Seems I have lost my Propeller Chip

greenman

I have 3 P8X32A boards, 2 Quickstart boards with built in USB and 1 schmartboard which needs a Prop Plug.
The schmartboard quit, everything looks fine. You can check the ID of the chip, prop plug looks fine in device manager.
Both Simpleide and Propeller say the code verified but Viewport says the code is not good. The chip will not do a blink
or send 'hello' to the terminal.

Is there anyway to verify the chip is dead. It has the standard 5 MHZ crystal. Could the prop plug be bad even though
it ID's the chip. I do have activity on RX and TX.
Can the reset from the prop plug be checked. I have tried to read it with a meter, my Oscope and my logic analyzer.
The meter reads 3.3 volt and stays there, the oscope wiggles and the analyzer shows a pulse occured but I do not know
the threshold for the pulse.

I have not used my 2 good boards as test subjects, hoping someone knows.
Any help would be great !!

Mike Green

Try a simple LED blink type program using the RCFAST clock rather than the crystal clock. You won't be able to do any serial I/O because the RCFAST clock isn't accurate enough or stable enough to do so. If that works, then you've blown the PLL output multiplexor. Poor bypassing or connecting to only some of the Vss/Vdd connections on the chip or a combination can cause voltage surges across the chip's internal busses that can burn out parts of the chip and the PLL output multiplexor seems to be the most vulnerable. The chip will identify itself and download programs since that uses the RCFAST clock, but you can't switch to a PLL-derived clock.

You can also use XINPUT and provide an external clock signal to XIN (any frequency up to 80MHz).

greenman

Thanks Mike, I will try that and post back.

Electrodude

If Propeller Tool can ID the chip, cog 0 definitely works. Did you try an LED blinker that uses RCSLOW or RCFAST instead of a crystal? I suspect either your crystal or your connections to the crystal is bad.

EDIT: ninja'd by half an hour as usual

greenman

When the chip quit I was blinking a single led on pin 16. By switching the clock to RCFAST the board blinks an led but not on pin 16.
It is dead, took me a while to switch pins (dah). There is one electrolytic 1uf between pin 8 and 17. No ceramic caps on the board at all.
Thank you Electrodude for your opinion, with a dead pin do you still think the PLL is OK ?

Mike Green

You really really need 0.1uF ceramic caps between each pair of Vss/Vdd supply pins, close to the chip. Having a 1uF electrolytic between Vss and Vdd nearby is fine, but not having the 0.1uF bypass caps is a sure recipe for PLL multiplexor damage. Remember that the chip runs typically on 3.3V and starts getting damaged at 4.0V. That's only 0.7V ... easy to get a noise spike on that order. If your regulator runs high (like 3.5V ... only +6%), that's only a 1/2V margin.

greenman

I agree 100% Mike, Not sure why Schmartboard does not know that. I really messed up by not looking at all the information on the different boards before buying one. Not sure what to do with this board now, maybe just write it off as a lesson. Putting the caps on is not a problem but adding a shield or breadboard would be a challenge because the spacing between rows of pins is not standard. If I supplied an external 80mhz would serial work or is it dependent on PLL. Thanks for your help !!

Mike Green

You could use an external 80MHz clock signal and specify XINPUT for the clock source. That doesn't use the PLL or its output multiplexor and would allow you to have stable serial I/O. Depending on the Baud you want to use, you could also use a lower clock rate (40MHz or 20MHz). Tracy Allen's version of the 4-port serial driver should be able to handle a single 115KB port with a 20MHz clock for example.

localroger

For that matter in my industry we rarely have a need for anything over 19.2 kbaud, and you can do that with the 5 MHz crystal by using it straight through instead of trying to multiply it.