A new version of the RP2350, the chip that powers the Pico 2, has been announced, but you will have to wait if you want a Pico 2 that uses it.

Soon after the Pico 2 launched an embarrassing fault was discovered. The GPIO output stage had a "fault" that resulted in a much higher leakage current in output mode than was usual.

This might have been acceptable and a minor irritation, but for the fact that the characteristic curve of leakage current v voltage was non-linear in a way that resulted in a stable state at around 2.2V. This stable state interfered with the resetting of the GPIO line when in pull-down mode. The line starts out at zero volts due to the pull-down but once another device pulls the line high it doesn't return to zero but to a stable point at around 2.2V. The problem is that 2.2V is a valid logic 1 and so the line appears to have locked up to logic 1 after its first transition to a high state. The line does vary between 2.2 and 3.2V as it is toggled, but this reads as a steady logic 1 to the input state.

Until now, the solution to the problem has been to use an external pull-down  resistor of around 10K or greater. This solves the problem at the expense of more GPIO current, but most applications can live with this.

Unsolved, the leakage problem is present in all modes and all GPIO lines and this sometimes results in some even stranger behavior if you are not aware of it - for example using a voltage divider with the "lower" resistor around 80k produces a similar lockup.

All of the Pico 2s in circulation to date have this problem and will have for an indefinite future. Raspberry Pi estimate that there are half a million faulty chips in use. Now we have an new "stepping" of the chip, the A2, and it completely removes the strange non-linear behavior of the input line:

The yellow line is the error and you can see that there is negative differential resistance - the current goes down as the voltage goes up - which causes the problem. The red line is the fixed GPIO input curve and you can see it is flat and very close to zero which is what it should be. Any external pull down resistors could be left in place but they might not be necessary any more. Other more complex input circuits could behave differently.

This fix might not be a fix for any circuit that was built to take account of the problem.

As a spin off the GPIO line is now rated as 5V tolerant so you might not even need the voltage divider any more.

As well as the big fix that affects everyone, there are also a collection of security improvements that deal with problems revealed in the last hacking challenge, see our report, Pico RP2350 Security Bounty Won.

To make sure that the fixes are an improvement a new hacking challenge has been announced:

And finally, we have another RP2350 Hacking Challenge, offering a $20,000 prize for a practical side-channel attack on the power-hardened AES library that underpins our decrypting bootloader.

Given some of the techniques used in the previous challenge, this is security beyond what most of us expect. If a hacker is going to go to the amount of trouble that was evidenced, all I can say is that they can have my secrets...

More seriously, there is still no support for the security features in the Pico SDK and this means that for most programmers the security features are basically off limits. They are a kit of parts that most of us aren't going to spend the time required learning to implement only to have the SDK catch up and make our work redundant.

The whole area of Pico security is best summed up as "it's all there, but it's up to you to figure out how to use it".

So can we now expect the next Pico 2 we order to arrive with a fully fixed chip?

Nope.

It will take time for the new chip to work its way through the production chain and there seems to be no way to ensure that the next Pico 2 you buy has the new stepping. I can't help thinking that if this was the motor industry there would be an expensive recall.

• Harry Fairhead, I-Programmer's hardware guru, is the author of the newly published Programming The Raspberry Pi Pico/W In C, 3rd Ed and Programming the Raspberry Pi Pico in MicroPython, Third Edition.  

More Information

RP2350 A4, RP2354, and a new Hacking Challenge

Related Articles

Pico RP2350 Security Bounty Won

The Pico/W In C: Erratum E9 Pull Down Problems

Pico 2W Announced But There Is A Surprise!

Raspberry Pi Beaten To It With Pico Plus 2

Pico 2 Faster With More Memory And RISCier

To be informed about new articles on I Programmer, sign up for our weekly newsletter, subscribe to the RSS feed and follow us on Twitter, Facebook or Linkedin.

Comments

Make a Comment or View Existing Comments Using Disqus

or email your comment to: comments@i-programmer.info