BA-3 Amplifier illustrated build guide

Jump to Latest
#802
andynor said:
One thing I have thought a bit about, is that a well built BA-3, adjusted for some sugar, have a beautiful midrange and soft and reasonably extended highs. So might be a bit too good an amp to use only for lows, or overkill to use another expression. Many Krells, after KSA-50 (maybe the 80 is an exception, I really dunno), use a modest amounts of feedback and deliver huge amounts of bass (wham, bam, thank you Dan!), but many times at the cost of some glare. So if you like the bass of the Krell, may I suggest you also consider a Vanilla F5 dual mono, or an F5 Turbo (without diodes should suffice). Parts for the BA-3, at least for balanced monoblocks, makes for efficient emptying of dollars from your pockets. The F5 and F5T strike me as less expensive, and with less unobtanium parts (J113 and K2013).

Another alternative, and the most obvious, is to try the BA-3 you are assembling with 3 pairs per side and see how it goes. If ZM sez your Altecs need no more, my guess is he is right - as always. Then adjust for minimal distortion and I am pretty sure you’ll have all the oomph you’ll ever need.

Regards,
Andy
Click to expand...

Having those amps and speakers (>100 dB efficient) I assure you the BA3 is more than capable. So is Amp Camp, for that matter. Or any of the first watt clan. M2 is very nice.

Russellc
 
#803
And if I am not mistaken you also run 1R source resistance? So there you are, Bergy! :cheers:

PS: Russel, you are the one that got me thinking the BA-3 might be too good to do only bass. A fantastic amp 🙂 Best not waste the joy of it’s mids and highs.
 
#804
andynor said:
In general you should get an amp with more control, especially in the LF’s, and more authority (citing Papa). Citing ZM, you may also get a bit less detail. And the devices will have to run at a lower IQ per device, which may also have an effect. I am contemplating the same build for my BA-3 iteration (see thread BA-3 revisited), but that project is on halt as I have decided to go for high efficiency speakers and singled ended class A instead - for now.

If you do this, I would think going for huge sinks to maximize bias current per device, is recommended. Meaning trying to get close to the 300 watts total dissipation as described in the manual.

Like ZM and Cubicincher say, it really comes down to what you are going to connect to it downstream.

Also, don’t underestimate Papas words about using prescribed parts in the front end. That means JFETs with 10mA IDSS and Toshiba MOSFETs. I can also recommend you push the FE bias high, and experiment with that. Just remember to sink the MOSFETs properly. Of all the (few) meaningful things I have done with my BA-3, maximizing FE bias had had the most dramatic effect, next to high OS IQ of course.

PS: I am a GH, so primarily listen to ZM and Dirk 🙂

Good luck!

Andy
Click to expand...

I don't know if 10 idss is all that magical, seems like when I inquired I was told 8 idss. anyway, they work just fine. They are Toshibas. I have some matched quads of 9-10 idss, but went with 8. Not saying its better or worse, just it work just great.

Russellc
 
#805
I don’t know either, other than that’s the specified IDSS to achieve 8mA bias current with the specified R3/4 values. Whether it is actually important or significant, I don’t know. I just decided to get the specified IDSS once I sourced original Toshibas. So I have quads @ 9.99mA now, but have not tested yet. You are probably right that there’s no magical difference, but mine bias at 5-6mA, so I guess there might be a difference - especially if the JFETs bias current affect the rest of the gain circuitry in the FE (can’t remember/don’t have the brains to say that in advance, but it is prolly in the manual. Will be exciting to find out once I get to it.

How is your J Zen coming along? I haven’t started.
 
#806
bergy said:
Thanks guys,
all your suggestions are really interesting to me.
Andy, you were spot on, I really really love bass performance provided by old Krell amplifiers, capacities that I never found in any other amp.
About Altec, the 416-8C (ferrite magnet) is a bit different than the most popular 416 alnico magnets, that I think Zen is referring, the 8C can easily eat many Watt and deliver beautiful controlled deep bass (if well drived).

PS
What you mean with "F5T (without diodes should suffice)" and what is Vanilla F5?
Click to expand...

IIRC it's rated about 75 watts RMS. Way before you get there you will have great sound pressure all right! I found it the most enjoyable of the drivers I've tried in my cabinets. I liked them better than the 515 8G I had in them originally. I understand the 515 8G needs some electronic compensation to get the best bass, but my system was a passive crossover, and the 416-8C was better to me like that.

russellc
 
#807
This discussion about bi-amping, changing resistors on output devices in exchange for tighter matching of mosfets, increasing rails, mono blocks, giant power supply etc has me reflecting.

Having built a few and having fiddled with them, increased rails, overkill power transformer and capacitor banks, monoblocked them I am no exception. The results all sounded good, but some better, some not so much when compared to the original spec.

Seems like odd as it may seem that PAPA knew what he was doing with the original spec! My mono block Aleph J and M2 sounded great....but when I had a power supply transformer fail in one, I had occasion to switch them back to stereo amps. Even after the replacement transformer came, they stay in stereo configuration. More simple yes, but more magic it seems?

Sometimes different magic is made, like Zenmods excellent contributions, but he has full mastery of harmonic distortion spectra, something I do not. So, I'm building stock or building Zenmod, best available giant shoulders are better than no giant shoulders, and I need the help lord knows.

My music would suck much more without you guys.

Russellc
 
#808
I agree, and that is the reason I have gotten hold of the specified parts for the BA-3 (to build to spec, in every way). Wrt PSU, I have guaranteed overdone it. But all choices have been founded on ZM and Papas advice, but added a bit of capacitance I’ll admit. Excited to see what it sounds like.
 
Last edited:
#817
If you have well-matched Mosfets the source resistors … can be reduced ... You will get more output power with smaller source resistors
Click to expand...

This makes sense, but I don’t understand the caveat “if you have well-matched MOSFETs”. Isn’t the degree of error the same no matter what, just with more amps if you use smaller resistors?
 
#818
to soaDMTGguy #817

Hello soaDMTGguy,


you always should have well matched MosFets in your outputstage.
Most important is, that the N-channel Mosfets are matched to each other and
the P-channel Mosfets are matched to each other. But also N-ch to P-ch.

Paralleled Mosfets work like a 'team'. They should have close values (Vgs, Idss,....). This means, that in a BA-3 outputstage 3 N-channel and 3 P-channel-MosFets are working together. If one MosFet has very different values, then you can get 'bias-hogging' or one MosFet gets much hotter than the others. Why?
One MosFet is doing more 'electrical - work' than the others. And that one
will quit his job very early.... :redhot::flame:

I suggest to buy matched MosFets from the diyAudio Store. Then you can

reduce the value of the bias resistors and run the Mosfets 'hotter'.

Don't buy a handful of Mosfets at any distrbutor and solder them untested / unmatched into your BA-3-outputstage!
Greets
Dirk
 
#819
In my experience, the store MOSFETs for the F4/BA-2 OS can have a bit poorer matching than 0V1 difference in VGS. I suggest you use 1R source resistors with those to ensure more even current sharing between devices, unless you make a jig and sort the closest triplets of N and P before putting them in circuit. I use 0R5 and had 11-19% discrepancy in bias/IQ. Though no crisis, not ideal.

Regards,
Andy
 
#820
You might also consider matching them yourself. The parts are cheap like dirt and its a skill worth having. Less worry you might actually fry your output transistors. I needed 24 N and P devices each, so I went that route and have plenty left-overs for the same money.

Socket:
Blocked

I needed some support for the socket, so no stress was on the part for most precise results.

I used these resistors:
https://www.mouser.de/ProductDetail...qLThgwXRvQhqX5gaJ/BWT9RNlzgVqL3NMZrHd4K/5kg==

Have fun :wave:
 

Attachments

  • Matching.JPG
    Matching.JPG
    372.2 KB · Views: 175
  • art_matching (1).pdf
    art_matching (1).pdf
    109.1 KB · Views: 94
You must log in or register to reply here.
Top Bottom