A little 5A 20V PSU CV&CC- Project

[bogdan2013]

I don't have any power supplies here so I decided to start a little project . Is not the most accurate PSU , but I think is ok .
Resolution :
2 mV / bit (2V range)
20 mV / bit (20V range)
22.5 mA / bit
OVP , OCP (in software) .
I limited the current at about 4.7A to don't use the regulator at max. current .

For AC input I will use a toroidal transformer 18V , 6.5A . I put some pots for calibration (15 turns) , 4 turns pot for current adj. and for voltage adj. .

[Jon86]

Not sure about everyone else, but I find that schematic pretty awful to read.

[bogdan2013]

I hope the A3 format is better .

[kxenos]

I also agree. When you split the design like that in the same sheet there has to be a logic sequence and not just random placement of the design parts. Also, the schematic is too simplistic to be considered a completed design. I think there are many flaws like for example using an LM317 for vRef, lack of decoupling etc. Also having low side current sensing is tricky. Have you taken account of the voltage drop in the current sense circuit?

[free_electron]

sigh. another hairball schematic..

Drawing style aside : here are some problems

- making a 2.048 volt reference with an lm317 ? ain't gonna work. that thing will drift under temperature...
- What is RV7 for ? that should be fixed.
- why have R11 ?
- what is Rv1 for ?
- what is the whole idea behind the current sink made with Q2 ? i'm hoping that is not how you intend to make current limiting ... if it is : you clearly don't understand how a current sink works.
- u7 u6 are lacking output capacitors
- u5 needs some caps too.
-your cpu may need a little reset circuit as well as an ICE connector...

[Jon86]

Okay that schematic looks a bit better, but still a pretty flawed circuit; and a pretty strangely designed one at that. 

[bogdan2013]

For constant current I was a bit confused when I tried to make the circuit , so just I used the drop voltage from shunt for adjust constant current  ... Those pots are for calibration , but I will replace with some 0.1% resistors . What do you think if I use an ADR292GRZ and I will put a voltage divider for obtain 2.048 mV reference ?
And with the constant current I'm still confused , could you help me a bit ?

[kxenos]

For constant current I was a bit confused when I tried to make the circuit , so just I used the drop voltage from shunt for adjust constant current  ... Those pots are for calibration , but I will replace with some 0.1% resistors . What do you think if I use an ADR292GRZ and I will put a voltage divider for obtain 2.048 mV reference ?
And with the constant current I'm still confused , could you help me a bit ?

Have a look at REF3020 2.098V ref. Also have a look at ZXCT1107 high side current sense. Also in the LM338 datasheet you will see many reference circuits. Have a look there for ideas. Also, did you say 5A? Please provide the power calculations you have made.

[bogdan2013]

power !? aprox. 100W , the current is not exactly 5A , 4.7 A , tomorrow i will rebuild the schematics

[kxenos]

aprox. 100W

OK, and what is the temperature raise above ambient for 100W dissipation for the package and the heatsink you use? You have to take into account the heat resistances involved and also the power derating.
The power calculation should look like Rtot = Rjc + Rch + Rha where
Rjc: heat resistance from junction to case (stated in the datasheet)
Rch: heat resistance from case to heatsink (about 0.2 if you use the plain white heat paste)
Rha: heat resistance from heatsink to ambient (stated in the heatsink's datasheet)
If you do the math you will get something like 300C temperature raise above ambient. (that is with a really good heatsink).

I think someone has to do a video about power calculations and explain these figures. Until then, google is your friend

[Jon86]

aprox. 100W

OK, and what is the temperature raise above ambient for 100W dissipation for the package and the heatsink you use? You have to take into account the heat resistances involved and also the power derating.
The power calculation should look like Rtot = Rjc + Rch + Rha where
Rjc: heat resistance from junction to case (stated in the datasheet)
Rch: heat resistance from case to heatsink (about 0.2 if you use the plain white heat paste)
Rha: heat resistance from heatsink to ambient (stated in the heatsink's datasheet)
If you do the math you will get something like 300C temperature raise above ambient. (that is with a really good heatsink).

I think someone has to do a video about power calculations and explain these figures. Until then, google is your friend

This.

You're either going to need a really good heatsink, something like a CPU Heatsink with a fan attached, that might be able to do 100W. Or you'll want to switch in different transformer taps. Either way, you need to have a bit of a think.

[strangelovemd12]

I think someone has to do a video about power calculations and explain these figures. Until then, google is your friend

Not sure if joking...

[kxenos]

Oh, nice! Not joking at all. This video is more than 3 years old...

[christopher iles]

Hello,

If you want a simple 5A variable voltage power supply that's bombproof, use an LM338 regulator; you need very few components and it will work first time!

Mine's over 20 years old now and still going as good as the day I built it.

There's loads of info on the net about this device along with the circuit.

No need at all for fancy circuits, all the features you need are built in to the regulator.

Hope this helps!

Kind regards, Chris.

[Jon86]

Hello,

If you want a simple 5A variable voltage power supply that's bombproof, use an LM338 regulator; you need very few components and it will work first time!

Mine's over 20 years old now and still going as good as the day I built it.

There's loads of info on the net about this device along with the circuit.

No need at all for fancy circuits, all the features you need are built in to the regulator.

Hope this helps!

Kind regards, Chris.

That's what he's using, but he's added in constant current and a microcontroller. I agree though, as a simple power supply, an LM338 will do well