NIGC 3875 Distortion?

[tkelley]

I recenetly built one of BrianGT's standard 3875 kits with great results. It is much more pleasurable than my "Super Store" amplifier. I was so pleased with the results that I decided to build a set of GC's with premium parts to see if it made a difference. Upon completion, I exchanged the BrianGT board for my own and the sound was awful. The amp works, but the midrange distortion is horrible. I believe I copied the BrianGT PCB topography 100%. The signal ground 'SG' and common ground 'CG' both terminate back to a star ground. I checked all of the connections and double checked the component values. Everything looks good. Is is possible to have a faulty 3875 chip. Any help of what to look for is greatly appreciated.

[llmobll]

well distortion can come froma variety of places so let me give you some points...


1. Check your grounding, most of an amps problems come from improper grounding issues.


2. Triple check your soldering points.


3. Clean off the excess flux.(Sounds stupid, but has solved soo many problems for me.)


4. Lastly... make sure your solderd joints are good, sometimes if a circuit isn't behaving correctly i take a iron to all the joints... remelting them and letting them settle again.

good luck

[Greg Erskine]

hi llmobll,

You did connect the LM3875 to a heatsink, otherwise SPiKe would have cut in, or have you mixed up the 220R and 22K resistors?

[jeff mai]

Your grounding scheme is different than BrianGT boards. If there are no obvious errors to fix, I'd start here.

[lazyfly]

Trying to figure out a layout for a board too and am a bit confused now Did a p2p 3875 and had Signal attached between the 220R and 22k resistors. 22k to ground, 220R to pin 7. Your image appears to have the Signal running through the 220R then the 22k is attached to ground.



Could that have anything to do with it because the p2p thing here works fine? Check the schematic in Brians User Guide to confirm.

*edit: Is chassis gnd connected to the chassis? Left that out accidentally and largely lost drums/bass which left mids and highs sounding a bit iffy.

Still don't understand all this so take little if any stock into what I say.
Good luck

[Greg Erskine]

Originally posted by jeff mai
Your grounding scheme is different than BrianGT boards. If there are no obvious errors to fix, I'd start here.

Good point. Power ground has to connect to signal ground somewhere, either on the board (like Brian's) or at the star ground.

[lazyfly]

Bugger, that resistor layout shouldn't matter either way, should it?

[tkelley]

I was connected to quite a large heatsink. I believe the resistors are correct. The thing that was interesting is that the volume level was identical to my previous PCB, just with distortion this time. It also did not have the usual amount of heat production that my previuos 3875 chip had while burning in.

[tkelley]

Originally posted by lazyfly
Did a p2p 3875 and had Signal attached between the 220R and 22k resistors. 22k to ground, 220R to pin 7. Your image appears to have the Signal running through the 220R then the 22k is attached to ground.

This is confusing to me as well. Following the traces on the BrianGT PCB it appears that it is wired as follows. Correct me if I am wrong, but this is the layout that I copied.

[lazyfly]

Okay I marked the image stupidly. My mistake, it's meant to be s-in at the bottom and pin7 at the top... whoops.

I'm assuming your reading it upside down/right way up Yes, your's is wired like that. signal > 220R > 22K - the schematic is signal > 22k > 220R. Thinking about it it shouldn't make a lick of difference (that's a question!) and yeah, your layout is the same as Brians pcb.

I like confusing the issue

Built a ESP P19 late last year using 3876's and that got really bloody hot for a few hours. The 3875 isn't getting hot at all and from reading past threads not all of them go through that excessive heat stage.

Rabbitz built Brian's kit and from memory that didn't get stupidly hot either.

Sorry for messing up that image