nrf pin drive strength

[dhalbert]

Currently nrf DigitalInOut uses high (strong) drive strength on the output pins, but none of the other peripherals for digital outputs are set to do that. The nrfx drivers all use standard drive strength,including, interestingly, I2C (TWIM).

We changed the atmel-samd to use high drive in the past, which is why I did the same for nrf DigitalInOut.

I could reconfigure the SPI, PWMOut, etc. pins to have high drive after initializing the devices.

Standard drive is typical 2ma, high is 9 or 10ma. rise/fall time is 9-25ns (15-50pf load) on standard, 4-8 on high.

However, Nordic recommends 15ma total from all GPIO pins, at least for nRF52832. This is not so much for safety as to prevent low voltages going out of range due to resistance (see below). I haven't found a similar limitation for nRF52840.
https://devzone.nordicsemi.com/f/nordic-q-a/20484/high-drive-mode-downsides-nrf52/79880#79880
https://devzone.nordicsemi.com/f/nordic-q-a/15800/gpio-sink-current-on-nrf52832/60290#60290
Quoting:

_ The reason you should not sink 15mA on multiple pins is that some pins may have common ground path. This means that if you sink high currents on pins that are close to each other you may see that the voltage on both pins (pads) rise above what is specified in Figure 4._

@ladyada @microbuilder any comments?

[uhrheber]

The main reason for configurable drive strength isn't power saving, but EMC. By choosing only as much strength as necessary, you can pass EMC tests, without having to add series resistors. This is especially important for high speed buses. But having a high drive strength doesn't necessarily mean, that this current flows through the pin. When driving bus lines, the current will flow only when the level changes, and only as log as you need to charge or discharge the line capacitance.

I think that everything that drives a bus line (SPI, I2C, I2S, etc.) should have a high drive strength by default, to avoid possible signal corruption.
PWM is another thing. When driving LEDs, that in any case have current limiting resistors, this isn't critical. But when driving e.g. power MOSFETS at high frequencies, their input capacitance will overload the output pin, without additional measures.

This is a typical use case for #1268

Inexperienced users will be OK with the preset values, while advanced users could manipulate the registers by themselves.

[dhalbert]

Thanks, yes. One thing I neglected to mention is that we already know Nordic recommends that some pins be used only at low-speed (< 10kHz) with standard drive for RFI reasons. The '832 data sheet is clearer on this than the '840 one:

Radio performance parameters, such as sensitivity, may be affected by high frequency digital I/O with large sink/source current close to the Radio power supply and antenna pins.

Table 23: GPIO recommended usage on page 109 identifies some GPIO that have recommended usage guidelines to maximize radio performance in an application ...

On the '832, pins P0.22 through P0.30 are denoted this way.
On the '840, it's P0.02-P0.03, P0.29-P0.31, P1.01-P1.07, and P1.09-P1.15.

[uhrheber]

I see your problem. Drive strength and speed limitations is something that a professional developer would consider beginning in the concept phase of a product, but a DIYer won't care about, so circuitpython should prevent him from doing any mistakes, that he won't be able to solve.
I don't have a simple answer for that.

[tannewt]

@dhalbert What do you want to do for this?

[dhalbert]

@ladyada Do you think we should try to make certain pins high drive strength (e.g. those assigned as I2C or SPI), or just forget this for now?

[ladyada]

id have em all be high drive strength - people like to connect LEDs and stuff...maybe at some point we can offer a way to configure the microcontroller.pin drive but we high strength default on samd already!

[dhalbert]

@ladyada Note comment above (#1270 (comment)) that Nordic says low-speed pins should use standard drive to avoid RFI to the radio. So there are arguments both ways: high drive so we can drive a lot of stuff easily; standard drive to avoid RFI. Maybe we'll have to document this, since if we respect the guidelines, it'll be confusing which are available for high drive.

[ladyada]

ok how about low speed pins we low drive (theres no lowspeed's on the feather nrf) and then high speed high drive :) we can document this in product pages

[bwshockley]

I've potentially got a board (Mini SAM Nordic version in the works) that would use a low speed pin (P1.10) to drive the builtin_led. Would this be configurable via board definition to set to high drive? Or is this not possible currently? I could just create a fork and modify to suite my needs, but I'd rather not fork CP too much if I can just define in the board files.

[dhalbert]

@ladyada That sounds good -- thanks!

@bwshockley We don't currently store any info labelling pins as low-speed or high-speed, but we could store that in the pin table and add some kind of setting or override mechanism.

The regular drive might actually be fine for your LED. What's the current draw you set via resistor?

[bwshockley]

Oh - by all means, there is no need to make changes on my account, just curious. 2ma should actually be okay - I tend to choose my resistors to set lower current anyway in case I'm on battery. But, I was trying to make sure that if I determined I needed a little more, I could get to it.

[tannewt]

@dhalbert and I chatted about this. We decided to punt it to 4.x or later when we can introduce a drive_strength property to DigitalInOut. Automatically deciding risks incorrect and confusing behavior.