• 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.

DD amp for ESLs...

HVfanatic

Member
2008-02-11 7:21 pm
Hi,

I've seen that some of you built DD amps.

My starting point were the Acoustat amps.

1. Why do they use SRPP and not ß-Followers? (Advantage / Disadvantage)

2. Are there any triodes around instead of TV tubes? Does this make sense? Are there triodes (not the 833!!!) that withstand 5KV?

3. Does anybody have good experience (sound/stability) with these Amps? Can somebody send me the schematics of his Amp???

Regs, Dirk
 
There has been discussion on this subject before..

diyaudio

I rebuild this amps nowadays with a tube frontend.
The only weak point, as mentioned before, is the HV transformer and sometimes the 6HB5 tubes.
The HV transformer can be replaced by a new and better one.
If a HV delay is built in, which kicks in the HV when all tubes are heated up, the 6HB5 tubes will be more stable.
There are triode replacements, but the disadvantage is the low gain.
The SRPP, altough not the best choice, works well in this arrangement.
If you like, I can provide a better schematic than in the link above.
 

HVfanatic

Member
2008-02-11 7:21 pm
Thanks a lot!

I already got the schematics. I also think, an all tube solution is superior. Did you only replace the op-amps? Or did you also replace the driver transistors?

I found Sashas rebuilt. Looks quite promising. My Idea would be, to use a triode differential stage (cascode like paravicini?) and a cathode follower as driver for the TV tubes.

What was your approach?

The acoustat looks strange to me in the elements C22/R32,
R41,C26

Is this some kind of phase alignment due to capacitive load???

How did you manage the task to prevent oszilation???

Would you prefer Sashas tubes? Or goto PL519???

Regs, Dirk
 
Hi Dirk,

yes, the driver transistors are also replaced to make it an all tube circuit.
Sashas rebuilt is indeed good, his tubes are better and easier to get, and cheaper to. PL519 is indeed also very usable, I tried it to.

My approach is a differential cascode, driven by a current source.
No need for the cathode followers as the penthodes are relative easy to drive.
There is no oscillation, but I reduced the NFB to about 10dB overall.

Don't forget the filter at the input. I use a passive HF filter, adjustable with the original HF balance pot.

Regs, Dick
 

HVfanatic

Member
2008-02-11 7:21 pm
Hi Dick,

Do you mean with "Cascode Differential Amp" that you have 2 stage differential??? Or that you use a Cascoded Differential Amp (1 stage???) Which tubes do you use???

I think of the following topology:

Output stage: PL519, mu follower like Allan Kimmel (upper Tube as Pentode, lower tube as "screen drive" Triode

Driver: Mosfet as follower instead of Cathode follower.

Input: Two stage differential Amp, cascoded with Fets for better Bandwith, lower Miller and more amplification. Both stages should get a bipolar current source.

Should the driver also operate with a current source in your opinion?

Where do you see the shortcomings in this design idea???

Regs, Dirk
 
Hi Dirk,

With Cascode diff amp I mean one stage, i.e. 2 cascodes where each cascode is one side of the diff. Amp.
I use ECC88 tubes, which have more than sufficient gain .

I think your topology works fine, but you don’t need the second diff amp.
The current source is not needed for the driver, as all stages are diff already.

Don’t forget a delay for the HV and be very carefull, these voltages are deadly.

Regs, Dick
 

HVfanatic

Member
2008-02-11 7:21 pm
Hi Dick,

I understand. But: If I look at your design, you use 3 tubes!???
Supposed the tube is a double triode, you would only need 2 tubes for the differential cascode. What is the third tube used for???

Regs, Dirk

P.S.: Funny: Paravicini uses the cascode solution as well, but with a cathode follower for driving the Pl509 in triode mode...
 

HVfanatic

Member
2008-02-11 7:21 pm
Hi,

I got it. You use the third tube for equalization circuitry??? What I read in other threads, that the HF needs damping. Looks strange to me, because an audiostatic has already compensation after the transformer... (?). Well, I made some thoughts about the circuit. Besides the HV, I also need heaters. These should be separate trannies - I guess I need 4 of them, center tapped, where the center tap is connected to the cathode. Wouldn't the capacitance between pimary and secondary harm me???? What type of transformer do I need to get the lowest coupling possible there???

I found another problem: If I want to use a pl519 sceen drive (triode like, lower tube), I only find data sheets to draw my loadline for less than 1000 Volts... Where can I get appropriate data? I'm not experienced with screendrive. Are there recommended values to set the grid voltage, and then fixing the bias with the screen (superimposing the signal voltage???) Or: Put the signal to the screen, no matter (within allowed ratings!) DC, and adjust bias with appropriate DC on the grid???

Regs, Dirk
 
Hi,

I like the circuit topology of Neil McKean and similar, using a current source loaded FET-stage.
My idea is to use a JFET-input stage and a MOSFET-driver, coupling to a HV-Triode( >4kV possible) working into a (voltage controlled) current source made from cascaded MOSFETs. This should give the triode the highest possible load (linearity) without the need of a floating heater for the upper tube and maybe even the possibility to get rid of the global feedback. At the moment I´m still a bit puzzled about the cascaded FET-current-source. Anyone schem suggestions?
The large amount of feedback and compensation is what I dislike about the Acoustat-amp.

jauu
Calvin
 

HVfanatic

Member
2008-02-11 7:21 pm
Hi,

the SRPP advantage is, that it is push-pull (less power wastage...)
The Fet proposal is nice, but limited voltage brings it down. I only know designs that have Fets in series with resistance voltage dividers. So you can not couple in a good manner. This means: Low bandwith. Do you have any idea (schematic!) how to build a current source you can modulate with Fets that does not suffer the described problems? Otherwise, I think we have to stick to a HV pentode as current source and build something like Alan Kimmels "power mu stage"...
Somebody outside that knows where to get / how to build a very low capacitance heater transformer???

Regs, Dirk

@Calvin: For voltages not exceedind 1300V a Fet solution looks appropriate...
@Dick: How can you solve the heater trouble with a single transformer? Wouldn't a strong capacitive coupling mean a "low load" for the lower tube??? - loss in amplification of the SRPP stage???
 
Hi Dirk,

Don’t worry to much about the heater transformer.
Don’t forget, de heater does not have to be connected to the cathode, as long as the potential differance does not exceed the limits !
I connected the heater with the cathode through a 10k 2W carbon resistor.
Look for a transformer with 5kV isolation between primary and secondary and you are done.
For each “upper” tube you need such a transformer.
When both “lower” 6HB5 share the same connection for the cathode , their heaters can be powered from the same transformer.

For 2 PL519 in SRPP penthode connected, powered from 5kV, use a G2-kathode voltage of 50V for both tubes.
Connect 100 ohm 2W carbon to the G2 grid.
Bias them at 12mA with adjustable negative G1 voltage on “bottom” Pl519. Use a pot between 0 and –50V.

The EQ I use is passive right after the input.

The original Acoustat X has indeed a lot of feedback and compensation.
I use no compensation and only 10dB of feedback. Works great.

I have seen a cascaded Mosfet HV amp, but it also dissipated a lot of heat and when demonstrated it smoked and died. Tubes are a lot easier to work with and they can withstand some abuse.

Regs, Dick
 

HVfanatic

Member
2008-02-11 7:21 pm
Hi Dick,

thanks a lot for your help. Now it's time for me to draw a first schematic which I will post here (takes some time... ) I think, I will draw 2 or 3 solutions, which I will post. I will highly appreciate your valuable opinion (and all others as well ;-))

Did you make distortion measurement with your amp? Would be very interesting...

Regs, Dirk

P.S.: Do you still have contact with sasha? I saw, he tried triodes instead of pentodes... Got some feedback?
 
Hi,

well SRPP is not necessarily pushpull. At least it ceases beeing balanced pushpull when the load impedance is not right. And in our case we have a widely varying impedance. So SRPP is imho not a good way to go for ESLs. Thats why I´d prefer a Kimmel-style current source loaded Triode-stage. I know about the voltage vise limitations of FETs and the risks of not properly cascaded ones. And Yes, it´d be quite easy to implement the ccs with a Pentode, but I like the idea of less heaters and higher linearity because of higher FET-ccs-impedance.

jauu
Calvin
 
djmiddelkoop said:
Hi Dirk,

Don’t worry to much about the heater transformer.
Don’t forget, de heater does not have to be connected to the cathode, as long as the potential differance does not exceed the limits !
I connected the heater with the cathode through a 10k 2W carbon resistor.
Look for a transformer with 5kV isolation between primary and secondary and you are done.
For each “upper” tube you need such a transformer.
When both “lower” 6HB5 share the same connection for the cathode , their heaters can be powered from the same transformer.

For 2 PL519 in SRPP penthode connected, powered from 5kV, use a G2-kathode voltage of 50V for both tubes.
Connect 100 ohm 2W carbon to the G2 grid.
Bias them at 12mA with adjustable negative G1 voltage on “bottom” Pl519. Use a pot between 0 and –50V.

The EQ I use is passive right after the input.

The original Acoustat X has indeed a lot of feedback and compensation.
I use no compensation and only 10dB of feedback. Works great.

I have seen a cascaded Mosfet HV amp, but it also dissipated a lot of heat and when demonstrated it smoked and died. Tubes are a lot easier to work with and they can withstand some abuse.

Regs, Dick


But Dick, the capacitance from sec to prim of that upper tube heater transformer can easily be several 100pF. That *is* a load on the amp that takes away usefull output current, isn't it?

Wouldn't it be a good idea to select xformers also on low capacitive coupling? I know that for instance in test amplifiers for high-voltage the upper heater xformer is a special one with the prim and sec at separate legs of the core.

Jan Didden
 
Hi Jan,
thanks for your comments.
But Dick, the capacitance from sec to prim of that upper tube heater transformer can easily be several 100pF
Yes, easily.
That *is* a load on the amp that takes away usefull output current, isn't it?
Yes, right again.
That's why I connect the heater to the cathode via a 10k resistor to isolate this capacitance. The value of 10k can easily be made 100k, it's just chosen because I had these resistors in stock.
Wouldn't it be a good idea to select xformers also on low capacitive coupling?
Yes, a good idea.

The tubes Sasha uses are a better choice than the stock 6HB5.
I didn't make distortion measurements, yet.

Regs, Dick
 
djmiddelkoop said:
[snip]That's why I connect the heater to the cathode via a 10k resistor to isolate this capacitance. The value of 10k can easily be made 100k, it's just chosen because I had these resistors in stock.[snip]Regs, Dick


Dick,

That seems a very smart thing to do. I haven't seen it done before. I have been looking at the circuits of the Fluke 5215A where they use very special, very low cap transformers for the upper final tube. Can it be that a simple resistor takes care of this??

Jan Didden
 
Well, yes.
The stray capacitance of the heater xformer is in parallel of the load, which is a capacitor to ( the ESL ).
By connecting this stray capacitance with a large series resistor it does not load the amp. stage anymore, while the voltage potential of the heater still follows the cathode, hence the heater-cathode voltage limit is not exceeded.
Anyway, it seems to work fine in the several HV amps I've made.

Regs,
Dick
 
djmiddelkoop said:
Well, yes.
The stray capacitance of the heater xformer is in parallel of the load, which is a capacitor to ( the ESL ).
By connecting this stray capacitance with a large series resistor it does not load the amp. stage anymore, while the voltage potential of the heater still follows the cathode, hence the heater-cathode voltage limit is not exceeded.
Anyway, it seems to work fine in the several HV amps I've made.

Regs,
Dick



OK. Not to want splitting hairs, just trying to make sure I get it:


There now is a lpf from cathode to heater. So, with increasing frequency, I would expect to see an increasing voltage difference between the upper tube cathode and its heater. Depending on the allowed cathode-heater voltage, one must take care that the lpf (10k resistor and the xformer cap) does not cause this max to be exceeded at max frequency and max output level.
Am I correct so far?

Jan Didden
 

HVfanatic

Member
2008-02-11 7:21 pm
Hi,

first I want to comment on Jan: I guess, that Dicks solution will exceed the heater cathode rating at HF, but luckily level will be low and nothing goes wrong...

So, I did some drawings:

A: "Dick pentode amp"
B: "Dirk screen drive beta follower amp"
C: "Alan Kimmel design mu stage"

What do you think of it? What's wrong? Should I replace the CF with fets? Should I even go for fets instead of the upper ECC88???

Better ideas???

For the transformers:I guess I take "X" types... Jan: Where can I get this stuff???

Regs, Dirk