Feat: Request PPS

[IvanLi-CN]

This PR simply supports requesting PPS voltages for a bit.

It's worth mentioning that I changed uom::si::u16 to uom::si::f32 in order to be able to use a voltage like 5.5V.

I'm just a hardware enthusiast and this PR has only been validated in my DIY projects, but I can't guarantee that it meets the specification.

[elagil]

Thanks for your PR!

Regarding the change of u16 to f32, that is not necessary because the voltage is represented in steps of 50 mV. So if you store 5.5 V in the standardized u16 format, it will be represented as 110.

[IvanLi-CN]

@elagil I have noticed _20millivolts, but when I define

uom::si::u16::ElectricPotential::new::<uom::si::electric_potential::millivolt>(7700)
// or
uom::si::u16::ElectricPotential::new::<_20millivolts>(350)

in my firmware, the result obtained is 7000 mV. Upon reviewing the final conversion request message, the calculated voltage is "((1661123644 & 2^20 − 1) >> 9) × 20 mV = 7000 mV", and the actual voltage measured on the line using a voltmeter is also 7 V.

The LLM informed me that this is due to the base unit being volts, which causes truncation. Although I find this quite surprising, I have not been able to successfully use u16 to obtain the desired voltage. I may need your assistance. If it is possible to use u16, could you provide some guidance based on the example I have submitted?

---

As a side note, I see that the PD 3.2 specification that came with your project says Output voltage in 20mV units for PPS.

I am not very familiar with this topic, so please do not mind if I have said anything incorrect.

[elagil]

Yes, you are right. Changing the storage size of electric potential only works via changing base units, which doesn't really help us. For now, let's go f32 then, until we find a better solution.

Please try and put the PPS request example into the existing STM32G4 example, e.g. toggle between requesting a fixed supply and the PPS. There is a lot of duplication otherwise.

[IvanLi-CN]

Okay, let me figure out how to merge the examples, I admit I made a separate copy out of laziness. 😜

[elagil]

No problem! In the H5 example, I also request (the same) power every five seconds. Maybe something like that, but switching between voltages.

[IvanLi-CN]

The example has been merged, and a small fix has been made for the floating-point precision issue. I think the PR should be ready.

[okhsunrog]

I think we shouldn't use f32 for that, operations with floats are imprecise, using u32 for mV would be a sufficient and better choise. You can take a look how it's done in Ralim's usb-pd lib if you can read C++:
https://github.com/Ralim/usb-pd/blob/609dda7d16a5d572b87794f4fd3b3c01f1bfd1f8/include/pd.h#L326

And here's an example of using it: https://github.com/Ralim/STM32-USB-PD-Demo/blob/main/Workspace/USBPDDemo/Core/user/PDUser.cpp

[i509VCB]

There is also the problem with using f32 and doing math on that value may involve rustc linking softfloat which can be an issue on targets without an FPU.

[IvanLi-CN]

I completely agree with the views of @okhsunrog and @i509VCB. I will study uom later to see if I can adjust the base units or use fractions for transmission. My goal is to reasonably use uom while avoiding the use of floating-point numbers.

[elagil]

I think we shouldn't use f32 for that, operations with floats are imprecise, using u32 for mV would be a sufficient

Yeah, I agree. Let's introduce an alias unit system like this:

// File: usbpd/src/protocol_layer/message/mod.rs
pub(super) mod cgs {
    ISQ!(
        uom::si,
        u32,
        (millimeter, kilogram, second, milliampere, kelvin, mole, candela)
    );
}

The above makes potential/current to be in base units of mV/mA. Power will then be in units of uW, which in u32 format still allows for a maximum represented value of 4295 W (good enough, I would say).

My tests run successfully with this, and also the derived units:

// File: usbpd/src/protocol_layer/message/mod.rs
#[cfg(test)]
mod tests {
    use uom::si::{electric_current::milliampere, electric_potential::millivolt};

    use super::_20millivolts_mod::_20millivolts;
    use super::cgs;

    #[test]
    fn test_units() {
        let current = cgs::ElectricCurrent::new::<milliampere>(123);
        let potential = cgs::ElectricPotential::new::<millivolt>(4560);

        assert_eq!(current.get::<milliampere>(), 123);
        assert_eq!(potential.get::<millivolt>(), 4560);
        assert_eq!(potential.get::<_20millivolts>(), 228);
    }
}

I suggest to get rid of any u8/u16 units and do everything in the new u32 system.

[IvanLi-CN]

Great, I applied elagil's cgs directly and made it public. Currently the code has been completely changed to use only custom cgs and remove the u8, u16, f32 featues. The examples have been changed as well and tested on my STM32G431 with no problems.

Thanks for your help!

[elagil]

Thank you, looks good to me! I will prepare a release.