MY LM3886 experience


2002-04-07 5:05 am
I just completed an amp using the LM3886 power op-amp.

I started out by making at pcb that would use 4 in parallel. I had .15ohm current sharing resistors along with a DC servo (I used an opa27 in mine). This config had some serious problems. Whenever more than one amp was enabled, the chips would draw LOTs of current (>2.5A) with a small input(<50mV). It appears the amps are not happy with each other. I think I am going to use 2 ohm current sharing resistors in my next attempt. The amps also liked to oscillate at about 1mHz when in parallel mode. Decoupling caps were right next to all power pins (I used 1206 SMT ceramic.) there was 10pF between inputs along with 10k+100pF in feedback loop (in parallel with the gain setting resistor). I tried other values for feedback caps and compensation caps but it still wanted to oscillate. I believe it was the interaction between the chips.

After than I wired 2 chips for a bridge mode. This config was excellent! SOunds excellent as long as you keep the chips cool. I have a layout where I use Pentuim II heatsinks and fans on each chip. I also reformed the leads so when chips where mounted with heatsink, the heatsink was parallel with the PCB. With fans running quiet, I could easily run the music very loud. I was looking at power peaks of 175Watts and the amps remained stable and clean. The temp near the chip got to about 50C. The noise of the fans was insignificant compared to how loud the amp was.

I did run the amp into a dummy load with a 20Hz sine wave continous at 100w into 8 ohm. with PII heatsink and fans, the temperature at the chip was 71C.

Overall I am impressed with the performance I got from the chips. 2 chips in bridge mode is very nice and is great for a general purpose amp.

I am going to investigate the parallel mode some more And get that to work right.

Thanks... great info ..

I'm about to make a 4 channel amp with 4 lm3886's.. these will be used in a Bi-amp system for the bass drivers...

I was hoping you could give me some info on the power supply you use ( cap's, rimple, hum at ouput, etc) ... do you use a dedicated brigde driver or are the lm3886 permanent brigde mode ? or is that possible..

well... I hope to hear from you soon...



2002-04-07 5:05 am
I tested it with +/-30v rails on a regulated benchtop supply (it was a linear supply not a switching supply). I could have probably squeezed more out of the amp if I would have high voltage rails. I also placed 50000uF on each rail. Its a bit overkill for what I was doing.

I have 700VA toriods that will allow me to have about+-37v rails. I have going to wire a couple units up with the 50000uF caps. I let you know how that turns out.

Overall, these chips are ncie to work with, you wont be dissapointed.
A couple months ago, on a related thread, I described building a parallel-bridged set up based on the National app note, but using both bridged halves in non-inverting mode. (The app note uses one half non-inverted, the other half, inverted.) This meant supplying a differential signal to a pair of non-inverted input opamps, which in turn, drive the LM3886 devices.

I paralleled two LM3886 for the + (non-inverted) signal and another two for the - (inverted) signal on one PCB. I planned on then paralleling 3 of these boards per channel. I built-up one board that seemed to test well and then built-up 5 more boards.

Now that I'm putting it together, it doesn't seem to work: it oscillates like crazy when bridged. When operating s.e., from either bridged half to ground, it works OK. So this is just the opposite of what you (IIR) found! The paralleled devices (2 of them), work OK in this set up and are not oscillating.

I've tried all sorts of remedies, such as cap-resistor filters to ground, caps in parallel with the feed-back resistor, and while the cap-resistor to ground (0.1 microfarad in series with 10 ohm resistor) does stop the oscillation, the wave forms at > 5 kHz have jiggles in the sine waves and there is a crossover distortion evident at all audio frequencies when these filtering attempts are used. (This is not affected by a load on the amp either.)

Any ideas?


2002-04-07 5:05 am
A couple of Notes:

When I tried to Parallel the devices they would draw lots of current from the supplies with a tiny bit of input. When I put a small sine wave in I did get some output. The output wave form had LOTS of cross-over distortion and severe positive offset.

In my setup I used all surface mount caps and resistors. .1uF cermaincs were RIGHT next to the supply leads. I also had a 50pF ceramic in parrallel with my feedback resistor (inverting setup). My gain was set with .1% resistors and was set to 27

I am going to try the paralell setup with 1 ohm current sharing resistors sometime soon.
I also tried varying the current sensor resistors, from 0.1 to 0.51 ohms per device. The circuit behaved slightly better with 0.51 ohm resistors, but still had the distortions/oscillations described above.

My signal was perhaps 0.2 or 0.3 V sine wave. The National app note used 0.1 with servos for 4 devices; this is exactly what I have; they showed really clean output. Curious, especially since it works non-bridged.