Adding a Crystal to a WP34s

[Gerry Schultz]

Okay, I've now got a 34s with a new overlay and it works great. As I understand it, to make the time functions accurate, it requires adding a crystal to my 30b. Has anyone compiled a parts list and where to put them in a 30b? Is the same for a 20b? I can get the parts and do the install as I assume it will be all surface mount components.

Thanks for any information you have.

Gerry

[Marcus von Cube, Germany]

The 20b and the 30b are identical in this respect. You need some SMD soldering skills as the capacitors look more like grain of sand then like electronic components.

Eric Rechlin has the values on his pages: 20b Hacker's Kit. You won't need the whole kit just the two 18pf 0603 caps and the 32768 kHz crystal. Once installed, the key sequence to activate the oscillator is ON+C (twice in a row). If the answer is OK, you've made it. Otherwise the calculator hangs.

[Marcus von Cube, Germany]

Just an interesting observation: The crystal seems to make the calculator faster, at least my 30b takes advantage of it. I tested it with the simple addition benchmark. The result was around 340.000 without the crystal and around 415.000 with the crystal enabled.

For the records: Both tests were made in decimal mode with the following code:

INC X

BACK 01

The program was started with 0 in the display and ran for about a minute. Integer mode is a bit faster.

[Paul Dale]

Want to try the eight queens benchmark with a crystal to improve the timings a bit :-) Source in the library.

I've added a crystal but no caps to one of my units. The caps I got were a bit too large to solder in place properly. Waiting for new smaller ones.

- Pauli

[Marcus von Cube, Germany]

Quote:

---

Want to try the eight queens benchmark with a crystal to improve the timings a bit :-)

---

Strange, I just get 2.1 and 2.4 seconds. Same result as yours.

Thinking twice it's not so strange any more: If you did use TICKS to time the benchmark, you will have got the same results as the ticker runs slower at the very same rate as the rest of the machine. To do real world timing, you need to run the benchmark 10 times and time it with a stop watch.

CLRREG is a bad command. It spoils automatic timing. I had to modify the benchmark to store TICKS in A just after the CLRREG so that LBL A; XEQ '8QN'; RCL- A; RTN could do the timing.

A command to just clear a range of registers would be useful. If it takes its limits from X it will be hard to make it work in integer mode. You need to introduce some scaling there.

[Paul Dale]

Quote:

---

Strange, I just get 2.1 and 2.4 seconds. Same result as yours.

---

Yes, I used TICKS for the timing. I half suspected it wouldn't alter the perceived timing.

Quote:

---

A command to just clear a range of registers would be useful. If it takes its limits from X it will be hard to make it work in integer mode. You need to introduce some scaling there.

---

I really shouldn't have to say this but, "read the manual" :-) R-CLR does what you want already.

- Pauli

[Marcus von Cube, Germany]

WP 34S is better than I suspected. ;-) We should change the 8QN listing to use this command.

Edited: 24 June 2011, 3:39 a.m.