HP 39gs and SD card

[rcpdvc]

HP 39gs has same PCB with SD socket footprint as a 50g. So, it is possible to mount socket and some components to get SD on 39gs. Of course, native firmware is not intended to work with SD, but it's implemented in newrpl (thanks to Claudio L.). Another advantage of SD is 50g-way flashing without Windows machine (real or virtual. Wine can be used to pass data over tty-COM, but flashing requires connection over USB driver).

To perform update save firmware file and "update.scp" text file containing firmware filename in card's root. Enter bootloader as usual (reset while holding "+" and "-"). Press "1" key ("1.UPDATE CODE"). Press "8" key (there is no hint for this on the screen, only "1.USB").

Missing resistors parameters are taken from 50g PCB photo, all in 0603 smd package.

PCB Nets Value Package
mark Ohm(F) mark

R61 7.DAT0 - Vcc 10k 103
R62 8.DAT1 - Vcc 10k 103
R63 9.DAT2 - Vcc 10k 103
R64 1.DAT3 - Vcc 10k 103
R65 5.CLK - CPU 10 100
R66 sw COM - Vcc 1k 102
R67 sw #WP - GND 100k 104 WriteEnable.
R68 sw CD - GND 100k 104
R69 2.CMD - Vcc 10k 103
C73 4.Vcc - GND ~0.1u - Decoupling. Accurate value not required.
C74 5.CLK - GND <10p -

I have omitted C74 at all. What the purpose of increasing CLK line capacitance? Noise filtering? Anyway it should be small enough to fit specification limits for total line capacitance (<40pF).
Confirmed working in bootloader (SD, MMC fat16 and fat32) and newrpl (SD fat16 and fat32).

[pier4r]

Cool. Would you write an article on it?

http://www.hpmuseum.org/forum/forum-14.html

[Claudio L.]

(10-29-2017 03:34 PM)rcpdvc Wrote:  Confirmed working in bootloader (SD, MMC fat16 and fat32) and newrpl (SD fat16 and fat32).

Wow, that's fantastic news! This should also apply to the 40gs as well, since they all share the same hardware.
Thanks for your work on this, we'd love to see a video or a step-by-step explanation for the not-so-technical. In any case, soldering those tiny surface mount components is not easy, but perhaps either you or somebody in the forum can make mods for other people. For newRPL it's a nice alternative to the 50g, now that it's hitting astronomical prices.

[matthiaspaul]

(10-29-2017 03:34 PM)rcpdvc Wrote:  R67 sw #WP - GND 100k 104 WriteEnable. Not required with newrpl or bootloader.

I haven't tried myself, does newRPL ignore the write protection switch? Doesn't sound like the best of ideas to me then...

Quote:C74 5.CLK - GND <10p -

I have omitted C74 at all. What the purpose of increasing CLK line capacitance? Noise filtering? Anyway it should be small enough to fit specification limits for total line capacitance (<40pF).

Yes, it is good practise to add some 10pF to the clock of SD cards. This reduces the emission of higher frequencies (harmonics) to avoid EMI problems. Otherwise you might "hear" your calculator in nearby radios.

Greetings,

Matthias

[rcpdvc]

(01-05-2018 02:34 AM)matthiaspaul Wrote:  I haven't tried myself, does newRPL ignore the write protection switch? Doesn't sound like the best of ideas to me then...

newrpl ignores write protection switch (at least 39gs version). And I think that's right. It's tiny mechanics and both lock failure and unlock failure are possible. Using file system facilities is way more reliable.

(01-05-2018 02:34 AM)matthiaspaul Wrote:  Yes, it is good practise to add some 10pF to the clock of SD cards. This reduces the emission of higher frequencies (harmonics) to avoid EMI problems. Otherwise you might "hear" your calculator in nearby radios.

I've got it. I need a radio to listen my 39gs before I place these 10pF.

[Claudio L.]

(01-05-2018 02:34 AM)matthiaspaul Wrote:  I haven't tried myself, does newRPL ignore the write protection switch? Doesn't sound like the best of ideas to me then...

newRPL does honor the write protection switch.

[matthiaspaul]

(01-05-2018 02:52 PM)Claudio L. Wrote:  

(01-05-2018 02:34 AM)matthiaspaul Wrote:  I haven't tried myself, does newRPL ignore the write protection switch? Doesn't sound like the best of ideas to me then...

newRPL does honor the write protection switch.

Great, thanks, I misred rcpdvc's statement. Obviously, he just found that the #WP line does not need to be weakly pulled down to GND. However, I would advise against not populating R67, in particular if the signal would go straight to the controller. (He does not state so, but readers could get the impression that, if it seems to work without R67, the resistor might not be needed at all.)

Charge may build up depending on various conditions and open inputs do not necessarily hold "low" all the time. Also, if the voltage level happens to float in the middle region between VCC and GND, this may increase the CMOS controller's power consumption considerably (even if the calculator is switched "off").

Greetings,

Matthias

[rcpdvc]

(01-05-2018 02:52 PM)Claudio L. Wrote:  newRPL does honor the write protection switch.

Yes, it's my mistake. It was floating #WP input. I placed R67 and it is clear now. But was that behavior of newrpl the same all the time? It seems to me that somewhere you mentioned about ignoring WP, may be I'm wrong. I've checked the oldest firmware I used (v0.8 build 862) and found that it scans WP switch too.

(01-05-2018 03:42 PM)matthiaspaul Wrote:  Obviously, he just found that the #WP line does not need to be weakly pulled down to GND. However, I would advise against not populating R67, in particular if the signal would go straight to the controller. (He does not state so, but readers could get the impression that, if it seems to work without R67, the resistor might not be needed at all.)
Charge may build up depending on various conditions and open inputs do not necessarily hold "low" all the time. Also, if the voltage level happens to float in the middle region between VCC and GND, this may increase the CMOS controller's power consumption considerably (even if the calculator is switched "off").

My #WP line wasn't connected neither to R67 nor to switch (socket to footprint mismatching) at all. Floating input. Obviously I have measured #WP line voltage, have got 0V and decided that newrpl ignores WP switch because it writes with #WP=0 (WE=0) and there is nothing more to do here. It's clear now #WP line was in high state and I just pulled it down with my meter.
I'll edit my first post. Sorry for misleading. Thanks for help.

[Vtile]

Hello. Any information about the part number or model of the SD card reader part. As a hobbyist, the identification of the part with correct footprint is pita.