• WARNING: Tube/Valve amplifiers use potentially LETHAL HIGH VOLTAGES.
    Building, troubleshooting and testing of these amplifiers should only be
    performed by someone who is thoroughly familiar with
    the safety precautions around high voltages.

SS phase splitters?

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.

Same concern. Minimum HFE=25+, might be a little low...
And they say nothing about capacitance, we need <25pF.

-----

They all give a current gain*bandwidth product = FT
that makes it seem like all these parts are really fast.
But this is pure current, operating like in a cascode.
Once you see 1 + 1 = 2 Voltage gain of a Concertina,
then Miller probably gets your way before FT does...

We could put an augmented cascode in the middle,
rather than a single lump of sand. Maybe that works
better?
 
Last edited:
MJE3439 is an ancient, slow, relatively low gain device, that was at one time offered as a TO-39 in its 2N3439 incarnation. I used to use them and their rough complement the 2N5416 30 years ago as VAS drivers out of desperation, as they were the only things I could readily find that would hold off voltage.

The transistors you want to use these days are the ones that were tailored for high voltage video drive duty. The Fairchild device may be one of them. I think Fairchild cloned them as KS from the original Sanyo 2S devices. I'll have more to say when I go home and look at my stock, as I bought some of these with VAS/hybrid tube circuits in mind. Mouser has them for the time being. I would advise buying quickly, as the need for HV video drivers is disappearing quickly as jug monitors/TV sets go away.
 
None of these are too slow FT(current gain bandwidth) for concertina
purposes. Current gain of last two mentioned is rather low, not slow.
If you need less than 25x current gain, they might be just fine.

Several are too slow in the Miller dept, but a few mentioned were <25pF.
At least half of them you can't tell cause the spec sheet doesn't say.
You have to download a spice .model to find the CJC specification...

Slow=Reliable?, Tell that to my lawnmower...
 
Last edited:
Since KSA1381 is suitable in every category including availability and price,
I'm not sure why I would need to lean on Darlington, Sziklai, or Cadcode?
Its worth my 49 cents not to screw with a complicated pairing.

Sure, pairs could work "better", but its a Concertina. It either works right,
or it doesn't. There's not much grey area to say one or other is better.

I think Fairchild numbers them both ways, 2SA1381 is same part.
http://www.alldatasheet.com/datasheet-pdf/pdf/38535/SANYO/2SA1381.html
Yeah, looks like Sanyo might have been the original maker...
 
Last edited:
http://www.mouser.com/ds/2/149/FQP3P50-110765.pdf
-500V TO220 PMOSFET. 9.5pF seems OK, as long as VDS>20.
Looks like they want a dollarseventeen.

Cap gets out of hand quick, if you don't keep at least 20V on it.
I think thats true of most MOSFETs, but this spec sheet has the
balls to show a truth that others conveniently omit...

Burn that graph into your eyesockets if you use MOSFETs.
Miller jumps to 600picos over the last 20 volts! A little
planning, and this area of misbehavior is totally avoidable.
Probably still more headroom than you get with a triode.
 
Last edited:
This Graph...
 

Attachments

  • CapSquirm.gif
    CapSquirm.gif
    45.9 KB · Views: 176
Turns out Cordell already made us a spice model for that PNP.

Cut one stage just to see if there was still enough open
loop gain to do something useful? It looks pretty alright.
I tweaked R15 for a closed loop voltage gain of 25x.

Strange consequence of the equal impedance split is that
it makes an inverted copy of any disturbance on the other
end. Thus, no problem that GNF is wired to only one end.

Not so useful, also makes an inverted copy of power supply
ripple appear at the bottom! I threw a filter cap at the prob.
It seems OK now, even with 5V of 120Hz ripple added to B+.
 

Attachments

  • minimal.gif
    minimal.gif
    47.1 KB · Views: 147
  • minimal.asc
    6.1 KB · Views: 28
Last edited:
Coolest stupid pet trick in the universe:

With the main input disabled, I open the loop
and apply fake GNF. Just to prove the splitter
mysteriously copies everything to both ends.
Yes, it does!

Instead of filtering the ripple coming into the
top, I could have injected ripple to the bottom.
And they'd probably cancel out, maybe. That
might be pushing at the edges of reality a bit...
 

Attachments

  • EQGNF.gif
    EQGNF.gif
    36.1 KB · Views: 146
Last edited:
Took me a while to figure how AB2 was going to work.
No followers! Let 30mA Concertina do all the driving.
Q1 will need a heatsink for sure...

The cheat is maintain bias by to pulling up only on the
cap in cutoff. Shhhh... Don't tell anyone.
 

Attachments

  • PNPAB2.gif
    PNPAB2.gif
    41 KB · Views: 135
  • PNPAB2.asc
    7.1 KB · Views: 27
Last edited:
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.