• 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.

Using mains transformer as output transformer

Hello,

revintage:
Why the extra tube? Can do the current mirror without it as well.

How about something like this. Throw the current mirror at the top.

R3 is selected so that Vcc-U1 > max anode swing.

C1 is the supply filter capacitor.

Can be no simpler, no? :D

Benefit is also that the saturation of the toroid initially is less because one direction current through toroid is only half of the bias current.


- Elias
 

Attachments

  • toroid_comp1.png
    toroid_comp1.png
    11.3 KB · Views: 600
Hello,

revintage said:
Your solution looks far out but interesting, I´m not sure it will work IRL. Anyway I will sim it for you to check.

The reason for using a second tube is that a cheapo tetrode do not need a heat-sink. It is also foolproof, would not trust sand with 300V over it to be so.

May I ask what is IRL?

I think voltage rating is no excuse as there are more than plenty of choices of BJT that can handle easily more than 800V.

The dissipation of BJTs are no problem as they do not take over whole supply voltage and because the fact that the Ibias is split into two. I calculated about 5W per transistor, so no heat sink needed instead can bult them to chassis.

If your simulation fails, I'll build a proto :)

- Elias
 
Hello,

I just realised my top current mirror concept approaches this very old idea from one patent. Mirrored current source imitates the choke. It was disscussed here earlier but I cannot locate that thread right now.

- Elias
 

Attachments

  • fig30qw.jpg
    fig30qw.jpg
    13.9 KB · Views: 534
The patent works with normal B+ because the ideal choke has infinite AC Z but minimal DCR. The effiency will theoretically be doubled but it will cost a lot since the choke will cost to much. Also something must be done to equal the DCRs and then you loose efficiency again.

If you want your circuit to work you have to double B+(6-700V :hot: ) for a triode-strapped EL34. Didn´t think so far when I said it wouldn´t work:clown: . Then you have to let the resistor above the triode take the heat. So in that sense it will work but what you win in less current adds in voltage. The efficiency will be the same as in my suggestion. So there are no free lunches......

In my eyes this is a lot more complicated. It might be worth if the halved current makes it sound better......

KenPeter: We could do like you suggest but it was not what we aimed for. What you suggest is a self-split PP and the &V& should then be substituted for another 2A3, but our aim was to make a SE. According to the sims I have done the uneven distortion doesn´t look to promising though.
 
Hello,

I think I tried to express the operation quite clearly:
Elias said:
R3 is selected so that Vcc-U1 > max anode swing.


revintage said:
If you want your circuit to work you have to double B+(6-700V :hot: ) for a triode-strapped EL34. Didn´t think so far when I said it wouldn´t work:clown: . Then you have to let the resistor above the triode take the heat. So in that sense it will work but what you win in less current adds in voltage. The efficiency will be the same as in my suggestion. So there are no free lunches......

In my eyes this is a lot more complicated. It might be worth if the halved current makes it sound better......


I think doubling the voltage is a little bit of exaggeration. Let's say for 8ohm load we want 6Wrms sinusoidal power, so with the given 6V:115V ratio anode swing will be +/-190Vpeak. Can put Vcc = 500V and select R3 so that U1 = 300V. Set anode current Ibias = 60mA. In this case power dissipation in R3 will be 6W, and also 6W is the dissipation in Q2. I think this amount of dissipation is still well within the manageable level. Anyway we are talking about class A SE tube amps here so... :D

Where is the complication? :) I think this is wery worthwhile as this reduces the initial magnetisation of the core by halfing the DC current, as it can have a huge efect on lowering the saturation of the core in the first place.

- Elias
 
revintage said:

KenPeter: We could do like you suggest but it was not what we aimed for. What you suggest is a self-split PP and the &V& should then be substituted for another 2A3, but our aim was to make a SE. According to the sims I have done the uneven distortion doesn´t look to promising though.

SE or close to it as your version already is. Only 2xSE...

Current change in the Pentode becomes entirely equal and
opposite that of the Triode. It is NOT a complementary triode,
but one with the curves bent strangely backward. Reinforces
rather than cancels any single-endedness...

In deliberately unbalanced common cathode amplifiers,
the device with the weaker GM becomes the master.
And the higher GM device simply does the opposite.
Also helps that the higher GM device largely ignores
any NFB from the plate, as Pentodes do so well.

Doesn't even matter which grid (Triode or Pentode or
both) the signal is injected, the weaker GM Triode
determines both the curvature and the anti-curvature.

If you substituted the 6V6 for another triode of more
closely matched GM, would become ordinary push-pull.
Not at all what I was suggesting.

Now the coupling cap might need to be a bit bigger than
470uF, to optimally bridge two such low impedance nodes.
but 470uF would be enough to verify the concept.

MJK might be able to give a better explanation.
 
Hello,

Hmm.. Also my first thought was that coupling cathodes together will make it as a self splitting PP.

The tubes being different makes the distortion behaviour very complex. I think regardless how you connect the tubes, much of the even order distortion is cancelled in the transformer when they drive it in a push-pullish way.

I believe it's difficult to achieve the SE 'sound' easily with that because there are so many affecting parameters. Afterall I think having that extra tube there makes things very complicated indeed ;)

- Elias
 
The Cathode of the Pentode is a low impedance Ground Follower.
Far lower Cathode impedance than the Triode paired with it.

You simply bypass the Cathode of the Triode to that virtual
ground node instead of to normal ground.

The magic is that any current variations of the Triode's Cathode
re-appear at the Pentode's Plate, thus completing the full circle.
Also doubling the effective SE power for no extra parts.

The current variations can't go anywhere else, the pair of CCS
prevent it. And the error of the Pentode is no worse than how
accurately it follows Ground * (Mu-1) of the controlling Triode.
The more grossly mismatched the devices, the smaller the error.

The lower GM Triode sets the current for the loop, the Pentode
is relegated to a mere Cascode, that can only pass that current
back up the opposite end of the transformer. One SE Triode in
complete control of differential current thru the whole OPT from
one end to the other, monitoring voltage and therefore load
reactance at that one SE plate only.

The Pentode screened to ignore variations of the load at its
own plate. Provided its GM is also very high, serves no signal
processing function than to follow ground...

---------------------------------------------------------------

Now the weird stuff... Suppose instead of following Ground,
the Pentode follows a Signal. Suppose instead of following
a Signal, the Triode follows Ground. ITS EXACTLY THE SAME.
Only inverted, but otherwise exactly the same...

The Pentode will impose the Signal onto the Cathode of the
Triode, with exactly the same result. The Triode is still the
only device that will be making audio processing decisions.

The Anti-Triode is absolute evil, after all... Fear is the
only sane and completely understandable reaction. I'm still
getting pea soup off the walls from our last encounter.
 
kenpeter: I simmed your (as usual weird circuit;) )and mating the 6V6 with the 2A3. Think you got something here. Have you actually tested it?

Anyway, the decoupling cap is critical. To high and you loose low-end(!) to low and you get a bump in the 10-20Hz region. Found 150uF to be optimal in this case. Also when doing this the distortion spectra is evenly falling like a good SE, no traces of PP even order products. Looks promising, but I would call it an Unsymmetrical Class-A Self Split PP:D.

Elias: Please advise PNP BJTs that can take at least 50ma at 400V at a reasonable price. Googling didn´t help me.

Compared to the "standard" topology you actually need to double B+ to get comparable results. Also it works best with the cap from above the resistor to have its low end connected to the triodes cathode. No cathode bypass to ground needed even if a small plastic type would help a little. Also a resistor of a few hunderd ohms should be inserted between the power BJT and the winding.
 
Elias,
Interesting! Another use for a centertapped mainstoroid could be be for a tube Circlotron as there are no current at all through the primary. Instead of OTL one could then get optimal loading for the tubes.......

Going back to your earlier suggestion, can you suggest any easy available, suitable PNP or P-MOS high voltage power transisistors?
 
High-V P channel

If you have an ON semi sales office, try the MTP2P50EG. 500V and 75W rated.

http://www.onsemi.com/PowerSolutions/product.do?id=MTP2P50E

PNP BJT with 400V AND 50mA is very tough (and inefficient), because second breakdown effects are at their worst in that quadrant. A vastly overrated part would be needed. But if you are determined to try that, use multiple SOT223 transistors in parallel, with a resistor in each emitter to prevent "hogging".

Zetex make the best parts of this kind, all the way to 500V -

http://www.zetex.com/3.0/3-3-2b.asp?rid=10

So, for instance you can handle about 300V/50ma with 5 x FZT958, or 400V/50mA with about 14(!) x FZT560.

To build, just get a chunk of copper-clad board and use a hacksaw to cut "tiles" into the copper. Use the Tab to connect the collector, so you can ignore the centre pin (see picture)
 

Attachments

  • sot223mnt.jpg
    sot223mnt.jpg
    80.7 KB · Views: 433