A while back there were 24VDC 6A switching power supplies everywhere for $15. I bought more than I should have with the thought of turning them into a bench power supply with some real oomph behind them.

I've been doing some drafting with the thought of discussion. I may build it, I might not.

I had a minor epiphany in that a simple high speed dual op amp could be used as controllers if the voltage to the op amp power supply regulators were higher than the voltage they were controlling. So I came up with this circuit...

This is a simple high frequency inverter that uses a hand wound toroid that will extend the power supplies to the op amps about ±5V beyond the base 24V 6A power supply. The op amps would be powered by 34VDC.

This will allow this simple current regulator to work...

Followed by this voltage regulator...

I did not bother to show the pins or power supply pins on the op amps, it is assumed. I also did not show the meters I would use if I actually build this sucker.

What can I say, my idea of fun on a Friday night. Discuss (but not my idea of fun).

 

So you're going to use the opamps to control the switching power supply module? Don't they already have a pot to control the voltage or current limiting?

Re the little inverter circuit, I'm assuming it is a standard 555 and not a CMOS version, so it would be good to use the output pin 3 as a current sink, not source. That way it can have better saturation and current abilities.

Expanding on that, if you use pin 7 (discharge pin) that is the same phase as the output pin3 BUT is an open-collector driver, so it can drive the transformer primary at 24v! (transformer primary between +24v DC and pin 7).

So your inverter efficiency just went through the roof.

Also you really need a good sized cap on the power pins of the 555 if you decide to stay with driving the transformer primary with 12v through the zener. Maybe a 0.33uF poly or better still a tantalum of a few uF.

If you drive the primary with 24v from pin7 then that's not so critical, but i would still use a 0.1uF cap across the 555 power pins.

 

The switching power supply is just a fixed 24V source, it could just as easily been batteries. The base circuit is pure analog. This would be a 0-20VDC 5A bench power supply, If I find a cheap 6V or 12V 6A switcher it might become a 26 or 32V variable power supply.

I thought about putting a cap on pin 3, I just need AC for the transformer. I'll add a filter cap on the 555 power too.

Thanks for the thoughts.

 

Comparing with your current limiter, my method was used easy way and has a little leakage current flows through the VR1, the DPDT switch was keeps for no current limited.

 

First image -

The transformer primary drive is an 11 V squarewave sitting on 5.5V DC (-ish). Can the transformer handle the DC component without saturating on the peaks?

Are the SD1-8 diodes in backwards?

Second image -

Because the reference voltage for the current limit set point is not independent of the voltage supplying the current being limited, there is a possible oscillation condition based on the speed of the CL opamp vs the control loop bandwidth of the bulk ps.

And a separate thought: Since the current requirement is basically just the static current of two opamps with 10K loads, couldn't you run the 555 through an 18 V zener instead of 12 V, and have the output drive two diode-capacitor voltage boosters? Compared to the first image, 2 more Cs, 4 fewer Ds, no iron.

ak

 

Comparing with your current limiter, my method was used easy way and has a little leakage current flows through the VR1, the DPDT switch was keeps for no current limited.

Mine should be perfectly linear and uber stable. Of course, I haven't tested it yet.

 

First image -

The transformer primary drive is an 11 V squarewave sitting on 5.5V DC (-ish). Can the transformer handle the DC component without saturating on the peaks?

Are the SD1-8 diodes in backwards?

Like I said, hand wound torrid. All I want is a few volts, which it should do fine. Even if the torrid does saturate, I suspect I'll get enough voltage out of it to use. That schematic is also obsolete, see post #3. I just haven't redrawn it yet.

As to the diodes, just double checked. Nope, it is a simple diode bridge, redrawn.

Second image -

Because the reference voltage for the current limit set point is not independent of the voltage supplying the current being limited, there is a possible oscillation condition based on the speed of the CL opamp vs the control loop bandwidth of the bulk ps.

The 24V is hard, not variable. All the reference that is needed for that circuit. The output of the voltage regulator will be both current regulated and voltage regulated, both being variable by a front panel control.

And a separate thought: Since the current requirement is basically just the static current of two opamps with 10K loads, couldn't you run the 555 through an 18 V zener instead of 12 V, and have the output drive two diode-capacitor voltage boosters? Compared to the first image, 2 more Cs, 4 fewer Ds, no iron.

ak

The power supply for the op amp (which the 555 provides) must be isolated. Vcc ends up being around +29VDC, Vdd around -5VDC). I could have used one diode instead of SD1-4, but I wanted to load both the source and sink of the 555 equally. The voltage to the 555 is 12VDC, a very comfortable voltage, and the output of the 555 is 10V p-p, which also happens to be the RMS since it is a square wave. I am pretty sure the outputs will be pretty pure 5V each.

I was considering adding more taps (secondary windings) to use digital voltmeters for the front panel display, but I haven't really decided on that yet.

The max pots (voltage and current) are a cheat to save parts, and are internal to the instrument.

 

Hmmm... Maybe I'm missing something...

In the first image, the bridge diode anodes are connected to the 100 uF +, and the cathodes are connected to the -.

Attached is a prelim schematic for the no-iron, double-bootstrap power supply I tried to describe. Diode types and capacitors can be adjusted for performance and/or inventory, but this covers the concept.

ak

 

Second image -

Because the reference voltage for the current limit set point is not independent of the voltage supplying the current being limited, there is a possible oscillation condition based on the speed of the CL opamp vs the control loop bandwidth of the bulk ps.

ak

I thought the same thing, but about the voltage control loop in the 3rd photo. I'd like to see a stability analysis to feel better about that circuit.

 

Because the darlington emitter follower does not add any voltage gain, the voltage loop usually is stable with just the compensation built into the opamp. Sometimes I see a small capacitance in parallel with the series feedback resistor (1/2 the pot in this drawing), particularly with large output capacitor banks.

ak

 

Hmmm... Maybe I'm missing something...

In the first image, the bridge diode anodes are connected to the 100 uF +, and the cathodes are connected to the -.

Attached is a prelim schematic for the no-iron, double-bootstrap power supply I tried to describe. Diode types and capacitors can be adjusted for performance and/or inventory, but this covers the concept.

ak

OK, you got me. I am wrong about the diode bridge.

I have never had much luck using voltage doublers for power supplies. I have personal spec of 10ma for this circuit, no reason other than an arbitrary choice.

It would probably work well.

 

Because the darlington emitter follower does not add any voltage gain, the voltage loop usually is stable with just the compensation built into the opamp. Sometimes I see a small capacitance in parallel with the series feedback resistor (1/2 the pot in this drawing), particularly with large output capacitor banks.

ak

I don't see that matters, as the loop gain can be high, depending on the setting of R6. Also depends on the load, which tends to scare me. I would also combine the voltage and current limits and just use a single pass element, but I think the OP is off this circuit now.

 

OK, the 555 inverter section, redraw.

Now I look for meters, and possibly draw the circuits to support them. LED and LCD front panel meters are picky about their power supplies, they absolutely require an isolated source. I may use mechanical types, it would be nice if I could use R1 as the shunt resistor.

BTW, The gain for the op amp on the voltage regulator would be X2. 12VDC voltage reference, 24 (or less) VDC out.

 

BTW, The gain for the op amp on the voltage regulator would be X2. 12VDC voltage reference, 24 (or less) VDC out.

Nope. It depends on the setting of R6.

 

Re the new 555 circuit, I'd make C2 at least the size of C3.

Or add a second cap; C2b 10uF across C2 0.1uF.

Otherwise your forward current into the transformer primary will be coming through the zener as peak current. Much better to draw from a large cap on Vdd, than draw high peak current spikes though the zener.

I still prefer my suggestion of driving the primary from pin 7 and 24v. That will be much better for circuit efficiency and only needs a small wattage zener now.

You don't need AC to drive the transformer you just need to pulse a magnetic field (build up current) and let it collapse into the secondary. Might even save you 6 of those secondary schottky diodes and some PCB space too.

 

What kind of IC is your switcher built around?

 

The switcher? Like I said, a commercial unit. They were pretty common around 2 years ago at a steal of a price from many sources. I suspect they will have short lives.

 

Nope. It depends on the setting of R6.

Yes, and the setting of R6 is such that the gain is 2.

ak

 

Nope. It depends on the setting of R6.

You speak as if the values are not known in this circuit. I even quoted what the values are to you. As I said earlier, rather than add extra resistors I just cheated and used a simple variable to replace all the parts.

 

OK, the 555 inverter section, redraw.

Now I look for meters, and possibly draw the circuits to support them. LED and LCD front panel meters are picky about their power supplies, they absolutely require an isolated source. I may use mechanical types, it would be nice if I could use R1 as the shunt resistor.

BTW, The gain for the op amp on the voltage regulator would be X2. 12VDC voltage reference, 24 (or less) VDC out.

If this is a pure hobbyist project. I would perhaps have looked in my boxes for a small transformer to use for the meters. And used a linear regulator approach. May make the meters more stable