 2011 6:01 am, edited 1 time in total.
User avatar
mediatechnology
Posts: 5444
Joined: Sat Aug 11, 2007 2:34 pm
Location: Oak Cliff, Texas
Contact:

Re: An Improved Servo for the THAT1510 and THAT1512

Post by mediatechnology »

I do like this one better than the first one. Wanna build it?

Image
THAT 1510 Servo CIrcuit Shown with Optional Low Gain Trim

Add a TP on the output of the OP07 for metering.
I'm being mean over 2 resistors.
Well, I saved you an electrolytic. :mrgreen:
User avatar
JR.
Posts: 3700
Joined: Sat May 24, 2008 7:21 pm
Contact:

Re: An Improved Servo for the THAT1510 and THAT1512

Post by JR. »

I'm afraid you're not finished yet... I want to see no trim pots.

I hope you are talking to the THAT engineers so the next generation part can roll some of this inside.

JR
Cancel the "cancel culture", do not support mob hatred.
User avatar
mediatechnology
Posts: 5444
Joined: Sat Aug 11, 2007 2:34 pm
Location: Oak Cliff, Texas
Contact:

Re: An Improved Servo for the THAT1510 and THAT1512

Post by mediatechnology »

I'm afraid you're not finished yet... I want to see no trim pots.
Then just ground the OP07's non-inverting input. Do I need to draw that?

Again, the trim can be eliminated if the minimum gain is around 15 dB or so. For the 1510 using a pot that's probably a good minimum gain anyway to get the rotational "feel" right. I think were all a little wrapped around the axle about that trim when 90+% of the mic pre apps don't need unity gain. At high gains the servo works fine without a trim.

Let's also not forget about the trim's tempco....
I hope you are talking to the THAT engineers so the next generation part can roll some of this inside.
I think it's called a THAT5171. ;)
jp-apb
Posts: 2
Joined: Sat Oct 18, 2008 2:43 pm

Re: An Improved Servo for the THAT1510 and THAT1512

Post by jp-apb »

mediatechnology wrote: I think they do. However, the OP07 is probably good enough when the servo is taken from the output as shown in the OP.

If the servo input is taken from emitters, as in "Phantom Menace I," I don't think the LT1012 is even good enough.
......

I easily got <<30 uV at high gains with the "improved" topology. Also advantageous is the decrease in servo Rin. Fortunately, the servo makes it's greatest errors as the gain is stepped downward so the decrease in gain as the dynamic servo error increases tend to cancel. You'll note that the 1570/5171 uses this same topology. It's not new, one of our members here have used it in a commercial design. I just didn't know it when I came up with it. Thanks JP for letting us use it. :)
Wayne,
Been away from the Forum for awhile, but I saw this thread reappear and thought I'd finally comment. Gary had emailed me after you initially came up with this servo topology, asking if we would mind if THAT published it since it was very similar to what we've been using in our consoles since 2005. We had no problem in saying OK. THAT has always been very supportive of our industry, in fact that's how Gary knew we were using the circuit. We typically ask the "THAT Masters" to review our Preamp & VCA circuit ideas to make sure we're not doing anything stupid.
In this case we had a new implementation that he hadn't seen before, but in other circuit reviews he's raised some concerns and had suggestions on how to do something better.
It works both ways...

For JR's later "request"- we don't use a trimpot for low-gain servo trim; that may make him a little happier...

JP
Post Reply