HP-15C LE bricked?

[Hans Brueggemann]

while experimenting with flashing a different firmware on an early HP-15 LE (serno. 003xx), i managed to make it totally unresponsive. the display went out after starting the erase sequence, which i expected. however, i can't see any response from the DBGTX pin when trying to send the firmware via the UART, and mysamba consequently says that it can't connect to the calculator.
i did a reset at some point and had to disconnect the power supply from the calculator, but of course, the erase worked "successfully" and the calculator is brain dead now. i'm not too desperate to get it fixed, but would be willing to change the processor as a last resort. any chance to revive it at this point before going surgical?

[dmh]

I was worried about this too but haven't had any issues with easily 10+ upgrades and downgrades to older and patched versions.

I followed the flash procedure from the WP 34S manual (shown in my downgrade video posted to the 15C LE downgrade thread) and it says if anything goes wrong you can just start again which made me think you can always recover.

I assume you've tried new batteries as the above warns about large power draw while the cable is connected?

Good luck and let us know if you manage to restore/recover it.

[Hans Brueggemann]

wasn't trusting the batteries, so went for a lab supply anyway. i'm seeing approx. 7.8 mA @ 3.3V when [ON], which seems to be fine. i repeated the suggested procedures multiple times already, but to no avail. sent bytes at RX pin of calculator look absolutely clean (115 kBaud), but TX pin on the debug port is silent, staying at logic "high" livel.

[dmh]

Maybe try with batteries. Can't hurt.

(07-22-2022 02:16 PM)Hans Brueggemann Wrote:  wasn't trusting the batteries, so went for a lab supply anyway. i'm seeing approx. 7.8 mA @ 3.3V when [ON], which seems to be fine. i repeated the suggested procedures multiple times already, but to no avail. sent bytes at RX pin of calculator look absolutely clean (115 kBaud), but TX pin on the debug port is silent, staying at logic "high" livel.

[J-F Garnier]

I don't know for the ARM-based 15C LE, but I often revived locked HP calculators by removing the batteries and doing a short-cut between the supply terminals to reset any spurious voltage from internal capacitors. You may try at the 6-pin connector pads.

J-F

[Hans Brueggemann]

thanks for the hints, gentlemen.
problem solved, probably just by chance: didn't try batteries, but on a hunch, i lowered the supply voltage to the calculator down to 3.0V (previously 3.3V) and tried the entire flash process again. lo and behold, WP34sFlash.exe did its job, and the calculator is functioning normally.
btw, i was using a JLINK ultra+ in VCOM mode (yeah, i'm a tool snob).

edit: find attached the STL file for a CR2032 dummy. couldn't post it as *.STL, so i added a ".txt" to the name. covered top and rim with self-adhesive copper foil and soldered two wires to it to make a q&d battery adapter for testing.

[vaklaff]

(07-23-2022 10:40 AM)Hans Brueggemann Wrote:  problem solved, probably just by chance: didn't try batteries, but on a hunch, i lowered the supply voltage to the calculator down to 3.0V (previously 3.3V) and tried the entire flash process again. lo and behold, WP34sFlash.exe did its job, and the calculator is functioning normally.

Congratulations, I'm happy for you :-)

(07-23-2022 10:40 AM)Hans Brueggemann Wrote:  edit: find attached the STL file for a CR2032 dummy. couldn't post it as *.STL, so i added a ".txt" to the name. covered top and rim with self-adhesive copper foil and soldered two wires to it to make a q&d battery adapter for testing.

Would you mind sharing a picture of the contraption of yours in action? In spite of being clumsy myself (or maybe because of it) I'm curious what skillful people do. Is it how you use the calculator regularly or was it just for the single occasion?

[Hans Brueggemann]

unfortunately, i don't have any pictures of the setup. to get a reliable electrical connection to the calculator, the 6-pin connector case for the back side was 3D-printed using TPU95A, a flexible material, which gives quite a good grip once inserted into the calculator. it needed quite an amount of trimming though to make it fit, hence no stl file for it, it's simply too ugly to be published.
contact pins for the connector are pogo-pins from digikey that i had on the shelf, don't remember the exact order code. the two buttons for RESET and ERASE were soldered onto a piece of prototyping pcb, and the whole contraption was then connected to the jlink with protyping cables fitted with dupont connectors.

i tried to investigate further on the voltage settings mentioned above, but couldn't replicate the behavior. hence, my conclusion on the difficulties i had with flashing the calculator is a wrong "timing" of pressing and releasing the buttons during the initialization sequence. maybe somebody else on the forum can give some hints on the timing that worked for him (her).

[dmh]

Having reflashed my 15C LE 10+ times I've had no issue with the timing of the button presses and you just need to follow the sequence as per the WP 34S manual which can be seen in real time in my 15C LE PSE bug fix video.

One observation that may help others, as the 15C LE keyboard is not as good as the original Voyager models and the ON and shift keys are worst, I found that the last ON before clicking download in MySamba did not always actually turn the calculator on and as there is no visible confirmation (the display doesn't show anything at this point) you can't tell if it is on or not but the software download will fail.

Subsequently I made sure the final ON before the download was a firm press to ensure it did power up. I've had no failures since.

[Hans Brueggemann]

ah, thanks, will give it a try!