K50 lm1875 feedback resistor distortion

[waz664]

I am building the K50 kit from electronics123.com. Seems to follow the datasheet pretty close so I didn't expect any issues. However I get a ton of distortion. It seems to have to do with the feedback resistors on the negative input. If I lower the gain ratio to less than 10 it sounds much better. But that seems to be inconsistent with the datasheet and every lm1875 schematic I've seen. I am using +/-18V supply and I tested that on a lm3886 board I have it was perfect. Everything I've seen seems to keep the gain ratio above 10 and most show 20. I am working off a protoboard and I wonder if I'm just inducing noise from the long resistor leads instead of the PCB. I just wanted to make sure the circuit was right before soldering everything. Has anyone else had a similar experience?

http://www.electronics123.net/amazon/datasheet/k50.pdf

[markiemrboo]

Originally posted by waz664
Everything I've seen seems to keep the gain ratio above 10 and most show 20.

You seem to be aware, but everything you have seen based on these chips has a gain of 10 or greater is because the datasheet specifies that the chips are unstable below a gain of 10. I have had first hand experience with this instability when I mixed up the feedback resistors by accident and ended up with a gain of just over 1. It's not pretty.

Are you absolutely sure you haven't mixed up the gain resistors somehow, so what you thought was a gain of less than 10 is actually more than 10, and your original gain setting was less than 10, when you thought it was ... say... 20?

I doubt very, very much it's got anything to do with the length of the resistor leads.




Do you have an oscilloscope you can check the output with? Oscillation can make things sound very bad apparently, though in my experience it just wrecked a test speaker.

If you don't have a 'scope, then you may be able to tell if it's oscillating by wiring a ~50W light bulb in series with the mains live before the transformer. This should flash bright briefly for a second or two and then fade and remain glowing dimly, or even not at all. If it stays bright, this could indicate oscillation. I would guess in your case, where it is producing sound, it is not likely to be indicating a short. It will also limit current in case of a fault.

[waz664]

Yes I've been going over this for days now. Rebuilt the circuit several times and had those resistors off the board countless times. Checking the color bands and double checking with an ohm meter. Such a simple circuit. I emailed the kit supplier and they suggested I just build the PCB and try it again. Not my favorite solution since I like to know WHY something should or should not work.

I do have a <VERY> old oscope. That's a good idea to just look around the board for noise.

The resistor the runs from the output pin to the + input can be left in place <Don't ever remove this with power on ; )> So with the resistor on the + pin to ground removed, the volume decreases. Doing the opposite causes horrible noise, can't even call it distortion at that point. So that simple test tells me that I am looking at the correct resistors.

With the resistors in place as indicated in the schematic, I can pull down the input to a level where the sound is good, it's just very quiet. By increasing the inuput I drive it to high distortion. Believe me this is the type of distortion that an 80 year old with a bad ear can easily point out. This is not just me being picky.

[CJ900RR]

Have you replaced the LM1875 itself? Just a thought.

[waz664]

That was my first thought. When something doesn't work, I check my wiring, which is usually the prob. Then the components themselves. I was shocked that there was no problem. Besides with the gain set below 10 it sounded pretty good. Just seemed a little weak on bass.

To give you an idea I used (2) 10K resistors for R4 and R5 and it sounded fine, just not a lot of volume. And this seems crazy since it's a gain of about 2.

[kaos]

I'd try reducing R5 to 22K and R4 to 2.2K and see if the problem goes away. Sometimes feedback resistors in the 100K range can cause stray pickup/capacitance issues.

[waz664]

Sounds like a good thing to try. I'll let you know what happens.

[waz664]

Well I decided to just assemble the PCB according to the schematic and use the 180K and 10K feedback resistors. Pulled the components right off the protoboard that I was using. And.... and... and... well it sounds great! My guess is that the protoboard was introducing capacitance that caused the chip to oscillate. At least that is what will let me sleep tonight.

Board sounds great. I really had that wow factor. Can't believe that $25 in parts, transformer, and PCB's can build a 15W/ch amp that sounds this good.

I'd recommend this to anyone looking to experiment with a chipamp and not spend a lot getting started. The transformer can be a large cost in a project like this. partsexpress.com has surplus transformers right now that are 26.2v center tapped with 2A rated output. This gives you 18V rails and 50W of supply power. Seems to be just fine for a little project like this. Oh and they are under $3 each!

Thanks for everyone offering help. It's good to get some advice when you are knocking your head against the wall.