[WP34s] Help Flashing!

[jayemel]

Hi,

I've only recently discovered the WP34s project and am trying to build one for myself. I own a number of HP machines (35, 21, 42s and 48GX) and am excited to add a WP34s to the family!

I'm an electronic engineer, so pretty familiar with the practical side of flashing a 30b into WP34s, but I seem to be stuck.

I have succeeded in erasing the stock firmware, but can't get it to program. I've tried a variety of USB to UART adapters (FTDI, Prolific and others), operating systems (Windows 10 and Ubuntu 15.04) and flashing applications (SAM-BA, MySamBa and wp34sFlash) and get the same result with all of them, that the programmer cannot connect.

I don't have a proper cable, so am holding wires firmly onto the relevant pads.

I'm powering the calculator from a lab power supply (at 3.2v) and I can see from the output current when the calculator is on and off, so the problem isn't battery related.

I have looked at the serial lines on an oscilloscpe, the computer is successfully sending three bytes with a few seconds gap between them, but the calculator never responds.

I've followed the instructions to a tee, and extensively combed the forums and tried everything that I can, but am getting nowhere.

Any ideas?
-- Josh

[BarryMead]

Many have reported that the software environment is pretty critical for a successful
flashing operation. The flashing occurs at 115200 baud, so the timing is more troublesome than the program/data transfer between the emulator and the calculator which takes place at 9600 baud.

Many have reported that linux/wine programming and Mac programming are unsuccessful. The most successful combination is Windows 7 with MySamba.exe

I have not heard any reports of how successful MySamba is under Windows 10.

If you can get your hands on an older Windows 7 machine, you might have better luck.

Remember to use the sequence Reset (on the back), On button (on the front) then transfer the file with MySamba. Also if you don't have the crystal use the calc.bin
or calc_full.bin not the calc_xtal.bin or calc_xtal_full.bin

Best of luck, Barry

[jayemel]

Thanks Barry,

I'll borrow a Windows 7 machine tomorrow and have a go.

Regards,
-- Josh

[walter b]

In addition to what Barry wrote:

• Forget SAM-BA, take MySamBa. The latter is less critical.
• Take an FTDI-based converter, forget the other stuff.
• Take a PC and Windows 7 or XP.
  

Not so sure about the baudrate but MySamBa controls it anyway. Well, and holding the wires onto the pads isn't optimum since 24s (flashing time) can be long.

Good luck!

d:-)

[jayemel]

Thank you Walter!

Regards,
-- Josh

[jayemel]

Update time:

I succeeded in flashing one of my calculators using the protocol described by Walter on a Windows XP virtual machine.

The other is still fighting me, although I noticed it is drawing much less current (~2mA v.s. ~15mA), so wonder if something else is wrong with it.

Cheers,
-- Josh

[walter b]

Congratulations, Josh!

For the second specimen: What's the difference to the first one during the flashing procedure now? You mentioned that both calculators didn't respond yesterday - but now one of them obviously did. Or what do I miss?

d:-?

[jayemel]

Dear Walter,

I have done some more digging, and come across a thread by John Galt that describes exactly what happened with my second calculator:

Quote:MySamBa reported a failure to acknowledge midway through the transmission. After the erase procedure the calculator had always drawn about 8 mA from the power supply. It now draws only 1.8 mA no matter what.

Sadly, it does not look like John resolved this issue, so I may have killed this 30b!

Regards,
-- Josh

[walter b]

Well, 'John Galt' was speculating then the error developed due to a power interruption. Does this concur with your observations?

d:-?

[jayemel]

I am not sure what the source of the interruption was, I think a UART interruption is more likely than power, but it is possible - all the connections were somewhat precarious.

-- Josh

[emece67]

I find the wp34s flashing procedure quite delicate when performed with DIY cables. In fact, as I reported months ago, up to now I've damaged (truly unrecoverable, even via JTAG probes) two machines during flashing.

I urge you to provide good connections for the serial interface; use fresh batteries (or supply the machine from the USB-serial converter); insert small (say, 100 Ohm - 1 kOhm) resistors in series with the RX/TX lines (see post by dewster here); follow strictly the flashing sequence given in some other posts...

Otherwise you have great odds to damage any (progressively scarcer at good prices) 20b/30b you try.

Regards.

[matthiaspaul]

(11-03-2015 05:06 PM)emece67 Wrote:  I urge you to provide good connections for the serial interface; use fresh batteries (or supply the machine from the USB-serial converter); insert small (say, 100 Ohm - 1 kOhm) resistors in series with the RX/TX lines;

I would like to second this. Both a proper and stable power supply and the series resistors are quite important for various reasons. One of them is to limit the potential current through the clamping diodes in the chip in case the voltage on the RX input pin would become higher than the microcontroller's supply voltage (or lower than 0V). The clamping diodes exist to protect the internal circuitry, but if they get damaged they will make the corresponding port pin unusable. This scenario is quite likely to happen if the batteries aren't full any more or if the calculator has been left without batteries for a while (so that the capacitors between supply and ground have discharged). In this case, the microcontroller not only gets reverse powered through the input (which may already exceed the current limit of the affected clamping diode), but if the batteries are empty or the capacitors discharged, it is basically like driving the serial input signal against a short circuit for a moment - definitely too much for the clamping diodes. A firm low-resistance connection between the I/O power supply that powers the source of the serial signal (f.e. the 3.3V output of a FTDI chip) and the supply plane of the calculator as well as a similarly good ground connection will prevent this for as long as these connections are established before the serial line is driven high. However, if this connection fails for some reason, it may likely damage the chip when logical high level is asserted on the serial line, at least when the series
resistor(s) aren't present.

The drawback of the series resistors is that they will limit the allowable length of the serial data line.

BTW. A low-resistance connection to the external power supply typically means that the calculator's batteries would be connected almost directly to the external supply. Since this may damage the batteries if the external supply voltage is higher (or lower), it is a good idea to remove the batteries from the calculator for as long as the calculator is powered externally or to add circuitry to decouple the battery supply path from the external supply path through a MOSFET switch (or - significantly less efficient - through an ultra-low drop Schottky diode).

Greetings,

Matthias