The Vcc pin has a 36V zener so I think I should be within specs. The minimum voltage should be 13V according to the datasheet (Startup Threshold).
Now I changed R75 for a 4.7Ohm 1/4W carbon resistor.
Here are some pictures of the real PCB (it's a mess):
So here are new measurementes with R75 being a 4.7 Ohm carbon resistor. This is on R75:
Pin 4:
Rectified AC:
Vout:
Should I try maybe reducing the 1mH inductance and also using some ferrite core with gap instead of the toroids I'm using now?
Now I see why you have only 283V on output, since you have 288V on input.
I feel that would be higher since you are saying you have 220Vrms..
Over current works ok, triggers about at 1.5v, you can change sense resistor back down to .5-1R.
Check your resistors again. Smd resistors are NOT rated for that high voltage. something like 125v for 0805. no where near your 280!
And inductor.. can it even store any amount of energy? Gapped core would be much better, of toroid that is made for this application
Last edited:
I'm not sure if I wouldn't put just to play with it, series resistor with inductor, pretty high value, say 100R? that should limit max current into inductor and would probably tell you that your inductor is the problem, since you could get different duty cycle on fet
Well no, if inductor is saturating(then yes, but you would need to see its current with series resistor) or not passing the current like big filter, lowering inductance will decrease its AC impedance
Last edited:
1206 resistors are absolutely perfect for this voltage. Much better than the clearences elsewhere on the PCB.
4R7 is obviously toooo much.
C37 is far from the place it should have been: as close to Pin6 and pin4 as possible.
Yes, reducing inductance is neccessary, however later you will find it will also increase heat, but now the task is simply make it work.
A through hole resistor has about 10 nH inductance. This with 0.25 ohm gives 40 ns time constant. This is what C37 has to compensate. So it is still too much. Cut it out!
4R7 is obviously toooo much.
C37 is far from the place it should have been: as close to Pin6 and pin4 as possible.
Yes, reducing inductance is neccessary, however later you will find it will also increase heat, but now the task is simply make it work.
A through hole resistor has about 10 nH inductance. This with 0.25 ohm gives 40 ns time constant. This is what C37 has to compensate. So it is still too much. Cut it out!
Yes, I will be changing the 4.7 Ohm resistor now, back to about 0.22 Ohm. Do you think my core is saturating? Are the calculations I did some posts ago correct?
I will place C37 even closer to pin 4 and 6. I don't have any SMD resistor of that value right now 🙁 Will try to cut my current through hole resistors as short as I can.
Can you please mark me where is the current loop? I'm really interested in this because I still don't quite catch the concept when I have to route my PCB.
The calculations are approx. good. But it would have saturated if the high current started to flow, which didn't happend yet.
which resistor?
Sorry, I have only minutes, on mobile. Follow the route: sense resistor, gnd, IC pin6, IC pin 4, series resistor, back to sense. Gnd connection is very long, and carries switching current. Therefore is far from equipotential.
I don't have any SMD resistor I can use for Rsense. I tried changing the inductor to 250uH but same result. Then I tried disconnecting the load and the output voltage gets to 400V and keeps raising, I disconnected everything when it reached 500V.
So the ground loop could be the problem here? It's firing the current protection?
This is what I have between pin 4 and 6:
It's far from 1.5V so the overcurrent protection shouldn't be firing right?
It's far from 1.5V so the overcurrent protection shouldn't be firing right?
that is ok. But you can still see, just after fet turns on, you have spike of current. That may need to be filtered or change layout
This is what I have between pin 4 and 6:
It's far from 1.5V so the overcurrent protection shouldn't be firing right?
During what circumstances? R75? Load? Input voltage? At least these info are neccessary to tell if operation is normal.
I suggest you to use 2 channels on scope whenever possible (or more if you have). Relative timing can tell very important things.
PS: Your building is quite nice for a prototype.
This is a peak current mode controller, so 1 thing is sure: impulse is terminated because of tripping threshold of current sensing. 1.5V is only the highest possible value, but during normal operation it is significantly less, and strongly variing.
R75 is 0.22 Ohm. The load are two 100W incandescent lamps in series as always and the input voltage is the same as a few posts before. The current protection threshold is variable because of the multiplier right? So I should look for the voltage on pin 3 and pin 4 at the same time considering that the voltage on pin 1 as relatively stable, will do that.
The ground plane doesn't work here? I should route everything to pin 6 of the IC and from there to the GND in the AC rectifier?
Output voltage is also important.
Current trip point is depending on input and output voltages.
Ground plane is never perfect, but in your case it is not a ground plane actually at all (has a thin neck, and doesn't cover neither high current nor signal current paths).
Current trip point is depending on input and output voltages.
Ground plane is never perfect, but in your case it is not a ground plane actually at all (has a thin neck, and doesn't cover neither high current nor signal current paths).
Do you refer to the thin necks marked in black? The one on the left shouldn't be so critical because on the right side it has a good amount of cooper to go trough or am I missing something?
I do see a thin neck on the lower part where the GND current from R75 should flow.
Ok, this is what I have on pin 2:
I think right there, there is something wrong. With my current output voltage I should have ~700mV on pin 2 that comes out from the voltage error amplifier. I checked the connections and they seem to be fine, maybe burned IC?
Here, the yellow trace is pin 3 and the blue trace is pin 4 (different time scales):
I see a lot of noise everywhere, probably because of the bad layout but also because I'm using the crocodile ground clips.
I think right there, there is something wrong. With my current output voltage I should have ~700mV on pin 2 that comes out from the voltage error amplifier. I checked the connections and they seem to be fine, maybe burned IC?
Here, the yellow trace is pin 3 and the blue trace is pin 4 (different time scales):
I see a lot of noise everywhere, probably because of the bad layout but also because I'm using the crocodile ground clips.
- Status
- Not open for further replies.