[SOLVED] HV boost - UC3843B not working

[Bianconero]

Hi, I'm a newbie here, but decided to write my first post on edaboard, cause I encountered a problem I can't handle. I designed a switching PS to drive some Nixie tubes. Here's the schematic I've tested in LTSpice and it worked like charm:



I made a PCB, soldered all the parts and obviously it's not working. The input voltage is 12 V DC, the output should be adjusted using pot R36 between 170 V - 210 V DC.

I measured a couple of values with a DMM and an oscilloscope:

- the reference voltage is 5.02 V DC (seems ok)
- the output voltage is 11.7 V DC (instead of ~190 V DC)
- on 4. pin (RT/CT) there's a 210 kHz signal of charging and discharging the cap C18 and it changes from 1.5 V to 3.1 V (but the datasheet says it should be between 1.2 V to 2.8 V, LTSpice shows that, too)
- the feedback pin voltage is about 400 mV DC (I guess it should be around 2.5 V DC)
- the compensation pin voltage is about 6.7 V DC (?)
- the output pin stays low, so the MOSFET is turned off and the current control pin is also at low state

I'd be really grateful if anyone could help me find out, what's wrong with this circuit. Is it possible that the IC is broken or is it more the problem of the PCB design itself? I don't have any ideas.

 

[ysba]

Hi, how much output power do you need (voltage and current)?

 

[Bianconero]

About 15 mA @ 200 V DC. I tested this circuit in LTSpice with a 20 mA current source as a load and it worked well, no problems at all.
However, I haven't tested my real circuit on PCB with any resistive load.

 

[Bianconero]

I changed the UC3843, replaced the timing cap to a higher value, but results remained the same, only the frequency of RT/CT signal dropped down.

 

[ysba]

I think the duty cycle is too high. From my calculations, it will be like 0.94, almost the maximum for UC3843B. I read something saying that for duty cycles above 0.5, this IC may be unstable, and the compensation network should be properly tuned. I've never worked with UC3843 before, so I can't tell if it's easy or not.

One thing I'd do is to put the circuit working with a lower output voltage (like 50V) by adjusting the feedback network, just to see if everything else is ok, and then begin to rise the output voltage.

 

[Bianconero]

OK, I'll try to do that tomorrow and let you know.

 

[FvM]

I think the duty cycle is too high. From my calculations, it will be like 0.94, almost the maximum for UC3843B. I read something saying that for duty cycles above 0.5, this IC may be unstable, and the compensation network should be properly tuned. I've never worked with UC3843 before, so I can't tell if it's easy or not.

It's normal operation that a SMPS controller goes to maximum duty cycle during startup and either if the compensation loop is stable or not, it will at least activate the output. This is a none-explanation.

The said duty cycle above 0.5 problem refers to possible subharmonic oscillations and is usually handled by so-called "slope-compensation". But there will be always a switcher output voltage.

The only reasonable explanation for the observed problems (output not switching at all) is either a wiring error or a defective UC3843. (MOSFET gate-source short would be another explanation, but causes high supply current and burn the UC3843 in return).

 

[Bianconero]

I've already replaced the UC3843 using the new one, the results were the same. I also replaced the current sense resistors to 3 parallel 3.3 Ohm resistors and double-checked the reverse biased Schottky diode on the OUT pin. Nothing's changed, the circuit doesn't switch the output. Every part on the PCB is cold, including MOSFET and driver.
I'll check the wiring again and let you know.

 

[mtwieg]

Based on what you've described (RT/CT pin is oscillating, COMP voltage high), you should be seeing the OUT pin switching. Something is almost certainly connected incorrectly, maybe the output pin is shorted somewhere. Or maybe you're using a 3842 by mistake, and the UVLO has not cleared?

 

[Bianconero]

I solved my problem thanks to your suggestions.
Of course it was an assembly problem. I soldered the D7 Schottky the wrong way, ie. it was forward biased, thus shorting output to the ground and damaging the UC3843. I replaced the IC, soldered D7 correctly and it started to generate about 210 V DC.
I checked the Schottky twice before, but I didn't see the problem. X-mas distraction...

Sorry for bothering and Merry Christmas to everyone