Buck regulator

[ronsimpson]

That explains it.
I know you put great effort in this but.....There is no good ground in your project. I can see you have more grounds than most people. There is a ground on each side of the ICs. The UC3823A is connected to ground on both the right and left side of the IC. It is a long distance from the left to the right side. The IR2110 has one ground connected to the UC3823A's ground but the other ground has a long distance. The power parts are too far away.

You are running slow, 40khz, but these bread boards are not built for high current, high voltage and fast edges. I think ground bounce is part of the problems.

 

[123duermed]

Yes I found out that the ground was very bad, so I have improved it. This made the circuit work very good up to 50V input. Then the pwm signal starts to act weird again. I have used a 40W xenon lamp as load and it draw 4amp without problem. The switching frequency that seems to be best is about 60Khz,with both 10uH and 30 uH. When the frequency gets higher, it takes lower input voltage to make the pwm signal to act weird. By weird, you can see the photo. This is the gate-source voltage. It seems to skip a period of the duty cycle.

 

[ronsimpson]

Your schematic shows one gate driver but your bread board shows a different part.
How is the IR2110 wired up? Are there inputs that are not connected?

 

[123duermed]

I have only on one gate driver connected. After breaking all of my IR2110s I had to use IR2117 again. Everything on the Ir2117 is connceted. Everything was also connceted to the ir2110. The weird acting pwm signal made the mosfet not swithing on and off property, so the mosfet overheated and breaked both itself and the ir2110.

Breadbord:


Schematic (not 100% but almost, some difference from breadbord)

 

[johansen]

**broken link removed**

I recommend prototyping like this. dremel the traces out, use one side as the ground, etc.

 

[123duermed]

The problem now is, when either the input voltage is high, or the load is high, it jumps into this weird mode. I can reach over 60V, but when I starts to draw more current it gets in the same mode.


Blue: Gate-sorce votage. Yellow: signal from gate driver (HO pin)


it seems to jump from DCM to CCM to only a spike signal.

 

[123duermed]

I think it has something to do about the inductor. because when i change from 3 to 1 inductor. The problem does not occur at the same voltage and current draw.

 

[ericgibbs]

I think it has something to do about the inductor. because when i change from 3 to 1 inductor. The problem does not occur at the same voltage and current draw.

hi,
Are those 3 series inductors all the same phase, the image shows one is reversed.??

E

 

[123duermed]

Thanks, changed it. What effect would this have?

 

[ericgibbs]

Thanks, changed it. What effect would this have?

hi,
Does it now run as you expected.?

E

 

[123duermed]

No not exactly. The Vout have a lot of spikes, so the signal to the pwm-controller have a lot of spikes as well. When the current increase the spikes becomes much bigger, and I think this starts the weird pwm controlling. I tried filtering the signal on pin1 on Uc3823a. I could get the signal to be a little better, but not enough i think.

 

[ronsimpson]

You are driving the gate with a 1k resistor? I would use a 10 ohm at the most. Some of your pictures show a slow turn on/off of the gate.

The 1n4001 is SLOW.

Interesting on how you are killing the MOSFET drivers. On the IR2117, connect the scope ground to pin3 (ground) and look at pin 6 (VS). Is it being driven negative far? I think the data sheet indicates it should not go negative very far. What is your big diode? (D1)

 

[johansen]

i'm guessing the source of that mosfet could dip 100 volts below ground on that breadboard, the leakage inductance between all the active components (input cap, mosfet, diode)... all that leakage inductance probably stores 100 times as much energy as does the combined 1-10 nf capacitance of the mosfet and diode.

edit: and op isn't even using the two power supply rails properly, one is used for ground, the power is on the other side of the board.
it certainly isn't impossible to get this circuit working on that breadboard, but it has to be done right.

the mosfet and the diode should be right next to each other, offset by one pin, then wrap the power and ground out to the rail right beside them, and put film caps on either side of those connections.
then put a capacitor in front of the mosfet and diode and stick the leads down the same lines of holes, that way you can get a combined length of about half an inch between all three caps. the inductor and output capacitor don't need low inductance paths btw, except to clean up the output ripple, which isn't important yet.

 

[ronsimpson]

Yes!
By remaking ground this supply is getting better. Slowly
Long wires have killed it.

 

[123duermed]

I'm not understanding what you mean. I have moved everything closer, so I only use the upper half of the breadboard. The load's ground is connected right beside the powersupply's ground.

 

[johansen]

fit the diode, mosfet, and three capacitors within one half inch square of the board, and then put the gate driver on the other side of the divide right next to the active components.

 

[123duermed]

Hey.
After getting the circuit on a PCB, things started to work good. But I just have some questing regarding the circuit. When I reach a voltage (in) around 130V the Pwm signal gets very little so it seems that the Pwm-controller skips some pwm-signal some times. Around 150V it's only gives every other signal. Around 175V every fourth signal gets high. I get the voltage that I want out, so it seems to work (I only gets bigger ripple of course). But would this behavior wear out the controller, or is this how it's supposed to work. I'm running with 40 KHz frequency, and when I get between 130-150V where it change to every other signal, the mosfet or inductor starts to buzz/hum. Is that a problem or is it just that the switching gets audibly? It also happens when i gets over 170V. However I always get 14V out.

Circuit:
View attachment Schematic Prints Voltage regulator.pdf

 

[123duermed]

Another (dumb) question regarding C3 C4 and C22 which are standing on Vout. I could use caps that only tolerate 100V or less. (up to 14V++) even when I can get 300v in. Just need to bee sure.

 

[johansen]

i think if you properly compensate the error amplifier these problems will go away.
that 1 uf cap you have basically nullifies the type three compensation network you have built.
try a type 1 or type 2 network and play around with it a bit.