cmoy sijosae rail splitter

[cachupito]

Hi, I built a Cmoy powered by 2x9v. Then I tried to add Sijosae's discrete rail splititer to avoid dc offset forcaused for uneven discharge of the batteries (tangentsoft advice). The problem is that 'm not sure if I did well. Can anyone check attached scheme to confirm that everything is OK?

Thanks

[jcx]

could work - use the same type Q for the diodes (short c-b) for better bias current control

there are several different ideas to keep track of

balancing battery drain

as shown if one battery V drops more than the other, the circuit will drain the other battery to match
the Qs can be very lightly biased so as not to draw much current when the batteries balance - in fact a little "deadband" from underbiasing is desirable as there will be AC signal caused V drop in each battery's internal resistance which you probably don't want to "balance"

active gnd/supply splitter

if you disconnect the battery midpoint then you have a "active gnd" where all of the headphone drive return current flows in the splitter Q - you probably want Class AB bias in this case - and lower emitter R
the batteries aren't "balanced" this way but they see exactly the same current drain - ps anti-reversal diodes would be a good idea though


I'm not really certain any balance circuitry is needed if you use the same manufacturer's batteries from the same production batch - after all there are 6 or more cells inside each "9 V" battery that you can't get at to balance
it would help if you never totally drained them - a low V cutoff circuit would maybe be a better use of componenets

[Mooly]

Unequal supplies (or batteries at different state of discharge) has no effect on DC conditions for this application (the CMOY).
Supplies of +6 and -9 or +20 and -4... the output of the opamp is still zero.

[cachupito]

I appreciate your quick answers. The CMOY works right with both 2x9v batteries charged. I tried then connecting one completely discharged battery and the other fully charged. The circuit went immediately very hot (it even started to smoke). I've not taken any measure of the output for different loads but I'm doing it next weekend. Thanks for your advice

[mutter]

If you look at the circuit as tangent redrew sijosae's original, you'll find that you did it a bit different:



In this case, the battery are in series.

[mutter]

I can't read the language, but sijosae's own post on the matter:

http://www.headphoneamp.co.kr/bbs/zb...esc=asc&no=159

[cachupito]

Yes, I know it's a bit different from sijosaes's design. I did it this way because I'm using an already designed pcb and I tried to modify the paths as little as I could. I'm not very electronics savvy and just wanted to know if this would prevent uneven discharge of the batteries or not.

[nelsonvandal]

High impedance virtual grounds don't sound good. Weird soundstage. High crosstalk. I don't think tangents schematic is correct. The capacitors are supposed to go from rails to virtual ground.

[cachupito]

Hi, I checked the output. Vout was 2-3mV DC when both batteries are charged. Vout is 6 V when there's only one battery (I took out the other one to simulate a uneven discharge). So there's quite high offset DC voltage that could damage the cans. Then I tried this modification:


Vout with both batteries: 2-3mV and I got same result wiht only one battery. No offset at all, so I think that sijosae's rail splitter is not needed with this configuration. What do you think?

[cachupito]

OK. I was wrong. It's clipping after 10 minuts. I've tried mutter scheme and It's working right.