0-30V 0-3A Latest Data Discussion

[deepak.roy.alchemist]

I have a question. In your particular circuit the maximum voltage is 33V (and including the -ve rail it is 34.3 volts). So will a CA3140 opamp work well as the max voltage stays within its limits. Actually a MC34071 is not available in my locality.
I will be eagerly awaiting your kind counsel.

 

[audioguru2]

I have a question. In your particular circuit the maximum voltage is 33V (and including the -ve rail it is 34.3 volts). So will a CA3140 opamp work well as the max voltage stays within its limits. Actually a MC34071 is not available in my locality.
I will be eagerly awaiting your kind counsel.

The MC34071 is no longer available in a through-hole package (it is only surface-mount now) so the TLE2141 opamp from Texas Instruments is used.
I think a 24V transformer produces peak rectified voltages of 32V at 3A and ripple reduces the minimum voltage to about 30V. That is why the circuit shown with the 24VAC transformer is rated for an output voltage of only 25V. The original project also used a 24VAC transformer and also could not produce 30V at 3A.

An old CA3140 opamp is very noisy and has a high maximum input offset voltage. Its supply voltage might be higher than its maximum allowed voltage of only 36V when there is low load current or no load. 

 

[elboa8]

I thought I would outline the voltage spike problems with the PSU. If anyone is not sure of the issues, I will try and outline them here. If for example, you use a 5v 100mA bulb as the load, then set the current pot to maximum and the voltage pot to 5v, all is well. You then switch off the mains supply, and because the voltage is controlled by the physical position of the pot there is NO spike it decays normally. BUT, using the same bulb, you set the voltage to maximum and set the current to 100mA, then switch off by the mains, things are very different! The -1.3 volts on the current control IC (pin 4) collapses almost immediately. The low output (pin 6) which was pulling down the input of the voltage control IC (pin 3) goes high and the output follows to the set level of the voltage pot, in this case maximum. The result is a rapid rise of the output voltage to its maximum. NOT good! Even with the transistor fitted to short the drive from the ouput of the voltage control IC at switch off it is not fast enough to stop the output spike. I spent a good while on this problem but as the supply spikes at switch on as well, and the PSU works very well in all other areas, I’m going to leave it, as is (no switch off clamp transistor) and look at a separate on off switch, perhaps using a FET. More later.

 

[liquibyte]

I thought I would outline the voltage spike problems with the PSU. If anyone is not sure of the issues, I will try and outline them here. If for example, you use a 5v 100mA bulb as the load, then set the current pot to maximum and the voltage pot to 5v, all is well. You then switch off the mains supply, and because the voltage is controlled by the physical position of the pot there is NO spike it decays normally. BUT, using the same bulb, you set the voltage to maximum and set the current to 100mA, then switch off by the mains, things are very different! The -1.3 volts on the current control IC (pin 4) collapses almost immediately. The low output (pin 6) which was pulling down the input of the voltage control IC (pin 3) goes high and the output follows to the set level of the voltage pot, in this case maximum. The result is a rapid rise of the output voltage to its maximum. NOT good! Even with the transistor fitted to short the drive from the ouput of the voltage control IC at switch off it is not fast enough to stop the output spike. I spent a good while on this problem but as the supply spikes at switch on as well, and the PSU works very well in all other areas, I’m going to leave it, as is (no switch off clamp transistor) and look at a separate on off switch, perhaps using a FET. More later.

Q1 solves shutdown transients for me, and I've tested at multiple settings and a few different load situations.  I've posted a PFET circuit that simulates well but I haven't tried yet (not much time) but the consensus is that it should solve the startup transient rather well.

I'm now going to split this conversation off to a separate discussion as to not pollute the finished projects posting thread.

 

[elboa8]

Hi, Yes, adding the “shut off” transistor helps, but it’s not the full answer. The collapse of the -1.3v rail is causing the problem and at the same time is trying to clamp the output via the “shut off” transistor.! Looking at the switch ON spike, I used the following test set up. I connected a 12v 500mA bulb to the supply. I set the voltage to maximum and the current limit to 50mA. At turn on you get a huge (repeatable) spike. (See att. Jpg) , before the current limit “kicks in”.

 

[liquibyte]

Hi, Yes, adding the “shut off” transistor helps, but it’s not the full answer. The collapse of the -1.3v rail is causing the problem and at the same time is trying to clamp the output via the “shut off” transistor.! Looking at the switch ON spike, I used the following test set up. I connected a 12v 500mA bulb to the supply. I set the voltage to maximum and the current limit to 50mA. At turn on you get a huge (repeatable) spike. (See att. Jpg) , before the current limit “kicks in”.

Like I said, try this and see if it works.  It should delay startup just long enough to get past the transient.  It's a hackish solution but the best I've been able to come up with so far without using a soft start on the primary of the transformer.  I've been told to do some research into the older HP supplies to see if something there might help but I just haven't had time lately.  I'm fairly new to the game so at this point it's still an experiment but in theory should work.  That's an excellent screenshot by the way, it shows exactly what I've been trying to see but can't because I have an analog scope.

 

[audioguru2]

When the load current is higher than the current regulator setting and the voltage is set high, the "on" voltage spike is caused by the time it takes for the negative supply to reach a low enough voltage (-1V?) for the current regulating opamp to pull down the voltage feeding the voltage output amplifier.

Here is my proposed fix: 

View attachment 42332

 

[redwire]

When the load current is higher than the current regulator setting and the voltage is set high, the "on" voltage spike is caused by the time it takes for the negative supply to reach a low enough voltage (-1V?) for the current regulating opamp to pull down the voltage feeding the voltage output amplifier.

Here is my proposed fix:

Yes,  I agree the proposed fix would likely give rid of a lots of parts and simplify the board.  When you add the two addtional transistors, is Q3 still needed or will the lower transistor connect to the led instead?  One thing I am curious about is how low the output will go with the recommended arrangement when the negative rail is deleted?
While I didn't build the suggested fix, a while back I disconnected the negative rail and connected Pin 4 of U3 to ground, the minimum output voltage  nearly doubled.

 

[audioguru2]

Yes,  I agree the proposed fix would likely give rid of a lots of parts and simplify the board. When you add the two addtional transistors, is Q3 still needed or will the lower transistor connect to the led instead?

Good point. Eliminate Q3, turn the polarity of the LED around so its cathode is at the collector of the new lower NPN transistor and its current-limiting resistor is between the anode of the LED and the +27.6V.

One thing I am curious about is how low the output will go with the recommended arrangement when the negative rail is deleted?
While I didn't build the suggested fix, a while back I disconnected the negative rail and connected Pin 4 of U3 to ground, the minimum output voltage  nearly doubled.

Another good point but there is no problem.
If the output is shorted then we want the output voltage to go to 0.00V. The new upper NPN transistor has its emitter connected to the circuit ground. Its collector is connected to the input of the output amplifier that must be +1.41V higher (the voltage drop for 3A in the current sensing resistor) for an output voltage of 0.00V when the output is shorted and the current is set to 3.0A. So the transistor does not need to saturate.
The output amplifier does not need a negative supply for its output voltage to be 0.00V because the output of the opamp drives the darlington-connected driver and output transistors so its voltage is two diode drops higher.

 

[liquibyte]

Wouldn't elimination of the negative rail mean that we could get rid of the 10V zener as well?  Something like this?

 

[audioguru2]

Wouldn't elimination of the negative rail mean that we could get rid of the 10V zener as well?  Something like this?

A 28V transformer might produce 29V when the project has no load. Then its 41.0V peak voltage is reduced to 39.6V by the rectifier bridge. It will be fine if the mains voltage is very stable.

 

[liquibyte]

I'm not sure how stable my mains is but I remember measuring it at 122V when I started working on this project.  Somewhere around here I wrote down the values I got out of the transformer(s) before the rectifier, after the rectifier, and after the filter caps.  I distinctly remember 43.7 volts after the filter caps because that's always been a concern of mine given the limits of the 2141's.

 

[elboa8]

Hi. Out of interest I set up the same test as before. Max voltage low current set into a bulb. Only this time I supplied the -1.3v via a permanent external source. Remember I’m not using a “clamp” transistor. The switch off and switch on were reduced but NOT eliminated. So it would appear reworking the –ve supply will help, but will not solve the problem completely! I’m going to stick with my original thoughts, the supply as it stands works well (except for the spikes). So I’m working on a separate DC toggle switch. I’ve attached, my final circuit, the switch on spike with permanent –ve 1.3 and my prototype toggle switch with  fast close down.

 

[elboa8]

Quick update on the toggle switch. After a couple of tweeks it's working fine. It defaults to OFF at startup. No spike! And it closes down fast at switch off. I'll play around a bit more and get some scope waveforms.

 

[elboa8]

Leaving aside the problems for a minute I’m very happy with the overall performance of the PSU. The two pictures show the supply with a 3A load, voltage at 5.51v and with no load the voltage is still showing 5.51v.
The two pass transistors are dissipating approximately 60W. Yes, you need a big heat sink and fan. I obtained these regulation figures by returning the voltage control IC 0v rail to the “other side” ( –Ve sense ) of the current sense resistor.
Footnote: The temp. stabilizes at about 53 c.

 

[redwire]

Arklite    take a look at this PS output at start up using current control.

 

[elboa8]

It looks good. It has moved quite away from the original. Perhaps not a bad thing. Still it's fun to experiment. I also changed the voltage reference on my circuit for the same reasons as you. Regards...

 

[liquibyte]

I've heard this from several people

One thing I would avoid which this design does is having one control loop drive the other.

which seems to be where this is headed and may ultimately be the cause of the problems in the first place.

 

[elboa8]

I went ahead and made a PCB (Diptrace) for the toggle switch. It all works fine. After some more tests I'll buy, or make a case for it. I found some nice 5 digit LED displays on eBay (where else) for the finished PSU.