Mixer External Power Supply - what is best practice for stability?

Help Support GroupDIY:

gnd

Well-known member
Joined
Jan 24, 2006
Messages
285
Mixer External Power Supply - what is best practice for stability and low noise?

I may need to redo PS for my mixer, 24 channels, stereo out and monitor section, some basic switching, in order to maximize mixer stability. Supply current cca 2-3 A, voltage 17V.

PS has the following components:
- external power toroid transformer,
- rectifier and power smoothing caps,
- DC lines to regulators
- boards with 317/337 regulators
- connection from regualtor boards to channels. Each channel is decoupled at LF with 220UF caps rails-ground, and all IC's are decoupled at HF with 100nF directly at IC pins rail-rail.
- mixer grounds star point, for four grounds (0F, 0IN, 0P, 0A)

I'm including skecth of my current setup (sorry for lack of artistic ability, I hope it is clear enough), with which I have problem with oscillations in mixer.

What is BEST PRACTICE to implement PS for such mixer, mostly maximizing STABILITY, avoiding oscillations as much as possible?

1) Power transformer doesn't fit in the box, so it must go external.
Is it better to:
a. have it at distance from mixer(say 2-3 meters) or
b. as close as possible to mixer, just far enought to prevent hum induction to mixer frame

2) rectifier diodes with say 4x10000uF smoothing caps and 100nF bypass, pre-regulator.
Better to have it:
a. as close to mixer as possible, to have ground and DC lines from CAPS to 317/337 regulators as short as possible, but in this case connection to mains ground is longer?
b. as close as possible to external transformer, to bring 0V point (between smoothing caps) as close to mains ground as possible?

3) DC connection (2x rail plus grounds) from smoothing caps to regulator boards 317/337 @ 17V
-Q1. should it be as short as possible? Is it more stable if shorter?

4) boards with 317/337 regulators
-Q1. Do these need to be as close to mixer channels as possible?
-Q2. Is it better to have one higher power board (317/337 plus boost power transistors on large heat sink), or better to have several lower power boards (just 317/337 on smaller sinks) feeding parts of mixer?

5) regulated DC lines from 317/337 to channels
-Q1. can these be say 1m long? Does the length of regulated rails influence mixer stability and noise?

6) mixer grounds star point
-Q1. Is it better to have ground star point:
a. as close to mains rectifier smoothing caps as possible, meaning that I take four ground cables say 3meter long, and connect them far from mixer, directly in between remote power smoothing 10000uF caps?
b. as close to channels as possible, meaning that I connect grounds inside mixer box, and then go with just one say 3metre long cable to point between 10000uF smoothing caps?

...
 

Attachments

  • power supply schemo.jpg
    power supply schemo.jpg
    163.3 KB · Views: 50

ruffrecords

Well-known member
Joined
Nov 10, 2006
Messages
13,931
Location
Norfolk - UK
You really need the transformer as close to the rectifiers and smoothing caps as possible and all together in a screened box. 30cms between them is asking for trouble.

Cheers

Ian
 

pedrocruz

Well-known member
Joined
Nov 13, 2009
Messages
73
ruffrecords said:
You really need the transformer as close to the rectifiers and smoothing caps as possible and all together in a screened box. 30cms between them is asking for trouble.

Cheers

Ian
Hi Ian, you really need to have the rectifiers closer to the Trafo as possible?
Ive just Diy'ed 8ch of SSL 9k and my PSU is like 3meters away from the Transformer, once the transformer is external to the PSU+SSL 9k Case...
I mean, the cable i run from transformer to PSU is like 3m long...
You can see some pictures here:
http://www.groupdiy.com/index.php?topic=33100.160

Ill be watching for your reply,
Thanks!

(Sorry invading this post with something that is nothing to do with the title)
 

ruffrecords

Well-known member
Joined
Nov 10, 2006
Messages
13,931
Location
Norfolk - UK
pedrocruz said:
Hi Ian, you really need to have the rectifiers closer to the Trafo as possible?
Ive just Diy'ed 8ch of SSL 9k and my PSU is like 3meters away from the Transformer, once the transformer is external to the PSU+SSL 9k Case...
I mean, the cable i run from transformer to PSU is like 3m long...
You can see some pictures here:
http://www.groupdiy.com/index.php?topic=33100.160

Yes, you really should. Remember that the peak currents flowing from the transformer to the rectifier can easily be 10 times the dc output current, and also of very short duration. This means they are a strong source of potential interference especially for low level circuits not to mention the voltage drop, the series inductance and  the possible instability this might cause in the remote regulators.

All these factors can be mitigated by using thick twisted pair screened cables and additional input capacitors near the regulators but the best way to to minimise them all is to have the transformer, caps and regulators all in the same box and just send dc down the cable.

Cheers

Ian
 

pedrocruz

Well-known member
Joined
Nov 13, 2009
Messages
73
ruffrecords said:
Yes, you really should. Remember that the peak currents flowing from the transformer to the rectifier can easily be 10 times the dc output current, and also of very short duration. This means they are a strong source of potential interference especially for low level circuits not to mention the voltage drop, the series inductance and  the possible instability this might cause in the remote regulators.

All these factors can be mitigated by using thick twisted pair screened cables and additional input capacitors near the regulators but the best way to to minimise them all is to have the transformer, caps and regulators all in the same box and just send dc down the cable.

Cheers

Ian

I see, im having some noticeable noise floor on all channels, specially if im using condenser mics with 48v on...
Im going to make a very small cable from trafo to psus just to see if i notice diference in this noise floor... Do you think this will improve that?
Thank you Ian!
All the best,

PS: in this case Trafo is 22v output and PSU output is 18V... 4V drop...
 

ruffrecords

Well-known member
Joined
Nov 10, 2006
Messages
13,931
Location
Norfolk - UK
pedrocruz said:
I see, im having some noticeable noise floor on all channels, specially if im using condenser mics with 48v on...
Im going to make a very small cable from trafo to psus just to see if i notice diference in this noise floor... Do you think this will improve that?

Can't hurt to try. I am surprised you have noise floor problems with condenser mics - they generally have a higher output than other types for the same sound level so tend to be less noisy but maybe the noise is getting into the +48V phantom supply. Then again you presumably have local smoothing of +48V per channel and if you have decently matched feed resistors and an input amp with a decent CMRR a little noise on the +48V should end up below the noise floor of the mic pre.

The other thing you could try is to get a meaty lab bench power supply (perhaps one with a variable output voltage) and set it for 22V and use that instead of the transformer etc. If the noise goes away then you know what is causing it.

Cheers

Ian
 

pedrocruz

Well-known member
Joined
Nov 13, 2009
Messages
73
It seems i was considering -98dB of noise floor, noise... but it seems is quite normal. This was read in RME Fireface digital scale so is it dBFS? no input load and gain at min...

The trafo cable length from 3 meter to 15cm, between trafo and psu, had no diference at all in the noise floor... but in the future ill keep psu near trafo in external case... Thanks for the warning Ian!
Cheers!
Best Regards...
 

ruffrecords

Well-known member
Joined
Nov 10, 2006
Messages
13,931
Location
Norfolk - UK
pedrocruz said:
It seems i was considering -98dB of noise floor, noise... but it seems is quite normal. This was read in RME Fireface digital scale so is it dBFS? no input load and gain at min...

If this is the level on the graph you get from RME and it extends all the way across the spectrum at this level then that will be the noise per root Hz of bandwidth so you would need to add 43dB to get the total noise power level in a 20KHz bandwidth i.e. it is really -55dBFS.

If it was the level shown on a bar graph type display then it probably really is dBFS already.

Cheers

Ian
 

guze

Well-known member
Joined
Mar 10, 2009
Messages
76
So, if I check with a frequency analyzer(voxengo span for example), and the highest peak of noise is around -98db, i would have to add 43db to that value?

I was thinking my mixer had a low noise, but maybe not...
 

ruffrecords

Well-known member
Joined
Nov 10, 2006
Messages
13,931
Location
Norfolk - UK
guze said:
So, if I check with a frequency analyzer(voxengo span for example), and the highest peak of noise is around -98db, i would have to add 43db to that value?

The voxengo has an RMS function in the statistics panel which should do that for you.

Cheers

Ian
 
Top