Variable trim in between 2 gain stage - ideal components value?

[ruffrecords]

Ian, I  don't see how that is, unless we assume the feedback impedance around the 1st amp has also been scaled up 10 times.  But if the 1st stage amp has been optimized for noise, it should be pretty low and will mostly swamp out that 10 X less load. 
Practically speaking, taking typical values used in 990 circuits, the difference will be closer to  2 X less load if the network is scaled up.

It is basic ohm's law: at any signal level, the current in a 1K resistor is 10 times the current in a 10K resistor. In a class B output stage  that current comes directly from the power supply.

Cheers

Ian

 

[Winston OBoogie]

Just contrary, I would say it is slightly better to use lower impedance pot if it isn't physically placed on the PCB between opamps,

This is also of consideration of course.  No op amp is bullet proof as far as oscillation or immunity from  rf.  Sometimes a small value cap is required from + input to ground and, using the resistance  values shown in post one, the upper frequency will change by around 1 octave wherever the pot is set.    Scaling up by 10 the pot and divider will make this moving around of upper end filtering around 9 octaves.

 

[Winston OBoogie]

It is basic ohm's law

Never heard of it

In your 10 X scaling up version, the point at which the transition through AB to B happens will still largely be determined by the feedback impedance in a well designed 990 circuit. 

A 10K pot and 4K99r are cheap as chips,  why not try both variants in your servo  amp Elskardio.

 

[moamps]

It is basic ohm's law: at any signal level, the current in a 1K resistor is 10 times the current in a 10K resistor. In a class B output stage  that current comes directly from the power supply.

990 opamp drives 25mA from a power supply. If only 20 mA is used for the output transistors, the question is how low should be the load to switch transistors into the class B at maximum output voltage swing. 

 

[Winston OBoogie]

990 opamp drives 25mA from a power supply. If only 20 mA is used for the output transistors, the question is how low should be the load to switch transistors into the class B at maximum output voltage swing.

The IPS and TIS are biased quite rich so the output transistors idle currents are a bit less than 20mA (circa 14mA).  A quick look at the schematic would suggest that, with the 3.9 ohm emitter resistors,  this is as close to perfect for minimizing crossover distortion as you could get.
Given that Dean designed the amp to be able to drive a reflected primary impedance of one of his transformers down to 75 ohms so as to compete with the likes of API et al., it might be safe to say that class B operation over a significant portion of the signal was anticipated.

Regarding crud noise from switching, complete schematics of mic amps from Jensen show 470uF bypass caps per voltage rail along with the usual small bypass.

 

[ruffrecords]

990 opamp drives 25mA from a power supply. If only 20 mA is used for the output transistors, the question is how low should be the load to switch transistors into the class B at maximum output voltage swing.

Is that what you are trying to achieve - ensure the 990 operates in class B? In that case definitely use 1K.

Cheers

Ian

Cheers

Ian

 

[moamps]

Also the correct work of the servo part should be checked when a pot is inserted in the input of the second 990 opamp. If the pot has high value, it may IMHO turn the servo and the whole opamp into (low frequency) instability, especially when the gain (DC gain also) is set to high.
The servo signal is definitely affected with the trim pot position. So the best pot for trim is no pot at all in this design.

 

[moamps]

Is that what you are trying to achieve - ensure the 990 operates in class B?

Not at all. I just try to say to you that the first opamp will work most of the time (maybe always in this design) in the class A for the load of 1k also, and there will not be additional current taken from the power supply.

 

[ruffrecords]

Not at all. I just try to say to you that the first opamp will work most of the time (maybe always in this design) in the class A for the load of 1k also, and there will not be additional current taken from the power supply.

You are probably right. I just looked back at the original question and no mention is made of the actual op amp being used let alone that is was a beast like the 990. It annoys me when people dribble out information over several posts that they should have put in the first one.

Cheers

Ian

 

[Winston OBoogie]

Is that what you are trying to achieve - ensure the 990 operates in class B? In that case definitely use 1K.

Elskardio shows an external load of 1K499 which, by itself, leaves 990 #1  in class A up to +20dBu.

To this add:  However much gain there is in 990 #2.  I'll ASSume it isn't zero gain or why bother adding it.  How about a nominal 10dB of gain (still bordering on why bother but...) , then we can then add that 10dB to our class A headroom. 

But, we also don't know the load on 990 #1 from the feedback network which will bring the load on 990 #1 down somewhat. 
Regardless and until or if ever we do know, it's hardly ensuring the 990 is operating in class B


As for feedback/gain  -  The better way is not really as shown in the twin servo scheme posted above.  And this isn't how Deane Jensen did it, nor how John Hardy does it either. 

 

[elskardio]

It annoys me when people dribble out information over several posts that they should have put in the first one.

Sorry if I annoyed you Ian. I was looking for a "general rule" hence the general question rather than something specific to the 990.

A 10K pot and 4K99r are cheap as chips,  why not try both variants in your servo  amp Elskardio.

Tested the circuit with 1k + 499R, 5k + 2k49, 10k + 4k99. All works perfectly fine and I can't measure any difference between them. Didn't check the change in current draw. To my understanding that's were they could differ.

Also the correct work of the servo part should be checked when a pot is inserted in the input of the second 990 opamp. If the pot has high value, it may IMHO turn the servo and the whole opamp into (low frequency) instability, especially when the gain (DC gain also) is set to high.

This is the first thing I checked after adding the trim control. The servos are perfectly fine with it. We're talking of +/- 10mV going from one end to the other of the Trim pot.

As for feedback/gain  -  The better way is not really as shown in the twin servo scheme posted above.

I know it creates a low pass filter and the higher the gain, lower the cut off frequency... but if you measure the frequency response at maximum gain, you'll see that it's way over 20kHz. And I doubt the user will have to use maximum gain. Even with vintage ribbon mic, the gain is at 3/4 max.

 

[ruffrecords]

Sorry if I annoyed you Ian. I was looking for a "general rule" hence the general question rather than something specific to the 990.

I which case I apologise. The answers I gave were aimed at the general case - op amps like TL072 or NE5532. The 990 is a special case where you can safely go down to 1K without problems.

Cheers

Ian

 

[Winston OBoogie]

I know it creates a low pass filter and the higher the gain, lower the cut off frequency... but if you measure the frequency response at maximum gain, you'll see that it's way over 20kHz.

OK yes, that's part of what I was implying, using a fixed series FB R with the parallel cap in lieu of how it's shown above.  FWIW,  1K and, I think, 680pF (it's been a while so it might have ended up differently) is what I used myself.

The other part is then being able to locate the maximum gain stopper resistor directly at the -tve input.  This then helps ameliorate any possible stability issues with running cabling or the PCB tracks to the gain pot.   

 

[Winston OBoogie]

Heya Adam
Hadn't forgotten about your PM, having had terrible issues with a dead laptop, and a tablet that started acting erratically (I've somehow managed to accidentally erase about 4 of the posts I made in this thread looking back on it) I finally saved the spondoolies and took delivery of an old, but new to me, MacBook Pro today so should be catching up here on out.

  By all means though push the first amp harder though (75ohm limit) and see what happens.. I don't think it'll get really exciting sounding, not with that particular opamp circuit in that application, but I'm curious about it.

I think you're correct in that.  I don't have golden ears, they're probably too old and abused now anyway, but I don't expect there will be any benefit at all in loading down a 990 to 75 ohms or by any more than necessary for good design house-keeping.
Despite the various opinions expressed here, I'm sure we'd all agree on that.

Again, without knowing the gain in 990 #2, I will reserve my right to still conclude that there's no real benefit in going higher than necessary either.    In reality, Elskardio is the best arbiter of what works best for him

 

[Winston OBoogie]

And I'm sure your ears/instincts are pretty darn good. 

Cheers and thanks Adam. 
From what I've read, your ears and instincts are absolutely on par with anybody's on here dude.  My only edge might be a few more recollections of distantly read or tried schemes. 
Catch up with ya soon my friend, curious to hear about your endeavours and successes