noise minimizing - discrete Elberg mixer

[lauritz]

Hi

I have got this mixer, which i am modding a bit to fit into my studio setup. It's is build by a danish engineer "Elberg" who has done a lot of work for danish broadcasting and also his own products -high quality stuff. It's all discrete opamps throughout, but it was originally intended as a live console, so some things needed to be changed for studio use. Here are some pics:

https://www.dropbox.com/sh/yk1n0i3as7a89cf/AADLvUk16lybYo76Zme28W2Sa?dl=0

Mods:
I have added transformers(lundahl) to balance the outputs and to the send/return.
Added a pad for hot signal(line level) and a switch to send signal on each channel to either direct out or stereobus.
Added impedance balancing(with resistors) to all direct outs(channels). 
Added mic input tranformers(and zobel network) to 10 channels that was unbalanced before.
Built a new power supply because i accidentially fried the old one by shorting the rails to ground. Current draw is approx. 0.8 A pr rail -running on +/-15 volts. So i made a PSU with LM317 and LM337 regulators.

well my question is...

I get around 50 db of signal to noise ratio when sending a signal through a channel and measuring on the outputs of the masterbus. All other faders are down. The direct outs a a lot quieter 78 db s/n. The noise floor with all faders down and masterbus faders up is -96 db relative to 0 db in Pro tools. When i pull all faders up the noise is pretty severe and easily audible.

How can i get it to be more quiet?  The noise is mainly 50 Hz and overtones/harmonics.

Seems like the noise is being picked up around the channel faders or maybe on the stereobus wires....but i'm not sure.

The case is a wooden flightcase covered by some conductive foil on the inside. I have checked the grounding several times and i think it should be ok. Maybe a regular metal case would help?

I have changed both transformer and psu board to ensure that the PSU is alright. I also added an additional CRCRC filter before the regulation.

Help and good ideas are much appreciated

best
Lauritz

 

[thermionic]

It could be due to a lot of reasons. Do you have a schematic, particularly with regards to topology and values for the summing section? If it has a high-ish impedance it will need careful screening and as short signal paths as possible. 

 

[lauritz]

Hi Again

Thanks for your reply Thermionic.

I did some more test and i think that i might have messed up the zobel network after the input transformers. When i got the console, it already had 6 transformer balanced inputs. With just these connected to the master bus -it is very quiet. When i add one of the channels where i have added a tranformer the noise is induced. The other mods have been done to all channels, so they seem to work alright.

Could the noise come from a bad layout of the zobel network...as i recall i did the testing of the tranformers without having it hooked up i the circuit, so i may have choosen some wrong cap/resistor values. Could that give rise to a lot of ground noise? Or maybe the transformers i have used are not suited for this particular job? 

I have added some pics of the inside to the dropbox folder above. I will try to draw up a schematic "asap", but i am not that fast with the schematics-program, so it might take me a little while. 

best
Lauritz

 

[lauritz]

Tested some more and the input transformers and zobel networks seem to be fine. I discovered that when i add more than 6 channels to the console the noise starts.

That makes me wonder if the PSU is not good enough, even though i thought it was alright. From the LM317, LM337 datasheet 1,5 A should be provided for each rail. I'm thinking that maybe my measurement of the current draw are not truthful and that the console needs more current than first assumed....

Could noise appear if the console is underpowered?
Another thought was that the power cable from the PSU is faulty, so that when a certain amount of current is pulled through it will start to be noisy?

It was no problem before i started messing around with the console, so i think the cable is ok -even though it is not shieded. (4 conducters: +15V, -15V, 0V, chassis ground)

best, Lauritz

 

[Andy Peters]

Tested some more and the input transformers and zobel networks seem to be fine. I discovered that when i add more than 6 channels to the console the noise starts.

That makes me wonder if the PSU is not good enough, even though i thought it was alright. From the LM317, LM337 datasheet 1,5 A should be provided for each rail. I'm thinking that maybe my measurement of the current draw are not truthful and that the console needs more current than first assumed....

Could noise appear if the console is underpowered?

Yes ... but ... the usual result of an overcurrent condition with the LM3x7-type regulators is that they turn off and then back on repeatedly. This is quite obvious if you look at the output on an oscilloscope.

Actually, the usual limiting factor is less the current draw than thermal overload. They need to be heatsinked properly or they will shut down even with a load current less than the spec'd maximum.

Also check to make sure that the voltage input to the regulators is high enough to meet the regulator drop-out specs.  For 15 V out at 1.5 A the data sheet says the drop-out voltage is about 2.5 V, so your input needs to be at least 17.5 V.

Another thought was that the power cable from the PSU is faulty, so that when a certain amount of current is pulled through it will start to be noisy?

There is some IR drop across the power cable, and that's of course a function of load current.  But unless the load is excessive and the cable is long, the loss is probably not a problem.

It was no problem before i started messing around with the console, so i think the cable is ok -even though it is not shieded. (4 conducters: +15V, -15V, 0V, chassis ground)

Cable shielding doesn't do anything for 50 Hz noise.

Now back to the original question.  You said you built the power supply.  This seems simple, but you have to mind the rules about board routing to make sure that the rectifier noise (ripple) doesn't end up in the output. This also means that you need sufficient capacitance after the rectifier and the regulators need output capacitance. Oh, also on the board you need some bulk capacitance at the power input.

 

[lauritz]

Thanks for the inputs, Andy. I have messed around with console some more...here's what i've done: 

- Changed the power cable going to the console with a shielded one, even though it wont help lowering the mains noise.

- Changed the power transformer to a larger one. -supplying 20v, 3.15A pr. rail.

- Removed the CRCRC board from the power supply...it acutally seemed to make the noise worse...think i used too large resistors (2 x 10 Ohms per rail)?  Just realized the it causes a quite large voltage drop with a current draw of about 1 A per rail(The old power transformer was about 25v AC at the secondaries.).  I think maybe that causes some trouble once the PSU gets loaded by the console...maybe i can put the filter back in with some more apropriate resistors. 1 Ohm, 5 W or something like that....

- Changed the PSU board to one designed by Silent Arts from this forum. It's also based on LM3x7 regulators. I think the layout should be alright on this board. I pulled some wires to the regulators and mounted them on the original (huge) heatsink. (Pics of the PSU before the changes are in the dropbox-folder above)

Results: The mains noise seems to be lowered significantly. I am not quite happy yet, but it's much better. Some more hiss(high freq noise) seem to have arrived as a trade off though...

Oh, also on the board you need some bulk capacitance at the power input.

This mod i had already done...3000uf caps between the rails and ground right at the entrance point of the power cable. Is that enough?

best wishes
Lauritz