Discrete Regulated power supply

[blackdog]

Hi Cliff Matthews,

I am currently busy with the work and administration involved.
Some measurements have already been made and of course it generates. 
But not at a frequency you would expect.

The power supply is easy to set up and to load with 1 or 2 Ampere, so that works.

But debugging takes a lot of time, and I don't have that available now, probably next week.
A bit of patience please 

Kind regards,
Bram

[schmitt trigger]

Blackdog;
I really like your prototype assembly techniques. They show you took the time to lay out things properly.

I specially like your "copper clad chassis". Very ingenious! And it makes a good ground plane too.

[GigaJoe]

So what the goal ? mediocrity PSU on 2 transistors ?  I may sound sarcastic , but really it not 80x  when find some opamp like to find a precious gemstone
for a zener need const curr, and diff amp need  const curr as well - to minimize pulsation

[blackdog]

Hi GigaJoe, 

I have already explained in this topic that with two opamps you can make a much better power supply.
But also that I like to work occasionally with discrete's, just for fun.

And not to forget, I am still learning from it!

I've already placed some extra "stopping" resistors here and there to avoid oscillations.
At a certain load and output voltage, oscillation occurs around 45 MHz! 
Touching the collector who is grounded from the differential amplifier, changed the frequency and level, total length of this wire 1.5cm!
Welcome in the world of small fast modern transistors...

A good reason to not do it this way 

Some pictures
Testing with a Heatkit dummy load.


The red circels are ground wires, all change the oscillation frequency.
Blue circle, wextra power decoupling off the board, can be removed again.
Small pink circle, collector is now connected via en small 100 Ohm resistor to ground.


Later more.

Kind regards,
Bram

[GigaJoe]

OK, got it, even pay attention on circuit, that i missed ...
does osculation happens when current limiter involved ?  That usually a problem with all compensating VR, current pushing regulated signal down, when in the same time a second half trying to compensate it ... that when oscillation happens.
so need to change signal on input of diff ampl.


 

[GigaJoe]

silly ideas:  dif amp - darlington,   output from a zener going to another zener - that would be a vref;

[mike_mike]

Hello,
Is there any way to add more than 2 power transistors in the attached schematic ?
I don't need more output current, just more power transistors in parallel. It is enough to add more transistors with emitter resistors in parallel or I need to make other modification ?
The maximum output current set by P3 will be 5A and maximum output voltage set by P4 will be 27-28V.

[not1xor1]

Hello,
Is there any way to add more than 2 power transistors in the attached schematic ?
I don't need more output current, just more power transistors in parallel. It is enough to add more transistors with emitter resistors in parallel or I need to make other modification ?
The maximum output current set by P3 will be 5A and maximum output voltage set by P4 will be 27-28V.

yes, you can 
that PSU is already so bad that there is no chance you can make it worse 

[Cliff Matthews]

@not1xor1 - If he adds one or 2 more, shouldn't he at least upgrade the PNP driver to a TIP42c or similar?? (Ic = 6a)

[IanMacdonald]

Yes, I would advise a TIP42C as a more substantial driver. You could also look at using 2N3773's instead of the 2N3055s. They are not much more expensive and are quite a bit higher rated.  Especially in voltage. (Which matters more than it might seem because the current handling of a transistor drops off quite rapidly as you approach its voltage rating - known as secondary breakdown.)

Other useful mod is that you can make P1 and P2 'bad wiper safe' by adding a high value resistor from the wiper to the minimum volts/current end. Say 220k. This will ensure that the supply goes to min output instead of max if the wiper goes open circuit.

[not1xor1]

@not1xor1 - If he adds one or 2 more, shouldn't he at least upgrade the PNP driver to a TIP42c or similar?? (Ic = 6a)

  NO because mike_mike wrote:

I don't need more output current, just more power transistors in parallel.

so the current required by the PNP driver would be approximately the same
the same because the maximum output current doesn't change, is just split among more BJTs
approximately because h_FE vs. I_C is not flat