Design files for 50g serial cable - Hewlett Packard

This is a discussion on Design files for 50g serial cable - Hewlett Packard ; In case anyone is interested, I have posted the design files (source code and pcb/gerber files) for the Allen & Eric 50g serial cable (commonly known as the hpcalc.org cable or simply Eric's cable) at http://allenwan.com/hpcalcserialcable/ A couple notes/warnings: -the ...

+ Reply to Thread
Results 1 to 9 of 9

Thread: Design files for 50g serial cable

  1. Design files for 50g serial cable

    In case anyone is interested, I have posted the design files (source
    code and pcb/gerber files) for the Allen & Eric 50g serial cable
    (commonly known as the hpcalc.org cable or simply Eric's cable) at
    http://allenwan.com/hpcalcserialcable/

    A couple notes/warnings:
    -the files provided are for the current hardware and software revisions
    which are HW:v3 and SW:ver.20080217

    -for the pcb/gerber files, we didn't actually use the silk screen or
    solder mask layers. I don't think there's any harm in using them, but I
    recommend double checking clearances.

    -for the source code, the HPCalcSerial.c contains all of the code unique
    to this project. It requires a few files/libraries from AVR-libc. The
    code was compiled with avr-gcc. The fuse settings used for the ATtiny24 are
    0xFF;0xD8;0x42
    extended;high;low

    -Allen Wan :-)

  2. Re: Design files for 50g serial cable

    Allen Wan kirjutas:
    > In case anyone is interested, I have posted the design files (source
    > code and pcb/gerber files) for the Allen & Eric 50g serial cable
    > (commonly known as the hpcalc.org cable or simply Eric's cable) at
    > http://allenwan.com/hpcalcserialcable/
    >
    > A couple notes/warnings:
    > -the files provided are for the current hardware and software revisions
    > which are HW:v3 and SW:ver.20080217
    >
    > -for the pcb/gerber files, we didn't actually use the silk screen or
    > solder mask layers. I don't think there's any harm in using them, but I
    > recommend double checking clearances.
    >
    > -for the source code, the HPCalcSerial.c contains all of the code unique
    > to this project. It requires a few files/libraries from AVR-libc. The
    > code was compiled with avr-gcc. The fuse settings used for the ATtiny24 are
    > 0xFF;0xD8;0x42
    > extended;high;low
    >
    > -Allen Wan :-)


    May I put my ignorance on display? What happens when omitting, between
    the two hp50g, the converter circuitry and using only copper
    (ground,RX,SX)? Is there a limitation on cable length, possible damage
    to controller or what?

    What is the difference between the SW:ver.20071007 and SW:ver.20080217.
    Is it only code regarding the autoshutdown mode switch?

    Robert Tiismus

  3. Re: Design files for 50g serial cable

    > May I put my ignorance on display? What happens when omitting, between
    > the two hp50g, the converter circuitry and using only copper
    > (ground,RX,SX)? Is there a limitation on cable length, possible damage
    > to controller or what?


    There are a couple of issues with that. First, I'm not aware of any
    cable that does that (i.e. a cross over cable for mini 4pin usb). The
    one we use, #150122 from Bestlink Netware would cause

    Vcc <--> Vcc
    TxD <--> TxD
    RxD <--> RxD
    GND <--> GND

    instead of

    Vcc <--> Vcc
    TxD <--> RxD
    RxD <--> TxD
    GND <--> GND

    Also one calculator's battery would attempt to charge the others. The
    batteries are literally directly connected to the Vcc lines on the mini
    4 pin USB socket. There's nothing in between so you actually get 6v not
    5v. But if you're going to build the cable yourself, you can easily fix
    that by building it without the power lines connected.

    Vcc XXXX Vcc
    TxD <--> RxD
    RxD <--> TxD
    GND <--> GND

    The remaining issue is that the 50g will drive RxD when the calculator
    is turned off or when the serial port is closed which could fry the
    calculators. You can easily work around this by not plugging in your
    cable until both calculators are turned on AND have their serial ports
    turned on (openio). And unplugging the cable before either closing IO or
    turning the calculators off. Also it'd be a good idea to put resistors
    on the data lines in case you forget.

    Vcc XXXX Vcc
    TxD <1k> RxD
    RxD <1k> TxD
    GND <--> GND
    where 1k is a 1k ohm resistor.

    This may all seem like a lot of trouble to go to but I can see the
    motivation given that the other available option runs around fifty
    dollars. The one hick-up would be getting the 4 pin mini usb cable which
    I'd probably be willing to sell you for five dollars shipped (in the U.S.).

    > What is the difference between the SW:ver.20071007 and SW:ver.20080217.
    > Is it only code regarding the autoshutdown mode switch?


    Yes, but it also coincided with us depopulating the switch that
    controlled the feature (there use to be a small dip switch inside the
    shell) and the software change made it so that it's on by default
    instead of off.

    -Allen Wan

  4. Re: Design files for 50g serial cable

    On Aug 22, 11:01*am, Allen Wan wrote:


    >The
    > batteries are literally directly connected to the Vcc lines on the mini
    > 4 pin USB socket. There's nothing in between so you actually get 6v not
    > 5v.


    Let me think. If you have alkaline batteries they have a nominal
    voltage of 1.5v.
    4 * 1.5 volts is 6 volts. But the nominal voltage for alkaline really
    only applies under no load or minimal load.
    With a reasonable load after having been used for a while they tend to
    put out closer to 1.25 v. 4 * 1.25 = 5 volts.
    You would only get closer to 6 volts if the calculator was off.

    For the attempting to charge each other reasons though there
    definitely could be issues.


    >But if you're going to build the cable yourself, you can easily fix
    > that by building it without the power lines connected.


    It is advisable to leave GND connected. (your diagrams show this, but
    you mentioned "power lines" plural, so clarifying is good. This
    ensures both calculators are measuring the data lines voltages from
    the same reference level.

    After all battery powered devices have no reference to true earth
    ground, and could certainly drift by several volts relative to each
    other.

    > Vcc XXXX Vcc
    > TxD <--> RxD
    > RxD <--> TxD
    > GND <--> GND
    >
    > The remaining issue is that the 50g will drive RxD when the calculator
    > is turned off or when the serial port is closed which could fry the
    > calculators. You can easily work around this by not plugging in your
    > cable until both calculators are turned on AND have their serial ports
    > turned on (openio). And unplugging the cable before either closing IO or
    > turning the calculators off. Also it'd be a good idea to put resistors
    > on the data lines in case you forget.
    >
    > Vcc XXXX Vcc
    > TxD <1k> RxD
    > RxD <1k> TxD
    > GND <--> GND
    > where 1k is a 1k ohm resistor.
    >
    > This may all seem like a lot of trouble to go to but I can see the
    > motivation given that the other available option runs around fifty
    > dollars. The one hick-up would be getting the 4 pin mini usb cable which
    > I'd probably be willing to sell you for five dollars shipped (in the U.S.).
    >
    > > What is the difference between the SW:ver.20071007 and SW:ver.20080217.
    > > Is it only code regarding the autoshutdown mode switch?

    >
    > Yes, but it also coincided with us depopulating the switch that
    > controlled the feature (there use to be a small dip switch inside the
    > shell) and the software change made it so that it's on by default
    > instead of off.
    >
    > -Allen Wan



  5. Re: Design files for 50g serial cable

    "Allen Wan" wrote in message
    news:E6idnYRAJ4pbSTPVnZ2dnUVZ_qHinZ2d@posted.oplin k...
    > The remaining issue is that the 50g will drive RxD when the calculator
    > is turned off or when the serial port is closed which could fry the
    > calculators.


    This seems highly, highly unlikely: Any decently-designed serial interface is
    fully tolerant of having "transmit" data lines completely shorted to ground,
    so I would sure *like* to believe that shorting TxD on an 50g (which is about
    the "worst" you can do to it -- having it look into RxD on a powered-off
    calculator is a little more forgiving) doesn't blow anything up.

    I am aware that in general the 50h's serial port is quite poorly designed, at
    least from the software point of view, though.

    > Also it'd be a good idea to put resistors
    > on the data lines in case you forget.


    This is good advice.

    ---Joel



  6. Re: Design files for 50g serial cable

    On Fri, 22 Aug 2008 17:01:51 -0500:

    > Any decently-designed serial interface is
    > fully tolerant of having "transmit" data lines completely shorted to ground


    Could be, but I'm suddenly reminded of the "serial port bug" of the HP49G,
    where early production runs of that model (including mine) have a serious
    circuit layout error, part of which was that one serial connector pin
    may have been routed straight to the battery,
    though there had been no intention to supply external power
    from the serial port!

    I don't know who designed that hardware,
    but I bet that Dave Arnett was thousands of miles from the scene.

    What's the quality of Kinpo's hardware engineering
    (or did HP do any of it themselves?)

    -[ ]-

  7. Re: Design files for 50g serial cable

    > This seems highly, highly unlikely: Any decently-designed serial interface is
    > fully tolerant of having "transmit" data lines completely shorted to ground,
    > so I would sure *like* to believe that shorting TxD on an 50g (which is about
    > the "worst" you can do to it -- having it look into RxD on a powered-off
    > calculator is a little more forgiving) doesn't blow anything up.


    In all fairness, I didn't say anything would blow up. And I did qualify
    it with "could." And this is why. When a 50g is in serial off (turn the
    calculator on, openIO then closeIO; yes, this results in a different
    state from just turning the calculator on!) the TX line is driven high
    and the RX line is driven low. So the "worst case" scenario you
    mentioned isn't just a hypothetical worst case, it actually is the case.
    The TX and RX lines are connected straight to the processor with no
    inline resistors [1] and I believe it's just a CMOS buffer with no short
    circuit protection.

    This won't necessarily kill the calculator. Eric's 50g appears to work
    just fine after we discovered this the hard way. As a side note, we
    stumbled upon this when we were taking power measurements and discovered
    that our test cable caused the calculator to draw more power when off
    than when the the calculator was on.[2] And so we know that the current
    drain during the short circuit situation is fairly substantial which is
    consistent with our belief that there is no short circuit protection.

    This is not something that you want to do your calculator on purpose and
    should degrade the functionality of the CMOS drivers even if it doesn't
    "fry them".

    -Allen :-)

    [1] Eric heard this from a reliable source but we haven't independently
    opened up a 50g and verified it.

    [2] Eric and I both vaguely recall this being the case but it's been
    over a year and we couldn't find our notes regarding exactly what the
    current drain was.


  8. Re: Design files for 50g serial cable

    "John H Meyers" wrote in message
    newsp.uga64iyknn735j@miu.edu...
    > I don't know who designed that hardware,
    > but I bet that Dave Arnett was thousands of miles from the scene.


    I'm sure you're correct... does he even still work for HP? How about Preston
    Brown?

    > What's the quality of Kinpo's hardware engineering
    > (or did HP do any of it themselves?)


    Given how all the keyboard problems and how someone thought it was a "good
    idea" to drop the switching regulator in the 50g -- saving a few pennies on
    production costs while eating batteries significantly faster -- I'd say
    nowhere near as good as the old HP.



  9. Re: Design files for 50g serial cable

    "Allen Wan" wrote in message
    news:G6CdnSo7bZJACzLVnZ2dnUVZ_q_inZ2d@posted.oplin k...
    > The TX and RX lines are connected straight to the processor with no
    > inline resistors [1]


    That definitely violates my "decently-designed" criteria. :-(

    > and I believe it's just a CMOS buffer with no short
    > circuit protection.


    Shorting a few CMOS buffers on a chip generally isn't harmful -- at least
    short term (seconds or minuets). But certainly not a good idea either.

    A cable like you describe (with resistors) is definitely the smart way to go.

    I apologize if I've caused undue commotion here; I just didn't want people to
    be so fearful of killing their $150 calculator that they became afraid to
    experiment with their own serial port cables.

    ---Joel



+ Reply to Thread