8 bit pic micro with 4 different PWMs with deadtime

mukeshp11 · Joined: 10 Dec 2008 Posts: 15

Hello everybody,

I want to generate 4 PWMs with deadband available between PWM1 AND PWM2, and dead band between PWM3 and PWM4.
I am trying with 16F690 or 16f887 in pulse steering mode but there also dead band is not available in pulse steering mode.

Also i cant use half bridge mode as i need 4 separate PWM which i will use as input-output in one state and pwm in other state.So i require 4 different PWM with dead time.

So its like:-based on some conditions

State A :- PWM1 and PWM2 as PWMs with dead time, PWM3 and PWM4 as GPIO.
State B:- PWM1 and PWM 2 as GPIO, PWM3, PWM4 as PWMs with dead time.

PWM switching frequency is same.

Can I get this feature in any 8 bit pic microcontroller with about 20 pins package ?

Kindly comment.

Mike Walne · Joined: 19 Feb 2004 Posts: 1785 Location: Boston Spa UK

This post has been here for some time now. That should tell you something.

Could you consider:-

(1) An external logic chip to do the steering?
(2) An external logic chip to create the dead band?
(3) A different way of doing what you want with available PICs?

Mike

temtronic · Posted: Wed Feb 22, 2012 6:32 am

Also you don't say anything about the PWM requirements.Details like frequency,resolution,etc. would be nice...

As for the original question ...it's very easy to get even a 16C84 to have 4 bits paired as 2-PWM, 2 I/O then 'flipped' as 2 I/O, 2 PWM as you post.

Ttelmah · Joined: 11 Mar 2010 Posts: 19569

If you think about it, you can't have two independant PWM's with 'deadtime'. The only way that deadtime can really exist, is with a single PWM, driving two outputs, where deadtime is the transition from driving one to driving the other. This is half bridge operation, which you say you don't want....
You do realise, that being 'half bridge', doesn't stop you using the pins as GPIO, any more than normal PWM's do?. You can either turn off the PWM, and then use the pins as GPIO, or on some chips you have output override for the PWM pins.

You could possibly 'synthesise' a behaviour close to what you want, by using a chip with four independant PWM's, driving two off timer1, and two off timer3. Then starting timer1, and waiting till it goes just a couple of counts past halfway, then starting timer3. Run PWM1 and 3 off timer1, and 2/4 off timer3. Then provided you never set the interval to more than 50% less twice the offset between the timers, you will have an apparent deadband between PWM1 and 2 (and the same between 3 & 4).
A chip like the 18F25K80 could do this.

However far easier (since it avoids the possibility of you setting the wrong value), to just use a dual ECCP, and when you want to drive the pins as normal just write 0000 to the CCPxM bits which disables the ECCP. Use the pins as normal, and re-enable the ECCP, when you want your pulses. The 24J11 for example, offers two ECCP's.

Best Wishes

Mike Walne · Joined: 19 Feb 2004 Posts: 1785 Location: Boston Spa UK

You've been asking essentially the same questions about pulse steering and PWM's since you joined the forum.

You've also had intelligent, helpful, and comprehensive answers from a small group of guys.

None of the responses seems to be satisfactory.

Maybe it's time to take a different approach, and pose different questions.

For starters:-

Can you explain, simply, what you are trying to do?
( I.e. at the block diagram level. Or is it a state secret?)

As it stands you're making us all guess at the real problem you/we are trying to deal with.

Mike