Almost Finished Lab PSU Schematic (Based On Mark III)

[xReM1x]

hi!

Specs:
0-25V
0-5A
Pre Regulated By LT1074
CC \ CV LED
Over Temp Protection
(no oscilloscope, cant measure much).

EDIT: forgot to mention that I'm going to use the MAX472 high side current sense that I got some time ago, and 5x 0.5R DALE Resistors. 3W each.


anything to change?

[c4757p]

A few comments:

You're putting 42V across a couple of your opamps. All of those are rated for only 44V, you're cutting it very close - and there will be quite a bit of energy available at 44V, if you get spikes or excess voltage (is the 30V rail regulated?) you could have some impressive holes in your opamps. Any chance of bringing the negative rail in a touch?

You might find you need more filtering on the buck converter output. Perhaps a second LC stage? You'll have to play with that in simulation though, as it will affect the stability.

Some resistance on Q8's emitter will be very desirable for stability.

You will likely find that a PNP pass stage without immediate, local feedback (for instance, you're not using a complementary feedback pair or anything) will have exceedingly poor ripple rejection, making the need for more filtering even greater. Perhaps a voltage-mode pass stage instead? NPN, or an NPN->PNP feedback pair, or an NPN+NPN Darlington. The NPN and NPN Darlington arrangements will be the most conducive to stability.

Where does the forward current for D7 come from? That comparator can sink a lot, but it only sources a couple hundred microamps.

You're not seriously biasing your voltage reference with 110mA, right? R31 is waaaaay too small.

If a negative voltage is applied to the output from outside, D5 and D4 will be toast. You will also have no way to regulate the current. I suggest an output fuse (I am adamant that every bench supply should have one!) to protect your pass transistor, and a different clamping arrangement to protect D5 and D4.

Also, dissipating that much power in your thermistors will lead to massive self-heating, you won't be able to trust them. Similarly, you don't need 30mA through your overtemp LED.

[xReM1x]

A few comments:

You're putting 42V across a couple of your opamps. All of those are rated for only 44V, you're cutting it very close - and there will be quite a bit of energy available at 44V, if you get spikes or excess voltage (is the 30V rail regulated?) you could have some impressive holes in your opamps. Any chance of bringing the negative rail in a touch?

You might find you need more filtering on the buck converter output. Perhaps a second LC stage? You'll have to play with that in simulation though, as it will affect the stability.

Some resistance on Q8's emitter will be very desirable for stability.

You will likely find that a PNP pass stage without immediate, local feedback (for instance, you're not using a complementary feedback pair or anything) will have exceedingly poor ripple rejection, making the need for more filtering even greater. Perhaps a voltage-mode pass stage instead? NPN, or an NPN->PNP feedback pair, or an NPN+NPN Darlington. The NPN and NPN Darlington arrangements will be the most conducive to stability.

Where does the forward current for D7 come from? That comparator can sink a lot, but it only sources a couple hundred microamps.

You're not seriously biasing your voltage reference with 110mA, right? R31 is waaaaay too small.

If a negative voltage is applied to the output from outside, D5 and D4 will be toast. You will also have no way to regulate the current. I suggest an output fuse (I am adamant that every bench supply should have one!) to protect your pass transistor, and a different clamping arrangement to protect D5 and D4.

-ye, just taught about that. I would most likely use the LM358 or something so I will have to lower the output.

for the buck converter, I could add lc stage but last time I checked I got more ripple with second stage lc filter. but I might did something really stupid. I will consider this.

for Q8 I will add resistance, how much you suggest?

what pass stage you suggest \ how can I implant voltage-mode pass stage? just changing it to darlington NPN?

 for D7 - I did not undestand what you said. if it matters I could use it to drive NPN for the led.

and for the LT1009 -> my mistake, I was just testing something. I will use 5-3k probably.

and as you said, I will add 5A fuse. if there is a need.

and for the thermistor -> should I use much higher value? like 100-10k?

[Kleinstein]

There are many critical points in the design:

The first is using a PNP as a kind of low drop regulator:
-  This type of regulator often has rather poor PSSR - so suppression of noise from the the preregulator is hard or poor.
-  stability is also generally not that simple - this can be tricky without a scope
   e.g. add something like a 1 mF low ESR cap to the output and this circuit will likely oscillate.

The second is using high side current sensing. This is the rather difficult way - not real need to make is that hard. The shunt is at a position where is does not help (e.g. like a resitor at the emitter of the power stage).

The output stage with just two transitors has a hard time to deliver 5 A - maybe 1 A is Ok for this configuration.

So I would really consider to restart with a different, more conventional topology. The standard would be a darlington emitter follwer with the shunt at the GND side.  The good thing is that you only want 25 V, so there is no need for extra voltage amplification - so just about the basic well proven circuit. This is likely simpler than fixing this mess.