WP34s Repurposing Journey PhotoJournal

[MarkHaysHarris777]

Whoohoo-- caps and crystals arrived today (snow finally let up)!

Thanks Eric!

Cheers

[MarkHaysHarris777]

The redesigned flash-clip with new reinforced brass Katie clips is ready to go. The heavier brass was much harder to work with; to get adjusted right (I've futzed around with it most of today), but finally in the end its going to be around for some extended flashing.



Again, I didn't see any sign of metal fatigue in this setup, but I decided to reinforce the tip of each Katie clip with a small blob of solder. This clip really 'snaps' into place and is royally solid. I'm totally regretting that I futzed around with the aluminum in the first place! :-}




Cheers

[rprosperi]

(02-04-2015 09:55 PM)MarkHaysHarris777 Wrote:  The redesigned flash-clip with new reinforced brass Katie clips is ready to go. The heavier brass was much harder to work with; to get adjusted right (I've futzed around with it most of today), but finally in the end its going to be around for some extended flashing.



Again, I didn't see any sign of metal fatigue in this setup, but I decided to reinforce the tip of each Katie clip with a small blob of solder. This clip really 'snaps' into place and is royally solid. I'm totally regretting that I futzed around with the aluminum in the first place! :-}




Cheers

Great photos Marcus, they really show the new design very well. I've been following the recent play-by-play as you build your ultimate 34S flashing station. Interesting stuff.

[MarkHaysHarris777]

When this penny was mint fresh in San Francisco President Nixon 'was not a crook' and was still in office... I was a sophomore in high school... and the HP35 was still a gleam in old man Hewlett's eye; the HP35 was launched in 1972. The two little objects that look something like grains of table salt are the 18pf caps that need to be soldered in place, along with the 32.768 Khz watch crystal at the right. I spent some considerable time watching Eric's video, and filing a very clean sharp tip on my soldering iron. Yes, I believe in meditation and prayer.



I used paste flux on the end of a tooth pick to prepare the places that needed the C3 C4 caps, and of course the two pads for the crystal. With surface mount stuff you pretty much get one shot, and don't screw up! One cap (C3) is slightly crooked (do I care?) NO! No cold solder connections, the machine booted up, and the system recognized the crystal... whoohoo--- if it ain't broke, don't fix it. (if y'all can do better, God bless you...)



You might have noticed that the COB blob is just about the same size as the penny... the entire system is in that blob of epoxy; astounding, really.

Cheers

[MarkHaysHarris777]

With the exception of the mA meter (3v Icc), the electronics of the Flash-box 1.0 is complete, and I have successfully updated my WP34s with calc_xtal_full.bin (albeit a back-level version). The last addition to the Manhattan board are the voltage monitors; amp (4) monitors the 3v line source and raises a red LED indicator warning should the voltage drop below 2.8 volts, while amp (3) monitors the voltage of the J32 reset pin on the wp34s-side of the J32 reset-ground-shot-diode. This is perhaps the real beauty of this particular project; because the wp34s gives no indication that it is 'on' and running during the 'erase' or 'update' process amp (3) monitors the reset line which goes 'high' when the calculator is 'on'. The voltage monitors are based on the following schematic:



Each of the four comparators works the same way (high gain op amp without feedback resistor, and open collector) in that the Q out transistor turns on (allowing LED to glow) should the non inverting input (+) drop below the inverting input (-). In the case of the 3v monitor the reference voltage is placed on (-) and if the line voltage drops below 2.8 volts the red LED glows. On the other hand, the reference is placed on (+) on amp (3). If the reset J32 pin raises above the reference then the wp34s is 'on' and the green power diode glows.

... one last look before we button it up ...



The following pic shows the completed Flash-box in the update mode with the power applied and the WP34s 'ON'. The only design plan not yet implemented is the mA meter (which is on the way... snail mail... )



The yellow LEDs in the upper right monitor the Rx Tx signals (dim rapid flicker, better than nothing) and the red LED in the lower right (normally off) signals a problem with the 3v line. There are actually two warnings that will occur should the 3v line have a problem... because the reset pin J32 will NOT raise high enough for the LM339 to turn on the LED unless the 3v line is above 2.8 volts (so we should have plenty of warning before beginning the flash).

During the power-on, erase, reset sequence the three green LEDs makes it very easy to remember where one is in the sequence (which otherwise is somewhat confusing).

Voila. ... and of course,
Cheers,

[rprosperi]

(02-05-2015 12:26 PM)MarkHaysHarris777 Wrote:  With the exception of the mA meter (3v Icc), the electronics of the Flash-box 1.0 is complete...

Cool photo-journal, thanks for taking us thru the design/build process Marcus. It's clear this is the most thoroughly instrumented flashing "cable" to date.

Seems bold to label it "1.0" with only 1 decimal place. I would have guessed something like 1.0.002.07.d, but I suppose that would require a much smaller font.

[MarkHaysHarris777]

(02-05-2015 02:11 PM)rprosperi Wrote:  Seems bold to label it "1.0" with only 1 decimal place. I would have guessed something like 1.0.002.07.d, but I suppose that would require a much smaller font.

Well, unless someone else builds one (there will only be one). Often my projects are cigar-box prototypes which never see the light of another work-bench nor make it to market. Like most of the software I write, I'm my own best customer... IBM taught me that... <ouch>

================
Ok, here are some operational notes some of you might be interested in:

1) I flashed twice today; once to see how things would work, and the second time was to update the unit to the latest realbuild of calc_xtal_full.bin/ the voltage monitors saw no problems, not even a flicker.

2) Something I was suspicious of, but now confirmed-- there is no point in monitoring the Rx traffic because there isn't any... nothing appreciable that is. During the flash operation the Tx light is quite active but the Rx light is silent (I know it works correctly because it flickers like crazy during loopback testing (with the loopback jumper in place). So there isn't much (if any) handshaking going on, and there isn't anything like making sure the comm happened... it either works or it doesn't. On the other hand, I did notice that I'm testing the wrong side of the cross-over resistor (on Rx) and that might be making a difference. I'll move it before the next test and let you know.

3) The update process is very smooth, no hiccups and the restore worked fine too.

4) The realtime clock does get reset (date and time) to defaults, so of course, I had to reset the date and time.

5) Well, that's it for a while (until I get another unit to play with). I am a firm believer in, "if it ain't broke don't fix it..." So, I won't be playing with 'flash' and this unit (my only one for now) because I don't want to brick it.


Cheers,

[MarkHaysHarris777]

(02-05-2015 02:11 PM)rprosperi Wrote:  ... It's clear this is the most thoroughly instrumented flashing "cable" to date.

Seems bold to label it "1.0" with only 1 decimal place. I would have guessed something like 1.0.002.07.d, but I suppose that would require a much smaller font.

Well, not quite. Turns out my voltage line monitor circuit has a little small snag that needs fixing... if the line voltage to the 34s falls below 2.8 volts (which can adversely affect the flash) for even a short time (milli - micro seconds) there might be a problem with the integrity of the flash. The voltage monitor needs a latch; consider the following modified schematic of the 3v line monitor:



The latch is used on 'output' this time, vs input; but its the same circuit as the reset and erase debouncers:

... if the 3v line voltage faults (below 2.8v reference) for even a short time the LM339 provides a ground shot to the pull-up input of the bistable latch. The latch toggles (and holds) lighting the red diode until the reset switch is pressed again. We will not know 'when' the fault event occurred necessarily, but we will not 'miss' the event even if very brief.

Just to be complete.

Cheers

[MarkHaysHarris777]

I received a copy of the 20b schematic today (thanks, Barry!) via email and discovered that an optional ADM3202AR I|O driver chip is available which would provide very nice optional buffering and ESD protection for the Tx Rx pins on the micro-controller (worth looking into). The chip would have to be surface mounted, and the J5 J6 zero-ohm jumpers would need to be removed. Might be fun.

I'm wondering if Marcus controls the RS232ON input to the Q2 BC856A transistor?

Anyway, I also noticed that TestPoint (TP11) is connected directly to the power switch; which means that if a usb-like connector brought the J3x pins out-bound along with the TP11 pin, the flashing|updating could be handled completely under automation, because the automated controller would be able to turn the calc on with a ground shot (ground TP11), similar to turning it 'off' by giving a ground shot to J32.

So, FlashBox v2.0 is in design consideration and will have all of the instrumentation of FB1 (including mA meter and voltage latches) but will also be entirely self contained (on-board battery) and fully integrated and automated. The flashing|updating will be under control of a Raspberry PI (from inside the FlashBox) and the flash image will be selectable by plugging in a custom thumb drive.

Coming sonn, stay tuned. FlashBox v2.0 will be portable, self-contained, and fully automated--- plug in the hardware (20b - 30b) and press GO...

Cheers,
marcus

[Marcus von Cube]

(02-20-2015 11:02 PM)MarkHaysHarris777 Wrote:  I'm wondering if Marcus controls the RS232ON input to the Q2 BC856A transistor?

I don't think so (but may not remember correctly). The problem with adding the RS232 level converter is that the original cable no longer works (it will probably cause harm). And who has an RS232 interface with that crude connector anyway? I think the provision for such a chip was for a reuse of the platform as an automation controller with custom firmware. I haven't heard of any implementation in this regard.

[MarkHaysHarris777]

(02-20-2015 11:37 PM)Marcus von Cube Wrote:  

(02-20-2015 11:02 PM)MarkHaysHarris777 Wrote:  I'm wondering if Marcus controls the RS232ON input to the Q2 BC856A transistor?

I don't think so (but may not remember correctly). The problem with adding the RS232 level converter is that the original cable no longer works (it will probably cause harm). And who has an RS232 interface with that crude connector anyway? I think the provision for such a chip was for a reuse of the platform as an automation controller with custom firmware. I haven't heard of any implementation in this regard.

Thank you, Marcus. Well, it finally dawned on me (sorry, I'm slower than the rest of you) this little micro controller could be 'repurposed' to|for other broad purposes besides being a calculator- duh. Like that line from Star Wars when Solo gets fed-up with 3-PO (3-PO is afraid he'll be shut down) and Solo says, "Nooo, I need you to tell me what's wrong with the Falcon... take the professor 'back' and plug him into the hyper-drive..." I got to realizing that with some soldering and a little programming I could use this thing to read, adj, and set the fuel injectors | timing on my Chevy Silverado... I call 'her' the Falcon too... Chevrolet has that stupid thing set to guzzle gasoline (if you lean it out a bit and adj the timing you can improve its mileage, reduce its pinging under load, and have better pick-up power... ) the trouble is that you have to adj it under power using a console that is proprietary to GM. Um, maybe not anymore. (I've been thinking along these lines for a while; Rasb PI, Arduino, but... this little puppy may be the ticket... maybe.

<sorry, I digress...>


Cheers,
marcus

[Dave Frederickson]

(02-20-2015 11:02 PM)MarkHaysHarris777 Wrote:  I received a copy of the 20b schematic today

The entire 20b Dev Kit, which includes the schematic, is available from Eric's site. If you ask around you can also find the original 30b firmware.

[MarkHaysHarris777]

(02-21-2015 08:39 PM)Dave Frederickson Wrote:  

(02-20-2015 11:02 PM)MarkHaysHarris777 Wrote:  I received a copy of the 20b schematic today

The entire 20b Dev Kit, which includes the schematic, is available from Eric's site. If you ask around you can also find the original 30b firmware.

Thank you, Dave. I'm wondering if those files could also be made available on Source Forge (or this site) for an easier find? Thanks, again.

marcus

[MarkHaysHarris777]

I have rebranded my prototype "WP34s Update Station" and have updated the main_board and schematics to reflect the final values of the LED current limiting resistors, component layout of the LM339N support components, and the on-line schematics (previous schematics are now updated). This is an above photo of the main board prior to voltage latch update, but reflecting final resistor values and placement of components (voltage latch update coming soon).



The fixed voltage reference for the Tx Rx monitor is in the lower right, and the variable voltage reference for the 3.3v fault circuit is in the upper right.

Cheers,
marcus

[MarkHaysHarris777]

WP34s Update Station

I finally received my mA meter today; however, as you can see instead of being a 20 mA meter at full swing it is a 50 μA meter at full tilt. ~rats. I don't mind making a nichrome inductive shunt for this thing and recalibrating it with labels, but hey, No. I'm gonna get the 20 mA meter I ordered.

At any rate, decided to post a pic as this thing develops (changed its name, and now there is an idea of what it will look like when fully instrumented. I will be adding the voltage latch this week, as well a 'test' switch (red dome moment button mounted next to the black buttons) used for checking the Vcc and voltage latch prior to plugging in the 20b-30b hardware.



PS. I do like this meter... its contrast is sharper (black bottom) while the meter I ordered is white all around... I'm going to have to wait and see--- which meter like better?

Cheers,
marcus

[matthiaspaul]

(02-20-2015 11:37 PM)Marcus von Cube Wrote:  

(02-20-2015 11:02 PM)MarkHaysHarris777 Wrote:  I'm wondering if Marcus controls the RS232ON input to the Q2 BC856A transistor?

I don't think so (but may not remember correctly).

I haven't looked this up in the sources, but according to one of Harald's photos showing how he connected the IR LED,

http://www.hpmuseum.org/cgi-sys/cgiwrap/...186#223186

you are using PC7 (that is, the port pin driving the "RS232ON" signal) to control the IR LED. Fortunately, PC7 / pin 94 is one of the few of the microcontroller's "high-power" pins able to source up to 4 mA - most of the other pins available can source only a fraction of it and would not be able to reliably drive a LED in the first place.

Still, in my own experience with other designs, it is better to avoid drawing much current from port pins in designs with weak power planes like this coin-cell powered device (and in the case of units with USB board even over long thin wires).

I think it would be better to use a transistor to drive the IR LED. This, combined with a small resistor (perhaps 10R) and a 10uF low-leakage & low-ESR capacitor to decouple the IR LED's anode from the VBAT plane, would most probably fix those reset problems some have observed when trying to print with a somewhat exhausted battery.

See also:

http://www.hpmuseum.org/forum/thread-2214.html
http://www.hpmuseum.org/forum/thread-1773.html

Alternatively, trying to support the VBAT plane by adding a 10uF low-leakage & low-ESR capacitor close to the CPU in parallel to C15 or C16 might help as well.

Quote:The problem with adding the RS232 level converter is that the original cable no longer works (it will probably cause harm).

Definitely.

Quote:And who has an RS232 interface with that crude connector anyway? I think the provision for such a chip was for a reuse of the platform as an automation controller with custom firmware.

I also don't think it would be a good idea to retrofit the ADM3202AR for as long as the calculator remains powered by two coin-cells only. The level translator should better be powered by the other side, either explicitly as through USB, or parasitically through RS232 hardware handshake lines.

Greetings,

Matthias

[MarkHaysHarris777]

Greetings,
Well, the first version of my WP34s Update Station (non automated cigar box prototype) is complete, functional, and fully instrumented including voltage latch (3.3v Vcc Fault), simulated current|voltage load test|latch, and mA meter. As you can see, the smaller white scale won out over the larger two-tone. I like this meter too, and I did not have to modify it! I am experimenting with an idea from Dave F. to see if I can get my photos formatted a little better; we'll see how it goes.



The following two pics are from an actual 'update' of my experimental model 30b. I have one unit that has an exceedingly 'bad' keyboard... so I've relegated it to the experimental pile (where I shall play with it ad nausea or until I brick it, which-ever comes first)! Ok, pics first, comment second:



The first file photo is the unit in standby waiting for me to press MySamba's go button... I have followed the 'update' procedure ON-D for debug mode, ON-S followed by S confirm to prepare the boot loader bit; notice the current - it is high for this unit, but not the 20 mA I was expecting... in the entire process the current never went above 10 mA (I'm suspicious its running in 'SLOW' frequency, but don't know for sure). The second pic is the actual 'update' happening from MySamba via the flash-box - WP34s Update Station. Again notice the current -- throughout the entire update Icc hovered around 6 mA, the Tx light flickered rapidly and the voltage monitor|latch did not trip / whoohoo! I had already run the Vcc current|load test (at 20 mA dummy load latch engaged) and the voltage latch did not trip during that test. This means that the FTDI USB to Serial converter 3.3v regulator is quite robust and has absolutely no problem with the update flash. Y'all will be encouraged and relieved to learn that the little beasty woke up, restored my latest backup, and is still ticking.


I have decided to place the automated version in a plastic box... I think its the first time ever that I have actually mounted a project in a 'project box'. A Raspberry PI B+ is going to handle the automation, and an Arduino Uno is going to handle the instrumentation. I'll keep y'all posted on that progress. In the mean time, my 43s prototype is underway and progressing nicely.

PS I'll post schematics later today.

PPS Seems to me that I set that machine in 'SLOW' mode some time ago. I wonder, does the machine stay at 'SLOW' mode if set explicitly (even for the update process?).



Cheers,
marcus

[MarkHaysHarris777]

Greetings,
Well, this is my last formal post to 'this' photo-journal; thank you all for your interest, it has been fun and a real pleasure. I am going to post two last hi res pics attempting to use the tip|(url trick) that Dave F is desperately trying to teach me-- trying to get the primary photos in a small space for clarity and space management, yet provide the hi-res detail behind that... again, first the pics (if you click the pics you should get the hi-res image), and then the comment:



I doodled this 'napkin sketch' while studying with my daughter on her campus this past weekend. She is studying physics this term (specifically resistors, caps, and inductors) and she asked how I know where I want a circuit to go, as it were (so we talked about it). There is an amazing amount of planning that goes into a Manhattan style bread-board.

Some of you are no doubt wondering why the transistor? "Why didn't the guy just drive the simulated current load directly with the inverter--- the one driving the base of Q1? All of my projects contain at least ONE 2N2222A NPN transistor; its my small lasting memorial to Bell Labs and the invention that makes ALL of my hobby (and profession) possible. Thank you Bell Labs.

For me the building of a Manhattan style bread-board (loved by all amateur radio 'makers' and other electrical engineers) is 'primarily' about sculpture. i build bread-boards for the same reason that my wife crochets, or my daughter needle-points; its art, and its creative release (only secondarily is it important that it 'functions'). The idea that we have some Platonic ideal that we are striving for and through our crude finger tips and a ghastly soldering iron out comes an Aristotelian representation of that ideal reality (sometimes it works, and sometimes it doesn't, but its art none the less).

This quad Nand gate provides the logic for the voltage fault latch, and for the simulated load latch (four gates, two latches). Both latches are reset by a line that runs directly from the reset switch to the pull-up resistors of each latch, which toggles the latches off. The voltage fault latch is toggled 'ON' by the voltage monitor (LM339N comparator). The simulated load latch is toggled 'ON' by the 'Vcc Test' switch seen in this photo left bottom.

Again, thanks for your interest and time.

PS (thank you Dave, if this photo thing doesn't work, believe me I gave it the old college try!)

PPS The board layout in the 'napkin sketch' showing the latch driving the Q1 transistor directly does not work; the base-emitter current grounds out the latch preventing it from actually latching--duh. The actual device has an inverter 'buffer' (as shown in the schematic) wired to drive the Q1 base. :-}

PPPS Well, I kept this whole project under five(5) chips and under twenty(20) gates. Generally, if a project is five or more chips and 20 or more gates its time definitely to consider a micro-controller instead. But, again, if the goal is spending time with your assistant building a functional sculpture, well forget the micro-controller and go for a Manhattan style bread-board!

Cheers,
marcus

[MarkHaysHarris777]

I've begun the work of reducing the layout size of my pics on this photo journal, to save scrolling space, and to save space in the cloud (thanks Dave F. for the suggestion, it was a really good idea).

marcus