New Power Supply

[alfa88]

My power supply went poof. Turns out the transformer was poorly assembled and the outer windings were able to short against the core. Anyhow I am considering more power. I came across this design. Credits go to Zamboniman60. Feedback anyone? I believe the positive is using a Darlington pair as a Voltage follower but I'm not sure what's going on with the negative

 

[Rleo6965]

I think that series pass transistor in negative side must be pnp and its emitter must be connected to load side.

 

[Harald Kapp]

I agree. The circuit should be symmetrical.
Harald

 

[davenn]

Im also looking at the weird arrangement of smoothing caps between the bridge rectifier and the the reg chips
you have 2 lots of 2 x .022F caps in series and paralleled with each other

you could more than halve the capacitor count and have more room on the PCB (in the case) just by using a single cap ( or maybe a pair in parallel) in each stage of the appropriate value, instead of 4 of them in each rail.

What current rating are you aiming at for each rail ?
rule of thumb is 1000uF for each amp of current drawn. So 10A per rail then 10,000uF per rail and you could up that a bit if required

Dave

 

[Rleo6965]

Was that "rule" for 60hz only?

How about for 20khz inverter bridge filter capacitor? What will be the capacitance per 1 amp?

 

[alfa88]

I'm aiming for 4 Amps. Transformers start getting pretty pricey beyond that. The circuit was supposed to be designed for 5 so a little tweaking in the current limiting area will be in order.

 

[Rleo6965]

I can't find current limit circuit on voltage reg U1 and U5. What's the function of 2 pcs LM317 U2 and U4. It look like constant current load for the power supply not current limit. This will make your power transformer and transistor hotter.

 

[jackorocko]

I can't find current limit circuit on voltage reg U1 and U5. What's the function of 2 pcs LM317 U2 and U4. It look like constant current load for the power supply not current limit. This will make your power transformer and transistor hotter.

some people use this as current limiting, but that is near impossible to do with a static resistance on the output of the LM317. You make a very valid point, this configuration is setup to be a constant current source. U3 and U6 are the current limiting devices I suppose.

aren't the capacitors at 25V rated to low for a 24VAC input? Did Zamboniman60 actually build this circuit?

 

[alfa88]

I found this at a robotics forum. It's around 6 years old and Zamboniman60 never related whether he got it working or not. For sure I'll be breadboarding this in stages. I've got plenty of 50V caps.

 

[davenn]

Was that "rule" for 60hz only?

How about for 20khz inverter bridge filter capacitor? What will be the capacitance per 1 amp?

yes for mains supplies 50/60 Hz

I have never seen a rule of thumb for any other freq's as in say your 20kHz

Dave

 

[davenn]

I'm aiming for 4 Amps. Transformers start getting pretty pricey beyond that. The circuit was supposed to be designed for 5 so a little tweaking in the current limiting area will be in order.

ok no probs, 4700uF would be ok but personally I would use 6800uF or so

and it wouldnt hurt to have a low value cap on the input to the regs either say ~ 0.1uF (100nF)... something else that is also pretty standard

Dave

 

[alfa88]

My latest iteration. I did away with the current regulation for now, took capacitor advice and added transistor labels. I selected GPs for Q2 and Q3. If memory serves correct Q1 and Q3 are the ones that handle the current. I'm not too happy with the cooling fan/ pilot light arrangement. I'd prefer a thermal switch for the fan and perhaps a neon light for the power lamp.

 

[davenn]

My latest iteration. I did away with the current regulation for now, took capacitor advice and added transistor labels. I selected GPs for Q2 and Q3. If memory serves correct Q1 and Q3 are the ones that handle the current. I'm not too happy with the cooling fan/ pilot light arrangement. I'd prefer a thermal switch for the fan and perhaps a neon light for the power lamp.

you have put the capacitors in the wrong places

the 6800uF on the inputs to the regulators and the 0.1uf caps on the outputs of the regulators

Dave

 

[Rleo6965]

Do you really need sw1 and R2 5 ohm 5w resistor? It's connected as shunt resistor to 48V output of bridge rectifier. Surely your R2 will go to smoke and also damaged your power transformer due to high current.

 

[alfa88]

Electronic design after cocktail hour is not a good idea. I would have caught that eventually. My bad Davenn.
R2 is a bleeder resistor. It wont be connected when the power is on. It does seem a little low in value though. I might change that to 1K.

 

[(*steve*)]

I might change that to 1K.

I would probably dispense with it unless I had a real need for it.

The voltages are not so high that bleed resistors are required to allow safety when working on the circuit.

 

[jackorocko]

Do you really need sw1 and R2 5 ohm 5w resistor? It's connected as shunt resistor to 48V output of bridge rectifier. Surely your R2 will go to smoke and also damaged your power transformer due to high current.

I said the same thing to myself. The way the switch is drawn it looks like when one part of sw1 is on the other part would be to.

What voltage range are you looking for alfa? I like the idea of using temperature to control the speed of the fan. How do you plan on implementing that, PWM?

 

[davenn]

R2 is a bleeder resistor. It wont be connected when the power is on. It does seem a little low in value though. I might change that to 1K.

just a further note on bleeder resistors, further to what Steve already said....
generally you will find them directly across the very large electrolytic caps, say when you may have 20,000uF and at 50V or greater when getting up to those ratings, they can store a heck of a charge for some time after the supply is switched off and catch out an unwary tech.
Usually the bleeder would be reasonably high value say ~ 6k8 to 10k and would be permanently soldered across the capacitor rather than switched. The hi value of the resistor means that the resistor doesnt affect the normal operation of the circuit and on power down it will discharge the cap(s) over a few sec's or so in a low current flow safe way rather than as a large fast hi current zap discharge, if you used a really low value as shown in your circuit.

cheers
Dave

 

[alfa88]

I'm soooo confused.

I want to have adjustable current limiting and I came across this scheme in a couple of different places. I don't see how this can possibly work. is this a hoax?

 

[(*steve*)]

There are many ways to implement a current limit, and that is actually one of them.

Note that it will have a voltage drop across it of several volts, so you need to place it before your voltage regulator.

Also note that it will dissipate a lot of heat if you short the power supply.

Many modern switchmode regulators include some ability to limit maximum current, and will do so without excessive dissipation.

The general method for variable current limiting is to compare a voltage drop across a current sense resistor with a voltage reference and prevent it from exceeding a certain value (thus applying a limit). The circuit above divides the voltage across a current sense resistor and essentially compares it with the Vbe of a transistor.