Electronics > Projects, Designs, and Technical Stuff

Classic linear lab power supply project

(1/1)

bogdan2014:
Since I have a bunch of transformers and a box full of 2N3055, I decided to build a linear lab power supply. Since I'm quite bad at linear circuits, I studied a few schematics until I came up with something that's close to what I want.
Below is the simulation of the main circuit. The main pass transistor is biased by Q1, and U1 and U3 are used for voltage and current feedback. V2 will actually be a TL431 5V reference and a potentiometer, while V3 will be taken from the reference, also adjustable with a potentiometer.

First, I would like to understand how Q3 and Q4 work in that configuration, it seems very strange that they're in series.
Another issue is with adding U2 into the schematic, which makes the output of the PSU unstable. It is used as a gain of 4 amplifier so that lower value current sense resistors can be used, and without it the output is stable. I suspect is some compensation issue, so I'd like to know how to solve it.

The final circuit will also have a tracking pre-regulator based on LM2576, but I couldn't find a way to simulate it.

Kleinstein:
Using the simulation is one way to test / optimize the compensation. There are mainly two way of doing this:
The more classical one is plotting the bode plot for the loop gain. For this one adds a source to the loop and calculates the loop gain as function of frequency. This is done separately for the voltage regulation and current regulation. By variations of some circuit elements one can "experimentally"  identify which part / corner of the curve is effected by the parts (especially caps).
The target is to get sufficient phase reserve at unity loop gain. As a complication one may have to do the test with different loads.


The second, more modern / heuristic version is looking at the output impedance (for voltage regulation) or admittance (current regulation). For this purpose a spice current source (voltage regulation mode) or voltage source (for current regulation) is used as a load. The impedance is that easy to calculate.  The targes here is to make sure the impedance has not too much phase shift, especially in the frequency rage a few decades below unity loop gain and above.  Try and error to see the effect of the difference caps / resistors is about the same as before and also independent tests for CV and CC mode. The advantage here is that one does not have to try different load impedances (e.g. extra external caps) - but still different DC loads. This method may not detect cases of grossly unstable circuits (e.g. those that would not be stable with essentially any load).

The next part is than to check the cross over between the 2 modes. This is usually done in transient mode, e.g. with a current source as a load and a parallel diode to prevent the voltage to go all the way negative. An otherwise stable design may still fail here. The tricky part is coming out of saturation. It may need more phase reserve in the frequency mode to get it stable.

Simulating the preregulator is usually done separately. Switched regulators can be quite tricky and may need a simulation in transient mode, which can take quite some time.

MosherIV:
Hi

I looked for ages for some linear desgin that is easy to understand.
Eventually I found this one:
http://www.microsyl.com/index.php/2010/03/31/bench-power-supply-0-25v-0-5amp/comment-page-2/

The voltage control side is a simple voltage follower.
The current control side is ORed via diodes so that when the current threshold is reached, the ORing pulls the output voltage down to maintain the current limit.
The current is read accross the high side shunt resistor by the ZTX device and fed into the current limit comparitor/op-amp.

xavier60:

--- Quote from: bogdan2014 on October 14, 2019, 02:03:55 pm ---Since I have a bunch of transformers and a box full of 2N3055, I decided to build a linear lab power supply. Since I'm quite bad at linear circuits, I studied a few schematics until I came up with something that's close to what I want.
Below is the simulation of the main circuit. The main pass transistor is biased by Q1, and U1 and U3 are used for voltage and current feedback. V2 will actually be a TL431 5V reference and a potentiometer, while V3 will be taken from the reference, also adjustable with a potentiometer.

First, I would like to understand how Q3 and Q4 work in that configuration, it seems very strange that they're in series.
Another issue is with adding U2 into the schematic, which makes the output of the PSU unstable. It is used as a gain of 4 amplifier so that lower value current sense resistors can be used, and without it the output is stable. I suspect is some compensation issue, so I'd like to know how to solve it.

The final circuit will also have a tracking pre-regulator based on LM2576, but I couldn't find a way to simulate it.

--- End quote ---
That circuit contains the typical mistakes found in most dysfunctional internet findings. It will be very difficult to stabilize, if not impossible.
Others have been down the same path.
I have been helping another member with this design, https://www.eevblog.com/forum/beginners/lm324-power-supply-with-variable-voltage-and-current/?action=dlattach;attach=854250
https://www.eevblog.com/forum/beginners/lm324-power-supply-with-variable-voltage-and-current/msg2736780/#msg2736780
It is mostly wart free.
One thing to be aware of is the implications of the low side current sensing which can cause complications when adding extra control and measurement circuitry.
It works fine as a basic bench supply regulator.

Navigation

[0] Message Index

There was an error while thanking
Thanking...
Go to full version
Powered by SMFPacks Advanced Attachments Uploader Mod