White Audio Labs B3 amplifier rebuild (Schematic included)

Joined 2013
Paid Member
Running into a problem - maybe you can assist. I have replaced the LTP on both boards, matched using a Chinese-made tester (average of 4-5 tests) and installed the closest matches, but the offset remained at 90mV and never goes below 75mV on either channel.
Then I disconnected each channel and checked the current draw on +/- of each channel. the Left channel worked fine, set the current at 800mA (800mV across a 1r 10w resistor on each rail) Then I get to the right channel and the positive rail is drawing about 300mA of additional current and fluctuates up and down by ~100mV (measuring 1.0V on the positive and only 750mV on the Negative rail)

So it's been about two weeks and today I tore apart the Right channel and replaced the remaining ZTX657/757s on the input board, checked all the outputs again (they all check good), reinstalled and the negative bias' fine to 800mA, but the right channel is 1.1-1.2mA and fluctuates ~100mA.
While I was replacing the transistors I pulled the 25k pot and checked it, seemed to work fine using a DMM to measure +/- adjustments.
Right channel still has ~55mV of offset even after all the changes.
Any thoughts why I would have larger draw on the positive rail, only on the right channel? I measured at the outputs and they all have almost identical voltage on the pins of the Lateral mosfets. This B2 version (4 pair outputs/channel) does not have any source resistors, but all the outputs are heating evenly and with a thermometer they are all show the same temp. as well as the super-scientific "finger" test.
If you're using the single ended inputs then I'd say the offset you're seeing is normal. I had the same issue. I ultra matched a set of LTP transistors. Normally when I do this I'll get low single digit offset measurements (like 1 - 2mV) in any amp LTP I put them into, but not the White amp. However when I connected the amp up using the XLR inputs the offset would drop to < 10mV but using the RCA inputs it was quite high. This was the case both on my Audio Precision system and with my Adcom GFP-750 preamp.

As far as your out of balance current draws. I'd say that if you have one channel with a high imbalance between the positive and negative rails then IMO it would imply a bad output(s) mosfet in that channel. If the offset is reasonably low in that channel, then you have to ask where is that additional current going because if it were going to the load then you should have a high DC offset? One thing that comes to mind, where are you measuring the current draw for the channels, is it between the respective channel rectifiers and the main caps or between the main caps and the channel module itself. If the former then the extra current draw could be due to a bad main PS cap, if the latter then it implies a failure in the channel module sinking extra current to ground and not the load. You may want to recheck the insulators on the output mosfets in the affected channel.

Maybe someone else will chime in with their opinion, but this seems like the most likely possibility to me. You could temporaily swap the predriver boards from channel to channel if you think the issue might be there, but I do not see this being the case.
Joined 2013
Paid Member
Thanks for the insights... so I tore it back apart and added (16) 0.47r 2w source resistors on both channels (prefer 0.22r, but didn't have any on hand). Reset the bias on both channels at 800mA per channel (used a 1r 10W resistor between bridge rectifier and supply capacitors. Still has about 0.250mV turn on thump and about -1.20mV turn off thump but otherwise appears working/heating as it should.

Current state (using RCA inputs):
Right channel is now down to 15-20mV (this channel has all new matched to-92s on the input board)
Left channel remains at 75-85mV (this channel has new matched TO-92s for Q1, Q2 and Q3) - so changing them and adding source resistors did nothing.

Listening for about 3-4 hours and sounds good, with a slight concern about the 80mV on 6.4r speaker (~12.5mA, or ~10mA more than Right channel)
Last edited:
Some observations/hints:

Resistors added at the source pins will address current sharing among the output FETs but, since they are inside the feedback loop, there's unlikely to be any detectable change in output offset behavior.

This amp doesn't have any input blocking cap, so be sure to measure DC voltage from signal source at the amp input, from Q1 base to GND1. In principle, any DC present at input will add with unity gain to the output. A blocking cap might be a good idea.

Remember output offset arises from Vbe mismatch, and bias current drops across bias resistors. Measure input bias currents by measuring drops across R5 and R12. Ideally, they will be equal. Swapping Q1 and Q2 and noting change can be insightful.

Be alert to possible leakage current in C1; just 1uA of leakage will add 10mV to output offset. [If C1 is suspicious, you can sub a low leakage 1uF film cap as diagnostic test. Low cutoff frequency will be inadequate, but the cap will keep the amp stable during bias check.]
Joined 2013
Paid Member
@BSST still digesting a lot of what's in your post and suggestions - but appreciate you taking time to add ideas and an explanation of how the parts in the circuit work - still learning as I go.

1) Measured the RCAs from the Pre powered on, no music playing and I get 0mV of DC on right and left cables
2) Measured across R5 input connected to Pre, no music playing (47mV on Left, 46mV on Right)
3) Measured across R12 input connected to Pre, no music playing (120mV on Left, 115.5mV on Right)
4) C1 is brand new Nichicon 35V 220uF - UKW (black foil/gold writing and not specifically a low leakage Cap like UKL)
Maybe that's my issue, because the datasheet does specify it has 77 to 231uA of leakage current, where the UKL specs at a maximum of 15.4uA?

Another finding, is that the right channel starts out at 15mV offset, but once the amp is fully warmed up (1-2 hours of music) it is back at ~60mV, where the left channel stays at 75-80mV the entire time (cold to warm)

I'm not looking for 0mV offset - but close to 25mV would make me feel better using the amp with my different speakers.

edited - added part about RCA DC from Pre
Try measuring the DCV drop across R11. I think there could be on the order of 0.7mV present, enough to explain much of the offset error.

I should have ask you to report some voltage readings as signed numbers re ground. So I believe Q1 base voltages are about -46mV and I think output offset is +75mV. And would you also report Q2 base voltage re GND1?

Last edited: