WP34S: Reset when printing statistical registers?

[walter b]

(07-03-2014 11:56 AM)Marcus von Cube Wrote:  Here is a short summary of the possible startup messages:
...
Memory seems to be intact: "Reset"

That's not what you wanted to tell us, is it? When I start my WP 34S and "Memory seems to be intact" it will display the last content of X. It does so consistently.

(07-03-2014 11:56 AM)Marcus von Cube Wrote:  "Reset" is an indication that the processor failed for some reason, be it a programming error (illegal instruction, stack corruption, etc.) or some physical condition (power too low for the selected processor speed).

Hmmh, now we're almost as informed as before, aren't we?

d:-?

[Paul Dale]

(07-03-2014 02:45 PM)walter b Wrote:  That's not what you wanted to tell us, is it? When I start my WP 34S and "Memory seems to be intact" it will display the last content of X. It does so consistently.

Pressing the on button doesn't actually start the device. It was already running. Start up in this context is the CPU's initial boot process.

Yes, the nomenclature is ambiguous

- Pauli

[jebem]

(07-03-2014 10:39 AM)rkf Wrote:  

(07-03-2014 10:30 AM)Paul Dale Wrote:  ... Most likely, we're drawing too much current from the coin cells and their voltage is dropping because of this. I've seen this too from time to time and sometimes the low battery indicator comes on briefly. ...

What a pity, just set ()DLAY to 0 again to provoke the Reset message, and watched closely the LCD panel while doing a print summation registers, trying to glimpse the Low Batt annunciator. Unfortunately, this time neither a Reset shows up, nor a broken I/O character is printed - no matter how often I try.

?!?

Kind regards, Ralf.

Interesting subject. A couple of questions...
- Did you get your WP-34S ready made, or did you install the IR LED yourself?
- How far can the calculator be from the printer and still get a good connection?

IMHO the IR LED current should not be too high when compared with what the calculator processor can consume.
I would say that the HP-30B can consume as much as 20mA in many situations, and the IR LED should only add a few mA more, maybe 2 to 4mA in a proper implementation and using one adequate and efficient IR LED type (however with that current values I would expect the working distance range to be small, like 30cm or about 1 foot at maximum).

But the real IR LED current consumption can be higher, depending on the circuit used to drive it. If the IR LED series current limiter is absent or is of a very low value (below 330 ohm), it can deplete the batteries in shorter time than normal, but you should be able to communicate at larger distances in this case!

From your description, the probable cause seems to be a voltage drop condition from the batteries you are using.
Sometimes the batteries recovers by themselves, yes, but in the end it will happen again soon.
And not all the battery brands are created equal...

I would try a new set of battery cells and see.
I get my batteries from China in the TAS for peanuts and the quality is more than acceptable for the asking price and free delivery.

As a final note, if you enjoy a little diy and have a multimeter, you may use an external 3Volt power supply (two AA batteries in series will do), a couple of cable clips to connect it to your calculator and measure the current consumption when printing to find out more about your calculator current consumption.

[pito]

(07-03-2014 09:29 AM)walter b Wrote:  

(07-03-2014 09:24 AM)pito Wrote:  Do we measure the battery voltage (BATT) under full load??
When not, the BATT does not show the proper numbers..

Do you want to switch the WP 34S to full load just for measuring the voltage by BATT?

d:-?

Of course YES! When you measure the wp34s current in the key_scanning_mode it takes maybe 150uA current, when you switch to let say 30MHz it takes for example 40mA current. Provided the internal resistance of the CR2032 is 10-1000 ohm (based on its capacity consumed, up to 100ohm at 60% used) you may easily calculate the difference between the measurements.. (done for you - see below for a single CR2032 cell, for 2 in parallel half the Rin).
PS: try to measure your car battery after a year of having it resting on the shelf in your garage - you will measure 14V when not loaded happily, but your car will definitely not start

[rkf]

(07-03-2014 11:08 AM)walter b Wrote:  ... Anyway, Ralf, if you can reproduce it once again, please SAVE your data before. I'd be interested in whether "Restored" will show up instead of "Reset" then. ...

Today I tried hard to reproduce it. First I interchanged some times the batteries of my both WP34S. Both have pairs of 3.02V each, with BATT displaying 3.0, but no way to reproduce the discussed behaviour.

Then I took the batteries of my PR Error HP15C LE. Those batteries had both 2.91V, and a BATT value of 2.8 - and voilá! The printer managed to print only the first line, and even did not accomplish printing the broken I/O symbol, because of a shutdown/restart of the WP34S - which ended in "Restored".

I hope this helps, Walter.
Kind regards,
Ralf.

[walter b]

(07-04-2014 04:37 AM)rkf Wrote:  I hope this helps, Walter.

Thank you very much, Ralf.

d:-)

[rkf]

(07-03-2014 10:13 PM)jebem Wrote:  ... A couple of questions...
- Did you get your WP-34S ready made, or did you install the IR LED yourself?
- How far can the calculator be from the printer and still get a good connection?
...

- I had the luck to get one WP-34S ready made
- I tried to measure this as exactly as possible. The maximum working distance I reached was 150 mm (for our U.S. readers: 5 29/32 inches, Happy Independence Day, BTW!)

Edit: Just tested my HP42S with respect to Infrared Printing. The maximum working distance of my 1991 model is 1100 mm, thus a good deal further.

Kind regards, Ralf.

[walter b]

Thank you for the calculations. We shouldn't forget that the WP 34S switches off for voltages <2.1V.

d:-)

[jebem]

(07-04-2014 05:31 AM)rkf Wrote:  

(07-03-2014 10:13 PM)jebem Wrote:  ... A couple of questions...
- Did you get your WP-34S ready made, or did you install the IR LED yourself?
- How far can the calculator be from the printer and still get a good connection?
...

- I had the luck to get one WP-34S ready made
- I tried to measure this as exactly as possible. The maximum working distance I reached was 150 mm (for our U.S. readers: 5 29/32 inches, Happy Independence Day, BTW!)

Edit: Just tested my HP42S with respect to Infrared Printing. The maximum working distance of my 1991 model is 1100 mm, thus a good deal further.

Kind regards, Ralf.

Thanks for your input, Ralf.
Somehow I have similar experience here.

I don't have a printer but I'm using my own IR receiver for testing purposes, and it seems I need to find a more efficient IR LED circuit for my 34S when I have the time and the mood for that.

I have one "as-new-in-the-box" HP-42S got the previous week, and it can reach larger distances than my 34S. But there are several possible reasons for this different behavior to be investigated if and when time and mood allows it.

[pito]

(07-04-2014 07:49 AM)walter b Wrote:  Thank you for the calculations. We shouldn't forget that the WP 34S switches off for voltages <2.1V.
d:-)

I've taken a fresh CR2032 (Vinnic?) from my stock:

Code:

1. Voltage without a load (actually 10Mohm, I = ~332nA) = 3.32V

2. Voltage with 200ohm load = 2.89V

3. Internal resistance of the battery is 

Rint = dV@Rint/dI@Rint = (3.32V - 2.89V)/(2.89V/200ohm - 332nA) = 29.75 ohm

So my above estimation was rather optimistic. Both batteries in parallel will have ~15ohm Rint when fresh. The load of ~15mA will create a ~0.22V drop with two fresh batteries in parallel.

When measuring with BATT not loaded with the "nominal load", you will measure ~3.3V, what is of course a "Hausnummer"

When 50-60% capacity consumed, a 10mA load (a calculation for example) will drop BATT below 2V (but the BATT will show you 3V).

[Marcus von Cube]

(07-03-2014 09:24 AM)pito Wrote:  Do we measure the battery voltage (BATT) under full load??
When not, the BATT does not show the proper numbers..

I agree on your conclusion but I can't help it. The battery monitor is more software than hardware. The brown out detection circuit is programmed to a threshold and signals if the current supply voltage is below or above the programmed value. To detect the actual voltage the software updates the threshold in regular intervals and monitors the result bit. The main problem is the time the circuit takes to stabilize after a change to the threshold. So the algorithm is:

1. Get the current bit value
2. Is the supply voltage below the threshold? If yes decrease the value, otherwise increase it.
3. Wait for the next interrupt to repeat the cycle.
   

The time between interrupts is needed for the output to stabilize. If I want to measure under load I'd need to switch to high speed whenever I adjust the threshold value, i. e. all the time. This would deplete the batteries unnecessarily.

You can measure the voltage under load by executing BATT in a programmed loop and let it run for a while. A running program is executed at full speed unless you set SLOW mode which roughly halves the speed.

[pito]

(07-04-2014 06:51 PM)Marcus von Cube Wrote:  

(07-03-2014 09:24 AM)pito Wrote:  Do we measure the battery voltage (BATT) under full load??
When not, the BATT does not show the proper numbers..

..
I agree on your conclusion but I can't help it.

You do measure the battery with BOD regularly _after_ this chunk:

Code:

 ..if ( GoFast ) {
                /*
                 *  We are doing serious work
                 */
                if ( ++GoFast == 5 ) {
                        speed = SPEED_HALF;
                }
                else if ( GoFast == 10 ) {
                        speed = SPEED_HIGH;
                        GoFast = 0;
                }
        }
    >>>>>>>>>>>>>>HERE YOU MEASURE BATT..

As we know quite well how a CR2032 behaves, you can predict the switching into SPEED_HALF (20mA) or SPEED_HIGH (40mA) will trigger the reset. So you may not to allow to switch into SPEED_HALF or SPEED_HIGH modi (based on the previous BATT values). That is maybe a small mod to the above code.. Just an idea..

[Marcus von Cube]

(07-04-2014 07:29 PM)pito Wrote:  As we know quite well how a CR2032 behaves, you can predict the switching into SPEED_HALF (20mA) or SPEED_HIGH (40mA) will trigger the reset. So you may not to allow to switch into SPEED_HALF or SPEED_HIGH modi (based on the previous BATT values). That is maybe a small mod to the above code.. Just an idea..

The speed switch itself is smart enough and does not allow SPEED_HIGH if the battery voltage is below a certain value. SPEED_HIGH is then treated the same as SPEED_HALF. The battery measurement algorithm is so slow that a single interrupt cycle makes no noticeable difference.

Disallowing SPEED_HALF isn't an option in my view. This is a situation were a shutdown makes more sense.

[pito]

(07-04-2014 07:54 PM)Marcus von Cube Wrote:  Disallowing SPEED_HALF isn't an option in my view. This is a situation were a shutdown makes more sense.

Another idea - maybe a quick BATT measurement "under full load" upon switching the calculator "on" may provide an info on the "real" battery status. When it shuts down during that measurement the user knows what is going on.. When the value will be at the edge you may slow down and indicate..

[matthiaspaul]

For easier cross-reference a similar thread can be found here:

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

Greetings,

Matthias

[ElectroDuende]

(07-03-2014 10:13 PM)jebem Wrote:  I get my batteries from China in the TAS for peanuts and the quality is more than acceptable for the asking price and free delivery.

It's a long time since this post, I was reviewing HP34S issues and went along it, and I though it may be interesting to know this.

Last year I got a pack of 20 CR2032 cells from a chinese auctioner, its expiration date is mid 2017. Some days ago one of my daugther toys with one of them in stopped working, and after opening I saw the battery had leaked (fortunately the toy was ok).

When I opened my spare baterries box, I saw that some of the unused chinese CR2032, had also leaked in the package!

So for future... no cheap chinese batteries in my calculators.