Voltage regulator circuit that doesn't regulate voltage. Am I doing it wrong?

[alec_t]

The thing that looks very strange on Mike's circuit (Post #17) is the 49 volts RMS at 13 Khz on the output.

That is odd, especially as the V(p-p) is only 68pV . Don't know which simulator Mike uses, so perhaps I'm not interpreting the readout properly?

 

[crutschow]

Strange. It simulates fine in LTspice. No sign of oscillation.

Alec, can you post your .asc file?
Thanks.

 

[Mikebits]

That is odd, especially as the V(p-p) is only 68pV . Don't know which simulator Mike uses, so perhaps I'm not interpreting the readout properly?

I simulated using NI Multisim, which uses the same Spice as LTspice.

NI Multisim (formerly MultiSIM) is an electronic schematic capture and simulation program which is part of a suite of circuit design programs, along with NI Ultiboard. Multisim is one of the few circuit design programs to employ the original Berkeley SPICE based software simulation.

For some reason without the bypass caps, I was getting a convergence error, adding the caps corrected that. Hmm I had not noticed the AC signals being present. I will try and retrace my steps. Also I will place a signal viewer (Sim scope) with the current sim file, and see if I can view anything.

 

[alec_t]

Here's my asc :

 

[evovi]

Many thanks for your schematic and testing Mike. I'll be able to get my hands on some 0.1uf caps tomorrow and report back with my findings. My old scope doesn't show any osculations, but I don't believe it rules out the possibility. I'm visiting a friend who has a better scope so maybe I can try his scope out aswell. I'll post back with my findings asap.

 

[Mikebits]

Many thanks for your schematic and testing Mike. I'll be able to get my hands on some 0.1uf caps tomorrow and report back with my findings. My old scope doesn't show any osculations, but I don't believe it rules out the possibility. I'm visiting a friend who has a better scope so maybe I can try his scope out aswell. I'll post back with my findings asap.

I think I suggested .01uf caps. One thing about scope probes is that they have capacitance (usually small 15pf to gnd), so while using the scope, the probe capacitance may be enough to stop any oscillation. So during probing keep a meter connected to the output and look for a change when you apply your scope probe to the circuit. Does that make sense?
Party on

 

[Mikebits]

I simulated using NI Multisim, which uses the same Spice as LTspice.



For some reason without the bypass caps, I was getting a convergence error, adding the caps corrected that. Hmm I had not noticed the AC signals being present. I will try and retrace my steps. Also I will place a signal viewer (Sim scope) with the current sim file, and see if I can view anything.

I tried to go back and duplicate the error I was getting, and could not reproduce it, so I dunno? The small AC signals showing up in the sim are probably just noise, if there is such a thing as simulated noise?

 

[evovi]

I think I suggested .01uf caps. One thing about scope probes is that they have capacitance (usually small 15pf to gnd), so while using the scope, the probe capacitance may be enough to stop any oscillation. So during probing keep a meter connected to the output and look for a change when you apply your scope probe to the circuit. Does that make sense?
Party on

Yep that makes perfect sense. Tomorrow I'll grab some caps and I will have access to a better scope than I own. I'll test it with and without the caps and report my findings.

 

[Mikebits]

Yep that makes perfect sense. Tomorrow I'll grab some caps and I will have access to a better scope than I own. I'll test it with and without the caps and report my findings.

So what's the word EVO?

 

[evovi]

Hey Mike, just got my hands on the cap's couple hrs. Beleive me I I'm itching to install the cap's asap. At friends place right now. I should get a chance either tonight or tomorrow morning to install them. Bought some new 24v zener as well. Update asap

 

[Mikebits]

Hey Mike, just got my hands on the cap's couple hrs. Beleive me I I'm itching to install the cap's asap. At friends place right now. I should get a chance either tonight or tomorrow morning to install them. Bought some new 24v zener as well. Update asap

Cool, let us know. Also as Alec suggested, double check your transistor pinouts especially the BD139, that one looks easy to flip backwards

 

[evovi]

I've triple check the orientation of the transistors all okay. Ive tried Mike's modified schematic. With adding 4x 0.1uf cap's, didn't solve the problem unfortunately I installed new transistors and a new 24v zener (measured 25.5v across the zener) so it's not doing its job for some reason still.

Edit: not sure if this info is worth anything but when I install new bc546's I can hear a quiet high pitch buzz for about the first 10secs of run time before the buzz stops. It's very quiet and almost not noticeable. Probably the transistors singing their last?

 

[Mikebits]

Try replacing the 220pf cap, it may be damaged. The cap is providing feedback..
Darlingtons are notorious oscillators. The cap should reduce gain at high frequencies, I think, can someone pls confirm if I am correct? Thanks

 

[evovi]

I can try that. But I'm wondering at this point if you have a suggestion for a voltage regulator circuit that is a little more reliable to build?

 

[Mikebits]

No, we are gonna get this to work, and we will both learn something in the process, I will stick it out with you as long as you do. I still think we have an oscillation, your layout (and no offense), really sucks, and may very well be the root of all your problems. If you can get some vero board, I would use that and rebuild. If you like I can make up a suggested layout using Fritzing (Cool software for breadboarding, I just got a copy the other day, You should grab a copy yourself). It is very handy and easy to use. Let me grab ya a link. https://fritzing.org/download/

I' also hope some others have some suggestions, because I have been known to make errors from time to time

 

[evovi]

Sounds good Mike. I agree learning from failed projects is sometimes the best teaching. And yeah the layout on the test board does suck. I'll have a look at that software. It's something I'd probably use a fair amount! I'll grab a couple new components to start fresh.

 

[Les Jones]

I decided to build the regulator because I could not see any reason why it should not work. Also some of the information did not make any sense. For example 32 volts across a 24 volt zenner diode that was later tested and found to be OK. I made some slight changes as I did not have exactly the right components. I used two 12 volt zenners in series as I did not have a 24 volt one. I used BC547s as I did not have any BC546s. Also I did not have the input filtering (The two extra 1000 uF capacitors and the two 22 ohm resistors.) The diodes connected directly to C3 and C4. I initially built it without the 220 pF capacitor between the base and collector of Q3. The DC input to the regulator was almost exactly 60 volts. The voltage on the emitter of Q3 was 23.77 volts. The output was 48.89 with no load and dropped by 0.01 volts wit a 1K load. (49 mA) The theoretical output should be (23.77 + 0.7 (Vbe of Q3)) * 2 = 48.94 This agrees pretty well with the actual output voltage. There was no sign of oscillation. I then added a 200 pF capacitor (As I did not have a 220pF one.) between the base and collector of Q3. Nothing changed. There was no particular care taken building the circuit. It was just rats nest construction on a piece of tag strip. Non of the component leads were trimmed. I think there must be an error in the building of the circuit or one or more faulty components.
It also occurred to me that as the circuit starts with a voltage doubler then a 24 volt positive regulator (7824) and a 24 volt negative regulator (7924) could be used to regulate to 48 volts.

Les.

 

[Mikebits]

Isn't the 220pf just for frequency compensation like an op amp? I would not imagine you would notice any change in the operation. I am a thinking perform an AC sweep of the circuit to see a change in BW. I could be wrong however, as I am not sure. So bottom line is that EVO's circuit most likely has a construction error?

 

[Les Jones]

Hi Mike,
As it did not show any signs of oscillation without the capacitor I did no expect it to make any difference.

Les.

 

[Mikebits]

Well, it simply sounds like the OP should just redo the board as the design should work so any problem is construction related.