How does this regulated power supply work?

[Toe Cutter]

Hello. I am interested in building a regulated laboratory-type power supply with variable voltage and variable current limiting. I've been looking around the net for schematics and ideas. I would like to avoid the LM317/ua723 style supplies and just use transistors and op-amps.

I found the schematics for the old HP linear supplies and I think they have some good ideas, but I can't quite understand one aspect of this design. I've attached images for the HP 6502B schematic and a simplified drawing from the HP app note 90B.

Most of it I can follow. What they do with this supply is have a floating reference/bias supply with a precision +/- 6.2 volts. The common of this floating supply is tied to the (+) output and one input of a differential amplifier. The output of the diff amp eventually drives the series regulator.

The +6.2 volt, regulated supply and a precision resistor (R12/R13) form a precision current source. This is tied to the other input of the differential amplifier and shunted by the voltage setting potentiometer.

I can sort-of, almost see how this scheme works to regulate the voltage, but I am missing something. Where does the reference current through R12/R13 go? It has to return to the reference regulator somehow. Does it flow through the R10 setting potentiometer? This is the part I am getting stuck at. Could someone help me better understand how the voltage setting and feedback works? Also, I found an IC, the MC1466, which uses a very similar concept. But this IC is long obsolete.

I find this idea, with the floating reference and auxillary supply appealing because it eliminates the requirement to use components rated for higher voltages. For example, I have a transformer that outputs 60VDC (rectified and filtered). Without an auxillary supply, I might need op-amps that have a > 60V rating. Thanks!

 

[RCinFLA]

The answer to your question is R10A and R10B.

You can build a similar circuit with a LM723 regulator I.C.

Be aware of the power dissipation in a linear regulator's series pass transistor if there is a large voltage drop across it with moderate current load.

Most new supplies with wide voltage range and several amps of capability use a tracking pre-regulating switching power supply that keeps the input to the final output linear regulator several volts above the output voltage. This gets the heat dissipation down while still providing a clean output with a final linear regulator.

A lab supply should have an adjustable current limiting capability.

 

[Toe Cutter]

I thought that R10 would be the only path for that current; maybe a little bit of that current flows into the transistor base. But, I still can't see where the programming current goes. There could be a voltage drop of 0-40 volts across R10. The programming current would then have to make its way back to the reference regulator, but I can't see the path it would take. Am I missing something?

That's interesting that manufacturers will use a switch mode supply as a pre regulator. Cuts down on the power dissipation. In the big, old HP supplies, I've seen an SCR pre-regulator that varies the input AC voltage. With the transformer I have, I could probably build a supply at 50-55 volts and 1-2 amps max. It might have to dissipate 100-120 watts, worse case. I think the pre-reg ideas are nice, but more complexity than I want. I have some big beefy transistors and heat sinks, so they will probably have to take all the heat.

 

[Diver300]

The return path is via ground to the negative of the main bridge, through the bridge, through Q7, R54 and connection no. 8.

You can follow it a lot more easily on the simplified schematic. The return path has to exist, but it can go up as well as down in voltage, if you have a power source.

 

[crutschow]

At work we had some old Power Design supplies that used a variac to reduce the voltage to the regulator and keep the dissipation proportional to current but independent of the voltage output. The Variac had a pot on the end of the shaft that caused the linear regulator output to track a few volts below the unregulated voltage from the variac output rectifiers.

I believe the variac must have had two windings to provide isolation, unlike a standard variac, which uses only one winding and does not provide isolation.

 

[Diver300]

I believe the variac must have had two windings to provide isolation, unlike a standard variac, which uses only one winding and does not provide isolation.

On a manually variable power supply that I used to work with, rated 0 - 5 V at 200A ac, there was a large variac, that didn't isolate, followed by a fixed 240 - 5 V isolation transformer. Could yours have had a second transformer?

 

[Reloadron]

I may be off base with this but something you may want to note is when they built constant voltage comparators using transistors the transistors like Q1 A & B were generally Matched pairs and used a little heat sink they shared for a common thermal environment. A Google of transistor matched pairs should get you more information.

Ron

 

[crutschow]

On a manually variable power supply that I used to work with, rated 0 - 5 V at 200A ac, there was a large variac, that didn't isolate, followed by a fixed 240 - 5 V isolation transformer. Could yours have had a second transformer?

I didn't actually look inside so they could have, but it would take about half the iron and less cost and space if they just added a second winding on the variac. So from an engineering (and KISS) point of view, I go for the second winding.

 

[Diver300]

I didn't actually look inside so they could have, but it would take about half the iron and less cost and space if they just added a second winding on the variac. So from an engineering (and KISS) point of view, I go for the second winding.

On the other hand, if variable auto-transformers and isolated fixed transformers are both off-the-shelf products, then you don't need to pay tooling charges.

There could also be a problem with making a low voltage variac. There can always be at least one shorted turn where the wiper moves from one secondary winding to another. If the secondary windings are large, that would cause a large short-circuit current. Also the adjustment step would be a large fraction of the voltage.

In the case of the 5 V, 200 A power supply that I used, I guess the transformer would give 1 V per turn, maybe more. So a variac would end up as a 0 - 1 - 2 - 3 - 4 - 5 V switch, which gives coarse adjustment. If a 240 V variable auto transformer is used, followed by a fixed 240 - 5 V transformer, the variac would also be about 1 V per turn, but that would only be 1/240*5 = 0.02 V, so the adjustment is much finer.

 

[crutschow]

On the other hand, if variable auto-transformers and isolated fixed transformers are both off-the-shelf products, then you don't need to pay tooling charges.

There could also be a problem with making a low voltage variac. There can always be at least one shorted turn where the wiper moves from one secondary winding to another. If the secondary windings are large, that would cause a large short-circuit current. Also the adjustment step would be a large fraction of the voltage.

In the case of the 5 V, 200 A power supply that I used, I guess the transformer would give 1 V per turn, maybe more. So a variac would end up as a 0 - 1 - 2 - 3 - 4 - 5 V switch, which gives coarse adjustment. If a 240 V variable auto transformer is used, followed by a fixed 240 - 5 V transformer, the variac would also be about 1 V per turn, but that would only be 1/240*5 = 0.02 V, so the adjustment is much finer.

Tooling charges are likely not a big factor for production quantities of a power supply.

That's a good point about the variac output voltage. But with two variac windings you could make the secondary fewer turns than the primary to give the desired voltage output at full scale. And the adjustment step size is not critical since the output goes to a regulator.

Here's a good article the discusses low voltage variacs with dual windings and also the problem of shorted turns by the brush.

 

[Reloadron]

While off topic I have a large 3 phase power supply (older style) we use for 0 to 200 VDC capable of 100 amps. We come in with 480 VAC 3 phase to a large motor driven three phase variac. That variac in turn drives three large transformer primaries with those secondaries feeding a large diode bank. The thing is a brute unregulated linear supply. You move it with a fork lift and actually it is hard wired to the 480 volt 3 phase mains.

Each large stud mount fuse has a small shawmut trigger around it. The triggers smack a small bar with micro switches. The idea being if any one of a number of fuses fails you get an instant shutdown. You do not even want to know what 600 volt rated 200 Amp stud mount fuses cost.

Ron

Ron

 

[bountyhunter]

Hello. I am interested in building a regulated laboratory-type power supply with variable voltage and variable current limiting. I've been looking around the net for schematics and ideas. I would like to avoid the LM317/ua723 style supplies and just use transistors and op-amps.

I found the schematics for the old HP linear supplies and I think they have some good ideas, but I can't quite understand one aspect of this design. I've attached images for the HP 6502B schematic and a simplified drawing from the HP app note 90B.

Most of it I can follow. What they do with this supply is have a floating reference/bias supply with a precision +/- 6.2 volts. The common of this floating supply is tied to the (+) output and one input of a differential amplifier. The output of the diff amp eventually drives the series regulator.

The +6.2 volt, regulated supply and a precision resistor (R12/R13) form a precision current source. This is tied to the other input of the differential amplifier and shunted by the voltage setting potentiometer.

I can sort-of, almost see how this scheme works to regulate the voltage, but I am missing something. Where does the reference current through R12/R13 go? It has to return to the reference regulator somehow. Does it flow through the R10 setting potentiometer? This is the part I am getting stuck at. Could someone help me better understand how the voltage setting and feedback works? Also, I found an IC, the MC1466, which uses a very similar concept. But this IC is long obsolete.

I find this idea, with the floating reference and auxillary supply appealing because it eliminates the requirement to use components rated for higher voltages. For example, I have a transformer that outputs 60VDC (rectified and filtered). Without an auxillary supply, I might need op-amps that have a > 60V rating. Thanks!

You never said what output voltage/current the power supply is supposed to have?

The schematic you posted is a cheap (and old) way of making a voltage rail to feed circuitry. There are much better designs for lab supplies, I have posted one before.

If you just want a lab supply with voltage/current regulation y6ou can leave out the over-temp and auto parallel circuitry in the middle.

 

[Toe Cutter]

The return path is via ground to the negative of the main bridge, through the bridge, through Q7, R54 and connection no. 8.

You can follow it a lot more easily on the simplified schematic. The return path has to exist, but it can go up as well as down in voltage, if you have a power source.

OK, I guess that is the only return path. But it still seems counter-intuitive to me, that the current would have to go down ~40volts potential and then back up ~40 volts. Maybe the fact that these two sub-circuits are powered by different transformer windings has me confused.

 

[Toe Cutter]

You never said what output voltage/current the power supply is supposed to have?

The schematic you posted is a cheap (and old) way of making a voltage rail to feed circuitry. There are much better designs for lab supplies, I have posted one before.

If you just want a lab supply with voltage/current regulation y6ou can leave out the over-temp and auto parallel circuitry in the middle.

Thanks for the schematic. I'll look it over. In one of my earlier posts (#3), I mentioned I have a ~60VDC unregulated supply. The transformer is 44 VAC rms (measured unloaded). I don't know the specs of this surplus transformer, but comparing the physical size to other known transformers, I'm estimating it is a 90-120 VA transformer. Maybe it can output 2 amps max. Taking into account ripple, regulator drop out, copper voltage drop, I was thinking I could build a supply that is 0-50 V and 0-1 A or 0-1.5 A.

I took a quick look at your schematic. 0-16 V with common parts: LM358, 2n3906 etc. If I was to use your schematic with a ~60 V supply I would need parts with a higher voltage rating, and probably have to change the resistor values. Not as many op-amps can work off a 60 V supply. I think this is the advantage of having an auxiliary/bias supply. I could use a LM358 (or any other common op-amp) to control a 60 V supply. Other lab supplies I've seen (HP, lambda, etc) go up to 100 V, 120 V, 250 V, etc. For a lower voltage supply 16 V, 20V, then omitting the auxiliary supply might be best, all the control circuitry can run off the main rail.

 

[bountyhunter]

I was thinking I could build a supply that is 0-50 V and 0-1 A or 0-1.5 A.

I took a quick look at your schematic. 0-16 V with common parts: LM358, 2n3906 etc. If I was to use your schematic with a ~60 V supply I would need parts with a higher voltage rating, and probably have to change the resistor values. .

You don't need 60V opamps. Just break the V+ line feeding the components and run them off 24V. Generate that using a small linear reg or zener plus an NPN transistor.

The component selection on the pass transistors and NPN current source would have to be adjusted.

 

[bountyhunter]

I mentioned I have a ~60VDC unregulated supply. The transformer is 44 VAC rms (measured unloaded). I don't know the specs of this surplus transformer, but comparing the physical size to other known transformers, I'm estimating it is a 90-120 VA transformer. Maybe it can output 2 amps max. Taking into account ripple, regulator drop out, copper voltage drop, I was thinking I could build a supply that is 0-50 V and 0-1 A or 0-1.5 A.

Too bad the transformer secondary isn't center tapped so you could separate it into a pair of secondary windings. A 22VAC (no load) winding is probably about 18 - 20VAC loaded, you could generate about 25VDC unreg supply. Allowing for ripple, you could build a pair of 0-20V outputs that could be stacked or parallelled as shown in my schematic. That's why I built duals, it allows stacking for more voltage or paralleling for more current. You also have two separate outputs which is very useful.

BTW: if a transformer secondary winding is rated for 1A RMS, it is only good for about 0.6A DC power supply. The RMS current in the transformer secondary is about 1.8X as much as the DC load current when the winding is connected to a FWB and capacitor filter.