PWM dithering

[MMitt.7]

I am using STM32F103C8 controller for my product desgin. I want to implement PWM dithering concept to increase my PWM resolution. after reading application note AN4507 following are my doubts

1. what is meant by 1-LSB duty cycle management
2. On page no 12. Can you eleaborate more point no.3 with suitable examples

Reagrds

Mudit Mittal

[S.Ma]

Haven't read the AN however temporal dithering means modulating the duty cycle on a period basis.

Imagine the PWM has 100 step (100%) with granularity of 1%.

If over 4 pulses period you generate say 45%, 45%,45%,45% 4 pulses, you get 45%

If over 4 pulses period you generate say 45%, 45%, 45%, 46% 4 pulses, you get 45.25%

The drawback is generation of a sub-harmonic modulation.

This is one of the used technique to increase nb of colors on displays by temporal dithering.

Using DMA on Timer output compare sweeping 8 values yields 8 substeps so 3 additional bits.

No DMA, use interrupt based update if the speed allows.

[AvaTar]

> The drawback is generation of a sub-harmonic modulation.

One specific use case for PWM dithering is spreading the EMI spectrum, to "flatten" peaks and comply with regulations.

[waclawek.jan]

Yes, but even then it's at the cost of low-frequency content. In other words, it requires to rethink the filters and the highest possible output frequency dramatically falls. And generating the optimal waveform is not entirely that trivial. One may also look at PDM as source of inspiration; it may be viewed as a PWM dithered by more than just one bit.

This may not matter for some particular purposes, but often comes as an unpleasant surprise, especially if the PWM is employed to spare down the cost of a DAC.

JW

[S.Ma]

We already know from display interfaces that if someone requires EMI reduction (which here may not be that critical) that spread spectrum maybe the typical next logical step besides good HW design.

Here the subharmonic typically is related to the case there is RC filter to output analog signal, and the time constant parameter will be the next talk...