Gainclone HUM, THUMP grounding problems - The solution

http://www.diyaudio.com/forums/showthread.php?s=&threadid=118735

There are numerous resources on the internet to build a Gainclone. I have successfully built 4 of them without any dramas, except on one area: Hum and Thump. These can be minimal but in a very quiet night can be very disturbing. My latest build had hum audible at 10cm which is low enough - but not perfect.

Thump is a more serious issue for me. I use the gainclone to drive speakers actively, and without protection to the tweeter. Thump means bye-bye tweeters.

So I researched the issue for about 3 months and it was quite a character building experience . So many times spent trying, untangling wire, ecetera which cost me lots and lots of valuable listening and family time. It has too, cost me a speaker (a test one thankfully).So now that I finally have a "black amp" so quiet that you could hear your neighbour fart , let's write it down to help beginners and non-engineers like me.

First of all there are excellent resources about grounding. I will list a few which had greatly helped me in this endeavour:

1. Peter Daniel's layout
2. Digi01's post about multi-channel amp grounding
3. Nuuk's (decibel dungeons) grounding
4. RJM Audio binary star grounding
5. LM3875's datasheet, very good ground loop explanation

For each of the contributors above I thank them for their insight about the topic. There is however, a lack of real wiring diagram for a complete, multi-channel, and specially non-PCB Gainclone systems.

So to cut to the chase, here is the final wiring diagram of a 4-channel Gainclone. Have a look and you may get the idea immediately. It is not overly complicated but there are minute criticalities which I will try to explain later.



Critical #1
'Separation' of signal ground and output ground. Notice that in the drawing there is a small ground point for inputs per channel, and one big ground point for everyting else. I use the term 'separation' loosely here as they are not really separated in electrical sense. There is a connection between 680R with the input ground, but it is not the same with connecting that input ground to the big centre ground point! This was my first mistake - thinking that connection points are all the same (thinking they measure 0 resistance on the ohm meter anyway)

Critical #2
Run the ground connection separately using cable to each relevant ground point. Yes there is a temptation to 'tidy up' and just use one cable and "hop" to the final ground point - it just won't work. Now, this may be different if you are using PCB as it may have beed designed well with large ground area etched.

Critical #3
Use different size of wire. Smallest for input ground, medium for output ground, and a really big one for PSU ground. This relates to point 1, and we try to make the signal flow from input ground to PSU ground - not the other way. Electric signal runs like water to the path of least resistance, and we want that path towards the PSU. During my experiment I found that using the same wiring topology, but also same wire throughout caused hum and more annoyingly loud THUMP when the amp is turned off.




In this photo you can see that my PSU is integrated with the first pair of Gainclone (my second pair would be mounted on top of this). The big cable come from PSU's ground.

Critical #4
Isolate input RCA ground from the chassis. Enough said, we want to 'control' the flow of signals to each ground point, and chassis is definitely not it. It took me quite sometimes to find RCA terminal with good isolation, found it at Dick Smith for $4 each.

Critical #5
Move the signal cables AWAY from any mains/transformes cables. I noticed that putting the signal wire really close to the amp's switch and cables produced noticeable hum. Good practice is to put potentiometers at the back, as close to the input as possible and use some kind of extention like Peter Daniel's chassis. But as you can see above I don't have the luxury of space by cramming 4 amps into such small chassis. I found out CAT5 cables for signal wiring is both neat and easy to tuck away from the mains/transformer.


Not so critical in terms of hum is that 5ohms resistor to AC ground. Some will question the wisdom of doing this in terms of safety, but it does help. I will leave this to you with disclaimer that yes, it is much safer to directly connect AC ground to the chassis.

There you go, hopefully my experience will help many, and if you're an experienced engineer please let us know if there are any inaccuaracy about the concept of this diagram. Thanks.