 Notice. New forum software under development. It's going to miss a few functions and look a bit ugly for a while, but I'm working on it full time now as the old forum was too unstable. Couple days, all good. If you notice any issues, please contact me. |
Forum Index : Microcontroller and PC projects : Hardware bug in RP2350
  Page 3 of 3 |
|||||
| Author | Message | ||||
TassyJim Guru  Joined: 07/08/2011 Location: AustraliaPosts: 6097 |
Code that uses internal pulldowns will not be portable. It is very plausible that some of the BITBANG routines use pulldowns internally. They all will need checking. High impedance sources are likely to need a buffer to get the required drive. Jim VK7JH MMedit MMBasic Help |
||||
| matherp Guru Joined: 11/12/2012 Location: United KingdomPosts: 9105 |
I've checked the Pico firmware source and the only place pulldowns are used is when the user explicitly requests them with SETPIN DIN or SETPIN INT(H/L/B) Accordingly, I don't intent to make rp2350 specific changes to the firmware so it will be up to the user to avoid pulldowns when using relevant releases of the chip. Who knows they may fix it sometime. |
||||
| TassyJim Guru Joined: 07/08/2011 Location: AustraliaPosts: 6097 |
That's good news. I will keep testing my collection of sensors as time permits but so far, all is good! Jim VK7JH MMedit MMBasic Help |
||||
Mixtel90 Guru Joined: 05/10/2019 Location: United KingdomPosts: 6778 |
Since internal pullups and pulldowns on the majority of microprocessors and microcontrollers are generally poor and only ever used over short distances it's not really a big issue anyway. When used in commercial applications (which is where RPi want to sell most of these) there will be external resistors (unless the design is *really* bad). The important thing is that pullups work. They are far more prevalent and used for very important stuff like on-board short range I2C and SPI. It will confuse beginners but I'm afraid that the literature aimed at them will have to be suitable. Most seem to use pulldown for buttons and switches anyway. Since this bug doesn't prevent the chip from being used in a commercial environment I doubt very much if they'll bother fixing it. Mick Zilog Inside! nascom.info for Nascom & Gemini Preliminary MMBasic docs & my PCB designs |
||||
| PhenixRising Guru Joined: 07/11/2023 Location: United KingdomPosts: 857 |
Thanks for the summary, guys. Precisely what I was about to ask for and kudos to Volhout for bringing the issue to our attention  |
||||
Grogster Admin Group  Joined: 31/12/2012 Location: New ZealandPosts: 9305 |
It must be something of a nightmare, designing a new die mask for chip production. We all know how easy it is to have bugs in your CODE, and that extends to "Bugs" in the silicon also. When designing a die, it needs to be 100% perfectly correct right from the off - a big ask, when you look at the complexity of a modern chip. Sure, just like us PCB designers, you do prototype runs to make sure everything is OK, but just like with PCB prototypes, you find and fix all the problems you THINK you missed.....only to find AFTER the thing goes into production, there was something else subtle you missed! Any PCB can easily have a new revision made to fix any layout error, but that is pretty much impossible with a silicon die after the prototype is released. Revision 2 etc dies do exist for chips, but as Volhort mentioned earlier.....cost big bucks! So the expense has to be worth it. I don't envy die designers!  Smoke makes things work. When the smoke gets out, it stops! |
||||
| Volhout Guru Joined: 05/03/2018 Location: NetherlandsPosts: 4222 |
Dies are going through several steps in production. Most of the steps( creating the P and N junctions) are processed on multiple wafers. Your die now basically is a huge box of P and N channel fets. The last few steps add the metal interconnect layers, to connect the fets into logic gates, adders, flipflops, and such...the final product. These steps are done for A0 ( first proto of the die), A1, second proto. So A0 and A1 have the same fets, only connected differen when bugs must be fixed. The chips we have are A2. So ot is the 3rd revision. Summary: not all masks are changed, typically only the metal layer masks. Volhout Ps, there are 100's of dies on a single wafer, maybe even 1000's. And a whole wafer is processed. Not independent dies. Edited 2024-08-31 23:07 by Volhout PicomiteVGA PETSCII ROBOTS |
||||
| robert.rozee Guru Joined: 31/12/2012 Location: New ZealandPosts: 2350 |
it looks like ADC inputs are also impacted in an 'odd' sort of way that MAY, UNDER SPECIFIC USAGE, see oscillations coming out of the pin: https://forums.raspberrypi.com/viewtopic.php?t=375631&start=50#p2249127 "Connect a low resistance pot (I used 5K) between 3.3V and ground and connect the wiper of the pot to pin 27 via a 1M resistor [...] When we examine the anomalous signal on the scope (200MHz Siglent SDS1202-XE) we see oscillations with a peak amplitude near enough to Vdd and a pulse width of 10ms but with a random interval between pulses which is at least 50ms but which can go out to several hundred ms - this is why the DVM only shows a relatively low voltage" looks like the bottom line may well be that all inputs need to be driven with a somewhat low-impedance source, in the case of analog inputs a buffer opamp wired for unity gain would be a good choice. cheers, rob :-) addendum: have added the words in red above: "MAY, UNDER SPECIFIC USAGE". it seems that ajay_m in the above linked post may be resetting the pin status between reads, which in the RP2040 has proven unproblematic, but with the RP2350 causes the behavior he is seeing. it occurs to me that, in the case of MMbasic, one solution may to, when a pin is configured for digital I/O, always configured for weak pull-up. this will ensure that unless a pin is explicitly pulled low by a suitable external signal, the pin will always return '1' and sit in a known state. Peter could add this as a permanent configuration to the firmware, ie hardwire the PULLUP parameter to SETPIN for all but analog/ADC configurations. cheers, rob :-) Edited 2024-09-01 22:08 by robert.rozee |
||||
| JanVolk Senior Member  Joined: 28/01/2023 Location: NetherlandsPosts: 144 |
Last week the manual for Raspberry RP2350 Datasheet was updated for the hardware bug GPIO RP2350-E9 on pages 1341 to 1343. Jan. |
||||
| Volhout Guru Joined: 05/03/2018 Location: NetherlandsPosts: 4222 |
All adc s have in their dataset a definition of input impedance. This guy uses a 1 meg series resistor. Not one ADC I know works flawless with this high impedance. If that impedance is required, put an opamp buffer between input and adc. Volhout PicomiteVGA PETSCII ROBOTS |
||||
| Mixtel90 Guru Joined: 05/10/2019 Location: United KingdomPosts: 6778 |
Agreed. It was a bit silly. The Datasheet says: "Over 100K" doesn't imply that the input impedance is anywhere close to 1M. It probably means that they tested it at 200K and it was fine. Mick Zilog Inside! nascom.info for Nascom & Gemini Preliminary MMBasic docs & my PCB designs |
||||
| Page 3 of 3 |
|||||
 Print this page |
