Thoughts on parallel LM3886s

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
I was looking at the National PA100 circuit with the infamous 0.1% resistors and was more concerned about problems at low and high frequencies. The attachment is a simulation with 10% error on one of the LF ecaps (ecaps are actually much worse than this).

As the sensitivity is about 1V rms, a maximum input at 20Hz will cause a differential voltage between the two outputs of 1.4V. NS suggest 0R1 output resistors so we end up with 7A flowing between the devices - the protection will cut in.

If we combine the two ecaps, this is cured but we now have a worst case differential dc offset on the output of 420mV instead of 20mV, so an offset null is needed.

At high frequencies the gain/bandwidth spread of the amps will become important. Matching LM3386s for bandwidth would be hard so I suggest putting a 0.7uH inductor in series with each of the amps so that the the rising reactance between the two output limits the error current.
 

Attachments

  • cap_error.pdf
    5.2 KB · Views: 93
parallel lm3886's

Say what??? your input voltage differential will be the V1+V2-D not the composite of both. Your Ce cap if placed before your gnd resistor will act as a dead short at frequency, therefor only effecting youre distortion at or around youre cut off frequency usually far below audible 20Hz if proper cap is used.
When the parallel chips work together they display an interesting effect, vanishing distortion, you can see this in some peoples THD test on successfull BPA units, the composite distortion is less then each chip indivdually. Electrons act kind of like water in to glass, if the two glasses are connected via a tube with a valve in the middle at the bottom of the glass one glass is filled 100% one 50% open the valve they settle at 75%. The Chips act the same way.
 
This is a bass problem. Even with larger values of ecap, the very poor tolerance can push this effect up to 100Hz.
Most domestic loudspeakers are at their lowest impedance at LF so they suck up the current. The time consant of the SPIKE protection makes the limiting cut in earlier with bass. Music can have high levels of low frequency - Pink Floyd, Leftfield, Bach organ etc.

Parlelled amplifiers should not cancel distortion. Bridged amps can theoretically cancel even harmonic.
Distortion should be reduced because each amplifier sees twice the load impedance.
What may also be happening is the current flowing between the two outputs is putting a class A bias onto the outputs
 
parallel lm3886's

I have never had a problem and I have built many BPA units, the one I am running now is a BPA 400 I have perfectly flat curve down to 4Hz and up to my set roll off of 35Khz.
None of my amps ever had current problems all sip current while idle and have a total DC offset of less then 20mv. My distortion curve is the highest at 20Hz ,but is still below .5% at 448 watts at 8ohms.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.