Jensen Twin Servo 990 - Design Ideas

[mitsos]

I was going to suggest adding an input bias circuit, but I see a few trimmers, I assume you have them in place? 

I think the idea of putting the second 990 to unity gain is good, even though it won't be a true "one stage design." JH once posted that, while the 990 is technically unity gain stable, it's the least stable at unity gain, so you may have problems with oscillation down the line. Test it and see though, cuz I think that would be the easiest and least "pop-py." 

What are you doing about the gain? How are you going to get 60dB of gain in the 1-stage design? Are you just going to switch out R6 to a 2R resistor or are you doing something else? 

 

[mrvision]

regarding the servos in the pre I coulcil to test the IC opamp's output and to see if these outs are saturated. If one or both the outs are saturated than two or one resistors should be modified to have the best reduction of DC on the DOA's out concerning the servo saturated.
- if in the first stage the servo's opamp is saturated than R9 must be reduced until the servo's opamp is not saturated with good margins.
- if in the second stage the servo's opamp is saturated than R15 must be reduced until the servo's opamp is not saturated with good margins.
However setting r9=6.8k and r15=100k is a good way to use several DOA's in the market, comsidering that this pre is designed around 990 so it migh not run with other DOA's.

True... I am suspicious about R15... it seems that the 2nd IC servo is having a hard time making the second DOA reach zero, the DC offset on the second stage is higher, even with the servo there. The first stage is working really good.

The Preamp is being built around the JE-990 but I will also study the circuit using other DOA's and find simple mods that will make the preamp more friendly with other Discrete Op Amps

As PPA stated it is possible to use DOA's as servos, but also a good precision IC will do the job at a fraction of the cost of a DOA. The legendary Jensen Twin Servo had IC's as servos and now days, some $3K clones out there have IC's servos as well.

I am working out of town right now but when I get back home I will test the second Servo IC for possible saturation and also will bring R15 to a more common value. I still have the PPA99F on second stage, most likely the problem with the second servo will be related to R15 as Pier stated.

I was going to suggest adding an input bias circuit, but I see a few trimmers, I assume you have them in place?  

I think the idea of putting the second 990 to unity gain is good, even though it won't be a true "one stage design." JH once posted that, while the 990 is technically unity gain stable, it's the least stable at unity gain, so you may have problems with oscillation down the line. Test it and see though, cuz I think that would be the easiest and least "pop-py."  

What are you doing about the gain? How are you going to get 60dB of gain in the 1-stage design? Are you just going to switch out R6 to a 2R resistor or are you doing something else?  

Yeah the trimmers are part of Ibias compensation circuits.

I agree with the possibility of finding some unstability problems with unity gain but I have to make some tests first.

The 60db gain... Dean Jensen designed the twin servo to give a combined gain of 54.4db, where each of the stages provide about 27.2db. My idea is to make the preamp selectable between 30db gain and 60db Gain. The 30db gain is enough to drive modern line level or instrument signals and the 60db option is good for mics. So is not really a selection betwen a 1 stage 60db gain preamp and another 2 stage 60db preamp. I would like to hear some input about this...What do you guys think is better?

 

[okgb]

60 x 60 ?

correct me if i'm wrong but i thought the idea is
gain headroom with a slight lower first gainstage
and then make up some of that & drive with the 2nd stage
more like 30 x 2

 

[mitsos]

I figured you wanted a 2 slightly different preamp choices, like M1 vs twin servo. Good luck, interested to see how this ends up.

 

[mrvision]

30x2 is correct (actually 27.2db x 2) Sorry If I did not explain myself in the last post... english is my 2nd language  ;D

There is some interesting info posted a while ago by John Hardy on gearslutz about how the twin Servo was designed by Dean Jensen:

http://www.gearslutz.com/board/495115-post58.html

I found the Twin configuration very useful with microphones, but with line level or instrument signals the preamp seems to be a little too sensitive, you have to lower the level on the source, thus increasing problems with signal to noise ratio, you amplify the signal, but at the same time you are amplifying a lot of electronic noise from the source.

One stage by itself works really good with line level signals.

Mitsos thought the design was a preamp that could work like a M-1 (1 Stage) or a Twin Servo (2 Stage), my original idea was having two different levels of gain, 30  and 60 db. The use of 2 stages is open to a lot of possibilities.

The question for you guys is:

What do you guys think is more useful in a studio?

1. A preamp that can work like a M-1 and Twin Servo.
2. A Preamp with 30db and 60db Gain.
3. Both

(Since the topic has changed direction more towards design ideas, I am also changing the title  to "Jensen Twin Servo 990 - Design Ideas")

 

[mitsos]

Is the gain problem sensitivity (gain increases too fast) or too high a minimum gain?
If the latter a slightly higher value pot will lower your minimum gain. Compared to the stock values a 2K2 pot will lower your gain per opamp from 6dB to  about 3.75dB, for a total minimum gain of about 14dB vs 18dB stock. Might make your gain knob a bit more sensitive though.  If you can find 2K5 you can get this down to 3.38dB per opamp.

But if you mean it is too sensitive, then I'd suggest going with a rotary switch

to make your life difficult I vote for 3.  ;D

 

[ppa]

True... I am suspicious about R15... it seems that the 2nd IC servo is having a hard time making the second DOA reach zero, the DC offset on the second stage is higher, even with the servo there. The first stage is working really good.

I agree

 

[ppa]

As PPA stated it is possible to use DOA's as servos, but also a good precision IC will do the job at a fraction of the cost of a DOA. The legendary Jensen Twin Servo had IC's as servos and now days, some $3K clones out there have IC's servos as well.

A DOA runs much better than the precision IC for sound performances but needs a trimmer for settings.
How I've stated the matter is that in twin servo pre its servos sound. In the M1 the servo is pratically "invisible", it does not sound.
If we change the IC in the M1 it's hard to hear sonic differences but isn't the same thing with the Twin Servo, so using a DOA or an IC opamp on the M1 is unproductive, but in the Twin Servo is an different thing and we can listen differences in sound.

 

[mrvision]

A DOA runs much better than the precision IC for sound performances but needs a trimmer for settings.
How I've stated the matter is that in twin servo pre its servos sound. In the M1 the servo is pratically "invisible", it does not sound.
If we change the IC in the M1 it's hard to hear sonic differences but isn't the same thing with the Twin Servo, so using a DOA or a better sounding IC opamp than the original on the M1 is unproductive, but in the Twin Servo is an different thing.

PPA: I have checked the original twin servo circuit with a scope in the past and and I did not see any significant audio signal passing through the Servo. the servo OpAmps are configured in Intregrator mode which will respond more to DC changes to its input than passing pure audio signal.

The M-1 has its place as a good design but we are trying to follow the original design created by Dean Jensen, I know that we are probably going to find all kinds of possible flaws but the truth is that as it is, that design once became famous... and it is still famous...

Mitsos: I like 3 too... I feel a little masochist... ;D. The gain is too high at minimun gain but only with line level signals,  the only thing with the pots is that finding a dual gang rev log pot is hard, 1k 5k or 10k are more easy finding values.

 

[ppa]

I have checked the original twin servo circuit with a scope in the past and and I did not see the audio signal passing through the Servo. the servo OpAmps are configured in Intregrator mode which will respond more to DC changes to its input than passing pure audio signal.

a good servo DC should not have signal on its out, but it's not mean that it does't work with signal.
in the twin servo the servos work as active filter to eliminate the AC audio signal on the DOAs' out, so the opamp inside them works with the audio signal the same, and on its job to reduce to zero the audio signal, this opamp makes distortion harmonics. Since these one are not visible in the oscilloscope we think that the opamp doesn't work with the audio signal, but the opamp works (a lot) the same with the audio signal.

 

[mitsos]

Mitsos: I like 3 too... I feel a little masochist... ;D. The gain is too high at minimun gain but only with line level signals,  the only thing with the pots is that finding a dual gang rev log pot is hard, 1k 5k or 10k are more easy finding values.

Go for it!

dual 2K2 rev log pots: (they are filed under "carbon" but they are Omegs so they are actually Conductive Plastic).
http://audiomaintenance.com/acatalog/om-01-073_extended_info.html

 

[ppa]

nice pots

 

[Kingston]

PPA: I have checked the original twin servo circuit with a scope in the past and and I did not see any significant audio signal passing through the Servo. the servo OpAmps are configured in Intregrator mode which will respond more to DC changes to its input than passing pure audio signal.

I urge you to play with different opamps as the servo. Since you already mention you hear differences between DOA's on the gain stage, you should easily spot the servo changes as well. Especially in this design, just as PPA mentions above.

+1 for DOA footprint on top of the existing IC servo. The board would become slightly bigger I expect. Just throwing the idea out there...

 

[mrvision]

Nice pots I might give them a try.

Thank you for sharing all your knoledge and experience guys. I will do some tests on the weekend and I will let you guys know what happened. I am out of town right now...

 

[ppa]

however, changing the existing IC opamp in the servos with other types may give more DC voltage on the DOA out, but generally with many IC opamps you have max 2mV of offset voltage on the DOA's out.

I think that is important to make the twin servo a pre for several DOA's in the market, so the existing dc servos should be modified.

 

[mrvision]

Forgot to mention... Adding a DOA footprint for those that want to use DOA's as servos will also require Input Bias compensation circuits.

There are a few options out there but which one do you guys think will work better to correct Ibias in this particular case?

Right now I am using the following input bias circuit on the (-) input on the DOA's and it works well to correct DC offset:

 +15V
    |
    |
  10K trim --- 100K---270K --  (-) 990
    |                     |
    |                1 uF cap
    |                     |
 -15V                 GND

PPA: I agree, I will revise the servo circuit with other DOA's.

 

[ppa]

for the servos that uses the 990 is required a trimmer for input currents of this opamp because the lm394 gives these ones equal but these are a lot greater than the AD706.

 

[ppa]

I will post the schematic very soon

 

[mrvision]

ppa:

I am using LM11CN's as servos, I have tried some other IC's and the DC offset goes down to about 2mV like you said, but the LM11CN brings it down to micro Volts. John Hardy seems to like the LM11CN too..

http://recording.org/tech-talk/40654-twin-servo-what-2.html

This might be a dumb question... but the last time working on the the circuit last week I was a little confused about the method of setting up the Input Bias Null circuit. I recall reading somehwere that the best way was taking measure of DC present at the output of the DOA at minimun Gain and then disconnect the servos and turn the gain all the way up and adjust the trimpot until the DC at the output of the doa was + - 1mv of the DC voltage at minimun gain. It ceirtanly worked for the first stage but with the Second Stage when I turn the gain up the trimpot reaches its limit and there is still a lot of -mV left to bring the DC close to the reading at minimun gain.

The good thing is that even though the Input bias calibration in the second stage seems to be "not enough" to bring the 2nd DOA to 0V at maximun gain when the servo is connected back the DC offset goes down to about 2mV, not too bad. So i guess the Servo works... but maybe the method of I bias calibration does not work for the 2nd Stage or maybe I need to modify the Input bias null circuit? What am I missing here?

I also want to clarify that I had the circuit with a 990 on first stage and a PPA99F on second stage.

I want to have a clear idea about the Input bias compensation circuits before I touch the circuit again this weekend. Specially because we are now considering including a footprint for DOA servos...

 

[ppa]

fet opamps have much more offset input voltage than bipolar opamps , but fet opamps has a much less input quiescent currents. The twin servo pre was designed only for the JE-990 that has an ultra low input offset voltage value but big input bias currents for to be an opamp (2.2uA), so to use others DOA than 990 this trimmer circuit should be modified. Could you post its schematic? , please.
To be clear, the input bias circuit used in the twin servo doesn't run with many DOA's in the market, but if it will be modified might be run with 90-98% of them.