My bench power supply

[ivan747]

I've finished laying out the board for my new bench power supply and I am looking for the panel mounted stuff.

It has current limiting, adjustable voltage from 25V down to 1.25, can supply 3A of current and the ripple is yet a mystery. One of the coolest things about is a comparator that selects the input voltage of the main voltage regulator. This way the regulator doesn't overheat when loaded with low voltages and high currents.

It switches a relay between the transformer's taps, giving 30V or 15V input. It triggers at ~12V and has ~200mV of hysteresis. I hope the filtering capacitor can maintain the voltage high enough for the switch to occur without problems. Also I hope I don't do an idiot mistake and make the thing to switch to the higher voltage tap when it is on low voltage mode.

It uses a low side current limiter since it is floating ground, so that isn't a problem. I simulated the response of the operational amplifier and ideally it only allows a very short surge of 8A in short-circuit, in the order of micro-sceonds. As an extra security feature, the voltage regulator also offers a over-current protection circuit.

The transformer is a 24V center tapped transformer (12V-0V-12V). The 12V regulator serves as reference for the comparator and the current limiter. I hope it is stable enough. Obviously, the peak current surge on the 12V rail will happen when the relay switches. I have added a 10uF capacitor next to it just in case. (I don't want the voltage to fall too low and make the comparator a weird relay oscillator).

The current shunt resistor is relatively small, both in resistance and wattage rating, as well in size. It is 0.1 ohms. This means that every amp makes the voltage to drop 100mV across the resistor.

By the way, does anyone know if those cheap panel meters can measure a voltage between two nodes that aren't connected directly to ground? Also, I've heard these panel meters need a completely separate power supply. If this is true, I will just add a 9V battery  .

 So far I have some connectors for the digital panel mount voltmeters. One is set to the 199.9V range (I loose a decimal point, so it sucks!), the other one, which measures current, is set to 1.999V, since the drop on the resistor goes higher than 199mV.

This supply isn't too hackable. I might add a digital interface of sort the next time I need one of these. You know, the classical FT232R with opto-isolators, a PIC microcontroller reading from the ADC, some DAC's to set the voltage and current and an LCD readout. Maybe even some sort of timer to have the thing on for a certain amount of time and even a clock to turn it on at a certain time of the day. And I think I could add an EEPROM to store pre-defined configurations, as well as a handy auxiliary output with an LM317 and a rotary switch to toggle between 1.8V, 2.5V, 3.3V, 5V, 9V and 12V.

I think the price for the parts I don't have is around $35 to $50. I might use the opportunity to order some test leads (banana to crocodile, funky names) and extra parts to have them at my bench just when I need them.

Ivan

[ivan747]

Ahh! I forgot to add the schematic!

All the 3 pin connectors are for potentiometers. The 2 pin connectors coming from the 12V line are for the panel meters and the power indicator LED.

The 3.3K resistors are there to load the regulator and make it more stable.

Here it is:

[johnmx]

Ivan,
You neglected several important aspects in the design.

I would like to mention just a few:
- No way the power supply can deliver 3A continuously. It can, but only in a very limited region. Check the LM338 datasheet, especially the “Thermal Resistance Junction to Case” parameter.
-R16 and R18 are in the wrong direction, rotate them 180º.
-C1 should be away of thermal sources, or its life expectancy will decrease a lot.
-My guess the current limiter circuit will oscillate.
-The return path of the ground signal (R11 pin) is too far from the C1 and B1. The 3A current will travel all over the board.
-Also the variable resistor R26 is placed badly. Its temperature will change a lot because it’s too close to the heatsinks, resulting in a variation of the output voltage.

The list continues but I leave the rest for other guys…

If this is your first board then I recommend you to make the PCB. Although it has critical problems this is the best way for you to detect the problems and try to correct them. We always learn a lot with our mistakes.

[ivan747]

Well, this is my first real high power, analog design.

- No way the power supply can deliver 3A continuously. It can, but only in a very limited region. Check the LM338 datasheet, especially the “Thermal Resistance Junction to Case” parameter.

Under investigation...

-R16 and R18 are in the wrong direction, rotate them 180º.

They are trimmer pots, so it isn't that critical or counter-intuitive since this is a one of project. Any way, thanks for pointing it out. I will flip them just to get it right. I'm always confused about what terminal goes into what pin.

-C1 should be away of thermal sources, or its life expectancy will decrease a lot.

Whoops... I didn't know it was that serious. Speaking of thermal design, should I have a fan in this?

-My guess the current limiter circuit will oscillate.

Mine too, if the hysteresis is not high enough. Maybe adding more hysteresis. It's easy, it is all through hole and I have plenty of solder wick and time . If it goes wrong I will just wire a switch and do it manually. It must be DPDT, this way I could tweak the current limiter's limit by adding a resistor in series.

-The return path of the ground signal (R11 pin) is too far from the C1 and B1. The 3A current will travel all over the board.

I am aware of this. That's the reason for those red jumper wires to be there. I will try to make the loop shorter or make the ground plane wider though.

-Also the variable resistor R26 is placed badly. Its temperature will change a lot because it’s too close to the heatsinks, resulting in a variation of the output voltage.

That's an easy fix!

Thanks for pointing out all of this. I hope it doesn't have any other mistakes.
Ivan

UPDATE: I will probably abandon this layout. I have found extremely difficult to move tha capacitor and the high power resistor. This design has had at least 3 failed layouts. I don't usually suck that much, but doing high current single layer design is not easy, specially with the comparator feature I added.

Now a question: can I improve the comparator tendency to oscillate if I use a zener diode instead of a voltage divider as reference. I guess I won't have problems with the voltage regulator compensating the relay switching if I have a zener diode in the reference pin.

I would do a breadboard circuit, but I need a 25V power supply! It is very ironical.
Maybe I can experiment with the actual transformer since I have it.

[noob64]

I have a question , on a power supply how much dose it matters putting the shunt resistor on the negative  or to put it on the positive after or back the power transistor.
I've seen something like braking up the ground loop by putting the shurnt resistor on negative(ground) is not so good.