Indestructible HV hybrid regulator!

[panos29]

After months of experimentation and trial and error, I finally managed to arrive at a working prototype of a comparatively new realisation of hybrid regulator circuit!

It is based on a HV opamp made by texas, OPA445. The prototype is capable of 1 A, yes 1A at 220V with less than 1,5mV ripple fully loaded (resistive load), which is more like random noise on my scope. Of course 1A is way too much but the tube is powerfull enough to withstand such a huge load for at least 10 seconds.

Anyway, one problem I noticed is that when the speed-up capacitor (470n) is out of the circuit the regulator is oscilating producing a steady 5-6KHz oscillation.

With the speed up bypass capacitor in the circuit it is completely stable and there is no measurable drop (some mV) in output voltage from no load to full load.

However, testing this regulator in actual circuits I noticed that when used this in powering a 6W SE tube amplifier at 238V/200ma at full output power, is that the AC component on the power rails, when sweeping from 5Hz to 20KHz is not uniform, but changes from 1mV@5Hz-1Khz to 30mV@20KHz.
Is this non uniform responce acceptable, or the regulator is suffering some kind of problem, is there any way I can fix this?

Finally, I really like this reg as benchtop power supply which can easily power almost anything, as I even tested this with a direct short and nothing happened (except a spectacular big bang)!

Any comments or suggestions are more than welcome.

[Eli Duttman]

The LARGE capacitance following the bridge rectifier causes considerable ripple freq. overtone energy to be present. Kill that "hash" by inserting a high current RF choke in the line leading to the 6C33. Bypassing the 220 KOhm resistor with a silvered mica cap. could help too.

[timpert]

Hi,

The observed increase in AC component at the power rail is to be expected when using an opamp as the error amplifier in a regulator. The vast majority of opamps have what's called a dominant pole, which causes the gain of the amp to drop proportionally to 1/f from a low frequency (typically around 10 Hz onwards). This is required to keep it stable in a feedback loop. The consequence is that the error gain decreases, and thus the load regulation. This is normal.

The regulator might oscillate due to the presence of the 4u7 capacitor, shorting out the bias current source at high frequencies. I think this defeats the purpose of the current source at HF, especially a cascoded one like you use. You have made a capacitive voltage divider with a ratio of roughly 1:20 with the 100 uF and 4.7 uF caps, injecting the AC on the output attenuated by a factor of 40 (the voltage divider formed by the 330k resistors and diodes) on the inverting input. Also, the negative supply pin gets this 1:20 attenuated AC on it. I can't quickly tell what happens on the output as a result.

In your regulator there are a lot more feedback paths present than you might initially realize. Eliminating them is desirable in order to obtain a well controlled behavior, removing the 4u7 cap might be a step in the right direction. Also, putting a 100nF decoupling cap on the inverting input of the opamp (between -in and the regulator output) greatly reduces the injection of output AC on the -in.

Good luck!

[panos29]

Thanks for the answers!

I have tried without the 4u7 capacitor and again I got oscilation. Same happens when i connect any capacitor from 100p to 100n to the inverting input. Maybe I got to try a simpler current source like 10M45s i got available?



Moreover, I gave the regulator a long listening period of more than 2 months. Compared to a simple regulator (with 6C33 and 6AU6-5651) on the same amp, the tubed reg sounds more "human" and liquid with a lot more harmonics to the music while the opamp regulator sounds sterile, flat and lacking in dynamics.

[timpert]

Hi,

I think the best configuration without extensive mods is to omit the 4u7 cap, put the 100n cap in (between inverting input of the opamp and regulator out), and leave the 470 nF cap in place. There's a whole lot to experiment with your setup, but as nobody here knows what iterations you've already gone through, it might become a dreadful experience when discussing this online with outsiders. If all works well, you are happy with the results and nothing is smoking: stop bothering and enjoy your work!

One general remark related to almost all cases of oscillation in amps and regulators: often only a schematic gets posted and commented here. However, even a perfectly good topology can often be made to oscillate by some unfortunate layout choices. Without layout info, there really is insufficient debug data!

[panos29]

You are very correct!

All your comments are spot on! I arrived to the present layout after much experimentation and trial and error, having said that it does not mean that there is no room for upgrade, the opposite. I will try the two mods you proposed and also try 10M45 CCS to see what happens.

What I really like with this regulator is that it got super stable output as far as regulation is concerned, very high current capabilities, it is almost imune to accidental destruction and on top it is kinda prototype in its realisation as I have never seen something similar yet.