Variable Bipolar Power Supply

[Hero999]

Here's the plans for a power supply I'm intending to build. By my claculations it should provide me with +-15.7V@1.5A. The mains transformer I'm using is a 15V-0-15V torroid rated to 2.67A which is slightly over-rated which is perfect as the regulators might not current limit untill 2.2A.

Anyway, I just thought I'd ask you lot if you can think of any improvements before I etch the PCB and build this.

 

[Hero999]

Not that anybody cares but I've nearly built this project. The PCB is finished and the componants are all soldered in place. The circuit works quite well the -V supply is alway within <10mV of the +V supply which is as close as fu*k is to swearing. My work's calibrated meter would show 0.01V difference on the 0-20V reading and 0.001V on the 0-2000mV setting, this is probably due to the tollerence in the meter rather than the circuit. I did consider using a better op-amp than the uA741, but then I thought I have no need to as it's only acting as a DC amplifier and the regulator IC will handle all the load transients, besides I've got loads of them in my junkbox.

I've drilled all the holes in the box and mounted the terminal posts. All I need to do is drill the holes for the power switch, LED, voltage adjustment pot, and power connector. When I've completed it I'll post some pictures providing I can find my digital camera.

 

[JimB]

Not that anybody cares ....

Well as I write this, your circuit has had 34 views, so various people have taken time to look at it, including me when you first posted the circuit.

You asked if there were any improvements to be made, I could not think of any, and presumably no one else could either.

So, take heart, you have built it, it worked in the way you expected first time, that sounds like a success to me.

Award yourself a few brownie points

JimB

 

[Hero999]

Point taken, it's had 35 veiws now.

I've made the hole for the power connector and drilled the hole for the potentiometer. Progress has been slow because I'm doing it at work. Hopefully I'll have it finished by the end of the week.

 

[Hero999]

I've finnished this thing last week but I've only got round to giving it a full test and I'm still quite satisfied even though it isn't quie as good as I first thought.

At the maximum output voltage the -V input is about 25mV lower than it should be, while at the lowest setting it's spot-on. This is because the inverting amplifier has a slightly higher gain than one which I is due to the componant tollerances. Even though I haven't bothered with a proper heat sink (I've mounted each regulator on a piece of copper and fixed it to the metal case which is being used as the heat sink), it performs quite well and the box doesn't get too hot even when short circuited.

I hope to get my digital camera back soon (I'd lent it to my brother), as soon as do I'll post some pictures.

 

[Dr.EM]

It shouldn't get hot short circuited as the chips shut down all together in that condition. Running at near the max (1A or so) it will do, but shouldn't overheat if your heatsinks are suitably sized. I', interested in the pics . I like the design, where they track each other. Mine has 2 controls and thats awkward, mabye I should change it to this design.

 

[Hero999]

It shouldn't get hot short circuited as the chips shut down all together in that condition.

In theory yes, but in practice it tends to oscillate at a low frequency. The chip gets too warm so it shuts down, it cools down so it turns back on agian. You're still right though I should test it at its maximum load.

Running at near the max (1A or so)

The datasheet says it sould shutdown between 1.5A and 3.4A but I haven't tested it.

it will do, but shouldn't overheat if your heatsinks are suitably sized.

There isn't a propper heatsink, it just dumps the heat into the case. Performanse will depend on the voltage, at lower voltages the current will be limited by the power dissipation and cooling of the chip, at higher voltages ripple might be a limiting factor.

I', interested in the pics . I like the design, where they track each other. Mine has 2 controls and thats awkward, mabye I should change it to this design.

It can be handy to have separate controls if you want to have a different negitive supply (which probably isn't that common). I decided that this wasn't an issue for me since I already have another monopolar LM217 power dupply if I need a different positive/negitive supply.

Anyway, you've convinced me I need to do some more performance testing at the highest and lowest voltage settings.

I have one more trick with an op-amp and a uA714 that can make the LM317 go all the way down to 0V. I'll post if you like, but I don't really need it.

 

[Hero999]

Oh well I'll post it anyway.

 

[eblc1388]

Have you done the ultimate test yet?

A direct short circuit between +ve and -Ve at maximum output voltage, done several times with a few seconds in between. If the power supply survives, then it has passed the test.

 

[Hero999]

Yes, that was my first test and of course it survived, the LM317 & LM337 just shut down.

When I've done the other tests I'll post the results.

 

[Roff]

Nice job, Hero!
Have you looked at the outputs with a scope? Have you done transient load tests? I would be wary of marginal loop stability (poor phase margin) in the negative regulation circuitry. If it is a problem, a cap (10nF?) from the 741 outout to its inverting input might help. If it's not a problem -

 

[Hero999]

That's all in the pipeline. I didn't think the loop stability would be a problem because the LM337's gain is in effect unity, remember it's only subtracting Vref from the output of the 741 - it's acting like an emmiter follower.

 

[Roff]

That's all in the pipeline. I didn't think the loop stability would be a problem because the LM337's gain is in effect unity, remember it's only subtracting Vref from the output of the 741 - it's acting like an emmiter follower.

Yes, but it adds another pole in the loop, which would degrade your phase margin.

 

[eng1]

Another thought: if you use the 120 ohm resistor the current is quite high: about 10 mA... an LF411 can be a good replacement for the ua741?

 

[Hero999]

Another thought: if you use the 120 ohm resistor the current is quite high: about 10 mA...

The 120ohm resistor keeps the LM337 happy because it's this minimum recommended no load current for the LM337.

an LF411 can be a good replacement for the ua741?

I know there are plenty of better alternatives to the uA741 (take the TL071 for example), but why would there be any advantage in using them? The uA741 is perfect for this application, using a better op-amp won't increase the performance much. High frequency response isn't improtance since the LM337 will handle load transients, CMR isn't important since the tollerances in the resistors will add a far greater error, noise doen't matter since the ripple from the mains will exceed any noise generated in the op-amp and it's configured for unity gain so negitive feedback will get rid of it anyway, I could go on. . .

Yes, but it adds another pole in the loop, which would degrade your phase margin.

I don't think so. Imagine you subsitute the LM337 for a PNP transistor, it wouldn't effect the phase margin since it's phase shift is practically zero and it's gain is slightly than zero. Exactly the same is true with the LM337.

The only reason I'm using a LM337 for the negitive supply is because it will shutdown if overloaded, it could just as easilly be a normal transistor.

 

[eng1]

The LF411 was my choiche when I was designing a similar circuit.

The ac parameter do matter in my opinion... the transients are handled directly by the op amp and the LM337 is just keeping a constant voltage (nominally 1.25 V) across its adj pin and out pin.

 

[Hero999]

The LF411 was my choiche when I was designing a similar circuit.

Which circuit did you build?

Please post a schematic.

The ac parameter do matter in my opinion... the transients are handled directly by the op amp and the LM337 is just keeping a constant voltage (nominally 1.25 V) across its adj pin and out pin.

Exactly the LM337 keeps the voltage constant, it's adj pin is being held at a particular voltage by the op-amp, if a load transient occurs the LM337 will just take care of it.

I have been slighltly tempted to add a low pass filter before/after the uA741 so load transients on the positive supply don't affect the negitive. I decided against it since the uA741 already has a small low pass filter built-in and the capacitors on the LM337 should take care of it.

Anyway, I'll need to do some tests to prove this, but I can't do them todat since I've left it at work.

 

[eng1]

I didn't build the circuit beacuse I was not satisfied with my design.. I started with the same idea: a voltage divider (I used two 12k6 1% resistors) and an op amp (the LF411). I was suggeted to add the resistor to ensure a minimum load for the LM337 regulator.
The inverting input is a virtual ground under steady state operative conditions; any change is handled by the op amp, so it should be fast, do you agree?
The main challenge was the stability. The op amp is used in the open loop configuration and it adds a low frequency pole. The loop gain rolls off with -40 dB/decade slope but I haven't thought of a compensation network yet..
Do you have the chance to simulate the circuit?

 

[Hero999]

I didn't build the circuit beacuse I was not satisfied with my design.. I started with the same idea: a voltage divider (I used two 12k6 1% resistors) and an op amp (the LF411). I was suggeted to add the resistor to ensure a minimum load for the LM337 regulator.

What value did you use?

The inverting input is a virtual ground under steady state operative conditions; any change is handled by the op amp, so it should be fast, do you agree?

I've thought about this again and I don't really know about this. Normally when you build an LM337 circuit, the LM337 handels the transients. However we have added another amplifier and the LM337 is in its feedback loop. I think that if the uA741 is too slow and holds Vadj at the same voltage then the LM337 will react and compensate for this.

The main challenge was the stability. The op amp is used in the open loop configuration and it adds a low frequency pole. The loop gain rolls off with -40 dB/decade slope but I haven't thought of a compensation network yet..

I think the LF411 is your problem here, I don't think it's completely stable when configured as a unity gain buffer.

Do you have the chance to simulate the circuit?

Kind of, the software I use doesn't have a a model for the LM317 and LM337 so I subsituted them for discrete transistor regulators, it worked so I assembled it.

Don't you have an exact schematic of your circuit to post?

 

[eng1]

I think the LF411 is your problem here, I don't think it's completely stable when configured as a unity gain buffer.

I don't think so. It is compensated for unit gain, like the ua741.

I've thought about this again and I don't really know about this. Normally when you build an LM337 circuit, the LM337 handels the transients. However we have added another amplifier and the LM337 is in its feedback loop. I think that if the uA741 is too slow and holds Vadj at the same voltage then the LM337 will react and compensate for this.

The LM337 has a local feed-back (the 120 ohm resistor) and it keeps the constant voltage accross the adj pin and the out pin, as we said. With this configuration it does no react to transients as it does when the adj pin is grounded. I mean.. if the output of the op amp were a sine wave, the output of the LM337 would be a sine wave + 1.25 V

Don't you have an exact schematic of your circuit to post?

At this moment I don't.
I remember I was trying to reduce the output current of the op amp and I found this solution: consider the required load current for the positive regulator. I used a LM317 and it needs 5 mA (max). If you use the op amp around the LM317 you can use a 220 ohm resistor (instead of 120 ohm) and the current will be 1.25/220 = 5.7 mA