Continue to Site

Welcome to our site!

Electro Tech is an online community (with over 170,000 members) who enjoy talking about and building electronic circuits, projects and gadgets. To participate you need to register. Registration is free. Click here to register now.

  • Welcome to our site! Electro Tech is an online community (with over 170,000 members) who enjoy talking about and building electronic circuits, projects and gadgets. To participate you need to register. Registration is free. Click here to register now.

Design a zero to 20v 20A lab supply?

Status
Not open for further replies.

Mr RB

Well-Known Member
Hi, this project has been on my to-do list for a while (for when I finally get some free time haha).

I have a nice 300W 24v toroidal transformer and a 20 amp analog panel meter, and would like to make a good high current lab PSU that goes down to zero volts. There are quite a lot of times I have needed to test high currents at low voltages, ie testing shunt resistors, wiring, DC motor stall characteristics etc.

Ideally these are the features I would like;
20v max, at up to 15A
20A from 0v to 15v
good voltage regulation right down to zero volts
analog ammeter placed before the voltage regulation
current limiting placed before the voltage regulation

I doubt if doing it directly with a switchmode circuit would work as the zero volts output (or very close to 0v) becomes a very short duty cycle and very high peak current.

So initial thoughts are to use a switchmode circuit as the main PSU followed by a very low voltage drop linear circuit dropping maybe 2v. The idea is that it's fairly easy to build a switchmode supply at 2v 20A using a conventional design.

It might be feasible to do it with just a switchmode voltage regulator, and a 30amp bridge rectifier on the output wired to act as a 3v 60Amp diode, with the voltage feedback taken AFTER the diode so the whole PSU can go down to zero volts.

I welcome any suggestions including op-amp or comparator type which are good down to 0v, of suggestions involving using a negative rail just for the regulator if that will be needed.
 
As far as I've seen all supplies that can go to zero volts use a linear regulator on the output. The more expensive kinds are hybrid and use a switching regulator to set things up for the linear regulator.
 
Last edited:
Linear supply

Here is a linear that would probably do if you have a pre regulator. I haven't built this yet, but it simulates okay. Let me know if you are interested and I can attach the files or mod it for your 20 amps at a few volts.
Maybe everyone could critique it before I build it.
 

Attachments

  • Linear.PNG
    Linear.PNG
    51.9 KB · Views: 269
Thanks for the feedback guys.

dknguyen- Yes it would be a hybrid, using a switchmode regulator for the bulk of the voltage drop and a small linear drop at the output.

ronv- Interesting regulator, but it's too complex for what I want seeing as I have to build a large high power switchmode regulator too.

I'm still keen on the possibility of using only one regulator (switchmode) with just a simple diode drop of 2v or 3v on the output to improve the duty cycle issues. It seems wasteful to have 2 separate regulators if it's not needed, and even could be problematic if they fight each other and get unstable.
 
I doubt if doing it directly with a switchmode circuit would work as the zero volts output (or very close to 0v) becomes a very short duty cycle and very high peak current.
I don't understand why the peak current will be very high for ~0V output... For a fixed frequency & fixed output current, the higher the voltage, the higher the peak (and ripple) current(s). For an ideal converter, 0v out will have a peak current equal to the output current; 2v out will have a peak current greater than the output current. I think it's possible to have a SMPS regulate down to ~0V.

For the hybrid PSU, the linear stage can be the standard diode-ORed square regulator (or whatever) and the SMPS IC controlled directly by the output of a [ground referenced] differential amplifier connected across the pass element; in this way, the SMPS regulates the voltage drop across the linear regulator.
 
Dougy83- The high peak current is really a problem. My big toroid mains transformer will produce around 40v DC, the power wiring and buck inductor being capable of 20 amps continuous are very low ohms, maybe 50 milliohms, maybe another 50 milliohms for the MOSFET. So 20A from 40v via 0.1 ohms will be a voltage conversion of 40v: (20A*0.1) or 40v:2v so the duty cycle is 20:1 which means peak current is 20*20A or 400 amps!

I want the PSU to be able to do high currents into low voltages quite a lot, as I have other PSUs for more typical tasks. So getting the duty cycle up and peak current down seems a good idea.

You said-
For the hybrid PSU, the linear stage can be the standard diode-ORed square regulator (or whatever) and the SMPS IC controlled directly by the output of a [ground referenced] differential amplifier connected across the pass element; in this way, the SMPS regulates the voltage drop across the linear regulator.

Yep that's what I had in mind. Instead of a proper secondary linear regulator, just using a dual diode drop from a big 35A bridge rectifier. This will be about 3v at 20A. Then if I also add in the 20milliohms from a 20A ammeter it looks more like 40v: (20A*0.12)+3v or 40:5.4 which is duty cycle of 7.4:1 and peak current now down to 150 amps which sounds a lot more reliable.
 
Last edited:
Hi MrRB,

how do you get 20V at 20A out of a 300VA transformer?

It should have a minimum rating of 470VA.
 
Last edited:
So 20A from 40v via 0.1 ohms will be a voltage conversion of 40v: (20A*0.1) or 40v:2v so the duty cycle is 20:1 which means peak current is 20*20A or 400 amps!
This is running discontinuous mode, I take it. Try running in continuous mode - the peak current should about the average current + half the ripple current, surely?? So for a ripple current of 1A, peak current should be ~20.5A...

Here's a simulation of ~17.1A into a 0V load. Peak current is ~17.2A, ripple is ~0.2A. Please ignore the fact the diodes are a little underrated...
 

Attachments

  • Continuous Mode.gif
    Continuous Mode.gif
    14.5 KB · Views: 218
Boncuk- I want the ability to get 20v (at whatever current) and get the full 20A below 15v. I did say that in post #1. :)

Dougy83- I apologise, you are right of course. If I run continuous mode (which should be easy enough now the duty cycle is not so severe because of the added output diode drop) the peak current gets very managable, but I think the ripple current will be pretty big maybe 60% or so as I want to keep the switching freq down as close to 20kHz as possible.

Now the question is going to be how to make a switching comparator that will work well around average 0v (with ripple!), and how/where to do the current sensing. I don't know if there are any 0v capable SMPS controller ICs available, in the past I've either rolled my own with opamps/comparators or used the common SMPS ICs like 494 or 34062 etc but they usually need 1.2v Vref sensing or more.
 
Dougy83- I apologise, you are right of course. If I run continuous mode (which should be easy enough now the duty cycle is not so severe because of the added output diode drop) the peak current gets very managable, but I think the ripple current will be pretty big maybe 60% or so as I want to keep the switching freq down as close to 20kHz as possible.
No need to apologise to me; I have oversights every day. If you're worried about the pulse width being too narrow, maybe try using a different control method, e.g. constant on-time. 60% ripple would likely require a horrible design. I don't see the desire to use a series diode in the output to be a nice solution.

Now the question is going to be how to make a switching comparator that will work well around average 0v (with ripple!), and how/where to do the current sensing. I don't know if there are any 0v capable SMPS controller ICs available, in the past I've either rolled my own with opamps/comparators or used the common SMPS ICs like 494 or 34062 etc but they usually need 1.2v Vref sensing or more.
I would not recommend using a comparator to directly measure the ~0V output; perhaps use an opamp with sufficient gain to feed into your comparator or the SMPS controller IC. As for current sensing, a simple very-low-value shunt with a couple of stages of gain should suffice. While there are a few common opamps that claim to have inputs able to go to 0V, I've not used one that has performed as I wished. As most decent cheap opamp require a negative rail, this could be produced using a simple diode pump.
 
If you're worried about the pulse width being too narrow, maybe try using a different control method, e.g. constant on-time. 60% ripple would likely require a horrible design. I don't see the desire to use a series diode in the output to be a nice solution.

I'm not keen on the idea of a fixed on time. That would reduce switching frequency and increase the current ripple even worse.

Why don't you like the idea of the series diode? I think this is an elegant solution to the complexity of a second (linear) regulator dropping a couple of volts and the predictable Vf curve of the diode should easily be compensated by the SMPS voltage regulation, and provide a better minimum duty cycle with none of the risk of two regulators conflicting.

With a cap before and after the diode it can also be used as a mild PI filter because of the soft forward curve of the diode, although obviously it won't be as good a ripple filter as a linear post-regulator.

60% current ripple (in the inductor) is not unusual for other high efficiciency SMPS systems I have designed. That's a +/-30% ripple at rated current. The efficiency gains of the lower switching speeds far outweigh the I2R losses from the slightly increased current ripple. It does require good inductors, but I have some very nice toroid cores left over from a 96% efficiency 15 amp buck product I designed a few years back.
 
This LINEAR Supply will give the Output you want.
**broken link removed**
 
I'm not keen on the idea of a fixed on time. That would reduce switching frequency and increase the current ripple even worse.
The frequency would reduce for reduced output power; it'd be high for heavy loads.

Why don't you like the idea of the series diode? I think this is an elegant solution to the complexity of a second (linear) regulator dropping a couple of volts and the predictable Vf curve of the diode should easily be compensated by the SMPS voltage regulation, and provide a better minimum duty cycle with none of the risk of two regulators conflicting.
I just don't think it's necessary and it wastes a decent amount of power (probably not an issue for you). You still need to regulate the output voltage and current which will require some opamps anyway. Obviously whatever you want to do, you'll do; I'd be interested in hearing how it works, and if it works with the series diode shorted. +/-30% doesn't sound so bad; I assumed you meant +/-60%...
 
Chemelec- Does that supply regulate right down to 0v? It looks like your voltage control differential pair is well above 0v but I might just be misunderstanding your voltage regulator. I'm reluctant to use a full linear supply because it's main usefulness will be at very low voltages and med-high currents, and my transformer makes about 40v DC...

Dougy83- The output diode may not be necessary as you said. At 0v output and a very low duty cycle the buck diode will be in circuit most of the time anyway, I forgot to add it's voltage drop into the duty cycle calc but even a good schottky will probably drop 0.5v at 20A. Thank you for the input, I do appreciate it.
 
Chemelec- Does that supply regulate right down to 0v? It looks like your voltage control differential pair is well above 0v but I might just be misunderstanding your voltage regulator. I'm reluctant to use a full linear supply because it's main usefulness will be at very low voltages and med-high currents, and my transformer makes about 40v DC...

YES, It Goes to ZERO on Both Current and Voltage.
(I have Two of these rated at 36 Volts and 8 Amps each.)
 
Status
Not open for further replies.

Latest threads

New Articles From Microcontroller Tips

Back
Top