Stacked power supply puzzler.

[Diabolical Artificer]

I'm knocking up a power supply to power 24v relays and 12v fans, so went down the stacked power regulated supply route using LM317's. The fans both on at the same time draw 120mA, one relay is on all the time and requires 60mA, but the 24v rail also will need to supply another identical relay in a fault condition.

So, simple yes? One circuit board with two identical LM317 12v regulated supplies, one stacked on top of the other. After an initial screw up where I soldered in the caps the wrong way round and tried to power to bridge rectifiers from a 15v 0v 15v wired transformer, it's working....ish.

Without a load I'm getting, 0v 12v 24v, with two 100r resistors in series as load the bottom supply goes up in value to 13.8v. After mucking about changing the adjust R the penny finally dropped, the load is acting as a potential divider and either feeding back to the bottom supply or doing what it's supposed to do, IE acting as an equal voltage divider.

After all that waffle, what is the best way to do this? Why not use a split rail supply using a LM317 & 337 to get +15v 0v -15v? The fans won't care as long as they see 12v and the relay will see 24v. But will I run into a similar issue?

Idea's on a postcard, message inna bottle or post a reply please, Andy.

[Refugee]

The LM317 will not sink current as you have found.
Neither of those parts need a regulated power supply and are unlikely to even need smoothing.
A capacitor will only be needed to stop any mechanical buzzing.

[vidjoman]

Put the fans in series then you only need a 24 volt supply.

[Lucien Nunes]

To understand why your existing setup isn't working it would be helpful to see the circuit. Certainly there will be trouble if any 12V load is powered from +24 to +12 unless there is consistently greater load from +12 to 0, due to the aforementioned point of the +12V rail not being able to sink current via a positive regulator.

But a split rail using positive and negative regulators giving +/-12V will accept load 12V loads from +12 to 0 and 0 to -12, and also 24V loads between +12 and -12 and should work fine.

I am not sure about series-connection of DC fans with electronically commutated motors. I experimented with this once, got some odd unpredictable voltage-sharing, and have not investigated further as to whether that was just an exception.

[vidjoman]

I only put 2 in series once about 25 or more years ago. It was in an Amstrad PC which had a little motor running very fast in a sort of tunnel and making a lot of noise. I found a second one and put one at each end and running more slowly, at half the voltage, they still shifted the air but we’re much quieter.

[frsimen]

You should be able to fix this problem very easily by changing and rewiring one resistor.

The voltage setting relay for the LM317 that is used for the 24V output should be recalculated to give 24V rather than 12V. Connect the lower end of the resistor to 0V (= negative side of the supply) rather than the 12V that it is connected to now.

Paula

[joebog1]

I like what Lucien said, and in fact its quite an elegant solution.

But a split rail using positive and negative regulators giving +/-12V will accept load 12V loads from +12 to 0 and 0 to -12, and also 24V loads between +12 and -12 and should work fine.


Joe

[Diabolical Artificer]

Neither of those parts need a regulated power supply and are unlikely to even need smoothing

They do if the only suitable tfmr you have is a dual 0v 15v, 21v DC into a low noise fan = dead, noisy fan. Also I wasn't using the 12v rail as a ground for a 12v supply, hence the 0v 12v 24v supply, but on the other hand I was with the resistor load, so good point. What I should have done to test was one 100r to test the 12v supply, two in series (but not connected i the middle) to test the 24v.

Put the fans in series then you only need a 24 volt supply

Good idea, though because the fans sit in separate housings,connecting them in series makes things tricky.

It seems to work fine with actual relay's and fan's Lucien, just not with a potential divider as load. Probably because both loads are reactive and not purely resistive. That's a good point, don't think I remembered to fit a reverse diode across the relay's; is one needed for reg IC rather than a tranny own as SW?

You should be able to fix this problem very easily by changing and rewiring one resistor.

Got it in one.

So, as Lucien says if two separate 12v supply's are needed a split rail supply is better. thinking about it valve scopes use a combination of the two, stacked & split. Stacked to get the various rails for the anodes, focus etc and split for the cathode. Which gets me thinking, Tek used stacked regulated supply's in the 500 series scopes, but they must have done it in such a way that various regulated supply's could sink current, unlike as per the 317. Might be wrong about that though.

Thanks all,learning a lot about PSU's this week, Andy.

[kalee20]

Circuit diagram would help massively, hand-drawn on a piece of paper and photographed?

Paula's suggestion, make an LM317-regulated 24V supply, and then power a further LM317 from that to get your 12V supply, on the face of it sounds good advice.

Originally Posted by Diabolical Artificer

...learning a lot about PSU's this week, Andy.

There's lots to learn... you'll get an idea why people can make an entire career out of power supply design!

[cmjones01]

Originally Posted by Diabolical Artificer

Which gets me thinking, Tek used stacked regulated supply's in the 500 series scopes, but they must have done it in such a way that various regulated supply's could sink current, unlike as per the 317.

Tek scopes frequently do have stacked regulators but they still can't sink current. They just arrange the loads carefully so that, for example, nothing much is trying to draw current from the +350V rail and return it to the +225V rail, other than perhaps a few bias networks which don't pass much current anyway and whose sourced current is swamped by other loads.

Chris

[turretslug]

Originally Posted by kalee20

Circuit diagram would help massively, hand-drawn on a piece of paper and photographed?

Paula's suggestion, make an LM317-regulated 24V supply, and then power a further LM317 from that to get your 12V supply, on the face of it sounds good advice.

Or to help reduce overall dissipation and improve efficiency, feed the regulator for the 12V rail from the transformer secondary centre tap, this makes use of 2 of the diodes in the existing bridge as a bi-phase full-wave set-up, this raw DC needs its own reservoir.

[frsimen]

Originally Posted by turretslug

Or to help reduce overall dissipation and improve efficiency, feed the regulator for the 12V rail from the transformer secondary centre tap, this makes use of 2 of the diodes in the existing bridge as a bi-phase full-wave set-up, this raw DC needs its own reservoir.

That is exactly what I had in mind when I replied.

Paula

[Diabolical Artificer]

Some good suggestions which I'll implement, thanks all, Andy.