Dual power supply with current limit

[rodd]

I need a circuit for a dual power supply where I could adjust the output from ~5 to 30V and limit the output current from ~100 mA to 1.5A.
The circuit will be used as part of a transistor curve tracer I am trying to build, based on the old Heatkit IT-3121.
I do not need lab quality regulation, but since te part count of my project is high, simplicity would be helpfull.
I found references for the positive part of the supply but NONE for the negative.
Thanks.

[russlk]

Look up the LM337T negative adjustable regulator. The data sheet shows how to hook it up. You won't be able to get 5 volts out at 1.5 amps because the regulator will overheat. The transformer should be 56 VAC CT or use two 28 VAC transformers. The LM317T is the positive regulator.

[rodd]

I consulted the National Semiconductor datasheet again just to make shure, but, as I thought, there is not an aplication like that in it.
I need to vary both the voltage and the current.
Do you have a different document?
Thanks

[Chris Foley]

Hi, Rogerio. From what I hear you saying, you need one or two sourcing/sinking switchable voltage/current sources, which can supply from -30 to +30 V in voltage mode, and -1.5A to +1.5A in current mode, controlled by a voltage/current switch and an external ramp signal? This is what you have in a standard curve tracer. Tall order. If that's what you want, you may want to look at power op amps such as the venerable National LM12. I believe you can power it at +/-35V regulated as long as you only need 1.5A, and you should be able to get your +/-30V. Quick and dirty would be to use one as a bidirectional voltage source, a second as a bidirectional current source, use a switch to transfer between the two, and use an external sawtooth ramp signal to provide the op amp inputs. Tall order, and LM12s aren't cheap (about 20 bucks a pop at DigiKey).<p>As an older, practical techie solution, have you considered using a Variac-controlled 25.2VAC transformer with diodes and series resistors to limit/steer current? For a quick hack curve tracer, that, two DC power supplies for biasing if necessary, and an oscilloscope with X and Y inputs (most scopes have them, but you may have to look in the back of the box) should be sufficient to characterize transistors. In days of yore trannies weren't all that reliable, and it was better to test 'em first. Not only that, but many engineers weren't very familiar with them, and a fair number of designs were based on optimistic betas. This used to be a popular way for technicians to do what you're shooting at. The scope display is a little different than you're used to with a curve tracer, and you're only going to get one curve at a time.<p>We can create our own tools from spare parts and the junk bin -- they may be "unique", but that's part of their charm. Happy hunting.<p>[ June 21, 2003: Message edited by: Chris Foley ]</p>

[grant fair]

Do you need independently variable voltages and current limits? Would a dual tracking supply work?<p>Grant

[rshayes]

The classic curve tracer was the Tektronix Model 575. The collector voltage was full wave rectified from a line powered transformer. Two windings were available, one with a peak voltage of 20 volts, the other with a peak voltage of 200 volts. Lower peak voltages were obtained using a variac on the primary of the transformer. Maximum currents were 10 amps and 1 amp respectively. This was in the same case size as the Type 545 Oscilloscope, and probably weighed about as much.<p>The base was driven by a step generator, which changed value each quarter cycle of the AC line.<p>Bipolar transistors are normally tested with a current source driving the base, while MOSFETs require a voltage source driving the gate.<p>Very few MOSFETs require a gate drive above 10 volts, and most have maximum gate voltage specfications of 20 volts or below. The main problem is supplying current to charge the MOSFET capacitances, and this can probably be supplied by an op amp with emitter followers added, since the rise time can be a hundred microseconds or so. Fifteen volt power supplies can give outputs up to at least 11 volts, and this would be adequate for most devices.<p>Bipolar transistors require a current drive with a compliance range of a couple of volts. Op amps may be a little awkward here, since a floating method of current sensing would be necessary. Possibly some form of open collector circuit could be used. Unless low beta devices are being tested at high currents, a few hundred milliamps should be adequate.<p>There is no need for these step generators to be bipolar. Four step generators (positive voltage, negative voltage, positive current, and negative current) could be used and selected by switches.<p>I believe that EICO also made a transistor curve tracer as well as Heathkit. These were both accessories to a separate oscilloscope, which made them much smaller and lighter than the Tektronix instruments. I suspect that they were also lighter duty instruments that could not test at high currents. Some of these show up occaisionally on eBay. The Tektronix curve tracers and similar instruments probably also show up on eBay, but the shipping would be a small fortune.

[rodd]

First of all, thank you for your replies to this topic.
Now, let me explain what I am trying to do a little further:
In the IT-3121 the VCE voltage is obtained from the secondary of a transformer, after a half wave rectifiyng section (no capacitor). The amplitude is controlled by transistors arranged in Darlington. The current is limited by a "bunch" of series resistors, selected by a rotary switch.
Elektor Eletronics Magazine published a circuit of a transistor curve tracer some time ago that used a very clever design using just two IC's for the the step generator (Vbe) and ramp generator (Vce).
The shortcomings are fixed Vce and IC max.
Also, it was not appropriate for testing FET's.
My idea is to use a variable power supply with adjustable output and variable current limit (instead of the series resistors)to power the circuit and add the circuit required to generate the current steps needed to test the FET's.

[grant fair]

An obsolete IC, the RC4194 (also SG4194) produces dual tracking from just over 0 volts to +_42 volts. It's only good for 200 ma, but pass transistors will up the current output. You can use a DA convertor to adjust the output. They show an 8 bit DAC but I assume a 12 bit version would give more range and resolution, and you could control the DAC with an old computer. I know I made a supply with a related chip, the RC4195, which includes potentiometer controlled current limiting, though I don't have the circuit. I can always look at the supply and work that out if you are interested. Seems like not bad technology for the 1970's.<p>Fairchild has a datasheet. They sold the chip tp Arrow/Zeus and state to direct all inquires to them. And Arrow/Zeus list Fairchild as a supplier, but without any stock!<p>If it would do the trick, I am sure an obsolete chip finder will have one for you, it will just take some looking on the 'net. <p>Grant

[rodd]

Hi folks,
In face of the difficulties, I am accepting to increase the part count for this supply.
I am thinking of using two regulators for each section, one configured as a current limiter and the other as a variable voltage regulator.
The drawback is the value of the sensing resistor, wich is always in the range of 250 ohms or less.
Since I would have two sections (positive and negative), this would imply finding a dual 250 ohm potentiometer, which I have never seen.
Is there any way to modify the standard current limiting circuit that is presented in the datasheets to use a more standard pot value (like 1k)?
Any suggestion are welcome.