6 channel LME49830 power amp

[pacificblue]

Maybe the test wires happen to introduce just the right amount of capacitance to work like snubbers.

Did you try to analyze the Demonstration Board Layout to understand, why certain things were done differently compared to your layout? The two most striking points are, the Demo Board has no planes and the Pout and Nout traces are as far away from the speaker traces as possible and dont run in parallel to them anywhere.

[cuibono]

You know, I really appreciate the people here. Sometimes I think 'oh, they are just spouting off some idea', but when the rubber hits the road, people really know what they are doing. Very helpful.

I cut the Nout and Pout traces near the LME chip, then tacked the resistors to the legs of the FETs, and then tacked on a couple of wires to connect them to the LME chip. Viola, problem solved (and it looks like the noise level went down too!) Nice, thanks Pacificblue!

[cuibono]

Okay, here is a question I'd like to throw out - assuming two pairs of FETs on the output is safe into 8ohms, if I double the number of output devices to four pairs, I should be safe into 4ohms, correct?

The LME49830 should have at least 56mA available to drive the output devices. How do I figure out if I need to add a driver stage in case the LME can't supply enough current?

[AndrewT]

what is the maximum current the FET will demand at the highest frequency and at the highest slew rate?
Can there be a circumstance where a higher current could be demanded?

[cuibono]

Andrew, I'm afraid I'm a total noob when it comes to this stuff. A fish out of water - but learning, and appreciate you're help.

I have no clue what resistance the LME chip is working into, nor how to factor for higher frequencies/slew rates. Beyond that, I don't know how to design the driver stage. Looking at others, there is a transistor per rail, but the resistors, their use.....

[cuibono]

Well, good news.

I finally got my hands on a square wave generator, and checked the amps out. A little ringing, but at least no oscillations!

Here is 10kHz, scope is 5V/div.

[cuibono]

Here is 2kHz.

[cuibono]

Here is 100Hz. Maybe I should check the LF roll off.

I'm pretty happy, I though this beast wasn't ready for square waves yet. I'll start final assembly, and give it a proper listen soon!

And what do I do about the ringing? Could someone point me in the right direction?

[cuibono]

Well, the good news is I have the whole thing assembled and tested, and it all looks good.

The not so good news is I can't seem to squash the ringing in the square wave. I tried a bunch of things, but after reading up a bit, I thought for sure the problem was the gate stopper resistors. AN-1645 is a must read - page 10 has a formula for calculating it, and it seems to suggest values in the 200-400 ohm range, which seems to be common, although one can go lower.

The amp as it is uses 22 and 10 ohm resistors which give f-3 values of 5MHz and 11MHz for the irfp240 and irfp9240 (if I remember right), which seems way high. The input capacitance for each device is 1275pF and 1400pF respectively, and values of 200ohm gives a f-3 of about 600kHz, which seems plenty high to me.

So I tried values in the 200 and 400 ranges - and the thing oscillates like crazy! The rails can't get to +/-30V before the thing starts swinging a large negative voltage on the output and oscillates. What gives? Larger resistors should have made the thing MORE stable, not less!

Any ideas?

[Yuli_35]

and a cap ???? in parallel ... 47pf ... or less depending on your resistor.