Modernizing the power source in an HP 35 – the first scientific pocket calculator.
A year ago, I finally added an example of the very first handheld scientific calculator – the HP 35 – to my collection. If you don’t insist on a pristine unit, they can be had for quite reasonable prices.
My calculator is a bit beat up, but works fine. The battery pack is missing though, so I didn’t even have the option of rebuilding it with new NiCd (or NiMH) cells. Well… What about lithium? In one of those pleasant cosmic coincidences, a LiPo cell’s 3.7-4.2 V matches the 3.75 V (3 × 1.25 V) NiCd pack almost perfectly. Almost.
I wasn’t comfortable running the HP 35 at 4.2 V. After all, the switching supply in it is early-1970s technology, and while I didn’t check how its output voltages responded to higher inputs, current draw rose with voltage – the opposite of how a modern design would behave.
Figure 1 HP-35 power supply (courtesy of Kees van der Sanden)
Not wanting to wait till I had time to build a proper solution, I threw together a LiPo battery “holder”, and for the last year, I’ve only lightly charged the battery (<3.8 V). It’s been a fun blast from the past (not that I ever had an HP 35 back in the day; my first serious calculator was TI’s excellent SR-56 programmable). If you’ve never used a LED-display calculator, well, there’s just something a bit magical about them. Or…that could be pure nostalgia talking. It’s hard to tell sometimes.
I believe the original battery pack was rated 500 mAH, whereas the LiPo I’m using is 1,000 mAH (and significantly lighter/smaller). Is there a Moore’s Law equivalent for batteries?
Figure 2 The beaver-hewn LiPo holder
As a relatively inexperienced woodworker, still growing my shop, I regularly make the stupidest mistakes. Not little things, but significant conceptual execution blunders. In this case, the battery was meant to be on the rounded side, with only the copper-tape contact strips on the flat side. Oops. Instead of starting over, well… I made it work as-is. I suppose I commit the occasional electronics blunder too. I wonder what’s going on. “Ah, it’s only a hunk of wood”, thinks my brain, and wanders off somewhere while my hands screw it up. Electronics gets more attention? Maybe.
A potential solution to this power situation would be an LDO (low dropout) regulator, with two key specs met: dropout voltage, and quiescent current. The need for a very low dropout is obvious: obtaining maximum battery life. And even once regulation is lost, maintaining that low differential will squeeze every last usable electron from the LiPo. Quiescent current is also key here, as the LDO is always connected to the battery. I wasn’t interested in modding the calculator to put the regulator inside the case, after the power switch.
A bit of searching led me to the LTC1844 and LT3021. Both can supply enough current (though the 1844 is just enough, at 150 mA). At 100 mA, the 1844 has a typical dropout of 60 mV, vs. 85 mV for the 3021. And the 1844’s typical IQ is 35 µA, vs. the 3021’s 120 µA. The LTC1844 wins.
I ordered a couple of the LDOs, plus a strip of 1 µF 0603 ceramic caps for input & output. Amazingly, I already had suitable voltage-setting resistors in my parts collection. The 3.75 V output is an easy match to the regulator’s 1.25 V internal reference. I used the 100k0 and 200k0 0603 chips in my bin.
Continuing with the suboptimal battery holder design, I decided to build the regulator circuit on copper tape, and stick it to the LiPo cell. Hey, you can learn from my mistakes.
It’s been a while since I’ve dead-bugged SMT parts, but surprisingly, the process went smoothly enough. The 0.95 mm pin spacing is not much worse than the easy 1.27 mm scale of a SOIC, but it felt pretty small. I probably need to use a smaller iron tip next time, to say nothing of my solder and 30 gauge wire, which were both looking pretty gigantic under the magnifier. I CA’d the IC to the tape, and strung the discretes as required.
Figure 3 The LDO dead-bugged on copper tape. The input cap is upper-left, output cap and divider at bottom.
Figure 4 The circuit assembled onto the LiPo. I added some spongy feet to keep the delicate components from hitting the case. Guess I’ve voided the warranty by throwing H.P.’s CAUTION to the wind.
Before final assembly, I measured battery current: 57.5 µA. Subtracting the 12.5 µA flowing through the resistor divider leaves 45 µA LDO IQ – somewhat over typical, but within spec. That means my 1,000,000 µAH cell will last 17,400 hours – about two years – just sitting there. I can live with that.
The output measures 3.768 V (0.5% out – I got lucky). I also made one measurement in dropout: 3.752 V input, 3.746 V output, but that was unloaded, hence pretty meaningless.
Figure 6 An HP family portrait: HP 35; HP 15C; HP 35s, introduced to mark the original's 35th anniversary; HP 39gs.
Finally, see it in action:
- Teardown: 1966 Programmable scientific calculator
- 1969 Compucorp calculator teardown
- Prototyping methods
- Does not compute! Everyday calculations
- AntEEques Roadshow
- Old faithful tools
—Michael Dunn is Editor in Chief at EDN with several decades of electronic design experience in various areas.