SG3524: Having issues achieving max PWM

Hello Douglas,
The SG3524 was one of the very first PWM integrated controllers available on the market and as you can imagine there has been many improvements since then.
As you discovered , its performance degrades as the oscillator frequency increases. This is because of propagation delays within the ic.
The ic parameters are specified at a switching frequency of 20kHz. Because the propogation delays are mostly fixed then as you increase the frequency to 200kHz the maximum duty cycle available will decrease.
The 45% max duty cycle rating means that at 20kHz the ic is guaranteed to have at least a maximum value of 45% duty cycle (or 22.5uS pulse width). You can expect this value to decrease as the frequency increases.
You should look at the UCC2808A-1 series of push pull pwm controllers.
www.ti.com/.../description
These parts have a rated max duty cycle of 48% (min), 49%(typical) 50% (max) at 194kHz (typical)
These may be more suitable for your application.
Regards
John

Hi John,
Thanks for the valuable info. We purchased some UCC2808A-1 chips and it made a great improvement. It still looks like the waveform is a bit asymetrical (looks like about .2 usec at 200khz) but it could be my imagination. Is there anyway to make it perfect? I have some other questions: Our application is very simple (no feedback) and I'm not sure what to do with the FB, Comp and CS pins. I have grounded CS and left the other two open and it works. Trying to run at 200khz requires a 330 pfd cap and an 11k resistor. According to the formula it should take about 20k which makes me think I'm doing something wrong. We would like to eventually be able to control the PWM with a pot but were not able to accomplish this by experimenting (applying voltages) with these 3 pins. I am using a regulated 10 volt supply to power the chip without a series resistor - is this acceptable? I would like to see examples of circuit designs other than the one in the Data Sheet to get other ideas as it is rather complicated. Could find none on the Internet but I know I'm not looking in the right place. Does TI have an easy way to view Application Notes on this device? This is a very important project for my company as we will be using this timebase (or whatever you suggest) to drive a large number of power Mosfets (via gate drivers) for a large inverter that my customer wants to place into production. Thank you so much for your valued assistance!!
Doug

Voltage Mode.docx Doug,

Glad to hear you are making progress.

The  oscillator for this part is running at twice the PWM frequency so if you want a 200kHz PWM then the oscillator needs to run at 400kHz.

That is why you need about 11k  here.

You are using the UCC2808A-1 and this requires  12.2 V as its turn on threshold. So it will not startup at 10V.

You should use the UCC2808A-1 which has a turn on threshold of 4.3V.

The applications notes for the TI parts are always on the webpage for the device and in this case it would be here:

However I can see that there is no real  information available except the data sheet.

In your application you need to generate an PWM ramp signal from the oscillator RC pin and use this as an input to the CS pin.

I am attaching an extract from the UCC28C40 data sheet which shows you how to do this.

What this does is connect a fraction of the oscillator ramp to the CS input.

If you connect FB to GND and leave COMP open then the PWM will be terminated every time the ramp voltage reaches 0.5 V.

If you make one of the resistor adjustable and with patience you should be able to adjust the PWM output  on OUTA and OUTB.

Regards

John

Hi John,

Thanks for all the info.  We are using the UCC2808AN-2.  I believe I indicated incorrectly that were using the -1.  In any case I'm still not sure I am using the correct chip for our application.  We are just trying to build a simple inverter with no feedback at the moment.  It needs to run at 200 khz and have a very small deadband (less than 100 nsec. or so).  We would also like to be able to control the PWM with a pot and it does not have to go to zero.  I'm still seeing the outputs of our  UCC2808AN providing asymetrical pulses.  Shouldn't the off time be exactly the same as the on time?  

Thanks again for your valued assistance.

Doug Mitchell

Hi John,

Using the UCC2808AN-2:  Upon further inspection, it looks like both the on times from pin 6 and 7 are the same duration.  However, when I connect a frequency counter to pin 6 I get 199.52 khz and when I connect it to pin 7 I get 199.24 khz.  How can this be??  I would think the readings should be identical.

Thanks,

Doug