1. Home
  2. Electronics Design
  3. Power supply puzzle

Power supply puzzle

Curious if the assembled gang can offer some suggestions for a power supply design.

The existing design includes a power transformer with outputs at around 460VAC RMS. This feeds into a vacuum tube rectifier and on into a pi (CLC) filter and on into a few RC sections. The power transformer mains are center tapped with the tap grounded.

Now I need -100VDC and +200VDC supplies as well. They'll feed to the cathode and plate of a vacuum tube cathode follower circuit. The key is to get the -100VDC onto the tube's cathode without messing anything else up.

The first thought was to pick the +200VDC off of the end of the main B+ supply (just add another RC section to drop down to

200VDC) and to develop the -100VDC with a simple full wave rectifier made from a pair of diodes mounted cathode toward the high voltage supply (connected before the tube rectifier). Turns out that, in the models at least, this fails miserably because of interaction between the + and - supplies.

The solution at present uses a second diode pair (anodes to the HV supply) to generate the +200V independent of the main B+. This seems to work well in the models. Down stream of the diode pair is a simple CRC setup -- the resistor to drop voltage down to the required level and a pair of caps to smooth things out a bit.

The current demand on both -100 and +200 supplies is expected to average 15mA (so we'll design for, say, 20mA).

There are all kinds of problems with the design as it stands -- not the least of which are that the series resistor in the

100/200 supplies has to drop hundreds of volts (I'm looking at 10W parts here) and if you operate the circuit with the tube removed, the current drops to essentially zero so the series dropping resistor doesn't drop and the filter caps have to be rated to withstand a huge no-load voltage -- so I'm looking at pairs of 350V caps in series and a huge amount of circuit real estate.

I've considered some other setups (combinations of series/shunt resistance to keep the no-load voltage low -- smaller/cheaper caps, but still big wattage resistors, a stack of zeners in series with a tap close to the required voltage, a second power transformer, a custom wound power transformer), but none seem elegant or cost effective.

So... If you had 460VAC already feeding a tube rectified DC supply and needed to add a +200VDC and -100VDC at 20mA, how would you do it? Any ideas would be greatly appreciated. -Todd

Reply to
Wakarusa
Loading thread data ...

In message , dated Wed, 13 Sep

2006, Wakarusa writes

It should not. You have something wrong somewhere. Are you using tube rectifiers? Put a schematic on a.b.s.e

For only 15 mA, you could use a half-wave rectifier fed from one side of the transformer secondary winding.

--
OOO - Own Opinions Only. Try www.jmwa.demon.co.uk and www.isce.org.uk
There are benefits from being irrational - just ask the square root of 2.
John Woodgate, J M Woodgate and Associates, Rayleigh, Essex UK
Reply to
John Woodgate

Since most mains isn't 2 phase what exactly are you going on about ?

You use a transformer with multiple or tapped secondaries.

What's the problem ?

Thus is trivially simple.

Graham

Reply to
Eeyore

In message , dated Wed, 13 Sep

2006, Eeyore writes

By 'mains' he means 'mains transformer secondary'.

[big snip]

He hasn't got one, and he doesn't need one. It's very simple to get lots of different voltages from one transformer. You could quite easily get 3 kV DC at low current from that transformer if you wanted it.

--
OOO - Own Opinions Only. Try www.jmwa.demon.co.uk and www.isce.org.uk
There are benefits from being irrational - just ask the square root of 2.
John Woodgate, J M Woodgate and Associates, Rayleigh, Essex UK
Reply to
John Woodgate

John Woodgate wrote in news:8h46v5Qnb+ snipped-for-privacy@jmwa.demon.co.uk:

Correct.

Also true. I'm looking for a simpler or more efficient solution than the one I have. The hope is that I'm missing something obvious and there's some clever way to avoid using the high wattage resistors and high voltage filter caps.

Reply to
Wakarusa

In message , dated Wed, 13 Sep

2006, Wakarusa writes

Please put your current schematic on alt.binaries.schematics.electronic

We can't help if we have insufficient input data.

--
OOO - Own Opinions Only. Try www.jmwa.demon.co.uk and www.isce.org.uk
There are benefits from being irrational - just ask the square root of 2.
John Woodgate, J M Woodgate and Associates, Rayleigh, Essex UK
Reply to
John Woodgate

John Woodgate wrote in news:vFM+ snipped-for-privacy@jmwa.demon.co.uk:

Did some more work on getting the +200VDC from the main B+ supply (by adding another RC section on the end) and doing a more thorough job of getting the R values right for the voltage divider. This works (and eliminates the separate +200VDC supply), but has some nasty start up characteristics. If I'm understanding this correctly, the issue is that the -100VDC supply, having a simple CRC filter, establishes its voltage much sooner than the

+200VDC supply at the end of CLC-RC-RC-RC. While the +200VDC is waiting for all of the preceding section's caps to charge up, the

-100VDC is already there and enough to pull what ought to be

+200VDC negative for a short period (about 120mS). The current model shows the +200VDC supply hitting a minimum of -5VDC during startup - probably not enough reverse voltage on the +200VDC electrolytics to cause troubles. The real problem now is that adding an extra 15-20mA of draw to the main B+ supply requires higher wattage resistors in the preceding RC stages -- so it's a wash between this method and the separate +200VDC.

To answer your question directly, the main B+ supply uses a tube rectifier, everything else is silicon.

The cathode followers that need the +200/-100 supply are part of an audio circuit. They're a low Zo driver between phase inverter and power tubes to allow for class AB2 operation. Seems that I'd need significantly improved filtering in a 1/2 wave rectified supply, but kudos to you for a very simple way to get the voltage down. I'll experiment with this later today. Wondering if pulling the -100VDC off of one transformer secondary and the +200VDC off of the other (both 1/2 wave rectified) will end up with ripple voltages in opposition that'll cancel each other out. If this works it should also keep the currents balanced in the transformer secondary.

Reply to
Wakarusa

One option: dump the tube and use a FWB from end to end. Rub: CT'd transformer has twice the end-to-end voltage you need.

It seems to me you WILL need regulation on that +200V rail, so you have little choice but to regulate it anyway. Even a resistor follower (i.e., resistor divider plus cathode or emitter or source follower) will provide much more regulation than a lone series resistor, and allow you to operate at lower bias current to boot (going easier on the -100V rail too).

Tim

-- Deep Fryer: a very philosophical monk. Website:

formatting link

Reply to
Tim Williams

This would be easy for a little switchmode suppply running from rectified filament voltage, but perhaps you don't want the switching noise or would rather stay completely with tubes. Tube-based buck PWM is certainly possible but would take too much space.

A simple off-the-shelf solution: get two 241-4-xx transformers from Signal Transformer, each with dual 117VAC primaries. Ignore the secondaries. Drive one primary of each from line. Take 117VAC off one second primary for your -100-volt supply, connect second primaries in series to get your 200 volt supply. These xfmrs are 1.37" x 1.25" x

2.375".

You could use beam power pentodes as series pass regulators, but that would take more space than the two little xfmrs. You could use

800-volt transistors as series pass regulators, but they'd need several watts worth of heatsink -- again more space than two small xfmrs.
Reply to
Don Foreman

If you use a capacitor to drop the voltage instead of a resistor, the heat dissipation goes down. Series capacitor from one arm of the power transformer, figure 20 mA into rectifier, which has circa 250V on the C-R-C pi filter on the other side... so roughly

2*pi *f*C*(460-250) = .020A

If you have 'mains', I guess f is 50 per second... and when I say 'roughly', (460-250) is the difference of RMS values of voltage, it should be the RMS value of the voltage difference (which is a different thing). Because the rectifier turns off sometimes, it's VERY different.

Reply to
whit3rd

Eeeeeh, the good old days..... 'Ello darling, do you want to come look at my beam power pentodes?'

Phwoar, ay?, phwoar.

DNA

Reply to
Genome

If you have 460 VAC it's probably from a center tapped transformer. Just add a IN400x diode to one side of the winding to generate -100 VDC. You can easily tap off the +200 VDC from the existing supply.

Reply to
Homer J Simpson

And now a bit of follow-up:

First, thanks to those who offered suggestions.

Using a capacitor to drop the voltage is an interesting idea. The problem is that most caps in the uF range required either can't tolerate the current or can't tolerate the high AC voltage or both (not to mention expense and size).

What I've settled on for the moment is a half wave rectifier (single diode) into a CRC section (10uF, 8K2, 10uF) that feeds a string of three zeners in series. Turns out that On Semi makes zeners with breakdown voltages in the range of 100-200V that are rated 5W and up to 31mA (whichever comes first ;) I can pick my -

100VDC (or +200VDC) rail from between a two of the zeners, throw in a largeish cap (100uF or so) for filtering and get everything I need.

The low capacitance values on the CRC input keep the caps relatively small and the 8K2 can be a 5W part. The circuit takes about 3 seconds to develop rated voltage and settle in - fine given that it's feeding a tube that can take 20-30 seconds to settle in, and maintains safe voltages if the tube is removed from the circuit.

Since the current demand is low (20mA max) I'm going to look at using a simple transistor (or dual transistor) series regulator, but think the high voltages may make finding an appropriate transistor more difficult than the simple string o' zeners.

Reply to
Wakarusa

ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.