HP Forums

I got a cheapo 11c which the user thought didn't work. It actually had life, the right side of the display had random segments if you pressed enough keys.

Took it apart, it has a single PCB, blue surface corrosion was most everywhere. Vinegar dip, dry, etc. and now the all segments of the right 5 positions work but the leftmost 5 & mantissa - sign are still blank. Pressing all keys show the entire keyboard does work (divide by 10^x a lot). calculations produce correct answers, so it appears only the - sign & 1st 5 digits are permanently off.

interestingly, when the batteries are out, positioning the display & the pink rubbery cables, random segments in all positions (including 1st 5) MIGHT flash. (static? charge left in the capacitor)

Has this happened to anyone? there is no obivious pitting of the PCB tracings, the corrosion was directly on the top of the gold colored tracings. viewing from the back of the PCB there is a large chip to the left, a position where it looks like a surface mount chip could be and to the right a smaller chip with the same footprint (pins & size) as the position in the middle. What's interesting, is the middle position does not look like a chip "fell off". All the "pads" where a chip could be have been contain solder & it appears all (or most, would have to look again) shunt one tracing to another. i.e. as if a chip should not be there. All the solder is nice & rounded as opposed to looking like a chip fell off somehow.

The rubbery floppy pink cables (zebra?) on the top/bottom edges of the display look OK & do have continuity at several randomly tested places when testing with an ohm meter.

should further testing be done or is this a "common" symptom? seen several posts about swapping displays with other units (i.e. cheap 12C's) but the discriptions don't match my exact symptoms.

thanks

Edited: 17 Aug 2012, 5:38 a.m.

Quote:
interestingly, when the batteries are out, positioning the display & the pink rubbery cables, random segments in all positions (including 1st 5) MIGHT flash. (static? charge left in the capacitor)

Residual voltage in the circuit and/or you is enough to
polarize otherwise unterminated segments causing such behaviour. Although it would have been useful to buzz out the interconnect
before removing the compression frame. Reliably reseating
the zebra strips isn't the most straightforward task.

Quote:
there is no obivious pitting of the PCB tracings, the corrosion was directly on the top of the gold colored tracings.

If you still have the compression frame released, it would
be useful to buzz out the continuity directly
from the LCD zebra pads
to the display controller to assure
no trace breaks exist.
Use a pair of pins and a low open voltage continuity tester.
You can also try flipping/rotating/swapping
the zebras as well. The elastomer can certainly take a permanent
set after a few decades and fail.

Quote:
..viewing from the back of the PCB there is a large chip to the left, a position where it looks like a surface mount chip could be and to the right a smaller chip with the same footprint (pins & size) as the position in the middle.

Mid production +/- voyagers use a common pcb for the series
which is depopulated for 11/12/16s, and stuffed with a second
ROM/RAM in the case of a 15c.

Quote:
should further testing be done or is this a "common" symptom? seen several posts about swapping displays with other units (i.e. cheap 12C's) but the discriptions don't match my exact symptoms.

This isn't common in my experience and I doubt your display
glass is defective. I'd suspect an break in a pcb trace or
lifted pad on the display controller (left, larger package).
Although quite rare,
you could have a defective display controller as
the right and left digits are controlled by two (independent
AFAICT) registers, so there is the possibility for independent
failure.

Sounds like the backplane LCD signal is missing for the left side.

Measure the voltage at pin 16 of the larger IC with the unit on. It should be approximately 1 volt ac at about 100 hz. If the signal is good, the trace is most likely open to the left most top and bottom connections of the LCD (as seen from the front). The failure is most likely under the IC and you'll have lift the IC off to find the open circuit.

If the signal is present out at the LCD connections, the LCD is bad.

If there is no voltage, the IC is bad and it is game over...

Thanks Umgawa/Randy.

Checked the items you mentioned.

The zebra cables have been interchanged, flipped end to end and rotated thereby exchanging which side hits the PCB. The display was tested after each of the 3 movements. No difference.

Looking down the display from each end and noting landmarks on the PCB & also carefully lifting the display shows the black portion of the cables probably are hitting the pads and sitting square to the PCB. Also there are water marks in the gold pads where the black part of the cable hits. Those marks correspond with the visual examintation. The top cable hits the pad almost in its entirety. The bottom cable is about 1/2 the black part below the pads, but don't appear to hit anything else. This visual appearance appears consistent across the entire length of the display (top & bottom).

Measured the voltage on the pin you mentioned. Presumed pin 1 was by the dot on the IC package and went counter clockwise. The chip has two groups of 12 pins along the bottom edge. Pin 16 has +1.2V. Turning the unit over the leftmost pads top & bottom join together and lead to a feed-through hole in the board. The pads & hole all have the same voltage. Removing batteries & checking continuity between the hole and pin 16 shows continuity (about .1-.2 ohm).

I did notice on the chip that pins 22 & 23 appear soldered together, I have no idea where they lead.

I have done the press & hold power, X (multiply), release power, release X and it goes through the self-test. You can see "running" (all segments) if you press on the retaining bracket evenly. The test pattern results in most segments lit up correctly. Yet a backspace and entering ten 8s show only the right 5 digits. (no matter how you press on the bracket)

??? is there something different that lights up the display during the self-test as opposed to the normal display of digits?

A consistently repeatable pattern is produced. The 1st digit shows the bottom and middle segments along with the upper right. I'll denote it with a V. The 6th digit is like a capital G (all segments but middle & upper right). Positions 2-5 show as a capital A (all segments but the bottom one).

test results:
-V,A A,A A G,8,8,8,8,

the annunciators are Grad D.MY C Prgm.

??? even though the self-test shows G in position 6, that digit lights up as 8 if you enter enough 8s

Just curious what this means.

thanks

Edited: 18 Aug 2012, 4:57 a.m.

Sounds like a zebra/LCD issue...

Quote:
I did notice on the chip that pins 22 & 23 appear soldered together, I have no idea where they lead.

Some of that board layout is rather interesting relative to
today's SMT PCB conventions (pads over vias, undersize power
rails, gang shorted pads, etc..) but this was 1983 or so.
The missing pins on the otherwise lqfp-44 NUT package almost
look to have been decided concurrent with ease of PCB routing
which is at least a little unconventional.

Regrettably I've been unsuccessful thus far attempting to
scrounge up any sort of a data sheet for an R2D2. If I ever
found such I'd be interested to draw up a reference schematic
for the legacy voyagers similar to the sam7 ARM version I
done a few months ago. There is also the board routing
difference between the display controller and glass in the
10c units relative to other family members which impacts
the firmware segment mapping and would be nice to capture
concretely.

Quote:
??? is there something different that lights up the display during the self-test as opposed to the normal display of digits?

No difference. Intermittents can be deceptive and I've seen
similar behaviour which appeared to defy logic.

So have you verified with all segments active (self-test completion)
you are seeing a square wave on all driven zebra pads (except for
the low batt annunciator)?

If you are convinced all zebra pads have a solid copper path to
the display controller,
something else I'd try would be to remove the zebras from the glass
and inspect that mating surface on the zebras as well as the
condition of the ITO glass traces. A gentle wipe down of the ITO with isopropyl would be another. Failing any resolution I'd
probably do a wholesale mutual LCD swap (glass, zebras, frame)
from a 12c and see what that yields for both the 11c and 12c.

[edited typos]

Edited: 19 Aug 2012, 3:08 p.m. after one or more responses were posted

I appreciate yours and Randy's comments. Its especially good they tend to lean towards the same place; display & connections. Possibly there's hope yet. Frankly I paid less than $10 for it, so whether I ever get it to work isn't that important. But I enjoy fiddling with stuff!

Square wave? I know what that means, but the most sophisticated instruments in my repertoire, other than my fingers & screwdrivers, is a multimeter. I have taken basic introductory electroncis in high school & college which was fundamental electricity/discrete component/tube stuff. Had one semester which was IC related being a 5 mile fly over of functions of various different chip types. i.e. not the architecture of the internals, but the functions of the pins on the chip. i.e. power, address/data lines, select, enable, reset, R/W, clock, etc. So, conceptually I have some understanding of what you're saying. I can follow a description of a schematic but trying to explain one myself likely would fail.

I enjoy your comments, as you can't learn unless you realize when you hear information & try.

Ya schematics of the PCB or the tracings for the segments in the LCD itself would be nice. I'll get out my magnifying glass & try to figure out what I can on the board.

thanks!

Quote:
Square wave? I know what that means, but the most sophisticated instruments in my repertoire, other than my fingers & screwdrivers, is a multimeter.

It kinda multiplies your $10 investment, but you really ought to get one of these or its bretheren (there are several brands which look virtually identical) for around $300. I've got the 60 MHz version of the Owon, and it is tons of fun. My only regret is I didn't get a 4-channel model.

Quote:
Square wave? I know what that means, but the most sophisticated instruments in my repertoire, other than my fingers & screwdrivers, is a multimeter.

With all segments active, all of the lcd drive lines should have
a 0/2.25/4.5 volt stepped
waveform. But just verifying some activity exists on the pin
will help rule out a bad driver in the display controller.

If you're interested in adding to the instrumentation war chest on the cheap, a rummage around evilbay will often turn up older (HP, Tektronix) analog scopes listed at spectacularly cheap prices.

Quote:
spectacularly cheap prices

True! That's what I got first - a (20 year old!) front line Tektronix for $100. If you are interested in that, send me an email.

Having only a multimeter I measured the voltage on all pads during the all segments self-test display.

BATT = 4.22V
on the front from left (i.e. normal usage view)
top pads:
leftmost = 2.11v
pad 2 = 1.74v
pad 3 = .57v
all remaining pads 2.11v
bottom pads: all 2.11v

internesting, when the calc is off:
top:
#1 = 1.4v
#2 = .79v
#3 = .25v
all rest = 1.4v
bottom:
#1 = 1.4v
#2 = 1.27v
#3-11 = 1.4v
#12-19 = 1.27
#20 (rightmost) = 1.21v

personally I was assuming there would be zero volts when the unit was off.

uhmgawa: "So have you verified with all segments active (self-test completion) you are seeing a square wave on all driven zebra pads (except for the low batt annunciator)?"

the question struck me as interesting, so I measured all pads. knowing there wouldn't be a way to determine wave shape I figured there would be a number of pins of +voltage & various pins of -voltage or ground.

The reason I thought so was the big chip doesn't have enough pins to provide a pin per segment and a common anode/cathode per digit position. So figured, with the two registers you mentioned, each 1/2 of the display was multiplexed & presumed 1/2 the digits had a "common bus array" per segment with a sequenced anode/cathode to light up each digit.

Assumed this per segment array would always be on & a series of per digit common opposite "pole" would blink on fast enough it'd just look ON all the time. i.e. all pads would likely have the same voltage during the self-test display.

So I measured all the pads after turning the unit on with a display of 0.00. Unfortunately I see in my notes the results were not written down, but as I recall, the results were the same or at least similar in respect all pads (other than pad 2/3 on top) had the same voltage. My assumption was after the 0's there would be a periodic sequence of every "n" pads to find one -voltage or at ground; i.e. the common annode/cathode shutting the digit off.

Is a multi-meter not sufficient to examine the situation? Is there a common cathode/annode per position? Is the waveform shifted (assume digits are multiplexed per controller) to produce a voltage potential path when needed to light a segment? Makes me curious since you specifically asked about the waveform as opposed to is there a voltage there.

I'm scouting around for a cheap-o 12c with the same display setup.

I apologize if this all is in a previous thread somewhere.

thanks

Quote:
Having only a multimeter I measured the voltage on all pads during the all segments self-test display.

Measuring via a multimeter won't reveal the AC characteristics
of the multiplex waveform. This vintage of voyager lcd controller
uses a three level bias of 0, Vbatt/2, and Vbatt to drive
the row/columns depending on active segments.

Quote:
personally I was assuming there would be zero volts when the unit was off.

It has been a while since I scoped the controller lcd output,
but I'm relatively certain
the driver outputs were low upon the controller
sniffing a DISOFF instruction. Each segment needs to have a
0V DC potential imposed upon it over time
and grounding of all drive
lines is typically used to assure this during blanking.
What type of multimeter are you using?

Quote:
..the big chip doesn't have enough pins to provide a pin per segment and a common anode/cathode per digit position.

The controller uses a multiplexed drive scheme
encoding three voltages
onto each line, which reduces the pin count required.

Quote:
So figured, with the two registers you mentioned, each 1/2 of the display was multiplexed & presumed 1/2 the digits had a "common bus array"

I'm really not sure what the logical mapping of the multiplex
scheme is. But it should be reverse engineer-able with a dual
trace scope for the motivated technical historian.
Note my mention
of the two internal display registers was only a stab in the dark
in the event a defective controller was a fault.

Quote:
Is a multi-meter not sufficient to examine the situation? Is there a common cathode/annode per position? Is the waveform shifted (assume digits are multiplexed per controller) to produce a voltage potential path when needed to light a segment? Makes me curious since you specifically asked about the waveform as opposed to is there a voltage there.

A multimeter doesn't provide much useful information.
Sometimes there just isn't any viable substitute for an
oscilloscope. No home should be without one.

Concerning the lcd drive, it is substantially more complex
than an led drive due to the requirement for all segments
irrespective of whether enabled, to see a 0V DC potential
over time. As such a time (phase) and voltage encoding
scheme used to carry as much information as possible in as
few possible conductors. To polarize (enable) an lcd segment
a minimum RMS voltage across its column and row lines
needs to be impressed, above
a specific threshold and within its persistence time frame.
In this scheme
the "off" segments will see some RMS voltage below their
polarization threshold and "on" segments an RMS voltage above
this threshold. The ratio of RMS off to RMS on
voltage roughly sets the contrast of the display. Also the
time multiplexing needs to occur within the polarization
persistence of the display or contrast will suffer and
flickering will become visible. It is an interesting approach
and frankly I'm impressed it works as well as it does
in practice.

Quote:
I'm scouting around for a cheap-o 12c with the same display setup.

I believe any S/N under 28xx should have an identical display.

I have a basic ACE Hardware type multi-meter. Took a peek and its a Gardner Bender GDT-311. I don't think there's anything special about it.

When mentioning 0V when the unit was off, I meant the on/off state of the calculator, not the digit being off. i.e. questioning why any voltage would be present when the calculator was off. I assume your response describes the calc being on and the segment off.

To describe my interpretation of what you're saying, in a generic fashion--

- each pad will have a squence of pulses (in whatever order they may be) at "low", "medium" and "high", for simplicity. (although I realize with engineering issues imprecision can make the dam break)

- the phasing of the pulses produce the potential difference needed to turn a segment on or a lack of sufficient difference to shut the segment off.

- a multimeter set to DC-V is incapable of discerning a fluctuating voltage versus an "average" of the pulse values (makes sense). Hence, a multimeter will show the pad as always having a voltage even when the segment is always off.

- there were things I don't think I entirely understand, but LCD's have requirements of their own, simply viewing a segment as ON or OFF is not a suitable description. i.e. there are time-based requirements of how long they are on/off (maybe due to multiplexing?) in order for them to function correctly, especially in terms of how your eye views it.

- the 2nd/3rd pads from the left, top row had different voltage than the rest of them. anyone know if that should be expected?

What they really need is super inexpensive electrical engineering classes designed for old people (with not a lot to do) who want to fix and/or understand their stash of old calculators.

Not a practical investment but... after hitting the coffee shop 5 times a day for years..... anything looks good!!!

thank you for your comments!

Edited: 21 Aug 2012, 1:40 a.m.

Quote:
When mentioning 0V when the unit was off, I meant the on/off state of the calculator, not the digit being off. i.e. questioning why any voltage would be present when the calculator was off. I assume your response describes the calc being on and the segment off.

The voyager design doesn't ever remove power from the R2D2(s)
nor NUT cpu unless the button cells have gone missing or are
depleted. So even when the unit is "off" as perceived by the
user, the internal circuitry remains powered. In this
state the controller is capable of driving the outputs
at a known level.

What does happen when the firmware responds to an "ON" key
power down is
turning off (blank) of the display controller and stopping of
the NUT LC resonator oscillation. This ceases clocking
in the NUT, ISA bus, and within the slave R2D2(s) which
results in the drastic current consumption reduction.

FWIW there is an intermediate power conservation state entered
when waiting for user key input.
Here the firmware allows the display controller to remain active
driving the glass, but the NUT has stopped oscillation as
above. The R2D2 however contains a simple RC oscillator
which is used as a timebase for multiplexing. In this state it
also clocks a 10min timer in the controller which upon expiration
kicks the NUT into wake up. Here IIRC the firmware
recognizes the event and blanks the display, returning to
sleep with oscillator disabled.

Quote:
To describe my interpretation..

That's essentially it. For more concrete examples
have a look at most any simple monochrome (not grayscale,
nor color) lcd controller data sheet (eg: st7565).

Duane Hess

uhmgawa

Randy

Duane Hess

Randy

uhmgawa

Duane Hess

Dave Shaffer (Arizona)

uhmgawa

Dave Shaffer (Arizona)

Duane Hess

uhmgawa

Duane Hess

uhmgawa