0-50V, 5A bench power supply - power dissipation

[TheMG]

I've been into the electronics hobby for a while and my field of work is also closely related, and I've been watching EEVBlog videos for a while. Decided I would finally sign up on the forums and poke the forum brains a little for some ideas. 

Believe it or not, I've gotten through all these years without a proper bench supply. I've literally got a whole pile of fixed-voltage power supplies I've been using. Cumbersome and a real pain in the butt.

Well, it's about time I get a proper bench supply. I could just go buy one, but what would be the fun in that?

Anyways, the specs I'm aiming for is 0-50V at up to 5A, linear. At these voltage and current levels power dissipation of a purely linear regulator will be an issue.

These are the possible solutions:

1) SCR-based tracking pre-regulator on primary side of transformer (kind of ugly, but it works, I've seen it done in some high current HP lab power supplies)

2) switchmode tracking pre-regulator before linear regulator

3) multi-tap transformer with relays to switch between taps

4) BIG BIG HEATSINK

5) Get creative?

A few considerations:

-I've got all the parts necessary to build several power supplies following option 1 or 2
-Option 2 would require careful design to avoid EMI/RFI being conducted through to the output.
-Option 3 would be more costly, as I would have to source and purchase transformers with multi-tap secondary
-Option 4 is absolutely ridiculous and there is absolutely NO WAY I will be doing it

As for "getting creative", I thought of perhaps using a series pass transistor between the bridge rectifier and filter capacitor. This transistor would be switched on until a comparator detects that the voltage on the capacitor reaches a certain level (Vout+5V for instance, arbitrary number for now), then turns the transistor completely off until the next cycle. Relatively simple, but in theory should be quite effective. Main problem with this setup is the large amount of 120Hz ripple at lower voltages and high currents, which would require a fairly substantial capacitance to keep the linear regulator in its "happy zone".

What do you think? I'm having trouble deciding which approach to take.

[PA0PBZ]

Talking about getting creative... how much space do you have? I've seen variacs with motor control 

[Dave]

I'd go for option 3. Having a toroidal transformer custom made doesn't cost that much.

[TheMG]

How much is "not that much"? Got any recommendations/links for such custom transformers?

Thing is... I already have two dozen 25V 250VA transformers kicking around (therefore, FREE).

[mariush]

I would just get 4x12v or 3 x 18v or 2 x 24v center tap transformers ... use relays to switch between taps

ex two of these would be just good enough to give you about 48v DC at about 6.5-7A after bridge rectifier and capacitors :

http://uk.farnell.com/pro-power/ctfcs200-9u/transformer-200va-2-x-9v/dp/1780903

^ this has two separate secondary windings, so if you want to go crazy you could also use relays to put them in parallel and get 9v ac rms 22 amps , so you could get 2 of these and have your psu doing 0-10v 24a , 0-24v 12a , 0- ~46-48v 6a

slightly worse 2x12v 150va : http://uk.farnell.com/pro-power/ctfcs150-12u/transformer-150va-2-x-12v/dp/1780900  ... not sure you can do 5a, but you only need two of these to get about 60v dc

so anyway with that 9v on each winding, you'd have in worst case scenario ~ 9v x 1.41 = 13v x 5-6a = 70w dissipated on the heatsink. Easily doable with a larger heatsink and a fan.

[Matje]

As for "getting creative", I thought of perhaps using a series pass transistor between the bridge rectifier and filter capacitor. This transistor would be switched on until a comparator detects that the voltage on the capacitor reaches a certain level (Vout+5V for instance, arbitrary number for now), then turns the transistor completely off until the next cycle. Relatively simple, but in theory should be quite effective. Main problem with this setup is the large amount of 120Hz ripple at lower voltages and high currents, which would require a fairly substantial capacitance to keep the linear regulator in its "happy zone".

What do you think? I'm having trouble deciding which approach to take.

I thought about doing something like that for a while. Mainly because the cheap chinese power supplies that just wouldn't reach the promised power output despite in principle being capable of it (problem is always power dissipation) pissed me off.

I thought of using a switched power mosfet as the pass transistor, I think you would also need an inductance there to limit current and voltage change per time over the filter cap. Max. ripple current through the cap is critical, may need multiple ones in parallel. Oh, and filter cap(s) after the rectifier before the pass transistor are needed too of course.

Some points that came of playing around with simulating that in spice (nothing was build in actual hardware):

- In principle this could work (I aimed for 0-30V, 5A) ;-)
- Switching frequency should be kept as low as possible (low KHz) so the linear reg is able to deal with the switching ripple
- Values of inductance and cap are difficult to set so that they work at both low and high load
- One critical point is at max. output current and max. output voltage, need to be able to load the filter cap fast enough
- Other critical point at very low output current and voltage, because voltage change per time over filter cap is highest and the energy stored in the inductance causes largish "overvoltages" over the filter cap (-> ripple on output)
- Also critical: switch on (I wanted to have an extra button to switch the outputs), current through the inductivity and the filter cap easily exceeds the values in steady state, this could be fixed by limiting the output voltage change rate, no idea if this is acceptable in all circumstances in a lab power supply

It just occurred to me that points 2-4 may be addressed by having the input voltage be substantially higher than usual for a linear reg, hmmm...

If you could come up with a reasonable (meaning it doesn't cost as much as buying an Agilent supply or so) design that reliably works I think a lot of people would be interested.

[Dave]

How much is "not that much"? Got any recommendations/links for such custom transformers?

I always order my toroidal transformers from a small manufacturer here in Slovenia. I shoot him an email with my requirements, he tells me the price and I order it. So far I have ordered 120VA transformers (23€ each) and 50VA (15€ each). This also includes two rubber insulators and a mounting bracket. The price probably depends on the complexity of the transformer (number or windings), but I can't say I really know how he calculates his prices.

You would need a 300VA transformer for your PS, so I'm guessing you are looking at around 40-50€.

[Kevin.D]

You dont need one big transformer ,do something like mariush said .Get 2 X24 with centre tap  , by switching windings to parrallel  when lower voltages are needed you can  get 2 X the current and so you still use the full power utilization of the transformer  . 
If  you  want real flexibilty instead of doing a single 0-50V @5 Amp output you could do  multiple smaller adj independant outputs that can be connected or switched in any parallel or series combination you chose (with tracking if needed).  So  lets say you have 3 independant smaller adj outputs of 0-24 V @2 Amp . Now switching these internally into parallel or series  then you would be able to  select all the various possible ouput combinations like ,3 independant 0-24 V @2amp outputs or, 1 x0-24V @ 4A +0-24V 2A ,1 x0-48@2 A + 0-24v@2A , 1 X 0-68V @2A ,1 X 0-24V@6A and the various +/- this would also create .  I don't know how practical the switching required would be because I haven't considered it in depth ,but it  doesn't sound to complex to do and offers great flexibilty .

[TheMG]

Well, I think I'm going to give my unconventional creative method a try. I have a rough schematic, gonna whack it together on a breadboard and see how it works.

I have a huge pile of those 25V 250VA transformers so i might as well try to use them before I go spending money.

In theory, if I slow the switching of the transistor down a bit to avoid a sharp edge on the waveform, it should keep RF noise to negligible levels and the series regulator should have no problem dealing with it, just like normal 120Hz ripple.

Kind of a cross between a clipper/limiter and a switcher.

Anyways, the idea here is not to design a marketable product. It is to try to build a usable and reliable power supply from mostly recycled parts. I have the aforementioned transformers, a whole pile of MJ11033 transistors, <1degC/W heatsinks, analog meters, potentiometers, rectifiers, filter capacitors, opamps, etc. Everything is pretty much there.

[Psi]

Most of the toroid transformers i've seen have the secondary winding on top.

It makes it quite practical to unwind the secondary and wind it back on with as many taps as you want.
A little time consuming but easy to do.

[TheMG]

Transformers I have are not toroids. They are fully enclosed E-I type Hammond transformers.

[Paul Price]

Just use one voltage control for two of these switchers and you got 6-amps. Series connect the two 25V transformers to get the 50V input to the switchers.

[robrenz]

I just bought a Power designs 6050A on the buy/sell page here that does exactly what you are talking about.  See the description in the attached pdf.  Notice up to 150% of rating is available over 105V input.  There is another one just listed for $70.00 includes shipping in the US.