Strange distortion problem

[Adam M.]

I am running an LM3886 per national's spec sheet with +-37v rails, 10,000uf cap per rail.

I have a slightly odd problem... anytime I get an output signal larger than 2v, I start hearing static through the speaker. Its not terribly smooth either, about 1.8 or 1.9v I can hear distortion, and then at 2v comes static.

I have the power coming into the chip via alligator clips from the transformer for now, and the ground leads of the capacitor also connect to the 0v point on the transformer, which I use as power star ground.

I have a short wire soldered to the ground pin on the chip, the mute pin is soldered to the bare end of the wire, and alligator clips from the -input and power ground both connect here as tightly as possible. I have tried an input through a pot and direct, my source is a wavetek signal generator.

DC offset is .045v

I don't have a scope to look at the output unfortanately...

Any ideas on what would cause this?

THanks,
Adam

 

[carlmart]

I have a short wire soldered to the ground pin on the chip, the mute pin is soldered to the bare end of the wire, and alligator clips from the -input and power ground both connect here as tightly as possible.

There should be a resistor, and preferably a bypass cap, between mute pin and -V.


Carlos

 

[Adam M.]

I have a 47k resistor between mute and ground, forgot to mention it. no bypass cap though. Would that cause the problem I am seeing, if so what value would you suggest?

 

[jan.didden]

The mute resistor must go to V-, it is there to bias the input stage. See the chip's internal diagram in the data sheet. Not having the input stage bias would explain the problem.

Jan Didden

 

[Adam M.]

I mis-spoke in the earlier post, I have mute going to v-, through the 47k resistor...to be specific I have the mute resistor soldered to the -Vin pin of the chip. The resistor I have to ground comes from the inverting input. (I made the mute resistor to ground mistake earlier and that caused 2 days of frustration trying to figure out it wouldn't work) THanks for the response though, this resource has been very valuable.

 

[jan.didden]

Adam,

The best you can do at this point is to sketch a diagram from your construction (don't look at the data sheet!) scan it and post it. We've all been down that road. Chances are you find the problem even before you finish the sketch.

Jan Didden

 

[Adam M.]

Attached is a .gif of the circuit drawn directly from the chip itself.

 

[jan.didden]

Hmm. looks OK. Not having a cap in the feedback loop explains the offset, but it's not too bad. Are you sure the supply voltages at the chip are OK?

jan Didden

 

[hughmon]

One possibility that comes to mind is an amplitude dependent oscillation. Basically as the output signal goes up, at some amplitude it reaches a sufficient magnitude to feedback to the input and osc. occurs. This tends to be a layout kind of thing, but can be bandaided with a capacitor to ground at the non-inverting input (if interested, try a 470pF cap to start with). Sometimes a 100 ohm resistor is needed between the actual input pin of the chip and the node of the cap and input resistor. Don't know if this will help or not, but it's what I'd try at this point.

 

[Adam M.]

Jan, I have tested the voltages on the power supply with reference to 0V at the middle of the transformer secondaries, and I get +-37.2V within a small margin on each rail, I checked this with the chip off before I hooked it up to make sure it was ok, and again with the chip in place. I have not checked the current it delivers.

hughmon, I'll give the cap from non-inverting input to ground a try and see if that helps.

Thank you for your time. If after some more troubleshooting I can't get it to work, I'll grab a digital camera and post a picture of the layout.

 

[tlmadsen]

Adam

Maybe it is some kind of osc. you are hearing. Try a 220 pF between pin 9 and 10.

Thomas

 

[analog_sa]

You should have a resistor between + and GND. And you may also need a Zoebel.

 

[Adam M.]

I tried the 470pF between +in and ground, 235 and 470pf between pin 9 and 10 (which reduced the DC offset to 3mv) and a 10k and 47k resistor between +in and ground, still having the same problem. The voltage the distortion comes on doesn't seem to be frequency dependent, I tried at 100hz, 500hz, 1khz, and 5khz, and it always shows up at the same place within a small margin of error. Changed the speaker to see if that was it, no luck.

Unfortanately, I didn't the components that seem common for a zobel laying around, so I couldn't try that. I am going to try to borrow a digital camera and get a picture of the circuit, as this is my first effort I am sure there are multiple layout problems.

Thanks again for the help, I am learning quite a bit about electronics troubleshooting, RF, ground loops....

 

[Nuuk]

Adam, may I suggest a couple of things?

Dispense with the crocodile clips and solder things up.

Make your power ground star the junction of the the two caps on the voltage pins of the 3886. Have a signal ground star that is as close as possible to the chip and join the two with a short length of wire.

450mV was a very high DC offset. The 470p cap seems to have sorted that though I am not sure why it was that high before.

 

[jan.didden]

From the description it seems that the cap solved an oscillation problem. Nuuks suggestion is a good one.

Jan Didden

 

[Adam M.]

With the cap across pin 9,10 I have minimal dc offset,(still around 3mv) and with the alligator clips out of the picture, the distortion is gone. Thanks everyone for the time you put into this thread... Time to get soldering!