Simple HT PSU design

[Don_Zalmrol]

Hi chaps and chapettes,

I'm working on a new simple HT PSU design to get an output of 250VDC for driving one 12AU7 and two 5687 valves with an output current of +- 100mA.

I've drawn a simple circuit to rectify the input current from AC to DC with some caps, bleeder resistors and a limiter resistor parallel with a choke.

But I want some input for the design, before I'll construct it.
Could you give me some feedback please?

Thanks!

[ukcol]

Hello Don

I'm not a design engineer and so you will get a more definitive answer from others, but a few things strike me.

The bleed resistor values are much too low, there will be more than 1 Amp flowing in each of them (if the source is up to it). Do you need them at all?

The limiter resistor needs to be in series with the output of the bridge rectifier.
If your AC is coming from the secondary of a transformer you may not need it at all.

The capacitors C1 and C2 probably have much higher values than strictly necessary.

The output D.C. voltage (off load), will be 1.414 (root 2) times the A.C input voltage and with those big caps only slightly less under load.

Colin M

[Don_Zalmrol]

Thank you for your reply Colin.

I have changed the value of the smoothing caps from 330µF to 220µF, perhaps even 100µF would suffice.

I've always learned & heard from others to install bleeding caps on smoothing/ buffer caps, to avoid shocks from the HT line. This will also even "stabilize" the PSU a little bit.

The RLim cap is not only for limiting the rush current. But also for stepping down the output voltage a bit more stable to something of a 250VDC. The valves can run on 300VDC, but the data sheets prefer 250.

For a choke I want to use a simple inductor, since it only has to filter/ stabilize the PSU a bit and I don't have room for filament choke. 10H would suffice or is it overkill?

https://www.distrelec.com/ishopWebFr...is/204133.html

[ukcol]

Choosing values for bleed resistors is always a compromise. If you arrange that the voltages on the capacitors cannot be touched than you can do away with them altogether. When choosing a value the compromise is between the time the resistor takes to discharge the capacitor to a safe low voltage and the amount of energy you waste in the resistor when the PSU is running.

I felt that your capacitors were probable getting on for a factor of 10 bigger than required, especially as you are using a choke and not simple R/C filtering.
Big caps also have implications for the discharge resistor values.


Yes you can also use the limiter resistor to reduce the output voltage but it will still need relocating as I said in post 2.

My guess is that a 10mH choke with those big caps is plenty but someone with design experience will be able to give you a well reasoned answer.

Colin M

[M0ALK Richard]

Googling 'power supply smoothing capacitor calculation' gives plenty of explanations.

The bigger the smoothing capacitor the less time the diodes have to charge it up during each cycle. Less charging time means the diodes have to be capable of providing a very high current for a short time.

10 uF should be adequate.

[G8HQP Dave]

Download Duncan's PSUD2 software. You can then play with different power supply ideas. A good way to check your own understanding.

Valve PSUs combine AC mains voltage with high DC voltages. Either can kill, so be careful!

[MrElectronicman]

The 200R resistors are much too low, as bleed resistors they should be more like 200K or higher. The capacitors would be alright at 10-20uF and the choke would be better at 10mH not 10uH

[glowinganode]

Even 10mH is too small to have any noticeable effect, you may as well not bother with it. 1H plus is more like it.
In the old days, chokes were cheaper than large electrolytics, so was a cost effective approach. These days large value electrolytics are available at comparatively little cost.
Look at commercial designs, the use of smoothing chokes went out in the 60's for all but specialised applications.
Rob.

[Herald1360]

First reaction is that 10uH choke is pointless- it's far too small to have any useful effect at 100Hz ripple frequency. 10H would be far more suitable.Then for decent ripple level ( <1V ) you will only need about 16uF for the capacitors.

However, a 10H 100mA dc choke isn't cheap, whereas capacitors are relatively.

So, forget the choke, increase the capacitors and just use a series resistor.....

If you use a couple of 100uF capacitors and a 220R 5W resistor, you'll still get <1V ripple. Using 100K 2.5W bleeder R will get voltages down to safe in less than 60 secs even with no load on the supply.

For the earlier stages of the amplifier, just add another 1K 1W / 47uF or so smoothing/decoupling stage in the HT feed, it'll only drop another 20V but it will clean up the remaining ripple nicely.

To keep repetitive peak current on the rectifiers down, use a 100R 5W resistor in series with the AC in to the bridge. With this the peak surge will be <10A and the repetitive peaks about 400mA. Well within 1N4007 ratings.

Remember that voltages will go up to 350V plus at switch on, before the valves warm up so use at least 400V electrolytics. A scrap PC PSU is a potential source of something suitable if it's not too old and tired.

Oh and stick some 10n 600V capacitors directly across each diode to kill off RF noise from the reverse recovery spikes from the 4007s.

Piccies show circuit with output volts, diode current and ripple volts at 100mA load.

[Don_Zalmrol]

So something like this?
Tonight, I will try and simulate this PSU in Duncan's PSU software

[Billy T]

You are still way off: the 100K series resistors in the feed to the bridge rectifier will leave you with very few volts out. The output bleed resistor doesn't need to be 2 watts either, 1 watt is ample. I get the feeling that your present level of knowledge is not adequate for the design of this equipment.

Billy

[Nicklyons2]

I think I must concur with Billy T above, the voltages you are using here are capable of killing and, unless you really understand what you are doing you should leave it alone. If you must try and construct something along these lines PLEASE use an ISOLATED supply, you need to derive your supply from a DOUBLE WOUND transformer not an auto-transformer. I really do not think you should connect a supply of this sort of design straight across the mains; I know equipment was made in the past which did but those of us who repaired it were used to it and even then, occasionally, we got 'stung'. The secondary of the transformer should be about 170 volt a.c. by the way at around 200mA or 170-0-170 at 150mA and you can use 2 diodes instead of a 'bridge' rectifier.

[Don_Zalmrol]

To clarify some things said. The tranny I'm using is a double wound transformer, but I just didn't draw it in the schematic. And have an isolated transformer by my side.
The tranny has only the following outputs: 0 - 220 - 240 - 260 - 280 and 320VAC. So 220 would be the logical input voltage and drives a current of 100mA. Plenty for 3 pre-amp valves.

So a two-diode setup wouldn't work here, unless I create an artificial ground between the two windings (0-220). And a bridge rectifier is easier to install.

220*1,414 = 311,08VDC

It isn't be powered directly from the mains
I'm not that crazy!

And the 100K resistor at the mains feed, is an error. Copy-paste mistake from my other resistors, these should be like Chris pointed out just 100R/5W. As you clearly can see in the previous drawing, there all named R1…

I'm not a novice, but aren't a pro also. PSU design is like Colin said, something special.

@Chris, which version of Duncan PSU designer are you using?

[Station X]

Isn't there something missing there? As drawn you could just use a 440uF capacitor in parallel with a 110K resistor.

[Herald1360]

Aha!

My circuit/plots print wasn't very clear which doesn't help. Also my description didn't quite tally with my circuit.

The circuit I simulated had 50R in series with each bridge AC input, but for simplicity and lower component count I figured that one 100R part in one line would do.

The 220R should be connected between the 100uF caps positives, the first cap goes straight to the bridge + output.

I've hacked your drawing (attached). I left the 220uF caps- they'll probably give you a slightly higher Vout and somewhat lower ripple. If you need to drop the Vout, increase the 220R, which will also further decrease ripple.

I'm not using Duncan PSU designer, my simulator is an old version of Simetrix. The guy that created it used to be a colleague back in the '80s.

http://www.simetrix.co.uk/site/index.html

You can download a demo version of the current release which has enough capability to simulate something like this, though it's crippled in various ways to make it useless for any professional work. Otherwise there are other genuinely free sims like TI's TINA which are pretty good.

[Nicklyons2]

A few further observations, some of which were mentioned earlier: if only about 100mA is required there is possibly more smoothing than absolutely necessary. There is need for some fault protection, something like 200 mA anti-surge(T) on the bridge input 200mA (F) on the DC side. In any power supply the 'prospective fault current' should always be borne in mind; in high voltage and mains derived supplies this is particularly important. 300 VDC with the source impedance of a 200uF elctrolytic will, for a split second, be capable of a substantial energy output if things went wrong. Increasing the smoothing resistor is fine assuming the load is reasonably constant; if the current varies considerably the output voltage may swing alarmingly, take care if there are long periods of low current that the output doesn't creep up too high. Finally, the circuit shown in post#13 shows the negative supply rail returned to Earth; do you really want/need this? it references all the DC voltages to ground making it more dangerous to handle; could the DC side be left 'floating'?

[Don_Zalmrol]

Hi all,

@Graham: Yes, I could use one 440µF cap or like Chris has drawn two 220µF with RLIM in the middle.

@Nicklyons: Something like this?
https://www.distrelec.com/ishopWebFr...w/is/true.html

Will redrawn them tonight and try out Simetrix and TIA.

[Nicklyons2]

No, the device you linked to above is a 'surge protector' not a fuse. Look for 20mm glass fuses 200mA and they should be available in F = fast and T = slow blow or 'anti-surge' types. You will also need a holder/clip for the fuse to fit in.
May I stress again the need for caution with voltages in this range; if you are not sure do not plug it in !

[Don_Zalmrol]

Ah, ok. So a normal slow blow fuse…

[Diabolical Artificer]

For a similar function I used a 230V primary 12V secondary reversed, fed by a 12V wall wart with the regulator removed into a hefty bridge rectifier out of a TV into a 32uF/32uF-450V cap with a 100 ohm 1W resistor between the two + terminals.

IT had a big, easy to get at, switch to turn the whole thing on with a fue and a switch ( I used an underated quick blow fuse when making changes so it would blow if I made a mistake) after the cap, and I made sure to discharge the cap through a 5W 220 ohm resistor to earth everytime I worked on the circuit. Not elegant,worked for me, Andy