The quest to fix my amp has begun!

[wintermute]

Hi All,

After a successfull mod to my sound card preamp today, I decided to use it as a buffer to allow me to test my trusty old playmaster series 200 amp using RMAA 5.2.

I'm not 100% sure about my methodology, but what I did was:

1. a loopback test on the sound card.
2. a test on the soundcard preamp. (so I have a baseline)
3. ran tests on the amp (with 7.5 ohm resistive loads). I took the input to the preamp off the resistors and had the preamp set to divide by 10 to limit the voltage getting to my sound card. The results were a bit disturbing

results of the tests on sound card and preamp here (including the improvement of changing the 10nf coupling cap on the preamp to 68nF, thanks Sreten).... http://home.swiftdsl.com.au/~tonywww...mpare_MKII.htm All looks fine.

some results of testing the amp here: http://home.swiftdsl.com.au/~tonywww..._whole_amp.htm

Note that the amp is an integrated amp, but I disconected the preamp stage and measured it separately to isolate where problems were coming from, the second set of results is running through the full amp.

The main thing I'm concerned about here is the very poor crosstalk result.

Can anyone make a suggestion where to look for the source of the crosstalk (I'm guessing power supply)?

I already asked some questions before about this amp here:
suggestions for improvement and haven't yet acted on the suggestions (mainly to replace all the electros as they are now 16+ years old).

The other thing I noticed is the pattern to the noise (ignore the 50Hz bump it's in the soundcard preamp), starting at 100Hz and every 50 Hz after that (up to about 4K) there is a spike, I'm assuming that this is comming from the power supply, I know its at a pretty low level, but I'd like to get rid of it if I can.

I previously posted about the PS here:
Power supply problem???

Tony.

[sreten]

I would presume the PA cross talk performance is due to the
channels sharing the resistance of a common ground return path.

The CD input looks like classic capacitive coupling behaviour.

I'm not too sure about your other results. You say the board
follows a star earthing topoloogy, but I suspect an innapropriate
Star earthing point, which should not be on the actual circuit
board. The board should have several seperate ground connections
back to the star point.

sreten.

[wintermute]

Well you pretty much guessed right on the way the star earthing is done. Now I have to decide do I butcher it or not!!!! Any parts in particular I should pay more attention to (apart from separating left and right!) Maybe some really big wire from the circuit board back to the earth point

The preamp section has a 56nF cap in it which I would say explains the slightly earlier rolloff at the low end than the sound card preamp, I'm not so sure about the high end rolloff though. The only cap in the signal path to the main amp section is a 1uF.

One thing that was kinda weird with the crosstalk was that it got worse as I increased the power (best result I got was about 65db) worst was 29db! (I'm assuming due to the common ground resistance) The strange part however was that when I bypassed the preamp, and fed the amp directly off the soundcard, the crosstalk was the same at 31.5db at low and high ampification levels (basically I used volume control in windows and used 1X and 0.1X settings on preamp.

Attached is a scan of the circuit board layout showing the "star" earthing.

Tony.

[sreten]

Not really, I thought you has seperate PA and preamp boards.

Doesn't look like there much you can do regarding grounding.

sreten.

[wintermute]

Nah I just disconnected the preamp section from the poweramp to do testing. What I am thinking though is at a minimum, I could cut the tracks on the Speaker Earth returns and run separate heavy duty wires back to the star point on the chassis.

I'm assuming that would be the major source of current flowing to earth in the amp. I reckon I can do that reasonably safely without stuffing anything up too majorly

Tony.

[sreten]

No !!!!!!!!!!

There should only be one star point.

You can't move the star point off the board without hideous rewiring.

sreten.

[wintermute]

Hmmmmmmm...... The rectifier and filter caps are on the base of the chassis with the ground on the filter caps going to the chassis earth point and the star point on the circuit board goes to the same earth point. (not obvious from previous posts).

Wouldn't cutting the track (so there are no loops) and wiring back to that same chassis point be effectively the same thing as having the current flow through the circuit board and and then back through the wire, to the same point on the chassis? Or am I missing something fundamental here?

Just looking at the circuit board diagram, I think I have just spotted a loop in the main earthing!!! (although both branches end up back at the same point...... probably nothing but may be worth fixing

Tony.

edit: attached the bit I'm talking about circled.

[wintermute]

Hmmmmm, I just rechecked the wiring diagram, and the main filter caps Ground is connected to the pcb ground then a separate wire runs from the pcb ground to the chasis....... Time for some experimenting me thinks......

Tony.

[wintermute]

Ok here is a pic of the PS wiring (very basic PS)

1st I think I'll run the earth from the filter caps to chassis ground instead of the pcb?

2nd assuming point one is valid should I have a separate wire from each cap to the chassis ground rather than the way it is wired now?

3rd what's with the 0.01 uF cap wired accross the mains switch? is it to filter switch contact bounce?

4th should the transformer center tap go to the caps as shown or to the chassis?

hmmm I keep editing....
5th should the mains earth be to a separate point on the chasis to the main ground point?

Tony.

[wintermute]

Ok I simply removed the lead from the caps to the star ground on the pcb, shortened the lead significantly and soldered it direct to the chassis earth lug. Didn't change anything else.

The result? Better noise performance especially over 4Khz! and slightly better distortion performance (but I suspect that this is simply due to the lower noise since the test tone is 1Khz and there were noise peaks at 2Khz, 3Khz, 4Khz etc).

Note that I also shortened the leads I had running to my load resistors from about 3M to about 10cm, I tested before and after doing this and the only change was that the crosstalk improved by about 6db.

Tommorow if I have time (and energy) after Christmas shopping, I'll buy some decent wire and maybe redo the rest of the power supply wiring. First uprating the wire from the star point back to the earth lug. The existing wiring is super flexible 512 strand wire which is basically as far as I know the sort of wire they use in multimeter leads. I'm going to get some 7 x 95/0.12mm strand 8ag OFC and use that instead.

Results of the latest test here: http://home.swiftdsl.com.au/~tonywww...earth_mod1.htm

Note that "amp only" means I bypassed the preamp stage and fed the soundcard straight into the 1uF cap at the input of the power amp.

Tony.