Go Back   Home > Forums > Amplifiers > Solid State

Solid State Talk all about solid state amplification.

Please consider donating to help us continue to serve you.

Ads on/off / Custom Title / More PMs / More album space / Advanced printing & mass image saving
Reply
 
Thread Tools Search this Thread
Old 14th December 2001, 11:52 PM   #1
diyAudio Member
 
Join Date: Dec 2001
Location: UK
Question Pick-up of charging spikes

I have built a 10-channel power amp using LM3886's on Rod Elliott's PCB. I have a big shared power supply (1kVA transformer, 112000uF, etc).

I am getting between 13 and 50mV pk-pk of sharp spikes on the output (sounds like mains related buzz), co-timed with the charging spikes on the supply rails (about 350mV pk-pk sawtooth on 32V rails). Channels where the noise is below 20mV pk-pk are almost acceptable, but the one channel that is up at 50mV is a real problem.

Does anyone know where this pick-up is likely to be happening, and what I might be able to do to lessen its effect? Power supply rejection on the LM3886 is supposed to be 85 to 100dB - I am certainly not achieving this. Any clues as to why there could be such variability between channels? (doesn't correlate with position in chassis relative to power supply)

Many thanks

Jamie
  Reply With Quote
Old 15th December 2001, 12:48 AM   #2
diyAudio Member
 
Join Date: Feb 2001
Location: Columbia, SC
Did you say 10 channels?
Did I miss something?
You're saying it's not power supply...at a guess, I'd look for a grounding/shielding problem. Be alert for cold solder joints.

Grey

EDIT:
Oh, yeah. Got a 'scope? Go hunting for oscillations. Chips are notorious for oscillating.
  Reply With Quote
Old 15th December 2001, 01:35 AM   #3
lohk is offline lohk  Europe
diyAudio Member
 
lohk's Avatar
 
Join Date: Apr 2001
Location: Northern Noricum, near the Limes
All the channels seem to be oscillating somehow.
Did you bypass the supply lines with caps close to each OPamp properly ?
Did you strictly obey to the "star ground law" ?
Did you keep a very tidy layout of all the wires ?

The difference between the channels relate probably more to the individual build quality of each amp. Power OPs are very sensitive for pcb layout (should not be a problem here), parts quality, not to forget the quality of the solder and the joints.

I have much better experiences with multi-channel amps using OPs by using partly separate PSUs (Shared transformer, one bridge and a pair of caps for each channel).

Try this, maybe it helps finding the faults:
Disconnect all channels but two and listen/measure again.
Put out one board and use it with a different PSU.

Klaus
  Reply With Quote
Old 15th December 2001, 09:54 AM   #4
diyAudio Member
 
Join Date: Dec 2001
Location: UK
Thanks for the rapid responses.

Yes, 10 channels. Wording may have been confusing - I'm using 5 of Rod's stereo PCB's, each sawn in half.

I'm using an analog 'scope which is how I found what I described above - digital 'scope wouldn't find it.

The circuit is slightly adapted from Rod's Project 19 suggestion to include more stabilization - I know that this can be counter-productive sometimes.

Positioning of decoupling on these boards isn't great, but I'm sure Rod would have changed it if it was a problem. For the 0.1u decouplers, I am using mylars because I had a pile of them lying around. I know people suggest ceramics - but I didn't think this would cause a problem. Also 220u electrolytics per rail on each module.

Wiring in a 10-channel amp is very difficult to keep tidy. The challenge of maintaining a low impedance to the power supply means heavy duty wiring which tends to go where it wants. The 10 modules are equally spaced around the power supply in a U, with all mains on the opposite side. Left on one side, right on other.

Grounding - I've put weeks of work into checking this. I now have ground-compensated outputs from my pre-amp eliminating inevitable loops in such a large system - didn't make a huge difference, maybe indicating my grounding was pretty good before. Implemented star-grounding as far as is possible with these PCB's. Shifted loudspeaker returns from star-point to amp modules just as an experiment - no difference. Connection from star to chassis/safety ground at one point via 100n/100R parallel network - no other 0V-safety ground connections in system.

I seem to have quite a bit of RF noise - maybe not surprising with the lid off, or maybe this is oscillation. Varies alot according to 'scope lead position.

Interested my problem might be oscillation - what are the symptoms of this? I'll do some experimenting today.

Again, many thanks

Jamie
  Reply With Quote
Old 15th December 2001, 01:29 PM   #5
diyAudio Member
 
Join Date: Dec 2001
Location: UK
... tried a 100n cap in parallel with each power supply rectifier diode to slow the peaks (Nelson Pass suggestion), but this has had no audible or measurable effect.

Also tried the worst module on its own. Noise is less bad, but still very noticeable.

Doug Self's site holds a few clues. Either provide separate decoupling returns to star (no longer much point in using Rod's boards if I do this), or try to minimize loops carrying half-rectified components to reduce induced pickup (these are probably extensive in my box - very difficult to resolve).

However, I can't help thinking these are palliatives, and my problem is pretty gross. Mind you, there's not much specific material out there on multi-channel amp design.

Jamie
  Reply With Quote
Old 15th December 2001, 02:40 PM   #6
diyAudio Member
 
Join Date: Feb 2001
Location: Columbia, SC
I confess that I'm not clear on how all this is laid out, but you might try running the power for each channel through a low value resistor (22-47 ohms) into another cap (say, 100uF), thus:
main power supply->resistor->cap->individual channel
As a test of this idea, unhook nine of the ten channels from the power supply, and see if the one remaining channel runs cleanly. If it does, try the above.

Grey
  Reply With Quote
Old 17th December 2001, 07:59 PM   #7
diyAudio Member
 
Join Date: Dec 2001
Location: UK
Conventional linear power supply (www.linkwitzlab.com/xo_eq.htm - 3886pwr.gif). V+, V- and 0V from each module to the power supply (tee'd off smoothing caps).

Amp circuit is more like 3886amp.gif (see above) than Rod's Project 19 (no cap in the signal path, fitted stability components that Rod left out). Though I have included Rod's local 220u decouplers.

Click the image to open in full size.

My pre-amp outputs are ground compensated (as per Doug Self's circuit) to avoid loop problems with so many channels. Shield connects to cold wire at amp module (soldered to PCB - no connectors) and this is subtracted from the signal at the pre-amp to remove any ground voltage difference. Moving to this from unbalanced made the problem better, but didn't fix it.

With only one power amp channel connected, if I terminate the input and use an external (bench) supply, the output is clean.

As soon as I connect my pre-amp (not powered), low level spikes appear (few mV). When I power up the pre-amp, they get much worse (~+12dB).

As I hook up other power amp channels, the spike amplitude increases, and I get more of them.

Taking Rsn / Csn out made no difference. Providing a dedicated return path for decouplers made no difference.

As I can still measure spikes with only one channel connected, I am not convinced that providing separate supplies for each channel will help.

Any clues as to how I determine if this is oscillation, pick-up or something else?
  Reply With Quote
Old 17th December 2001, 08:20 PM   #8
diyAudio Member
 
Join Date: Feb 2001
Location: Columbia, SC
I find it suspicious that the amp begins to misbehave when the preamp is hooked up. Any chance of trying the amp (either one channel or more) with a different preamp? Incompatibilities between preamps and amps aren't common, but they do happen.

Grey
  Reply With Quote
Old 18th December 2001, 06:44 PM   #9
The one and only
 
Nelson Pass's Avatar
 
Join Date: Mar 2001
There are several possibilities here:

Grounding

Oscillation

Transformer field

I think the most probable is the transformer field
inducing noise into ground loops.

Take the transformer out of the chassis and hook
it up a couple feet away and see what happens.
  Reply With Quote
Old 21st December 2001, 09:20 AM   #10
diyAudio Member
 
Join Date: Dec 2001
Location: UK
Nelson

Many thanks for the suggestions.

Attached is a drawing of the AC component of my +ve supply rail (-ve is mirror of this) loaded only by a 3k resistor (no amp modules). The peaks are slightly rounded.

This component grows, as expected, with load. I am worried about the spikes, or the rapid initial decay after the peak, as it is these spikes that seem to be getting though to my outputs. I would have expected to see a more regular sawtooth on the supply rails.

I have got 100n caps across each diode in the bridge.

On your point about the transformer, I didn't suspect this because I assumed that the spikes would be introduced by the bridge or the smoothing caps.

Is a waveform like this normal, or does it indicate a problem?

Many thanks

Jamie
Attached Images
File Type: jpg supply rail.jpg (1.9 KB, 310 views)
  Reply With Quote

Reply


Hide this!Advertise here!
Thread Tools Search this Thread
Search this Thread:

Advanced Search

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are Off
Pingbacks are Off
Refbacks are Off


Similar Threads
Thread Thread Starter Forum Replies Last Post
charging capacitor joeyd71 Car Audio 4 4th April 2009 08:24 AM
Help with charging Li-ion cell EWorkshop1708 Power Supplies 14 29th May 2008 03:59 PM
charging 3x 12V batteries in series. RichJones Solid State 7 24th September 2003 10:10 AM
Self Charging Capacitors?! Rarkov Parts 4 6th August 2003 07:06 AM


New To Site? Need Help?

All times are GMT. The time now is 12:27 AM.


vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2014 DragonByte Technologies Ltd.
Copyright 1999-2014 diyAudio

Content Relevant URLs by vBSEO 3.3.2