Importance of layout

[Pars]

I've been messing around with a Counterpoint preamp I have owned for several years on and off. It's always had a slight amount of hum, not real bad, but it bothered me. Over the years I've managed to make it worse, and then much better, but never quite gone completely. After dealings with a couple of former service techs on diyaudio, I finally decided to try once again to rectify this.

I was prodded to look at the PSU, particularly the heater supply (I'd already done some work and parts replacement in the HV section). Looking at the heaters, I had recalled seeing some ripple on them back when I rewired the umbilical (both before and after).



About 50mV of ripple at 2x line frequency (120Hz). 50mV/division (x10 probe).

After being pointed in the direction of board layout, I noticed the following (once it sunk in):

• C is the lead to the input of the LM350K regulator. It is wired right next to the input (+) from the bridge rectifier, putting raw DC in. Moved to past the filter cap (C75).
• 5 is the output (-) to the tube heaters in the amp chassis. It is also wired prior to the cap (C75)

One more change not shown in the pic: R72 (green rectangle shows location) needed to have the bottom end rotated 90º counterclockwise so it returned beyond C75. After making these changes, the net result:



Same 50mV/division, timebase is extended out a bit in this pic. Designer seems to have a problem common to many: PSU design. And this is a high-end preamp from the '80s, lauded by Arthur Salvatore, yada yada.

Such simple errors, but it surprised me as the changes involved moving components/wires along signal paths that are unbroken and electrically connected. The previous connections weren't wrong, they just weren't really right. Noise was being fed into the regulator because its feed was taken before the filter cap, not after it. The output (pin 5) was also being taken before the cap, not after. Same with the adjust pin return resistor (R72). One other thing not done originally: per the datasheet, the addition of a 5 cent capacitor across the adjust resistor buys you 15dB better noise rejection. Implemented now.

I haven't listened to it yet, but anticipate that it should be gone

Edited January 3 by Pars
refreshed pictures

[DouglasQuaid]

You mean to tell us that autorouter isn't infallible? BLASPHEMY!

Happy to hear you solved your ripple problems. I like this little lesson. It reminds me of people who think ground is ground, until they finally deal with some nasty mixed-signal devices really sensitive to noise and offsets.

[luvdunhill]

Never run heater traces on the PCB. Period. End of subject. Pass the scotch.

Edited February 17, 2011 by luvdunhill

[Pars]

Sorry, fresh out of scotch

Probably good advice. Point to point isn't likely to happen in this given the

• form factor (think '80s Threshold preamps with their brick PSUs; then think of stuffing a full tube pre and tube rectified PSU into that form factor) and
• resale value,

so I pretty much have to deal with what I have.

Corrected (for the .1 version) schematic

Edited January 3 by Pars
refresh picture

[bhjazz]

Jebus. Nice work, Pars!

And BTW, that PS board is quite the trainwreck.

[luvdunhill]

No not p2p. I did a balanced Pre rexntly. Just do the heaters with pads right next to the so sockets. Just keel the traces odd the pcb and use twists wire there.

[kevin gilmore]

Singlepower ran dc filament current all over the boards.

Trouble was the significant AC content due to the diodes

got into just about everything. The diy T2 does absolutely

the minimum amount of filament on the board for this reason.

[Tyll Hertsens]

Physics (and good sense) FTW!

Nice shootin'.

[morphsci]

Well done sir, well done.

[Voltron]

What he^ said!

[Pars]

Still trying to decide whether to swap out the PCB with one I did a couple of years ago. Looking it over again, it has a couple of minor errors on it, but I think I can work around them. I didn't check the stock tube layouts in Eagle, and this turned out a bit small on the pad circle. This socket will fit, but I'd rather use the Azuma socket that is in there now, and it doesn't fit. I also don't think I will put the two 500V Cerafines like the one pictured in, but will use the Panasonics pictured on the board. The old board is coated with Q-Dope... not dirt

http://i883.photobucket.com/albums/ac31/Pars1746/Counterpoint%20SA5/new_PSU_PCB.jpg

http://i883.photobucket.com/albums/ac31/Pars1746/Counterpoint%20SA5/PSU_PCBs.jpg

Edited January 3 by Pars
refreshed pictures

[Pars]

Tested it out this morning. A slight amount of tube hiss (barely audible with head < 12" from speaker), but no hum. Gone at all gain settings. And this is with the tubes cold, so it should decrease after a bit of warm up.

I need to publicly thank Alan (VivaVee on diyaudio) for his help in finding this. He is the same person who helped me fix my brother's power amp as well. Good guy.

Edited February 19, 2011 by Pars