The HP15c is in my opinion the best pocket calculator that was ever
made. It might not be the one with the most functionality and it might
not be the fastest but it was well designed. It can do
everything one needs to do with a pocket calculator, it has an excellent
keyboard and a good size.
HP produced the HP15c again for its 30'th anniversary as HP15c-LE (limited edition). The HP15c-LE comes pretty close to the original HP15c but has besides
a number of minor faults two more severe problems:
The PSE function works only once per program execution
The battery indicator does not come on and the battery life time is rather limited
I would like to investigate the battery and power consumption problem a bit.
The big spike: 20mA current draw at button press.
The hp15-le runs with 2x CR2032 batteries that are connected in parallel.
An oscilloscope shows a bit under 2mA current consumption at normal
operation and 20mA at button press. The spike at button press is normally
about 100ms long but if you hold the button then the calculator will
consume 20mA as long as you hold the button. That means we load each
CR2032 with 10mA at button press. That is a rather heavy load for
such a coin cell.
Now is that a hardware fault or a software problem?
The hp15-le has a self test mode and you enter it by holding the g and enter keys
and then press on. The display shows 1.L2.C3.H and then you can press
1 or 2 or 3 to test different parts.
No spike while pressing a button during the self test. The average power consumption is 2-4mA.
It's interesting that there are no spikes in the power consumption
when button are pressed while executing the self test. Not even when
running the keyboard self test.
To me that indicates that the 20mA at button press is really a software
fault. It seems the AT91SAM7L128 Atmel ARM processor that HP used
for this machine is not properly going back to its sleep state when a button
press event was detected and it's looping until the button is released.
Now what is wrong with the low battery indicator? I have read various speculations about the way it works. Some say that the battery status is only checked
every 50 buttons press events and the CR2032 might loose at the end of
its life the capacity rather quickly.
The 15C-le should show a "*" in the display at low battery. Let's connect it
to a regulated lab power supply and then see how it behaves.
Empty battery simulation with a regulated lab power supply.
I took out the batteries on the back and used small crocodile clamps to connect the power supply.
The hp15c-le shuts off if the supply voltage is only at 2.0V and one presses
a button. It resets itself then and when you turn it on again at a higher
voltage it will display an error indicating that it has lost its data.
Therefore let's try 2.1V. It seems to be able to still function at that voltage.
I have let it run for an hour at that supply voltage and I have pressed
all kind of buttons more that 100 times. The low battery indicator will
never every show up in the display.
The low battery status indicator functionality is either not implemented
at all or totally faulty. It does not work at all.
The hp15c-le is still a nice calculator but it has not been built with
the same quality and passion and testing that went into the original HP15c.
There are microcontrollers for which Atmel promises 100 years
data retention on the flash memory that holds the firmware but
for the AT91SAM7L128 that is used in side the hp15c-le they promise
only 10 years. It's not sure
that the hp15c-le will see it's 30th anniversary. My original
HP15C is however still working and it might still be good at it's 60th
anniversary. To be fair I should however mention that all
of the CPUs for calculators produced today are based on a similar flash
firmware storage technology with an expected life of 10 years. Maybe
they will last 20 years but not much longer.
I would like to see a company that produces again a real hp15c. A calculator
known for it's quality, functionality and ergonomics. http://www.swissmicros.com/ started to do that but
the keyboard is rather small (the whole calculator is too small) and the
keyboard is not nearly as good as the one from the hp15c or the hp15c-le.
http://www.finetune.co.jp/~lyuka/interests/calc/hp15cle/, a partial fix for the high power spike problem: add a 100uF ceramic capacitor in parallel to the batteries (requires opening of the case). Note: this extra capacitor may extend the battery life but the low battery indicator will still not work. The calculator will simply shut off at low power and show "Pr Error" at the next power-on. All programs and registers will be lost.
