TM4C1294NCPDT: Dead-band for PWM in TI-RTOS

Mohammed Zakariya said:

I would like to use the dead-band generator to keep the three PWM's different from each other.

Vendor has provided fast & valid PWM detail.

However - my small Tech Group has a (much) different take:

• Such 'Deadband' is most usually employed when the PWM Signals are 'Complementary' - and  'Drive a, 'Power-Stage' - via an 'efficiency technique' known as, 'Synchronous Rectification.'
  • Under this condition - the deadband imposes guaranteed OFF-Time - upon both PWM-A & PWM-B (w/in the same PWM Generator.)
  • We do not believe that the MCU Manual clearly notes that such 'Complementary' PWM Mode (may only) be achieved by, 'Enabling the Deadband Generator!'
  • Complementary PWM forces 'PWM-A & PWM-B' to be 180° out of phase.   Software then drives only one of those PWM outputs - the remaining output becomes, 'Inversely Slaved.'
  • Deadband IS required in most all Power Stages - as most exhibit, 'Differing Power Semiconductor Turn-On & Turn-Off times!'     Thus it becomes possible (often likely) that (both) the High-Side & Low Side Power Devices are, 'On together!'    This is well known - and is noted as, 'Shoot-Thru' - usually resulting in (near) unlimited current flowing (almost entirely) thru both (Hi & Lo) Power Devices.  (And NOT to the Load!)    Destroying the Power Devices - and often (even) damaging the pcb & other components.
  • Deadband attempts to 'compensate for those different Turn/On/Off times'  (encountered while the  'Complementary PWM Stage' is 'Switching') - by enforcing a, 'Programmable Guaranteed OFF Time' - upon both the High & Low Side Power Devices.   In this manner - the chance of both (Hi & Lo Sides) being on together - is greatly reduced.    Even though - and especially though - 'ON/OFF times' of the PWM-Driven Power Devices differ!

• Now your application is not seen as involving such a High, 'Power Stage.'    Thus there appears 'little harm' - should (the usually dreaded) 'Shoot-Thru' occur.    And - in your case - 'Deadband proves a liability - offering NO Advantage.   (indeed - it (even) renders one of your PWM Outputs (essentially) worthless!)

  • And in fact - as Vendor's Charles notes - a single PWM Generator can drive 2 Leds - with (either) the same - or completely different 'Duty Cycles!'
  • Note too - you seek to drive 3 Leds - thus a 2nd PWM Generator will be required.   (yet only one of its two outputs)   And again as Charles states - you may adjust that 2nd PWM Generator to the same frequency as the 1st one.   You thus wind up w/3 distinct, Programmable, PWM Outputs - each one capable of 'Independent Duty Cycle!'

As the above (hopefully) illustrates - 'Deadband' proves 'Not your friend' in such application!   (as deadband creates an 'Unwanted Linkage' between 2 PWM Outputs - which is outside your stated goal of 'PWM Independence!'

Tag: Poster's Goal was fine - his 'method' ... (pardon) perhaps, 'Not so much!...'

Thanks for the screen captures and the detailed explanation, I understand that using a dead band is not going to help me in controlling the PWMs to drive the LEDs individually. With your earlier reply you mentioned about a Shoot-Thru, this is the exact reason why I am trying to achieve different PWMs. The query raised is with a simple use case, but the actual application is power heavy.

So going forward without the dead-band generator being used, How do I bring in the control between the three PWMs?

My friend - you have  (substantially)  "Moved the cheese" - have you not?     This forces additional time & effort upon Staff & myself - and we have already 'labored mightily' (and we, along w/this vendor, believe effectively) in your behalf!      Should we detail further - what prevents your (then) freshly 'Motorized Cheese' - from 'Driving even more distant?'

Staff & I are strong proponents of 'KISS' - the specificity demanded proves invaluable (both) in design/development and in 'Diagnosis.'    We note your new request has escalated to 'Power Heavy' - which is 'far outside' your initial  (3 Led) description - and as vendor's Charles has noted - the (now critical) 'Deadband' Control - appears 'unavailable' under this vendor's 'RTOS!'    

My small tech firm services many far larger firms - both they & our investors 'demand' that we, 'Choose ARM MCUs & Development/Operating Systems which are 'Best Suited to Task & Multi-Vendor Welcoming!'     (Thus never/ever confined to (any) single vendor!)     We have thus never considered this vendor's (so restricted) RTOS - as it proves contrary to our clients' 'Pursuit of Optimization' - likely & most always achieved by 'Casting a (very) Wide (i.e. multi-vendor) Net.'

If you wish further tech support from my team - kindly 'reward' my post(s) here - and issue a new posting - with the requirement for vendor's (so restricted) RTOS absent.    (FREE RTOS  falls w/in our 'multi-vendor capable' specification.)

In addition - the new requirement, "application is power heavy" - is devoid of (any) necessary detail - always required by 'KISS!'    Again - it is believed that a, 'New Posting' - adequately detailed and 'Direct in its Request' - proves very much to your advantage...

TAG:  Cheese suddenly moves - that's not acceptable...

Sorry mate, the cheese had to become bigger. Your efforts and quick responses are highly appreciable. I understand the discussion of a Shoot-thru is for another thread, but with this thread the feasibility of a dead-band in a controller running TI-RTOS got discussed which was one of the queries.

As understandable from your responses the dead-band is not the way to go if I want to drive the LEDs independently. If running them from different generator blocks, is it possible to bring in a sense of control over the highs and lows between the PWMs leaving aside whether the device is run power heavy or not? I hope this can be considered a discussion for this thread.