Console/mixer frequency response

[lauritz]

Hi

I'm currently working on some upgrades to my console - handbuilt fully discrete 16 channel Elberg (danish audio engineer/technician) console from the 80ies. Tweaking it to be even more useful as a studio tool. I've added seperate switchable line/mic transformer inputs and also output transformers + some more gentle mods. Put in a lot of vintage Lundahl input/output transformers that i got from old swedish broadcasting equipment.

It generally sounds really great and i did some testing of the frequency response on a channel running through the mixbus and out of the main outputs. It's is looking good and very flat all the way through, with a slight hi/lo cut. Approx. 1,5-2 dB rolloff at both 30 Hz (bottom) and 20 kHz (top). So i'm curious how that compares to the frequency response of popular transformerbased consoles such as Neve, API, Harrison, Trident etc. Are they complety flat in the full audible spectrum or would they have some inherent "transformer warmth" (read. hi/lo rolloffs).

Does anybody have frequency response data that they want to share? I reckon it would be part of a console service manual or generel data provided by the manufacturer.

I've attached a few pics if anybody is interested to see the Elberg mixer - and a channelstrip

all the best
Lauritz, Denmark

 

[ruffrecords]

At Neve, back in the 70s, when there were plenty of transformers in the signal chain, the top end of a channel amp like the 1073 was no more than 3dB down at 75KHz. For this you need a good low driving source impedance for output transformers and a transformer design with low leakage inductance.

The danger with adding output transformers to an existing design is that one or more of the above criteria may not be met.

The spec for a Neve console as whole ( entire signal chain -typically 5 to 7 transformers long) ) was +- 0.5dB from 20Hz to 20KHz.

P.S. I like the descrete op amps in the channel srtrip. Do you have a schematic for them?

Cheers

Ian

 

[lauritz]

At Neve, back in the 70s, when there were plenty of transformers in the signal chain, the top end of a channel amp like the 1073 was no more than 3dB down at 75KHz. For this you need a good low driving source impedance for output transformers and a transformer design with low leakage inductance.

The danger with adding output transformers to an existing design is that one or more of the above criteria may not be met.

The spec for a Neve console as whole ( entire signal chain -typically 5 to 7 transformers long) ) was +- 0.5dB from 20Hz to 20KHz.

P.S. I like the descrete op amps in the channel srtrip. Do you have a schematic for them?

Cheers

Ian

Thanks for the input Ian! I don't have a schematic ready at hand. It's the same opamps all the way through the console, in the preamp, eq section and also on the outputs and aux'es. They are labelled M6 - and i don't know if it is an original design or some old common discrete design? The transistors used are BC550b and BC560b. I've added a couple of pics of a spare one i have. Without knowing the details of these amps i would expext them to have low output impedance.

Just checked the main output section (master section) and realised that there's a coupling capacitor (16 uf) after the output opamp that i forgot to pay attention to when installing the output transaformers. I think that might be the reason for the low rollof. I have added another capacitor before the output tranformer (220 uf), but the 2 in series would mean a low capatitance making up a high pass filter with the primary coil of the output transformer (?). Oops. I will take a second (and better) look at what else is going at the master module!

 

[ruffrecords]

Two capacitors in series can never be greater than the value of the smaller one. You need to replace the 16uF with 220uF.

Cheers

Ian

 

[Murdock]

Nice console! I don’t know if it comes from a broadcast background, but a lot of broadcast consoles were intentionally limited in their frequency response. Especially high frequency. I think It’s not uncommon to have high class vintage broadcast consoles with not much above 20Khz.

 

[lauritz]

Two capacitors in series can never be greater than the value of the smaller one. You need to replace the 16uF with 220uF.

Cheers

Ian

Hi Again Ian.

Yes, that's was in fact what i tried to say Lower capacitance - higher cutoff frequency. I've now tried to replace the coupling capacitors (remove the original 16uf) in right side of the master module and did some testing which is attached below. Instead of adding bottom end the cutoff actually became greater (around 3 db now) - i've read that the series capacitance before the transformer will give rise to a resonant peak around the cutoff frequency, so maybe that peak has now been removed now by raising the cap value (or pushed down below 20 Hz at least). Resulting in less bottom end...(?).

Also did some testing of the transformer direct output on the channelstrip and that looks a bit better. I think i'll have to investigate further what is going on at the stereo master module.

The tests have been done in REW with a stepped sine signal 31 measurements across the spectrum, so not super high resolution, but i think it's good enough to show the general behaviour.

 

[lauritz]

Nice console! I don’t know if it comes from a broadcast background, but a lot of broadcast consoles were intentionally limited in their frequency response. Especially high frequency. I think It’s not uncommon to have high class vintage broadcast consoles with not much above 20Khz.

Thanks! I thinks it's very nice too, which is why i'm bothering to make the modifications. I think it has been used as a live mixer before, but it has a tape-out on the master module, so also intended for recording. It was completely unbalanced before, except for 6 of the channelstrips which had mic transformers on the input.

I will take a closer look at the boards and look for any frequency filtering along the way. I'm no specialist on opamps, but it seems strange to me if they would have a limited band response themselves.

all the best, Lauritz

 

[Matt Syson]

The op amps themselves will not have a significant frequency response on their own so largely DC to 100KHz but probably some distortion at higher frequencies. Transformers also introduce phase errors such that if a desk has group modules that can mix back into the main stereo bus (compared to direct channel to stereo routing) they can actually be out of phase at 15KHz or more! I was testing broadcast mixers destined for the BBC (MK4, GP desks) and discovered this slight anomaly but never commented at the time as they were in production for several years beforehand so obviously not noticed by the people using the desks. Also note that a low vamlue capacitor when feeding a transformer will 'ring' so the caps need to be a LOT larger unless you want a LF boost (hump). Audix used 2200uF caps (or even 4700uF to drive the larger transformers) especially as the transformer gave a 6dB level boost meaning the standard test load of 600 Ohms 'looks' like 50 Ohms. Remember broadcast desks were deliberately rolled off above 15KHz because FM radio has a brrickwall filter above that ( to keep clear of the 19KHz pilot tone for stereo, and then cut at the bottom because some transmisssion chains used a tone of around 5Hz (I think) which was part of remote transmitter testing so if the tone was absent a fault indicator would come up in a control room to signal a problem.
AM radio failed to broadcast above 5KHz as they were on 10KHz channel spacing!

 

[FIX]

So few people remember those rules! The original API 550 and 550A had a filter switch that was 50 and 15K, people think it's just a HP filter...

Transformers have a great amount of HF and LF phase shift, but remember, if all the channels are the same, then the relative phase shift is meaningless. This is one reason why "in the old days" people used a console to make a record, with all the same Mic Pres, els and summing. With the evolution of the outboard gear availability of a million different options, the coherence if many songs has taken a hit...

 

[ruffrecords]

The tests have been done in REW with a stepped sine signal 31 measurements across the spectrum, so not super high resolution, but i think it's good enough to show the general behaviour.

Have you created a cal file in REW for your interface?

Cheers

Ian

 

[lauritz]

Have you created a cal file in REW for your interface?

Cheers

Ian

Almost... I did a test with the computer output routed directly to input and the response was very flat all across the spectrum. Same 31 point measurement. My converters are Apogee AD/DA 16x...old, but still good interfaces i think. Guess i should do a proper cal file, but i don't think would account for missing amplitude of the order 1-3 dB in top and bottom.

I will do some more testing on the console, also bypassing the transformers...to see if that's the source of the limited bandwidth.

best, Lauritz

 

[lauritz]

The op amps themselves will not have a significant frequency response on their own so largely DC to 100KHz but probably some distortion at higher frequencies. Transformers also introduce phase errors such that if a desk has group modules that can mix back into the main stereo bus (compared to direct channel to stereo routing) they can actually be out of phase at 15KHz or more! I was testing broadcast mixers destined for the BBC (MK4, GP desks) and discovered this slight anomaly but never commented at the time as they were in production for several years beforehand so obviously not noticed by the people using the desks. Also note that a low vamlue capacitor when feeding a transformer will 'ring' so the caps need to be a LOT larger unless you want a LF boost (hump). Audix used 2200uF caps (or even 4700uF to drive the larger transformers) especially as the transformer gave a 6dB level boost meaning the standard test load of 600 Ohms 'looks' like 50 Ohms. Remember broadcast desks were deliberately rolled off above 15KHz because FM radio has a brrickwall filter above that ( to keep clear of the 19KHz pilot tone for stereo, and then cut at the bottom because some transmisssion chains used a tone of around 5Hz (I think) which was part of remote transmitter testing so if the tone was absent a fault indicator would come up in a control room to signal a problem.
AM radio failed to broadcast above 5KHz as they were on 10KHz channel spacing!

Many thanks for the insights! Since i'm lacking a little top/bottom i guess such a frequency hump would not apply to the problem on my console. But I'm thinking that maybe since the transformers stem from broadcast gear they could be limited in their frequency response, due to the points your making. I think i will do a test on them outside the circuit, but with a similar load.

They are called LL2801 and SR 503. Both Lundahls, so i assumed them to be of high quality. Not much info to be found online except for pinouts ans the fact that they are output transformers.

best, Lauritz

 

[ruffrecords]

They are called LL2801 and SR 503. Both Lundahls, so i assumed them to be of high quality. Not much info to be found online except for pinouts ans the fact that they are output transformers.

best, Lauritz

Hi quality does not necessarily mean right for this application. If the 2801 is anything like the 2811 it has a relatively low primary inductance which will explain your poor LF response. The 2811 is really designed to be driven by and active feedback circuit.

Cheers

Ian

 

[lauritz]

Hi quality does not necessarily mean right for this application. If the 2801 is anything like the 2811 it has a relatively low primary inductance which will explain your poor LF response. The 2811 is really designed to be driven by and active feedback circuit.

Cheers

Ian

Good point! I will try with a different output transformer where info is less sparse. Have some other types lying around. And also do some measuments with no transformers at all...

I hope i can get it to work a little better with the vintage lundahls though...i have quite a few of those that i bought some years back

best, Lauritz

 

[ruffrecords]

Also it would really help if we had a schematic. We have no idea of the limitations or capabilities of the output stages.

Cheers

Ian

Cheers

Ian

 

[13engrsapper]

Since the manual is on my desk...
With the filters switched off a Studer 963 (these are transformer coupled in and out) the specs are: +0.5 to - 1.0dB from 4.5Hz to 40kHz.
Not too shabby for 40 years old, non-discreet, and transformers.

 

[lauritz]

Also it would really help if we had a schematic. We have no idea of the limitations or capabilities of the output stages.

Cheers

Ian

Cheers

Ian

Yes, of course. Don't have the real schematics for the console, but i pulled out the master module last night and i've drawn a quick diagram of the channel input sections and the master output section. See attached pdf. There's a few resistors where i could'nt see the values, because they are hidden under the opamps, but it think the general topology should but be clear the drawing.

Also did some tests with input signal injected straight to the preamp opamp and that did'nt change the response, so i think input transformers are good and working well. I've attached a few pics of the output section to reference. The opamp with the blue ring is the one feeding the main output.

all the best
Lauritz

 

[lauritz]

Since the manual is on my desk...
With the filters switched off a Studer 963 (these are transformer coupled in and out) the specs are: +0.5 to - 1.0dB from 4.5Hz to 40kHz.
Not too shabby for 40 years old, non-discreet, and transformers.

Thanks for sharing

 

[lauritz]

Yes, of course. Don't have the real schematics for the console, but i pulled out the master module last night and i've drawn a quick diagram of the channel input sections and the master output section. See attached pdf. There's a few resistors where i could'nt see the values, because they are hidden under the opamps, but it think the general topology should but be clear the drawing.

Also did some tests with input signal injected straight to the preamp opamp and that did'nt change the response, so i think input transformers are good and working well. I've attached a few pics of the output section to reference. The opamp with the blue ring is the one feeding the main output.

all the best
Lauritz

Small correction: The input coupling cap. at the channel input is not polarized - or it's actually 2 x 33 uf electrolytics back-to-back.

 

[ruffrecords]

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