Convert 0-30V 3A PSU to 5A or more

[mixos1]

I measured the AC voltage across the transformer at

1V/3A the output was a distorted sine with 32V amplitude

and it goes till:

1,4V/4,4A where the output was also a distorted sine with 31V amplitude

Is the transformer goes to current saturation?

At these outputs except the Q4 also the rectifier diodes becomes hot.

 

[Staigen1]

Yes, the transformer is rated at about 72 VA, and you try to take out about 90-100 VA! Is there any place on the internet that describe rectifiers, somewhere? I have difficulties explaining exactly how a rectifier works, i know how it work, but explaining it in a few words is harder. And i mean rectifiers with its asociated transformer and reservoir condenser, also under load.

Shortly, the transformer have to be rated about 1,5 times the actual load, in this case 24 Volts times 3 Amps times 1.5 = 108 VA! Thats a 4.5 Amps transformer.

Then we have the ripple. This is because the current you draw from the reservoir cap when the voltage out from the transformer is lower than the actual voltage over the cap. The current is unloading the cap during that time and the voltage is dropping. The remedy for this is to heavily upgrade the cap to about 10-20 times the normal value. Also the rectifier diodes have to be at somewhat higher rating.

Now to something else. How much shall we upgrade the PSU to? I have a little suggestion here, how about 0 to 30 Volts and up to 5 Amps with good regulating and ripple and 7 Amps at somewhat lower voltage, also with good regulation ripple and 10 Amps intermittent but not moore than about 25 volts output with good regulation and ripple. I belive this is possible. At least i am going to build the PSU that way, if we success!

//Staigen

 

[bogdan2]

ok. i have a suggestion. what if the 2N3055 transistor could be replaced by 2? this will lower the dissipation ok each transformer.
but the drop on the transistors means that the output will not be as high as 30V....
or, another suggestion. put a resistor in series with the transistor. and a relay with a control circuit. when the output voltage is more than 15V(let's say) then the resistor is shortcircuited by the relay and so you can get the maximum voltage. when at the output is lover than 15V then the relay is OFF and some voltage gets dropped across the resistor. this way you minimise the power dissipation ok the transistor.
and for the heatsink...i think that mounting a small fan, maybe one similar to a computer cpu fan coud be used. having an artificial air flow cools the transistor better than a large heatsink. and for getting 12V for the fan, i would go for a 7812 regulator. it is cheap, and for the current that a fan will draw, less than 200mA it is very good.
well, hope that my suggestions will do some good.

 

[Staigen1]

Hi there Bogdan, shure they do! But we come to that later. Of course the pass transistor must have headroom, and i am also going to suggest multiple pass tansistors. Also fan cooling for the heatsink, in order to use a cheaper one. But the relay and its control circuit can easily be moore costly than a few transistors.

//Staigen

 

[bogdan2]

well, yes, the realy and resistor and control circuit(a low power transistor, a zenner and some resistors) can be more expensive than the transistor itrself. but what about adding the heatsik? you will not need a heatsink for the resistors. yet, i think that the power resistors can be more expensive.
what about using something else in place of the resistors ? like car lamps or other types of lamps....?

 

[Staigen1]

Adding a heatsink? The heatsink is already there. Do you mean that we shall use the pass transistors without heatsink? I dont think that will work. Of course a bigger heatsink costs moore, but you only buy it once, but the PSU you use year out and year in, so if it cost i little moore in the beginning, but lasts longer i belive its worth it.

//Staigen

 

[bogdan2]

nope, you did not understand me right.
i was suggestiong to use the heatsink and the transistor that are already there and another resistor in series with the transistor, wich will be short-circuited(the resistor) when the voltage required at the output of the power suply is highter. what i was trying to say is that for the resistor, a heatsink is not required.

 

[Staigen1]

I did understand you from the beginning, but i dont think the idea is so good, it complicate things, and it can make the PSU less reliable. The idea is not so dumb, it is much power dissipated at low output voltages and high currents, and that is transferred to heat, so if you can dissipate that in a resistor there will be much to gain, but it have moore disadvantages to, for example, where to put the trip point etc. And lamps dont have constant resistance, they have heavy positive temperature constant.

//Staigen

 

[bogdan2]

yes, lamps are not more reliable.
i think that the best option here is to either put tu 2n3055 transistors, but it think that tha will require having a 30V a.c. transformer to obtain 30V at the output, or to use a highter power transistor.
or, maybe what about using 2 2n3055 transistors in paralel?
i have seen this in many schematics. maybe it will work better than having 2 transistors in series?

 

[Staigen1]

Of course the transistors have to be connected in parallell. And i will suggest moore than 2 transistors. I dont think we can rise the transformer output to 30 Volts, the op-amps would not stand that voltage.

//Staigen

 

[mixos1]

Which is the relation between the number of the output transistors and the output voltage and current?

If we put for example 2 output transistors in parallel, the output current we can draw is doubled (theoritically if the rest of the circuit can affort it)? shall we change the rectifier diodes or something else to something more powerfull?

And what about the voltage?

 

[Staigen1]

There is a relation, but it is not that simple, you must also doubble the heatsink! If you put the 2 transistors on the same heatsink you cant double the current. Also, you cant just parallellconnect the 2 transistors, they are not going to share the current equally, you must also put a small resistor at the emitters, one at each emitter.

Then we have the transformer and the rectifier and its asociated parts. All you who is going to build this PSU, either the original PSU or the upgraded version we are discussing here, read this! First, decide the current output you want, than multiply this with 1.5! Thats the right rating of the transformer. Mixos have tested his PSU and he is using a transformer rated at 3 Amps, as stated in the projects text, and his transformer/rectifier brooke down at about 2 Amps. The rectifier must at least stand that current too, but here it is better if you can use a rectifier of a higher rating, the reason for this is that a bigger rectifier have a lower forward drop. Than theres the reservoir cap, usally you use 1000 uF/Amp, but in this PSU the cap must be much bigger, 10 to 20 times bigger, and the reason for this is that the ripple is raising when you take out large currents if the cap is too small. Usally when you design a PSU of this type the transformer is designed to deliver a voltage equal to or slight above the maximum output voltage, but in this PSU the maximum output voltage is higer than the transformer voltage, hence the heavy rectifier. Here i will suggest to move the whole rectifier and the reservoir condenser off the pcb and use a rectifier of the type that is bolted to the box of the PSU, also use a cap that is also bolted. The cap shall also be a "high grade" or "computer grade" type with very low ESR. Also the transformer must be of a very good type, with a very low ouput resistance, therefore i suggest a toroidal type. They cost a little bit moore, but nowadays not much moore, and the rise in performance is considerably higher!

//Staigen

 

[bogdan2]

yes, i think that a 10 or 15A rectifer could be used....
why not use a bridge from schotty diodes? expensive, but more performant.
a good transformer would be a toroidal one, probably a 30V, 200W would be suitable.
i think that having 30V a.c. output rather than 24V. if we use 2 transistors in paralel, than a resistor must be bonnected to the emitter of the transistors. cannot use them directly.
also, 2 heatsinks must be user, or maybe a bigger one for the 2 transistors. like mentioned before, the putting the 2 transistors on a heatsing will not increase tha current. since the power lost as heat on the transistor is dependent on the current, having half of the current troug one transistors means that only half of the power is disipated on that one. and there are 2 transistors, you still have the same total power didpated.
anyway, i would be curious of something. say that i will cool a 2n3055 transistor with water and maintain the temperature at a low value, less than 50 degrees, do you think that i could make a current more than 15A pass trough it? i think that it is possible...but not tried it yet.

 

[Staigen1]

Have you thougt of the IC:s? They cant stand to much voltage!

//Staigen

 

[Staigen1]

Oops! I think i missed some of your reply. First the rectifier bridge, yes, and even higher rating, the difference here in sweden is minor between at 10 amp bridge and a 35 amp bridge ( 18:- Sek and 22:- Sek ) ! Schottky rectifier bridges i dont have any experience of, so there i can not give you an answer, but i think that i read somewhere that they dont perform so well in a regular rectifier at 50 or 60 hertz, or maybe it was that you did not get much advatage by using them so. Regarding the transformer, look above. And then the pass transistors and output current, of course you can buld it for moore and moore current, but i belive that is beyond the discussion here. You only have to get higher ratings of evrerything thats related to to it( transformer, rectifier, resevoir condenser, the number of pass transistors, the heatsink etc etc). Than the watercooled heatsink, of course that can be implemented, it is your money, you must in the end get rid of the heat anywhere, so you must cool the water somwhere else. Then the number of pass transistors, i will suggest using moore than 2 transistors, not moore than 1 to 2 amps per transistor, the higher maximum output current the lower current per transistor. This also depends on the heatsink. And here i think you also should think of safety and the SOA of the pass transistor(s)!

//Staigen

 

[bogdan2]

what do you mean my SOA???
well....i forgot about the ic's. maybe they can be replaced by something else?

soooooo....
in conclusion the idea is to put 2 transistors in paralel....right?
and change the bridge....
yes, you are right, there is a small difference in price between a 10A and a 35A bridge, but still, i think that a low drop one should be used....

 

[Staigen1]

SOA = Safe Operating Area, its in the datasheet of the 2N3055. Its often overlooked in constructions of this type, the constructor only see the 117 Watts of dissipation possibilltys.

//Staigen

 

[bogdan2]

so................... ??? ???
isn't anyone still inerested in modifying the suply? ???

 

[Staigen1]

Im probably going to build this one , but i dont have all the comonents yet . The reason for the delay is that the PSU Mixos have and do some testing on, brooke down. He has now repaired it, and hopefully he can do further tests this week.

//Staigen

 

[ante1]

I have been following this project for some time now and I think all the problems with heat, SOA, big expensive heatsinks, double fans and so on can disappear by going switchmode. This is more economic, much higher efficiency, lover component count and a lot less heat. And it will fit in a smaller case too. :