The circuit is near enough an SRPP. The problem is that you will exceed a number of ratings. Simply having two heater windings won't solve the upper Vhk problem. I assume you want to swing significant voltage on the upper valve. That means Vhk will be exceeded, and that's why I say it won't work. Sorry.
Hi Lars,
I believe the screen is only rated for about 450V. I'd say that if it worked, it would be on the hairy edge. A short from one of your ESL's would probably cause fatal havoc with the amp and speaker.
Have you considered driving the panels differentially between the plates? You could then regulate your screens and reduce the risk of a catastrophe.
-Chris
I believe the screen is only rated for about 450V. I'd say that if it worked, it would be on the hairy edge. A short from one of your ESL's would probably cause fatal havoc with the amp and speaker.
Have you considered driving the panels differentially between the plates? You could then regulate your screens and reduce the risk of a catastrophe.
-Chris
anatech said:
Hi again,
Forgot to tell you. The idea was, that this was only half of a bridged amp with both sides of the stators connected by capacitors. This would give around 1100V rms on the stators.
Here is the bridged mode:
www.revintage.se/EL342.pdf
Have made an alternative connection of the screens to get a DC of ca 400V on the lower EL34 (that is AC-connected as a triode). Do not know which one is the better.
Have used the AC connected triode-mode in a preamp with EF86 and it works there.
About Vhk: Why wouldn´t two separate heater-windings with one side attached to each of the upper cathodes help the problem?
Hi Lars,
R3 causes the Vg2 to be lower than the anode voltage, this restricts the maximum output voltage swing.
The maximum possible signal swing (sine wave) is less than 2*Vg2.
The maximum Vg2 voltage specification is often determined by the max power dissipation of g2 under penthode operation. In penthode mode g2 power dissipation increases drastically if Va < Vg2. But in triode mode Vg2=Va so max power g2 dissipation is reached at a higher Vg2.
The question is if Vg2 breakdown (arcing) to any other structures is reached before max power dissipation of g2. I never checked this for the EL34 but in triode mode (Va=Vg3=Vg2) Vg2 can go to Vg2_0 which is rated at 800V. But pushing it over 1kV might be too much.
Looking at the abuse NOS EL34s can take before failure you might be able to get this circuit working, but I wouldn't bet my NOS EL34s on it.
A solution could be running the bottom tube as penthode. You'll also have to modifying the upper tube circuit so it determines the bias voltage. Supply a fixed voltage (750V) via a high value resistor to g1 and connect g1 via a capacitor to the anode of the bottom tube.
A direct drive ESL design can be found on http://ultranalog.com/
Corne
R3 causes the Vg2 to be lower than the anode voltage, this restricts the maximum output voltage swing.
The maximum possible signal swing (sine wave) is less than 2*Vg2.
The maximum Vg2 voltage specification is often determined by the max power dissipation of g2 under penthode operation. In penthode mode g2 power dissipation increases drastically if Va < Vg2. But in triode mode Vg2=Va so max power g2 dissipation is reached at a higher Vg2.
The question is if Vg2 breakdown (arcing) to any other structures is reached before max power dissipation of g2. I never checked this for the EL34 but in triode mode (Va=Vg3=Vg2) Vg2 can go to Vg2_0 which is rated at 800V. But pushing it over 1kV might be too much.
Looking at the abuse NOS EL34s can take before failure you might be able to get this circuit working, but I wouldn't bet my NOS EL34s on it.
A solution could be running the bottom tube as penthode. You'll also have to modifying the upper tube circuit so it determines the bias voltage. Supply a fixed voltage (750V) via a high value resistor to g1 and connect g1 via a capacitor to the anode of the bottom tube.
A direct drive ESL design can be found on http://ultranalog.com/
Corne
Because the upper tube's cathode must swing more than Vhk(max). If the heater voltage reference is fixed, you'll exceed the Vhk rating. If you tie that reference to the signal, you have added all sorts of stray reactances and the circuit will be difficult to stabilize (not to mention having bandwidth problems).
corne said:
Hi Corne,
The whole idea of this was to get to make the bottom pentode to work as a pseudo-triode to make it more linear. You get this at the cost of sensitivity. Think you need around 70V rms in make it clip.
One could also look at the bottom tube as a pentode with 100% AC-feedback to g2.
Lars,
I realise that you want to use pseudo-triode but I don't think it's possible to do this with EL34's and the voltages you require. My suggestion is solution that might work.
As I said in my previous post, you'll limit the output swing to 2*Vg2 of the bottom tube. You can't get any higher output swing because Vg2 must go negative (which it can't) to do so.
If you want to respect Vg2_0 the maximum voltage swing is limited 800Vpp per section so 1600V for the bridge circuit.
Corne
I realise that you want to use pseudo-triode but I don't think it's possible to do this with EL34's and the voltages you require. My suggestion is solution that might work.
As I said in my previous post, you'll limit the output swing to 2*Vg2 of the bottom tube. You can't get any higher output swing because Vg2 must go negative (which it can't) to do so.
If you want to respect Vg2_0 the maximum voltage swing is limited 800Vpp per section so 1600V for the bridge circuit.
Corne
Tying the heater reference to the upper cathode CAN work if some care is taken. You'll need a heater secondary that can handle the large signal voltage swings without breakdown, and one where the capacitance from secondary to primary is low. Since you are already driving the capacitance of your ESL, make sure that this additional capacitive load is considerably lower. Once you embark on a direct-drive ESL amp, you must be prepared for numerous challenges. It's almost as hard as a good RIAA stage!
For low voltage duty (in other kinds of amplifiers), I have used split-bobbin transformers where the measured primary-to-secondary capacitance values were on the order of tens of picofarads. You might be able to find such a transformer capable of these higher voltages. In any case, you can place a resistor of a few kilo-ohms in series with the cathode-to-heater-supply connection. This provides a slight isolation of the supply's reactance from the cathode without adversely affecting real-world Vhk concerns, unless you plan on hours of full power 20KHz square-wave testing. The resistor is probably not needed if you've done well with the transformer selection, but it "feels right".
In a rather extreme case, I used a floating 833 tube in a similar SRPP bridge arrangement to yours. This tube has a directly heated filament cathode, but the concept is similar. I wound my own filament transformers using toroidal cores from 300VA "Variacs". Eighteen turns with 15KV silicone wire gave me the 10V at 10A that I needed, tweaked spot-on by using a sliding-tap power resistor in the primary circuit. I don't have the measured secondary-to-primary capacitance figures at hand right now, but I think they were in the tens of picofarads range also, well less than the load presented by the Quads themselves.
Hey, the whole notion of building a direct-drive ESL amp is adventurous (if not crazy
) to begin with. If you're set on doing it, then why not jump in with both feet?
However, after having done this as a challenge myself, I now would recommend designing a conventional tube amp, using a mild step-up output transformer, perhaps with a turns ratio in the 1:3 range, give or take. The Hammond catalog won't show these. On another thread here, we mused about the possibility of using an AM modulation transformer that might have step-up ratios in this range, or you could wind your own. This oddball OPT design challenge would probably be less daunting than the full-out DD amp. Just my 2 cents...
For low voltage duty (in other kinds of amplifiers), I have used split-bobbin transformers where the measured primary-to-secondary capacitance values were on the order of tens of picofarads. You might be able to find such a transformer capable of these higher voltages. In any case, you can place a resistor of a few kilo-ohms in series with the cathode-to-heater-supply connection. This provides a slight isolation of the supply's reactance from the cathode without adversely affecting real-world Vhk concerns, unless you plan on hours of full power 20KHz square-wave testing. The resistor is probably not needed if you've done well with the transformer selection, but it "feels right".
In a rather extreme case, I used a floating 833 tube in a similar SRPP bridge arrangement to yours. This tube has a directly heated filament cathode, but the concept is similar. I wound my own filament transformers using toroidal cores from 300VA "Variacs". Eighteen turns with 15KV silicone wire gave me the 10V at 10A that I needed, tweaked spot-on by using a sliding-tap power resistor in the primary circuit. I don't have the measured secondary-to-primary capacitance figures at hand right now, but I think they were in the tens of picofarads range also, well less than the load presented by the Quads themselves.
Hey, the whole notion of building a direct-drive ESL amp is adventurous (if not crazy
) to begin with. If you're set on doing it, then why not jump in with both feet?However, after having done this as a challenge myself, I now would recommend designing a conventional tube amp, using a mild step-up output transformer, perhaps with a turns ratio in the 1:3 range, give or take. The Hammond catalog won't show these. On another thread here, we mused about the possibility of using an AM modulation transformer that might have step-up ratios in this range, or you could wind your own. This oddball OPT design challenge would probably be less daunting than the full-out DD amp. Just my 2 cents...
Attachments
If you do a PP amp, why not try a Mains Toroidal as a stepup- people have reported success driving dynamic speakers with them.
Using a fairly high voltage toroidal (240v/3=80v secondary, 120v/3=40v), massively oversized, with the 'secondary' wired as the primary.
Would be interesting. Esp. if you exceed the insulation ratings (which are ~4000v IIRC).
James
Using a fairly high voltage toroidal (240v/3=80v secondary, 120v/3=40v), massively oversized, with the 'secondary' wired as the primary.
Would be interesting. Esp. if you exceed the insulation ratings (which are ~4000v IIRC).
James
Yes, that's an idea. A couple of thoughts: The high-frequency roll-off or peaking behavior of the power transformer's leakage inductance and shunt capacitance interacting with the ESL's capacitance will be a bit of a gamble. And, if used full-range, I wonder if the primary (formerly known as "secondary") inductance will be high enough. Low rp triodes might be in order. Still, an idea worth pursuing.
If you are into the transformer idea I think a 2*115V primary with two 115 secondaries wired at each end as a centertapped 1:2 autotransformer would be the thing to go for.
But my idea is still a bridged SRPP with triode in the bottom and a pentode at the top. As EL34 connected as pseudotriode can take 600V (if Vamax is 15W) I will try that for my 500Hz and up ESLs.
But my idea is still a bridged SRPP with triode in the bottom and a pentode at the top. As EL34 connected as pseudotriode can take 600V (if Vamax is 15W) I will try that for my 500Hz and up ESLs.
As I already have the step-up transformers (1:70) for my Scripts maybe these could be used together with triode-connected EL34s (at 600V with Wamax 15W). This is the idea:
www.revintage.se/ScriptEL34.pdf
In this bridged-mode the former primary could be used for feedback. What I don´t know is if the transformer can take the DC-current of 25mA/side?
www.revintage.se/ScriptEL34.pdf
In this bridged-mode the former primary could be used for feedback. What I don´t know is if the transformer can take the DC-current of 25mA/side?
Interesting. So, if I understand your plan, you intend to use the former secondary of the step-up transformer as a push-pull plate-load choke for the final tubes, right? I would expect it to be extremely sensitive to plate current imbalance due to the huge number of turns. Any imbalance is multiplied by thousands of turns. What is the DCR? Can it handle the IR dissipation?
Also, because your transformer is in the feedback loop, the flaws ("sound") of the transformer will be impressed upon the final output signal. A 70:1 transformer is far from an ideal transformer. I think the balanced plate choke idea has merit, but I might be tempted to scrap the transformer feedback and either have no feedback, or return resistive feedback to the grid of the EC99 in the lower right from the appropriate EL34 plate. There may be other "gotchas" that I haven't caught here yet, but I salute you for "thinking out of the box".
Also, because your transformer is in the feedback loop, the flaws ("sound") of the transformer will be impressed upon the final output signal. A 70:1 transformer is far from an ideal transformer. I think the balanced plate choke idea has merit, but I might be tempted to scrap the transformer feedback and either have no feedback, or return resistive feedback to the grid of the EC99 in the lower right from the appropriate EL34 plate. There may be other "gotchas" that I haven't caught here yet, but I salute you for "thinking out of the box".
Brian Beck said:
Thank you Brian, nice to have your input on this idea. About the "DCR" (not sure about this) I think you refer to the current source for the PI and it just handles about 20V and 20mA.
Maybe some kind of current-balancing device is needed for.
Another thought was to use 300B for this application but its Va is limited to 450V. Can it maybe take more if the current is only 20-30mA?
Cheap 6BX7s or 6BL7s could maybe also be used if parallelled and Va reduced to 500V.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.