power supply for my single ended amp

[jhaible]

I'm designing the PSU for my SE KT66 amp. 2 channels, approx. 150mA, Ht something between 440V (max. for fixed bias) and 480V (max. for cathode bias).

I want the PSU to have a low output impedance down to low frequencies, and I want superb smoothing.

I don't want a voltage regulator, though, because of this:
When the mains voltage gets a little low, the regulator might loose its "headroom" and lets a nasty ripple thru to its output,
or - if designed with lots of headroom - I'm wasting a lot of power under normal conditions.

So the idea is to use a source follower, with a passive 3-pole filter at its gate for good smoothing. This control voltage is derived from the reservoir cap with a voltage divider (trimpot integrated in the bleeder resistor string). So when the mains voltage gets low, the HT will also get low, but will still be smooth.

The first draft of the circuit is here:
http://jhaible.heim.at/se_kt66/se_kt66_psu_idea.pdf

Please let me know if this is a good idea, or what should better be different. It simulates good, but I'd love to get some feedback before I actually build it. I'm a little worried about the high impedance of the filter chain. In theory the MOSFET has a high enough gate impedance, of course, but maybe some parasitic effects will ruin its behaviour?
In simulation, I get the ripple down to a few millivolts, which seems too good to be true.

The lower part is a delay for the HT, derived from the heater voltage. I've breadboarded that part - works great. Gives a delay of 30 seconds. Is this long enough to prevent cathode stripping?

Comments welcome.

JH.

[yagoolar]

1. How does it respond to fast voltage spikes in the power line?
2. How fast does the voltage on the output rises after switch-on?

--Yag

[jaymanaa]

Hi, Your last cap is labelled "motor start". I have never used one, but have heard much negative talk about them. Maybe a motor "run" oil capacitor would be better. Nice delay circuit, I think 30 seconds will be fine. Jay

[jhaible]

Originally posted by yagoolar
1. How does it respond to fast voltage spikes in the power line?

Ahh ... I never thought about that!
Thanks for pointing it out.

What exactly do I have to watch out for? The spikes bypassing the FET via parasitic capacitance?

2. How fast does the voltage on the output rises after switch-on?
--Yag

After a delay of 30 seconds (relay), it rises up in ca. 500ms (gate filter).

Is this too long? (overheating of the FET ?)
Or too short? (thumping in the speakers ?)

JH.

[jhaible]

Originally posted by jaymanaa
Hi, Your last cap is labelled "motor start". I have never used one, but have heard much negative talk about them. Maybe a motor "run" oil capacitor would be better. Nice delay circuit, I think 30 seconds will be fine. Jay

I have browsed thru the forum, read about "start" and "run", but thought it was just different descriptions of the same thing - my bad.

I have no idea for which application it is - the description just read "Motor Kondensator".

Here is what I bought:

link

It's polypropylene - I thought this was good ...

JH.

[yagoolar]

Originally posted by jhaible


Ahh ... I never thought about that!
Thanks for pointing it out.

What exactly do I have to watch out for? The spikes bypassing the FET via parasitic capacitance?





After a delay of 30 seconds (relay), it rises up in ca. 500ms (gate filter).

Is this too long? (overheating of the FET ?)
Or too short? (thumping in the speakers ?)

JH.

AFA spikes are concerned watch for very narrow ones which appear on the output. And they may also origin from rectifiers while recovering.

Overheating is a matter of UxI on turn-on, so you should check both current and voltage characteristics in time domain - the most power is emitted at the intersection of U and I. I hope you know what I mean.

[jhaible]

Originally posted by yagoolar


AFA spikes are concerned watch for very narrow ones which appear on the output. And they may also origin from rectifiers while recovering.

Overheating is a matter of UxI on turn-on, so you should check both current and voltage characteristics in time domain - the most power is emitted at the intersection of U and I. I hope you know what I mean.

I know what you mean in general.
And I know how to calculate a heatsink for permanent power dissipation. What I don't know is how much I can undersize the heatsink for a much higher power dissipation that only lasts for half a second.

Let me see:

I may have 50V * 150mA = 7.5 W permanently.

On power-on, very roughly, 300V * 150mA = 45 W, but only for
half a second. (45 W is the maximum power from spec - I guess it means maximum permanent power, but with an "ideal" heatsink.)

I intend to fix the MOSFET (Rth to case = 3K/Watt) directly to the enclosure, which will give a few more K/W in total. Should be enough for the permanent power. What I don't know is whether the short time of higher power will hurt or not.

JH.

[dhaen]

Short term dissapation is a non-issue. Things take a long time to heat up (and cool down) - thermal inertia.

Your power supply design is an interesting concept which I will watch with interest.

[Klimon]

I have browsed thru the forum, read about "start" and "run", but thought it was just different descriptions of the same thing - my bad.

Two words for the same thing; polypropylene (or paper) in oil = very good.

Simon

[jhaible]

Originally posted by Klimon

Two words for the same thing
Simon

I see - thanks!

JH.