• 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.

Corona: An Ultra-Low Distortion A2 DHT SE Amp Prototype

With a high gain circuit, small changes in any grid or cathode voltages will result in drift for sure. It feels like you will likely need a DC servo or a DC feedback loop to keep things stable. You may want to build and see. I’d add the screen DC feedback as suggested, you won’t go wrong and have also a stable Screen supply; so you will kill two birds with one stone
Cheers
Ale
 
In all of the amps I have ever built, I have included a circuit that monitors output tube cathode current and shuts the amplifier down if the average current exceeds a threshold. In my experience, the time constant doesn't have to be very long. This successfully protected all of the electronics in the amplifier I built for my brother when a KT88 failed and started arcing regularly. Never even blew the OT fuse.

This is personally how I would protect the 801As. I'm averse to series resistance in the grid driver of an A2 amplifier because it is a significant distortion generator.

But there are many ways to skin this cat. If you keep the series resistance, the feedback will try to compensate for the voltage drop across the resistor, but it will never be able to get distortion down to where it would be without it. Also, be aware that when the voltage drop across the resistor starts becoming significant, the plate of the driver will drive even harder to make up for it. You have to plan for that in choosing the power rating of the resistor. If your system doesn't spend much time in the positive grid drive region because of sensitive speakers it may not be much of an issue.
 
In regard to the stability of Vg2 in the CCS fed pentode stage, can the screen voltage simply be regulated? Since it was suggested to lower the source follower B+ to spare the 801A grid, this Maida regulated supply could easily be lowered to 100V, decoupled, and fed to the EF37A screen as well, fixing it to 100V, resolving two issues in one go. Too simple perhaps? I still have much to learn.

In my current breadboard setup, I have a series resistor/zener string from main B+ with a cap bypass. That worked well.

A dual-purpose supply is always a good thing, in my book.
 
Member
Joined 2019
Paid Member
Thank you gentlemen, I'll give your suggestions some thought and report back.

I do have a few square waves from the amplifier as originally built in my first post. The measurements were taken in haste, I will be more thorough later this week.

Here is a 1kHz square wave from the OPT secondary. There is some ringing, perhaps from the OPT leakage inductance (?). I will compare primary to secondary on my next round of measurements.

bn0CxJm.png


And 10kHz sine wave with 33pF in parallel with the 50K FB resistor


nktQKBI.png


ZDxBxUh.png


I measured the 10kHz square with two 33pF capacitors in series to see if I could get faster edges, but it results in ringing.

WC3bAnS.png


Below is the FR, there is some HF rolloff, but much improved with the addition of NFB.

bzAGWd1.png


The first version of this amplifier used a CCS loaded 841 as the input and the LL9202 OPT was wired for 11K:8 as opposed to 6.5K:8 in the NFB implementation. This resulted in significant HF rolloff and an output impedance of ~4ohms, which prompted the NFB experiments.
 
Last edited:
Member
Joined 2004
Paid Member
The HF behaviour is due to the high ratio OPT.
Very few 10-11k OPT goes further than 20-25kHz ... Lundahl is not among them.

The figure below shows similar amplifier's (CCS loaded 841, source follower, PSE 801a, 5k:8 custom wound OPT, 1W) freq. response.
As you can see, the medium ratio transformer easily reached 50-60kHz HF bandwidth (at 8W 50kHz).
 

Attachments

  • 801a PSE 1W sweep.jpg
    801a PSE 1W sweep.jpg
    170.4 KB · Views: 145
Member
Joined 2019
Paid Member
Yes, the bandwidth is compromised by the high internal resistance of the 801A combined with the parasitic inductance / capacitance of the high impedance OPT. I had hoped to reach 20kHz, -1dB @ 20kHz would have been acceptable, but that was optimistic. Using a medium ratio OPT requires a PSE output stage, or else the output impedance is too high to damp my 8ohm speakers. You have answered many of my questions euro21 :) prototyping this amplifer has been very educational, I now understand the pitfalls of a high ratio OPT.
 
Member
Joined 2019
Paid Member
No not yet sorry, I plan to on the next go-around of measurements which will be more thorough, have been very busy lately unfortunately. I am waiting for some parts to arrive Wednesday for a few minor circuit changes, will have more variety of cap values on hand for the FB loop as well. This is from the Lundahl LL9202 50mA 100H wired for 6.5K:8.
 
Member
Joined 2019
Paid Member
I found some time to take more thorough measurements of this plate-to-cathode FB A2 design, I'd appreciate any thoughts / input.

First off, with what I had on hand, I decoupled the V+ to V- supplies of the source follower / source CCS with a 3.3uF film capacitor. I made some minor bias / feedback adjustments to the EF37A driver as well, with an approximate 190V / 100Vg2 / 3.3Vk bias point.

Here are square waves at ~1W output with 33pF in parallel with the 50K feedback resistor. The ringing artifact does not seem to be caused by the Lundahl LL9202 OPT, it is present on the primary.

100Hz primary

RHeW9FZ.png


100Hz secondary

awc3LGt.png


1kHz primary

jXG21HI.png


1kHz secondary

slQobdX.png


10kHz primary

KDc7UFF.png


10kHz secondary

IaRE6Kx.png


I checked a 1kHz square wave at the primary with no feedback compensation capacitor, I didn't pursue any other square wave measurements as it appears to be unstable.

uK24DSR.png


Seems something is oscillating / ringing, tamed by adding some capacitance to the FB loop, but not altogether fixed.

Unfortunately, due to the limited HF bandwidth of the Lundahl LL9202 50mA, even adding a small amount of capacitance to the FB loop compromises the already compromised high frequency response.

FR with no feedback compensation capacitance: 20kHz -1dB

tgpITvC.png


FR with 33pF feedback compensation: 20kHz at -1.45dB

PbpuZH1.png


Here is THD at 1W with 33pF feedback compensation: 0.083%. Noise floor is quite ugly, due at least in part to the prototype being built on a piece of plywood with long sprawling alligator clips every which way.

MwMsK6A.png


So where to go from here? Some thoughts.

1) I need to identify the cause of the ringing / oscillation seen on my primary / secondary square waves with 33pF FB capacitance.

2) I should consider changing my OPT, perhaps to a 5K primary with better bandwidth. The limited bandwidth of the Lundahl LL9202 50mA has been pretty disappointing and I cannot spare any high frequency response to stabilize the feedback loop. A shame, as the transformers sound fantastic otherwise.

3) I am seeing very stable screen voltages with a 1Meg resistor from my Maida regulated B+. At very high outputs, the screen voltage is rock solid, I see a drift in my EF37A plate voltage of a few volts close to clipping on the 801A output stage.
 
Last edited:
I'd spend some time looking more at the first stage. How does it perform with the output tube removed from the circuit? It's going to have a lot of gain, but how do square waves look there?

When I was experimenting with CCS-loaded power pentodes and shunt feedback, I built two CCSs, one good for 900V and another good for 1700V. I noticed that the 900V CCS (10M90S/DN2540 cascode) was much better behaved at high frequencies than the 1700V CCS (IXYS 1700V part/DN2540 cascode). I was getting 400kHz out of the circuit with the 900V CCS, no ringing on square waves, and no capacitor in parallel with the feedback cap. With the 1700V CCS, I got ringing on square waves and had to put a cap across the feedback resistor and restrict bandwidth to 200kHz.

I guess what I'm trying to say is that some devices are better than others in the CCS, so I'd be curious to see how the devices you are using are performing. In my 826 amp, I'm using the 10M90S/DN2540 cascode since that combo performs so well.

I also keep looking at the screen supply in the first stage. When the amp is presented with a square wave, we will see sudden jumps in voltage at the cathode of the input stage which will couple to the screen through the capacitor. There is a time constant there and I wonder if that is coming into play a little bit? I've referenced my screen supply to GND, which gives a tiny UL effect and reduces gain a bit. I like the idea of referencing the screen to the cathode, but maybe it is messing with stability a bit? Or maybe not. I don't imagine there is a lot of gain for a screen voltage change to affect things. It'd be easy to try the other way, anyway.
 
Member
Joined 2019
Paid Member
Thanks SS, that's good advice. I've been using the IXYS IXTP08N100D2 and IXTP08N50D2 as a cascode CCS in other designs, but none have been this finnicky and prone to instability. I'll look into it and probably try some other FETs.

I did try referencing the screen to ground rather than the cathode in the EF37A since it is an easy change, it looks to have actually exacerbated the ringing and some loss of the screen voltage stability, doesn't seem to be the culprit. I'll isolate the pentode stage and dissect it this weekend, take a look at some square waves and see if the instability resides there.

Yeah the 1Meg screen resistor will likely need to be changed to a more robust supply, trying to find the time to have to look into Ale's suggestion using the DC feedback loop, no doubt that would be an improvement.

I appreciate the help!
 
Nice design, I like it. :)
I still have driver boards fabricated for the Gubernator-71 that was very similar. But the first pentode was not CCS loaded, it had feedback divider in it's anode. The driver tube was 12LH7, the output tube was GK-71 with 850 V on anode, 450V G2, and 56V G3. The same 56V was used for the drain of the driving MOSFET loaded on a CCS. The amp produced 45W on onset of soft clipping.
 
Nice design, I like it. :)
I still have driver boards fabricated for the Gubernator-71 that was very similar. But the first pentode was not CCS loaded, it had feedback divider in it's anode. The driver tube was 12LH7, the output tube was GK-71 with 850 V on anode, 450V G2, and 56V G3. The same 56V was used for the drain of the driving MOSFET loaded on a CCS. The amp produced 45W on onset of soft clipping.

Is the Gubernator-71 a commercial product you are keeping a lid on or do you have a schematic released to the public somewhere?

GK-71 looks like a fun tube! I've thought about buying a Chinese 211 to adapt to this amp and see how it goes. I think it would work really well. Of course, I'd have to buy some bigger output transformers first.
 
Is the Gubernator-71 a commercial product you are keeping a lid on or do you have a schematic released to the public somewhere?

GK-71 looks like a fun tube! I've thought about buying a Chinese 211 to adapt to this amp and see how it goes. I think it would work really well. Of course, I'd have to buy some bigger output transformers first.

Well; it supposed to be a commercial product, but I don't think that I will manufacture it in the nearest future. I do not have a complete schematics, just a fragment. I hope it helps.
 

Attachments

  • Screenshot from 2020-09-01 21-54-30.png
    Screenshot from 2020-09-01 21-54-30.png
    88.5 KB · Views: 665
Well; it supposed to be a commercial product, but I don't think that I will manufacture it in the nearest future. I do not have a complete schematics, just a fragment. I hope it helps.

Hi Anatoliy,
Very clever design, hats off. I bet it sounds great! I like the way you feed the follower from the G3 supply. How did you derive the G3 voltage level needed (e.g. measured, listening)?
Cheers
Ale