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 »

Thanks JP. It was very gracious of you to let me post it. Hope I can return the favor. :)
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. »

Thank you... I really dislike trim pots.

One product I inherited when I went to work at Peavey in the mid '80s was a 4 track cassette deck... this unit was finished and in the early stages of production, when the lead engineer quit... This was a difficult product to get out of final assembly because of the time it took for highly skilled technicians to set them up... I probably don't recall all the trims it had... There was one for record bias (x4), then an adjustable bias trap to remove that bias (x4), from leaking into the playback side and messing with the NR, Dolby NR record and playback level trims(x4 x2) , there may have even been a playback speed trim (it had a front panel +/- speed adjustment) (only x1) ... and if a tech finds anything wrong with the build at a later stage of set up, it's start over again after the unit is repaired.. arghh..

Of course I had my boss's boot up my butt, to get more units out (we were seriously back ordered), and there was precious little I could do without major overhaul, and even then it was the nature of magnetic recording that you will never be completely trim less. IIRC I doubled the set up stations, but that was lots of expensive test equipment, and the need for skilled warm bodies who weren't very available at this factory built 25 miles away from Meridian and the main factories on purpose, to hire cheaper labor...

One of several things I don't miss about that old job, and magnetic tape recording.

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 »

JR. wrote:Thank you... I really dislike trim pots.

One product I inherited when I went to work at Peavey in the mid '80s was a 4 track cassette deck... this unit was finished and in the early stages of production, when the lead engineer quit... This was a difficult product to get out of final assembly because of the time it took for highly skilled technicians to set them up... I probably don't recall all the trims it had... There was one for record bias (x4), then an adjustable bias trap to remove that bias (x4), from leaking into the playback side and messing with the NR, Dolby NR record and playback level trims(x4 x2) , there may have even been a playback speed trim (it had a front panel +/- speed adjustment) (only x1) ... and if a tech finds anything wrong with the build at a later stage of set up, it's start over again after the unit is repaired.. arghh..

JR
Welcome to the world of the 20th century studio maintenance tech or assistant engineer. Imagine this with 24 tracks and having to do repro alignement for both the sync and playback paths.
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. »

Glad to be in the 21st century now...

JR
Cancel the "cancel culture", do not support mob hatred.
hazmo
Posts: 19
Joined: Wed Dec 23, 2009 7:31 am
Location: Germany

Re: An Improved Servo for the THAT1510 and THAT1512

Post by hazmo »

Hi Wayne,

digging up this thread, since I'm about to build some of these preamps based on your draft on page 1.

My question is concerning the servo, why did you set the corner frequency that "high"? Looking at other circuits using servos that way, it is usually set a good bit lower. It is also higher than what the typical cut-off frequency would be defined by the phantom-blocking capacitors and the input bias resistors. Any insight?


Thanks, Volker
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 »

Volker - Thanks for re-visiting this. I'm not sure page one's figure is the final word.
Having said that, bruno2000 used it with good results in his 12-pack preamp.
viewtopic.php?f=7&t=339

I think we may want to try this one instead. ricardo had me re-visit it and this seem to be a good improvement.
I haven't built the one below but would be willing to do so.

Image

I do think you would want to increase the C or input R a bit since it's around 7 Hz.
I'd experiment.

The reason why I went on the high side of a LF cutoff was for it to recover more quickly with gain changes.
My pot and switch rotation tests (in degrees/sec) was pretty fast - far faster than anything you'd do real-world.
FWIW The worst case of "catch-up" for the servo is going from low gain to high.
So you might lower it.
Maybe 100-470 nF? Or, 220K to 1M?
I wouldn't get too much lower for Rin than 220K since the amount of output current the OP07 has to slew for a given cut-off becomes much larger.

JR has also pointed out that an additional passive pole in servos either at the servo output or input might serve this circuit well.
However, the attenuation of the servo output is quite large being ~2M2/Rbias.
But there's almost an equal amount of maximum gain.
That thread is here: viewtopic.php?f=6&t=419
I found that with the OP07 220K-1M worked well for Rin.

The second part of your question was:
It is also higher than what the typical cut-off frequency would be defined by the phantom-blocking capacitors and the input bias resistors
Yes it is. It doesn't have to be.
The biggest considerations in determining input cap values are a balance of LF common mode rejection and stored charge.
That drives value selection more IMHO than differential LF cutoff.
You want the caps big so the match is less critical.
But you don't want them too big or the duration of the phantom fault current peaks - and total energy - makes fault protection more difficult.
In blowing up tube after tube of THAT1510s, SSM2019s and INA217s I found that values above 100 uF were just too problematic.

The bias resistor values may be too low for some tastes and I wouldn't go above 10K/leg.
With 1K5 Rbias the differential input impedance is 3K||(Rphantom*2).
1K5 (and even lower) is what THAT likes to recommend to reduce 1/f noise.
Rbias can be higher.
At some point increasing Rbias has diminished returns (in terms of Zdiff in) since it will appear in parallel with phantom.

You can reduce the input C values and makeup for the increased sensitivity to mis-match by using T-bias:
viewtopic.php?f=6&t=191
The above link is a single-supply circuit but could be used dual supply.
With T-bias however you need to provide two servo outputs independently to each leg with one being an inverted copy of the other.
That's because the shared bias resistor converts a single-ended servo output voltage from differential to common mode.
Driving the other leg with the inverted copy restores servo operation.
hazmo
Posts: 19
Joined: Wed Dec 23, 2009 7:31 am
Location: Germany

Re: An Improved Servo for the THAT1510 and THAT1512

Post by hazmo »

Thanks for the long and detailed response!

Regarding the choice of topology: is there any major disadvantage with your first circuit, other than the usage of one more opamp? As a matter of fact the pcb layout is actually ready to send to the fab...


Ok, increasing the servo input resistor is what I was gonna do, thanks for the verification. Bigger polypropylene capacitors get too bulky quickly.


Re the bias resistors/input capacitors: with your paper about the phantom menace in mind and following THAT's recommendations, I was going for 47µ and 1.2k, so resulting in a differential input impedance of about 2k. Would you go for something higher? Load-wise it should be ok for practically any microphone I would guess. Is there anything else to consider concerning the noise besides Johnson and bias current? I don't have enough experience to quantify the noise penalty without actually calculating.


Thanks again,
Volker
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 »

Regarding the choice of topology: is there any major disadvantage with your first circuit, other than the usage of one more opamp? As a matter of fact the pcb layout is actually ready to send to the fab...
Go for it. I've PM'd bruno to post his sch which is being used in his remote truck. It's the same as page one's schematic. We know it works at the cost of an extra op amp.
Ok, increasing the servo input resistor is what I was gonna do, thanks for the verification. Bigger polypropylene capacitors get too bulky quickly.
I just looked over THAT's original and they used 100nF/510K. (with a Bifet). That puts the -3dB point at ~ 3Hz vs. the 7 Hz I originally showed.
Do be aware that you need a low-bias current (bipolar) op amp to run it up that high. The OP07's bias current is low enough.
Re the bias resistors/input capacitors: with your paper about the phantom menace in mind and following THAT's recommendations, I was going for 47µ and 1.2k, so resulting in a differential input impedance of about 2k. Would you go for something higher? Load-wise it should be ok for practically any microphone I would guess. Is there anything else to consider concerning the noise besides Johnson and bias current? I don't have enough experience to quantify the noise penalty without actually calculating.
I'd stay with what you have.
If you do go with higher value bias resistors the source impedance will usually dominate anyway.
Open-circuit input noise will be higher as Rbias rises and I think that drives THAT's concerns to make them low-ish in value.
At low frequencies the reactance of the caps will add to the source impedance however.
At 20 Hz/47 uF you'll see ~170 Ohms/leg in series with the source.
Is there anything else to consider concerning the noise besides Johnson and bias current?
Capacitor leakage when phantom is applied is not trivial.
Good ones are 1-3 uA (DC) after minutes of "warm-up."
It's not just DC - there can be some significant noise current depending on the state of electrolytic formation.

Send us baby pictures when you get a board built-up.
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 »

Here's the link to bruno2000's build of the mic preamp: http://www.ka-electronics.com/images/pdf/WAYNE17A.pdf
hazmo
Posts: 19
Joined: Wed Dec 23, 2009 7:31 am
Location: Germany

Re: An Improved Servo for the THAT1510 and THAT1512

Post by hazmo »

Go for it. I've PM'd bruno to post his sch which is being used in his remote truck. It's the same as page one's schematic. We know it works at the cost of an extra op amp.
Only the engineering mind is the one that is a little unsatisfied ;). Cost wasn't much of an argument anyway, I get OP07s in single units at 0,29€ from a German distributor. And having four 8-pinners on the board makes for a nicer symmetrical layout...
Capacitor leakage when phantom is applied is not trivial.
Good ones are 1-3 uA (DC) after minutes of "warm-up."
It's not just DC - there can be some significant noise current depending on the state of electrolytic formation.
That sounds funky. And it's one of those things I would never think about myself unless someone told me about it. Where are those ideal parts we were told about in classes? :D
Send us baby pictures when you get a board built-up.
Sure, will do! And thanks for bruno2000's schematic. Reassurance is always a good thing.
Post Reply