BA-3 Amplifier illustrated build guide

Member
Joined 2019
Paid Member
So "standard" BA3 rails at 24 volts (18 + 18 transformer) and 6 mosfets per side as well?

Got my eye on this. Hope you play with P3/ REW/Focusrite before turning it into a mono block.

Just wondering,

Russellc

I suppose I could have used a better word than "standard". :D The essentials are:

1) PSU - 2x400VA transformers with 18V secondaries with the DIYA Universal PSU with the diode boards. At 120VAC, they deliver right at about 26.5V unloaded. When I was dialing it in yesterday, I was using a variac set to 120V - it was delivering about 24V5 loaded. Right now, they're set up in a standard dual mono configuration. Later, after conversion to monoblocks, they can be run parallel for an effective 800VA.

2) For now, 6 devices per "side". 0A5 at each device. Plans are to give this configuration a good bit of time to settle. Heatsinks got to ~42C, but I did not take careful measurements. Using pretty tightly matched FETs and 0R22 source resistors. Plans are to try both 6 and 12 devices per "side" in both stereo and monoblock configurations at various bias points along with trying the SE output stage. I have not determined the exact order. Itsmee provided a fabulous idea for mounting 12 devices (as have others), but that would require me re-building one set of boards. I think I'll use wowbagger's idea for the complimentary and Itsmee's for the SE.

3) re: REW - DEFINITELY. That's one of the reasons I wanted to start slowy with this configuration. Less complexity, and I get to play with H2. :D

Firing it up now. I'm going to uninstall / reinstall REW and the Focusrite Control software to see if I can get it to behave. It worked flawlessly for about 3 hours. A computer restart/sleep/Windows update seems to have scrambled its eggs. I may use the Mac, but there is a goofy thing going on there too. For clarity... I never blame software, when it could easily be a noooooob user error. :D I just want to get it going properly again. I have some wiring to get the rig set up.
 
Member
Joined 2003
Paid Member
I knew what you meant by standard, I that's what I was wondering about. Number of outputs and rail voltage.

I did 31ish rails like 6l6 did in his BA3 build guide. This pushes it to around 50 watts class A. I love this amp.

In this mode it drives speakers that test the limits of Aleph J and M2. Lot of projects ahead. But eventually I would like a half and half mono blocks with switchable power supplies, half SissySit, and half M25.

Russellc
 
Member
Joined 2019
Paid Member
Check my head... and my math :D

Well, I am completely loving the BA-3. Shocking, right! :D

I had a whole sequence of "experiments" planned, some of which I may still carry out. However, I may take a change in tack.

Here's where my head is. I have been after current all along. My quest for current was loosely based on some assumptions regarding my speakers and some back of the matchbook math. I have no clue if I interpreted some of this correctly - or if I made too big a leap in one place or another.

Assume for a moment, however ridiculous it may be, that I'd want to be able to hit 98dBSPL if even briefly at my listening position and that my speakers meet their advertised spec of 98dB @ 2V83 (2W) @ 1m. 4 ohm impedance. Also assume that they dip to 2 ohms (unverified) somewhere in the bass region. Assume, that I'd like to be able to deliver 98dBSPL from 20Hz to 20kHZ at the listening position.

If I understand correctly, I will lose ~14dB moving to a 5m listening position, correct?

So, to make up that 14dB - I need roughly 5x (2.83) the voltage, correct?
...V..........R........I..........W=VA.....dB
14V15......8....1A769......25.03......98
14V15......6....2A358......33.37......98
14V15......4....3A538......50.06......98
14V15......2....7A075......100.11....98

If I want to be able to have the same SPL in my assumed bass region that dips to 2 ohms - I need a circuit that can "move" ~7A per channel, correct?

If I also assume that I'd never go above 98dBSPL at the listening position, then I'd never need above say... 15V0 rails. Is that correct?

So, could I bring the rails down (different transformer) and bring the bias up while keeping the total watts dissipated at the heatsinks the same for my purposes? Trade voltage (which I don't seem to need) for some current (which would be nice). Conceivably, I could run 15V0 x 1A5 across 6 devices with no issues. Heatsinks should take it, no going over 35W per device, I wouldn't exceed 1/2 VA of transformer etc...

How high does the voltage need to be on the front end / output stages for them to work properly? I thought I had read 18V0 minimum, but I can't seem to find my reference.

I also may never go to balanced monoblocks - if I am picking up what's being put down... the only gain is voltage swing, which I don't need. The cancellation of H2 may also be a detriment. Was my quest for balanced foolish and/or potentially for the wrong reasons?

Should I be expecting a room with padded walls in the future, and/or am I thinking this through in a logical way? :D
 
Official Court Jester
Joined 2003
Paid Member
let's stay at 18Vdc for OS rails as minimum; in that case 18Vdc is sorta enough for FE, if you don't need exactly rail to rail swing, but few volts more are nice; though, for simplicity and sane engineering case, you can stay at 18Vdc rails for both FE and OS and be happy till Doomsday

even if source follower, it's nice to try keeping mosfets with enough voltage (ya know the story - bad capacitances etc.)

yes, if chasing current, bridged config means just more hassle and heat,especially if you don't deliberately want vanished 2nd and dominant 3rd

of course that intentional dis-balance of symmetry is possible ( my Babelfish XJ and numerous newer Papamps are proof that same is possible), but that's different story
 
Member
Joined 2019
Paid Member
let's stay at 18Vdc for OS rails as minimum; in that case 18Vdc is sorta enough for FE, if you don't need exactly rail to rail swing, but few volts more are nice; though, for simplicity and sane engineering case, you can stay at 18Vdc rails for both FE and OS and be happy till Doomsday

Wonderful to know. If I get a bit more advanced in the future, I can consider powering FE with different supply and/or removing FE altogether. All of my pre-amps will deliver appropriate voltage at sufficiently low output impedance. I am just a big sissy to try to remove. I considered trying with F4 first with 300B amp, but haven't had the courage to try the crippled.

even if source follower, it's nice to try keeping mosfets with enough voltage (ya know the story - bad capacitances etc.)

This I will have to learn, but for now I trust....

yes, if chasing current, bridged config means just more hassle and heat,especially if you don't deliberately want vanished 2nd and dominant 3rd

I'd certainly be willing to try balanced, that was my original plan. Now, I lean toward reducing the voltage, and taking the single-ended output stages for a spin. At some point before I try any of this, I will be trying to hook up a scope to see if I ever (in practical use) get anywhere near voltage clipping. At this moment, my whole premise is based on my limited understanding and potentially faulty math. :D

of course that intentional dis-balance of symmetry is possible ( my Babelfish XJ and numerous newer Papamps are proof that same is possible), but that's different story

It is fun. I measured the balance on the BA-3 with REW and set it currently to "the middle". I will leave it alone for some time and then vary the amount to see if I like the results. I have Papa's H2 device out now, and I am playing with that. One of my fascinations is how the whole balance of the harmonics makes it through the entire chain (per previous conversation). Then I get to try with F5 and Ishikawa on M2x. I am very interested to see results on the M2x, b/c M2 has been said to have a dominant 3rd. So, will my impression change if I play with the distortion profile? :D

Thank you as always, I am learning in small bits.
 
Last edited:
Member
Joined 2019
Paid Member
Yes, your math is correct, but as Zen says, you really do want more voltage (I.E., the designed 24v rails...) to make everything work in the sweet spot. :D:D:D

AHA! Well, there ya have it. Never second guess, Papa. :D

Also as I was digging and and looking at whether I could move from theory to practice....

Commonly available products either drop me too low (12V secondaries) or don't drop me low enough to potentially realize a ton of benefit (15V secondaries).

So, while I do love to learn, and the thought exercise was enlightening - from a practical standpoint, I'll continue to run the 18V secondaries with ~24V5 loaded rails. They are "Burning Amps" after all. Plus, I still have a bit of room for more heat. I think realistically, with my existing heatsinks I can possibly achieve 7Aish total bias current per channel at 24V rails. I still am not sure how much heatsinks have to be "de-rated" in practical use. Some time in the future, I'll see how much I can get out of them. :eek:

Note - I have left everything alone since my initial settings of 0A5 per device with P3 "dead center". It'll stay that way for at least a few more weeks. Just speculating. :D :D :D
 
Member
Joined 2019
Paid Member
I know this thread isn't "Amplifier Basics for Dummies", but I'll try to keep it on topic.

More reading and re-reading... more learning... some enlightenment... some confusion...

For the BA-3 (complimentary output stage), would the current available in Class A (regardless of its importance) be double the bias current? Then, once that point is reached, the amplifier moves to Class B operation, correct? The peak power limit would be reached in Class B when I "hit the rails", and voltage clip, correct? I think I may have done it incorrectly earlier.

I know there is no, "klunk" :D I am simply trying to understand the basics.

I can't wrap my head around how the Class A bias current could be more than the sum of the bias current for each device (6 x 0A5) => 3A0

I've visualized the handoff etc. but that's still 3 devices handing off to 3 devices.

I may have misinterpreted Papa's article, but I've read it a zillion times in the past year. This is one of the relevant passages.

"Push-pull amplifiers generally operate in Class A mode up to a point where the output current is twice the value of the bias current."

Where have I gone wrong? :confused: Brain has reached saturation, and a few cups of coffee and thinking on it for a few days has not helped. :ashamed:
 
Official Court Jester
Joined 2003
Paid Member
.....

For the BA-3 (complimentary output stage), would the current available in Class A (regardless of its importance) be double the bias current? Then, once that point is reached, the amplifier moves to Class B operation, correct? The peak power limit would be reached in Class B when I "hit the rails", and voltage clip, correct? I think I may have done it incorrectly earlier........

pretty much correct

Klunk!!!

:clown:
 
Member
Joined 2019
Paid Member
You're a sly one Mr. ZM! :joker:

I always trust Papa, but ... I still can't wrap my head around, why???

If I have each of 3Ns and 3Ps at 0A5, how can I get 1A2 of current without leaving Class A?

Brain melting!

Oh well, school is for another time in another thread. I don't need to understand it fully yet. Maybe one day, it will "click". :clown:

I have an Iron Pre to finish getting into a proper chassis to go with my BA. :D
 
Official Court Jester
Joined 2003
Paid Member
You're a sly one Mr. ZM! :joker:

I always trust Papa, but .......
....

well, that's the Root of All Our Problems :rofl:

see, M2 OS, same thing you're looking to understand , source follower

observe Iq, currents through upper and lower mos and current through load

it'll be easier if you, for start, take as granted that lower and upper mosfet are working together as those two Lumberjack guys with push-pull (see :rofl: ) -saw, to supply the load, and none pausing as long they're in Class A

that's pretty much definition of class A, no part pausing for coffee break
 

Attachments

  • sim sshot.png
    sim sshot.png
    31.4 KB · Views: 534
Last edited:
Member
Joined 2019
Paid Member
ZM and 6L6 :worship:

The explanation and the graphic made me move from "klunk" to "click". :D :D :D

I had the picture in my mind of once the "smooth handoff" was made, one guy stops running. The people sawing in push-pull or one leg pulling up on a bike pedal while the other is still pushing down and vice versa... now that clicks! So, when biking with clips, I'm pedaling in Class A. :joker:

:cheers:

Back to drilling holes :D