Avoid Harsh Noises From Regulator Spikes

fubar3

Member
2011-10-25 3:56 am
Toronto
Amps I have tried with 317-337 regulators powered by a 2-wire wall adapter, have unpleasant noises when power is applied. This is usually solved with relays. But what can be done without relay designs.

Main cause of this problem is half-wave rectifiers which power the regulators alternately. I tried a rail-split configuration because it happens to charge the rail caps together. Also, I used a bridge rectifier for improved filtering of 120hz compared to 60hz. The split rails drive the regulators which minimize virtual ground effects.

Another consideration is that 317 and 337 are different beasts, as some say. They are complimentary but not matched for performance like certain transistor pairs.

Finally, the amp IC chips may have quirks when the regulator falls into it's non-deterministic dropout state. I found that each regulator output needs a 470u cap decoupled with a schottky diode.

The included schematic features a transistor switch which trickle-charges the rail caps for 2 seconds before saturating. This further reduces the inrush current without impeding normal operation. But this circuit is an experiment which may not be needed in every case.
 

Attachments

  • bridge1.png
    bridge1.png
    35.5 KB · Views: 154

fubar3

Member
2011-10-25 3:56 am
Toronto
317-337 investigation

Sounds like as good a reason as any to use a cap on the adjust pin. At least that ought to make a difference... (Note: include parallel diode for discharging on power-down.)

An adjust bypass cap did not help my test. However, there needs to be 0.1uf bypass on output pin to prevent oscillation.

I forgot to mention using the BUF634P for this test. I have another amp with O2 ICs and 317LZ-337LZ (TO92). This one has half-wave rectification and a 10ohm resistor before the diodes. There are no harsh noises. So, unless you are doing class-A bias, the LZ regulators are worth a look.