Console/mixer frequency response

[lauritz]

The 46.7 ohm resistor in series with the op amp in the direct out could well be contributing ti the LF loss. It is needed for an unbalanced out to prevent op amp instability when driving capacitive loads but is not necessary for driving the transformer. (Is it really 46.7 ohms??). I notice this resistor is absent on the output section.

Cheers

Ian

Thanks. Yes, the value of this resistor about right. It is in fact 46,4 ohms (specified by the color code, 1 % tolerance), but i measured it with the ohm-meter to 46,7 Ohms. There's another 46,4 Ohm series resistor (+ 50 uf coupling cap) between the output of the eq section and the channel fader - i guess that would add to lowcut as well then? Or would it not be more of a low pass filter? (series resistor + some sort of capacitance to ground).

This resistor is indeed absent on the main output section. I just checked again.

I found a few mistakes in the schematic and made updates for a version 2. I had the input impedence resistors connected to the wrong input for the preamp opamp and also remembered that i changed the value of the mic + line resistor values to better suit the newly added input transformers, so those values have now been updated. Still within same order of magnitude (1k to 4.3 k and 22 k to 47 k). Frequency responense is just about the same anyway. I did test both with the original values and these updated ones.

The mic transformer LL1527 datasheet specifies 3-4 kOhms termination for best squarewave response and the line transformer (Lundahl SR 502 30k:30k line input bridging) i reckonened 22 kOhm was a bit too low. Looked at some Jensen transformer datasheets and their 10k:10k input bridging transformers are recommend to be terminated with a least 10 kOhms (or more). So i took that as a guide.

PS. added a couple of closeups of the input channelstrip i'm testing.

 

[ruffrecords]

Assuming the channel fader is 10K ohms then the series 47 ohm resistor will not affect the LF response (but it could for the much lower impedance of the output transformer).

From what you have measured so far it seems that the input transformers are not the problem.

Is the frequency response to the direct out the same as to the main out? If they are then the series 47 ohm resistor is not the problem.

It is odd that they seem to be using E96 series resistors. When was this mixer made?

Cheers

Ian

 

[lauritz]

Assuming the channel fader is 10K ohms then the series 47 ohm resistor will not affect the LF response (but it could for the much lower impedance of the output transformer).

From what you have measured so far it seems that the input transformers are not the problem.

Is the frequency response to the direct out the same as to the main out? If they are then the series 47 ohm resistor is not the problem.

It is odd that they seem to be using E96 series resistors. When was this mixer made?

Cheers

Ian

The mixer is from 1985 and all faders are 10k ohms. The pcb have that year printed and i assume the mixer was build around the same time. I have written the guy that build it (Elberg) for documentation on the mixer, but my email bounced back. Will try again, but he is an older gentleman at this point

Direct out is better than the main output - in the bottom end. Less than 0.5 dB down at 20 Hz. Both direct out and main out have the same dampening in the top end though. In one the first post i've attached a graph of the responenses of the direct out, and the main outs also in another graph. One is showing the direct out i've measured on both line and mic inputs, with the line input having the best response.

Guess i could pull out the 47 ohm resistor and test the direct outs again.

 

[Matt Syson]

the frequency response can even be affected by the level of your test signal ad the transformer may begin to saturate (lower effective input impedance) and drive the op amp into current limiting so you need to consider other effects not just frequency response. the output should also be terminated correctly.
There are so mant y pitfalls to looking at one aspect of a circuit only.

 

[lauritz]

the frequency response can even be affected by the level of your test signal ad the transformer may begin to saturate (lower effective input impedance) and drive the op amp into current limiting so you need to consider other effects not just frequency response. the output should also be terminated correctly.
There are so mant y pitfalls to looking at one aspect of a circuit only.

Yes, thanks for the reminder!, Speaking of Ievels i think that i may have been driving the mic input a little too hot in earlier measurement, as compared to typical mic levels. Have'nt used any impedance conversion box or anything like that, just the line level turned down. Just made a new measurements of the channel direct out with the level down to a more reasonable level.

Also i bypassed the 47 ohm series resistor and the low end is now much better - almost flat for the line input. Top end seems slightly better too, now less than 1 dB at 20 kHz. I've attached the updated measurements. The blue graph is the loopback test for the soundcard. Green is mic input, and yellow is the line input. Both taken from the direct out.

I have'nt terminated the output transformer secondaries, so they are just terminated by the load of the soundcard input. I read that sometimes it may be a good idea to terminate the outputs with an additional resistor, is that what you mean by terminating correctly? I looked at some Jensen (1:1) line output transformers and their suggested output circuits all seem to have no termination, but i guess i can differ for other transformers.

Thanks for the feedback, i'm learning a lot doing this project

 

[ruffrecords]

Looks much better to me.

Whether or not you need to terminate an output transformer depends on the transformer. Some older ones do but many modern ones do net, Some need an output Zobel network to tame the supersonic response and some do not. Unfortunately much of this info is scarce.

Cheers

Ian

 

[lauritz]

Looks much better to me.

Whether or not you need to terminate an output transformer depends on the transformer. Some older ones do but many modern ones do net, Some need an output Zobel network to tame the supersonic response and some do not. Unfortunately much of this info is scarce.

Cheers

Ian

I think it looks much better too

Now I'm curious what would happen if i bypassed the additional 47 ohm resistors going to aux sends, and the ones between preamp and eq section + eq section and channel fader. Even though the direct outs are looking good the response should be the same as before at the mixbus.

I will also check the squarewave response at the outputs. I already did zobel networks for secondaries windings at the inputs. But not for the outputs at this point

 

[lauritz]

I don't think the channel direct outputs need zobel network. Just fed a 800mV RMS squareware through the line inputs and measured with the oscilloscope at the secondaries of the output transformer. With a 10kOhm termination at the secondaries. Looks just about the same unterminated. Heres are some pics of the oscilloscope at a few test frequencies 1kHz, 10kHz and 20 Khz.

 

[Matt Syson]

The 47 ohm resistors are also there to decouple the op amp output from the capacitance of the circuit traces and other wiring otherwise there is some risk of the op amp oscilmlmating at some loading or gain setting. IF the op amp was only used at low gain and has a negative feedback capacitor of a suitable value then you could be quite sure it won't oscillate, but having several variables is never a good idea for a reliable product.

 

[lauritz]

The 47 ohm resistors are also there to decouple the op amp output from the capacitance of the circuit traces and other wiring otherwise there is some risk of the op amp oscilmlmating at some loading or gain setting. IF the op amp was only used at low gain and has a negative feedback capacitor of a suitable value then you could be quite sure it won't oscillate, but having several variables is never a good idea for a reliable product.

Thanks for your inputs Matt.

Sounds like the advice would to maybe not mess around too much with these resistors. The one resistor i "bypassed" is only bypassed when switching to the direct output - when switch back to the rest of the circuitry it's still in place. I just moved the switching point.
As far as i understood from comments earlier in this thread the transformer itself will act as resistance to make sure that no oscillation will take place.

The reason why there's no resistor on the last opamp on the main output section - is that because it's a different opamp configuration that is not likely to oscillate? (See hand drawn schematic v2 in earlier post).

/Lauritz

 

[lauritz]

I did some simulations in LTSpice to further investigate the damping in the high frequency spectrum. The simulation response is very close to the measurement i've done on the channelstrips. The capacitor (C1) seems to determine the cut-off frequency in the top end. The value is 100pf and if i lower it to for example 50pf the cutoff moves further up and gets almost invisible at 20kHz. I dont know the purpose of this capacitor in the circuit, so my question is - can it be altered safely without messing something else up?

I've seen similar preamp circuits where there is no capacitor at all in this position.

Attached is some screenshot from LTSPice - the circuit and the frequency sweep

 

[ruffrecords]

Whether or not you need a capacitor in this position depends on the op amp and whether or not you want to curtail the HF response. Some op amps don't need it (TL072 for example) and with others you have to have one (NE5534 for example). Their principle function is to maintain stability at low gains which works by deliberately making the response fall off at supersonic frequencies. If you use a larger value you still get the stability but you may also affect the audio band response. I suspect that, since this is a broadcast mixer, it has been used to deliberately curtail the HF response. I would change the cap value to 47pF.

Cheers

Ian

 

[lauritz]

Whether or not you need a capacitor in this position depends on the op amp and whether or not you want to curtail the HF response. Some op amps don't need it (TL072 for example) and with others you have to have one (NE5534 for example). Their principle function is to maintain stability at low gains which works by deliberately making the response fall off at supersonic frequencies. If you use a larger value you still get the stability but you may also affect the audio band response. I suspect that, since this is a broadcast mixer, it has been used to deliberately curtail the HF response. I would change the cap value to 47pF.

Cheers

Ian

I will get some 47pf caps and try it out then Thanks for all the help - it's much appreciated.

 

[lauritz]

Installed a 51 pf i had lying around and did this measurement on the channelstrip, line input + direct out. That helped the high end response as predicted and it's now down to less than 0.2 db at 20khz. I'm happy now and will move on to the rest of the channels for the same mods. Also the opamp still seems stable with the change of the capacitor. No oscillations, noise etc.

Attached is a measurement with the original 100pf cap and the new 51pf cap.

all the best, Lauritz