problems with SG2525

[zilog]

I cant get my basic circuit based on a SG2525 to oscillate. The aim is to give as much pulsewidth as possible, and only let the deadtime-resistor decide on how long the off-state should be.

I have used this schematic to get started, the problem is that both pin 3 and 4 are silent.. what have I done wrong?

[luka]

Hi

All I did to get is oscilating is.

- Link 16 with resistor (3600R) to 2, 1 open
- R and C for oscilator, no deadtime in my case, so short 5 and 7
- Unused output via 8.2k to gnd
- 13 and 15 to vcc
- 12 to gnd
- no softstart
- 10 to gnd

And thats it, only 11 pins used
Only diffrence I see is I hae R between 16 and 2, but more important, R as load to SG output and pin 10 business, probably should have some resistor to gnd instead of C

[zilog]

Originally posted by luka
Hi

All I did to get is oscilating is.

- Link 16 with resistor (3600R) to 2, 1 open
- R and C for oscilator, no deadtime in my case, so short 5 and 7
- Unused output via 8.2k to gnd
- 13 and 15 to vcc
- 12 to gnd
- no softstart
- 10 to gnd

And thats it, only 11 pins used
Only diffrence I see is I hae R between 16 and 2, but more important, R as load to SG output and pin 10 business, probably should have some resistor to gnd instead of C

You have grounded pin 10 (shutdown), what else? (you talk about 8.2k resistor). What do you mean with "resistor to gnd instead of C"?

/Daniel

[luka]

Hi

I don't have floating unused output, I use only one, which drives fet(boost converter), other output of SG is not floating but "drives" 8.2k resistor to ground

Yea sorry about "resistor to gnd instead of C", I looked at wrong pin. That is softstart pin not shutdown pin. You should have pin 10(shutdown) connected to GND or via resistor if you will use it

[zilog]

I have now got the thing oscillating. The problem is that it only oscillates with Rd <= 100 ohm, and that produces a deadtime of 2.5 us, while I nead on-time = 9us, and off-time = 10us. What shall I do? The circuit most obviously doesnt work as described in the datasheet.

Any other ideas on how to get the waveforms I need, and the ability to drive a half-bridge with it? Logic circuits, and some sort of UV-lockout logic to prevent goof-ups when the power comes on/off?

[luka]

Hi

Ammm your... what is your freq. now, so that I don't have to look into datasheet? I think you C and R are a bit odd, well I wouldn't say really odd, but I use smaller C and bigger R, and Rd very high way more then 1k with no problem.

As I understand you need pulse to be width as possibe? So 50%(max for SG3525 from one output)Then you should have 0 dead time to max, so that SG will still work, but you won't be able to regulate D.C. from 0 to 50%, not with dead time

Edit: ahh I didn't read all again... so max D.C. for this SG is somewhere from 45 to 49%, so you have freq if 50kHz, I use 1nF and xxk for this, and Rd can be any from 0 to more then 500, so you will be able to get those 9us or more...

[zilog]

Originally posted by luka
Hi

Ammm your... what is your freq. now, so that I don't have to look into datasheet? I think you C and R are a bit odd, well I wouldn't say really odd, but I use smaller C and bigger R, and Rd very high way more then 1k with no problem.

As I understand you need pulse to be width as possibe? So 50%(max for SG3525 from one output)Then you should have 0 dead time to max, so that SG will still work, but you won't be able to regulate D.C. from 0 to 50%, not with dead time

I want to drive a half-bridge without feedback with 9us on, 10us off for each phase. That is, the outputs should be on for 50% of what is maximally possible with a half-bridge. The reason for letting the transformer relax for 50% of the time, is that both phases drive the Xfmr in the same direction of flux.

The frequency should thus be 26.5kHz, and each phase should be the double of that, with slightly less than 50% duty cycle (~9us on, ~10us off)

[luka]

Hi

Amm if each fet is 9us on, 10us off, you have period of 20us(1us dead time), that is 50kHz...so does it work at that time intervals?

[zilog]

Originally posted by luka
Hi

Amm if each fet is 9us on, 10us off, you have period of 20us(1us dead time), that is 50kHz...so does it work at that time intervals?

DT and on-time add, so the total period should grow from added DT.

[luka]

Hi

I don't think so, Ton will change with dead time, I think I saw that...constant period...but you could be right, never was lookng at that to hard