Raspberry Pi’s new RP2350 microcontroller household continues to be the main focus of appreciable troubleshooting this week, as builders more and more run into the side-effects of a design flaw often called erratum RP2350-E9 — with the end result that some are cancelling designs constructed across the new chip, at the least till a brand new stepping is launched with the fault fastened.
Raspberry Pi unveiled the RP2350, a quad-core dual-architecture successor to the favored RP2040 and its first half to characteristic RISC-V cores, again in early August, having supplied early entry to a variety of {hardware} companions to design their very own boards to sit down alongside the Raspberry Pi Pico 2. Sadly, all has not been easy with the system’s launch: the invention of a {hardware} fault, often called erratum RP2350-E9, which causes general-purpose enter/output (GPIO) pins to “latch” at across the 2V mark has been the reason for appreciable consternation amongst designers and engineers.
The fallout from Raspberry Pi RP2350 errata E9 continues, with Bus Pirate 5XL and 6 manufacturing halted till a workaround may be discovered. (📷: Ian Lesnet)
“[I] discovered a silicon bug,” Harmful Prototypes’ Ian Lesnet defined of the problem, following his report of the identical to Raspberry Pi and its recording of the flaw as a proper erratum. “When a GPIO [General-Purpose Input/Output] pin is an enter with the pull-down resistor enabled, it acts like a bus maintain. We use the pull-down on the button, which connects to three.3V when pressed. Through the self-test urgent the button works, however then it by no means goes low once more, it sits at 2.15V…”
Lesnet had been working with the RP2350 for the Bus Pirate 5XL and Bus Pirate 6, constructing on his earlier RP2040-powered Bus Pirate 5 design. These, nonetheless, have now been shelved, at the least in the meanwhile. “RP2350 based mostly Bus Pirate (6 & 5XL) manufacturing is paused,” he explains in an replace posted to Mastodon this week. “E9 errata seems extra expansive than reported in [the] datasheet, [and the] first batch of RP2350 Bus Pirates are most likely DOA [Dead On Arrival]. RP2350 inputs appear to stay at ~2.1 volts with out pull-down/load of ~9k ohms/0.33mA.”
The flaw impacts a variety of operations, from fundamental digital output to PIO-driven SPI — corrupting the sign. (📷: NNNILabs)
“BP makes use of pin pull-downs with open collector bus varieties as a result of it simplifies PIO [Programmable Input/Output] applications that run by way of 1.2-5 volt bidirectional buffers. I tried to right for E9 based mostly on what was recognized on the time: by disabling pin pull-down and utilizing an exterior 100k pull-down resistor. Sadly, this most likely wasn’t sufficient. As reported by […] others, the latch up would not appear restricted to when pull-downs are enabled. Any enter pin not held low with [a] ~9k resistor will stick at 2.1 volts after going excessive in a number of exams.
The issue, simpler to set off than the official erratum within the datasheet would counsel, has been confirmed by others together with CircuitPython core developer Dan Halbert and maker Tod Kurt, who has acknowledged that “I presently have a number of Pico-based initiatives that can’t use Pico 2/RP2350 due to this downside” — whereas Agustín Gimenez Bernad has cancelled a work-in-progress effort to port his RP2040-based logic analyzer to the RP2350 as a direct results of the flaw. “In accordance with the official errata the lock solely occurs when the pull downs are enabled, you enter a excessive stage worth after which the GPIO begins outputting 2.1V,” he explains. “That does not sound too dangerous however the actuality could be very totally different.”
The flaw impacts all gadgets based mostly on the RP2350, together with these from third-parties like Adafruit. (📷: Adafruit)
“I have been testing it and even forcing the pulldowns to be disabled,” Bernad continues, “the PIO triggers the lock. On this state, the RP2350 is ineffective if you might want to use the GPIOs to enter any information. The one workaround supplied is to disable the pins and allow them when you’re going to learn and disable them after it to reset the pin standing, however as you possibly can think about with the PIO that is unimaginable, and even when it was attainable the seize velocity could be decreased a lot that the analyzer could be completely ineffective. Sadly I need to cease the port to the Pico 2 till this case is solved.”
Raspberry Pi engineer James Hughes has issued temporary remark within the firm’s discussion board, confirming that the corporate is in discussions with the IP provider it blames for the fault within the chip — however, on the time of writing, the corporate was sticking with the outline of the flaw because it stands within the datasheet, and had not commented on the timing and even probability of a brand new stepping, which might resolve the issue in {hardware} somewhat than in documentation.
events can monitor the progress of the investigation through the Raspberry Pi Pico GitHub Situation.