[chbot] PWM challenge

[Charles Manning]

Thanks all

I think the idea I had of trying to "overlap" two PWMs for better precision
would only work if I was using a predivider to divide one dwon to a lower
frequency. That won't work for my application as I want to drive with at
least 30 to 40kHz (preferably more). I think the only way I can achieve a
higher precision would be by going for a higher clock speed (ie. a PWM
driven from at least 120MHz).

The other option might be to do some sort of dithering, but that will add
lower frequency ripple.

I've been looking at various interesting micros recently. In STM32, there
are the newer U5 series which run up to 160MHz which should give more
resolution than I need as well as a few other features that look
interesting for motor control etc (eg. proper security to stop people
tampering).

At the other end, there are a bunch of RISC V micros coming out of China.
144MHz or faster, less than $1.

THanks

Regards

Charles


On Mon, Feb 12, 2024 at 2:28 PM Robin Gilks <gb7ipd at gmail.com> wrote:

> A lot will depend on the SOC or microcontroller you are using.
> I'm playing with an stm32f303re (Nucleo board)  at present and I see that
> Timer 2 is a 32 bit timer and can be clocked at 144MHz (i.e. system clock *
> 2).
> Another approach I've used for generating interesting PWM shapes is to use
> an SPI MOSI output being given data under DMA from memory. The above mirco
> can do SPI at 18Mb/s. Set the DMA  for circular operation and change the
> RAM data on the fly
> Yet another method if you are straying into the analog domain with a PSU
> is to use a DAC - same micro, 2 12 bit DAC outputs, can be driven by DMA
> and a maximum rate of 1MS/s for single bit output change (I've been using
> one at 1.6MS/s and it seems fine!!). Linearity and offset might be an issue
> with the DAC with errors of up to 4 and 12 LSBs respectively with 12 bit
> codes.
>
> On Sun, Feb 11, 2024 at 8:49 PM Mark Atherton <markaren1 at xtra.co.nz>
> wrote:
>
>> The resolution of a digital based PWM is always(?) going to be limited
>> by the associated highest available clock. You may be able to use
>> additional switched delays to increase resolution. AFAIK similar work
>> has been done in a similar vein by various extremely high resolution
>> FPGA based time-to-digital converter designs.
>>
>> However in the analogue domain using a ramp generator, a voltage source
>> and a comparator almost limitless resolution is available. Accuracy
>> however will (likely) be limited by thermal drift etc.
>>
>> What are you hoping to achieve, prey tell ??
>>
>> -M
>>
>>
>> On 2/11/2024 7:37 PM, Charles Manning wrote:
>> > Hi All
>> >
>> > I have an idea I'd like to bounce around to see what others think.
>> >
>> > I want to get a high precision PWM going to drive a buck switch mode at
>> > a high enough frequency.
>> > Using a single PWM channel limits the precision I can get.
>> >
>> > So I was thinking: maybe there is a way I can combine two PWMs to
>> > generate a single high precision signal.
>> >
>> > For example, maybe one PWM can run at a slower frequency and provide
>> the
>> > "most significant bits "and another can run at a faster frequency and
>> > provide the least significant bits" and combine them with a flip flop
>> or
>> > such.
>> >
>> > Has anyone tried similar trickery in the past or have any other cunning
>> > tricks?
>> >
>> > Thanks
>> >
>> > Charles
>> >
>> >
>> > _______________________________________________
>> > Chchrobotics mailing list Chchrobotics at lists.ourshack.com
>> > https://lists.ourshack.com/mailman/listinfo/chchrobotics
>> > Mail Archives: http://lists.ourshack.com/pipermail/chchrobotics/
>> > Meetings usually 3rd Monday each month. See http://kiwibots.org for
>> venue, directions and dates.
>> > When replying, please edit your Subject line to reflect new subjects.
>>
>>
>> _______________________________________________
>> Chchrobotics mailing list Chchrobotics at lists.ourshack.com
>> https://lists.ourshack.com/mailman/listinfo/chchrobotics
>> Mail Archives: http://lists.ourshack.com/pipermail/chchrobotics/
>> Meetings usually 3rd Monday each month. See http://kiwibots.org for
>> venue, directions and dates.
>> When replying, please edit your Subject line to reflect new subjects.
>
> _______________________________________________
> Chchrobotics mailing list Chchrobotics at lists.ourshack.com
> https://lists.ourshack.com/mailman/listinfo/chchrobotics
> Mail Archives: http://lists.ourshack.com/pipermail/chchrobotics/
> Meetings usually 3rd Monday each month. See http://kiwibots.org for
> venue, directions and dates.
> When replying, please edit your Subject line to reflect new subjects.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.ourshack.com/pipermail/chchrobotics/attachments/20240212/19c573f4/attachment.html>