BA-3 Amplifier illustrated build guide

Jump to Latest
#981
to SoaDMTGguy

Try to keep your DC-offset below 50mV (=0,05 V). Below 10mV is good.
Below 1mV is excellent.

DC-offset in the BA-3 frontend always fluctuates a little bit - that's normal.
If you blow over the JFets - the DC-offset will change a bit. Very sensitive
to temperature-changes.
Cheers
Dirk 😉
p.s.: P3 should be at mid position
adjust P1 and P2 slowly - especially at higher bias levels
bias of Q3 is measured over resistor R10 - this means from one 'leg' of R10

to the other 'leg'. Set DMM to mV.
measured voltage (over R10) divided by the value of the resistor R10
(22 Ohm) is the current / bias running through Q3.
U:R = I (measured in: Volts divided by Ohm = Amperes)
 
#982
SoaDMTGguy said:
Ok, so +probe should start at the top of P3 *or* R12, and the negative should be on ground.

Is there a reason to do this instead of messing *across* R12? Is that equivalent, or different?
Click to expand...

Measuring across R12 will not give a reading of DC offset. In fact, I think it will read almost nothing.

Probe either at input end/top of C3 and other probe to audio gnd, or on C3 side leg of R12 and other leg to gnd. To complicate matters, lid must be on to find null point as IQ/offset will drift with temp, as Dirk points out.

Setting this FE takes a bit of time.

A tip: warm up, set, ensure offset is approx 100mV positive. Put on lid, and you will see the offset reduce towards zero. If it goes negative, you need to go even higher than 100mV before putting on lid.

Quite rewarding in the end, the road can be annoying 🙂 If one has little time that is. Patience is key. That said, it sounds good with a bit of offset too.

Regards,
Andy
 
#984
Update: I've been making many adjustments and readings on the front in board with the output disconnected.

1. I can match Q3 and Q4 *or* zero the offset. Which should be more important? Deviation between Q3 and Q4 to zero offset is ~10-20mV, or, offset with Q3 and Q4 matched is ~<30mV

2. Feeding a 1KHz sin wave into the front end, the output voltage differs by ~2.5% (~0.50v difference between ~22v outputs from a ~2.8v input). I can balance it, but I have to shift both sets of Q3/Q4 out of balance to do so (by ~200mV between each pair). I can affect the output via P3 and P1/P2, but any time I try to find a balanced configuration it swings back apart.
 
#985
Good work! :cheers:

1: I don’t think I would worry about such smallish differences between q3/4. you won’t be able to have them all equal without serious matching, not worth it. I always aim for lowest offset, but I kinda doubt you’ll hear a diff, with such low values. For example, I believe the original Aleph 3 had an offset at the outputs of sub 50mV. So you are definately well within the ball park.

2: @2V8 I believe you are clipping the amp. Try 1 volt and see what happens, then 2. still imbalance? Dennis asked about tolerance of your R13 resistors. They @5% or lower? And also: can you hear difference between channels? It is not certain you will. Pots may even have more diff between chans than 2,5%, so maybe it is not so important? Or?

Regards,
Andy
 
Last edited:
#986
SoaDMTGguy said:
Update: I've been making many adjustments and readings on the front in board with the output disconnected.
Click to expand...

Remember, once you fire up the output stage, bias it to cooking state, and put on lid, front end bias will go up and offset probably shift quite a bit (so much that you have to readjust IME).

But getting the good hang of it now makes it all that much easier later. Be prepared for quite significant offset and IQ/bias drifts with and without lid. I always cook good, high OS IQ, and diff in front end IQ with and without lid can be ivo 5%, meaning quite a bit of offset drift. In my case, depending on which probe you put where, to the negative side. Hence my recommendation of leaving offset at pos side before putting on lid.

On the positive side, output stage sets nice and easily so long as you have significant range to set IQ and null offset.
 
#987
Ok cool, I’ll shoot for zero offset. Thanks!

2.8V is clipping? I thought, to hit 25W output into 8 ohms, 25W = V^2/8 ohm = 14V? But, I will try 1V and compare.

R13 are 0.1% 😀

Yes, I can hear difference. This whole dive into adjustments started because the diff was bothering my while listening.

Yeah, I’m familiar with the temp effects. I’m trying to get a sense of what changes will achieve what results, then I can dial in the final values with everything hooked up and cooked up good. I think I have a pretty good handle on the output stages, it’s just the front end that’s driving me batty.

Thanks Andy! I owe you a beer at this point!
 
#988
SoaDMTGguy said:
2.8V is clipping? I thought, to hit 25W output into 8 ohms, 25W = V^2/8 ohm = 14V? But, I will try 1V and compare.
Click to expand...

SoaDMTGguy said:
2. ... ~22v outputs from a ~2.8v input
Click to expand...

I never know anyone's level of expertise, so humble apologies if anything below seems like I'm "talking down". Below is how I might work through it.

You said 2V8 was the input... and 22V was the output, correct?

You've answered your own question mostly kinda sorta maybe, but by getting the wrong answer to the right question... Assuming an amp of unknown origin cannot output a voltage higher than 14V (it clips at 14V), if you're inputting a 2V8rms signal... with an output of 22Vrms... it'll clip. Right? So... the question is... how can an amp that clips at 14V output 22Vrms? It can't, right?

Edited here to add clarity just in case... and correct an error on my part. Using your hypothetical calc, the 14V would be 14Vrms not 14Vpk. So, the hypothetical amp would clip at ~20Vpk. Still, I think below is useful.

I'd suggest the following exercises for fun... you can post back if you'd like showing your calcs, but if this is crazy to you, then no worries.

What are the rms and peak voltage output of a 25Wrms amp into an 8ohm load?

Check the designer's expected gain of the BA-3 against your expectations and your real-world build.

Calculate the expected maximum peak voltage output (where it clips) for your BA-3 based on your personal build. Assuming it can source enough current and deliver the full power (max voltage) into a reasonable speaker load (it likely can if built properly), what are the peak and rms power outputs of your personal BA-3? Under what load will it achieve its max power output?

Calculate the input voltage necessary to drive your BA-3 to its peak output voltage (clipping) aka the input sensitivity. Show rms and pk.

Either way... if you were inputting a 2V8rms signal into your amp, that's pretty potent. 2V83 output into 8ohm (1W) is generally considered a nice place to do some test measurements. Maybe that's why you chose it... but it needs to be the output voltage, not the input voltage. So, continuing the exercise... what would your input voltage need to be to achieve a 1W output into 8ohm on your personal BA-3? What will the output voltage be with a 1V input signal? What power delivery does that equate to into an 8ohm load? Don't forget to use rms.

Check your calcs against your actual amp. If they match, you're golden. Then... at a reasonable output... perhaps worry about the channel imbalance. If you're hearing it, and it bothers you, it's likely more than 3%, but I truly don't know.

Good fun. Hope you get it sorted out. Others please chime in (ANDY! 🙂) if I'm just rambling on about nothing or have most of the methodology incorrect.
 
Last edited:
#989
Argh! I was thinking of the output voltage, not the input voltage!! I deserve all the talking down you can throw at me 😛

Thanks for the smack on the head. Hopefully I'll get some better results today.

If you're hearing it, and it bothers you, it's likely more than 3%, but I truly don't know.
Click to expand...

I did notice a channel imbalance, and then checked the amp and found the voltage disparity. It's 3% both at the outputs and at the front-end. It's possible there are other factors at play, because I agree that's pretty low. Hopefully I can have some effect on it today and learn more.
 
#991
Yes, but with your quality parts, I would say no.

Have you verified all resistors are in the correct position? Did you measure them all before soldering them in?

R3-4 and 8-11 and R13 important.

Jfets all from same source?

Regards,
Andy
 
#992
I did not measure all the resistors before I installed them. I have pulled and tested R8-13 on the left channel, they all test good. I'm fairly certain I didn't install anything in the wrong place. JFETs are from the diyA store, matched quad. MOSFETs are the original Toshiba's, matched quad purchased from Prakit on this forum.

Not sure if it's relevant, but last time I powered down the amp I checked the resistance of both R13's in-circuit; The right channel measured as expected, 330.5 ohms. The left channel measured a value ~200 ohms which steadily increased to ~230 ohms. I pulled one leg off and measured it, and got the same 330.5 ohm reading as the right channel. Is there any particular reading why the in-circuit measurements would differ like that?

I'm at least gaining a much better understanding of how this circuit works, and the way the adjustments affect things. Through everything, various accidents, rechecking things, etc, I've damaged the board in a few places, and two of the MOSFETs are running on patched legs. I'm planning to rebuild the board from scratch with all/mostly new parts at some point in the near future, so that will be a good sanity check.
 
#993
I wouldn’t stop now, this is possible to fix.

When you have time, voltage drop across your R13s may prove useful 🙂 with shorted inputs.

Drain source voltage @ jfets and mosfets would probably not hurt either. But I am faring in unknown waters now, but those measurements may be useful for proper greedy boyz wanting to help 🙂

Edit: Hell, I might even give it a shot myself.

You centered P3? Please also measure voltage drops across R3 and R4, both sides.

Edit 2: Didn’t see you damaged stuff. That’s just typical. Either way, finding the culprit is a good idea. Might save you a headache for the next board.

Regards,
Andy
 
Last edited:
#994
I do think this board is fixable. The worst is the damage is, as I said, the broken legs on two of my MOSFETS that I’ve fixed with soldered-on legs. I’ve also damaged some of the holes on the left channel so they don’t wanna hold solder properly. Haven’t quite reached the point of needing to patch the traces. It’s a combination of things I’ve had to repair and lessons I’ve learned that make me want to build a new one.

I’ve rechecked the balance issue with my biological auditory sensors, after ensuring my listening position was dead center. Seems ok.

I’m going to proceed with dialing in P3 and zeroing offsets. I’m adjusting P3 by ear and balancing P3 across both channels by level-matching R4. I’ll readjust the bias levels once I’m fairly happy with a position.

What would I do with the information of drain-source voltages on the FETs?
 
#995
The suggested measurements were intended/hoping to provide a possible answer to why you measure an imbalance. But if it doesn’t bother you, maybe not worth the effort.

How do you like the sound? It opens up after a while, but is pretty good to begin with.
 
#996
The sound is fantastic! I’m currently just about balanced on P3, just slightly on the lean side. I keep sitting down to listen “just for a bit” and not getting up 😀

This experience has been a lot of fun. I think I’m going to do a preamp next. Can’t decide if I want to do the BA-3 pre, or something with tubes.
 
#997
To drive the BA-3, not much gain is needed. So for that amp I would choose differently than BA-3 pre (perfect for F4 though, same OS as your amp sans JFET buffer.

I found waynes linestage very, very good. ACP+ would probably rock it too, as well as B1K. For BA-3, B1 buffer would even do the job marvelously.

Toobz is toobz, have no experience with them. But if sugar is your fave, B1K strikes the balances quite nicely, trusting rumour, reviews and manual 🙂
 
You must log in or register to reply here.
Top Bottom