HP 45 High Precision Timer

[Sadsilence]

Hello all,

as mentioned in another thread ( (Re-)Print labels on back side of classics) there is a more or less easy way to bring hidden HP 45 timer feature an accurate clock generator, despite the fact, that there are are no 784 kHz crystals available any more (see HP-45 Crystal Modification).

As reaction to some interest I ordered material for 10 units. Seven units above captive are created, still available and just need calibration. I would give them away together with a step by step tutorial for feedback. Just drop a line.

Best regards,

Werner

[Etienne Victoria]

Very interested in finally getting some accuracy to my Hp-45 timer !

So I gladly drop a line :-)

Thank you for making this possible !

Etienne

[cruff]

(09-20-2015 06:44 PM)Sadsilence Wrote:  … that there are are no 784 kHz crystals available …

Well strictly speaking, if you an afford to order the minimum batch size, many crystal manufacturers will custom manufacture some for you. :-) You can order a 7.8432 MHz crystal off DigiKey, run it through a divide by 10 and only be slightly fast.

[Sadsilence]

(09-20-2015 09:48 PM)cruff Wrote:  Well strictly speaking, if you an afford to order the minimum batch size, many crystal manufacturers will custom manufacture some for you. :-) You can order a 7.8432 MHz crystal off DigiKey, run it through a divide by 10 and only be slightly fast.

Of course you are right. I checked such opportunities, too. In the end, they are either too expensive or too complex or needs too much power or a combination of all. If there is a real demand of 784 kHz crystals, there would be a lot of manufactures. A customs design in 100 pieces range is beyond any reasonable budget.

Anything else but a matching crystal requires additional electronic parts.

When using preprogrammable oscillators an output level adaption would be necessary. But it is like they try to tease HP-45 owners ;-). Free available chips with preprogrammable frequencies below 1 MHz are rare. The parts in right frequency range are either in wrong supply range, have housings, not manageable for private use or require > 10 mA.

Programmable devices, I include your base frequency divide idea in that category, requires even more parts. Counter/Shifters with a lot of pins for frequency dividing, a precise voltage reference for VCOs or even a microcontroller for PLL stuff or oscillators with a programming interface and, when one chip is not able to handle 3.6 - 4.5V, a voltage regulation circuit. Besides limited space inside HP-45 you have to develope a special PCB...

Another problem is extreme accuracy when it comes to time measurement. 0.1% or 1% tolerances of devices (resisitors etc.) are normally regarded as very good. For a everyday timer 0.01% is regarded as "ok".

Of course, I would be glad, if there is an easier solution or I missed a 784 kHz crystal source or I understand HP-45 curcuits wrong. Feedback is highly appreciated.

Best regards,

Werner

[cruff]

Just remembered that ceramic resonators can also be ordered for specific frequencies, but none of the standard stocked values are 784 kHz. :-(

[Sadsilence]

(09-21-2015 05:58 PM)cruff Wrote:  Just remembered that ceramic resonators can also be ordered for specific frequencies, but none of the standard stocked values are 784 kHz. :-(

Indeed, there are some in 455 kHz range or between 2 and 20 MHz. Moreover accuracy is only about 0.25-0.5%.

Today I come along some MEMS oscillators, programmable between 32 kHz and 100 MHz with a tolerance of only +- 20ppm and "only" 6 to 9 mA load. Voltage levels do not match with HP-45 supply and DFN housing is difficult to handle, but at least it is worth thinking about. Accuracy would increase even more and time consuming resistor selection of current solution is not necessary any more.

[Duane Hess]

Received mine today. Thank you very much!

Arrived safe & sound.

[Sadsilence]

Thanks for your feedback and thanks to all others spending HP 45 calculators and time to give curcuit a chance.

As far as I know every installation reported back was sucessful in the end (thanks to patience of some users) and no calculator passed away. I got several improvement hints, too.

Achieved accuracy is between +-100 ppm (1 s every 3 hours) and +-300 ppm (1s every hour). That is not bad and several hundred times better than normal 100000 ppm (10%) tolerance but misses aim of +-100 ppm by a small margin. Nevertheless, keep human reaction time for pressing start/stop button or mean time buttons in mind it is more than enough for an accurate measurement in some hours range. (Besides the fact, that is more funny than reasonable using an HP 45 as stopwatch ;-)).

On the other side any further accuracy improvement based on that technology is mere luck, due to limited temperature compensation of used oscillator chip and adjustment resistors as well as some voltage supply influcence and output signal jitter.

Anyway, at the moment there are no curcuits left (no more tiny PCBs).

With still a few inquiries and maybe some more interest I am currently testing an improved version (HP 45 Timer 2.0 ;-)), based on a preprogrammable MEMS oscillator with +- 20 ppm. Same PCB, same connection scheme to HP 45 (existing 1.0 version can be replaced without complete disassembly/desoldering wires from HP 45 mainboard) , no time consuming selection of adjustment resistors for me, better output signal and another 10 times accuracy improvement over whole temperature range and power supply range.

First 24h test showed a derivation of only -0.46s (-5,3 ppm). That is in range of human reaction time. So not a very precise measurement, but good enough to say derivation is less than 1s a day with moderate temperature variation and supply voltage variation.

Disadvantages:
- difficult to solder (no real user problem, as I test every unit before shipment anyway)
- 8 mA typical supply current which is 10 times higher compared to first version. HP 45 itself has 120 mA - 150 mA typical supply current. Overall consumption increases by 5-7%. In other words: When your brand new 2500 mAh replacement battery pack is good enough for 15h of wireless operation time you will loose about one hour.

Before building a second batch, several tests with a second prototype unit and a second HP 45 are necessary together with a revised guide. If for any reason V2.0 does not work, I will create some more V1.0 units. How many depends on your interest ;-).

Thanks for your time.

Werner

(10-11-2015 11:18 AM)Duane Hess Wrote:  Received mine today. Thank you very much!

Arrived safe & sound.

[TASP]

Werner wrote:

Before building a second batch, several tests with a second prototype unit and a second HP 45 are necessary together with a revised guide. If for any reason V2.0 does not work, I will create some more V1.0 units. How many depends on your interest ;-).




Tasp responds:

yes, I be interested.

[Sadsilence]

Version 2 is ready to build and material for second batch is here.

Precision is really good. So far I had no unit with a derivation of more than 1 second within 72 h :-). Again 10 to 20 times better than version 1.

Additional current load is up to 10 mA, so I will still generate version 1 at request with only < 1mA typical additional current load.

Revised tutorial is attached. Feel free to contact me in case of any questions or when you decide to give curcuit a try.

[Etienne Victoria]

Was first to ask for V1....and probably the last one to implement it :-)

Thanks Werner, it works like a charm !!!

Gluing the PCB was easy, as well as soldering GND & CLK.



The PCB fits nicely.



The little tricky step is to route the red wire under the display & ICs



Then every thing rests easily back in place



Timer precision for my sample is even better than specs given by Werner.

Thank you again for this very elegant mod.

Etienne

[Sadsilence]

Hello Etienne,

thank you very much for your detailed feedback. Really glad, that you are satisfied. Of course you are invited to test V2 ;-).

Best regards,

Werner

[Sadsilence]

One forum member asked for schematics of HP 45 High Precision Timer versions. Thank you because up to now I had only hand sketches...

   
   

Unfortunately Si501 programmable oscillator of version 2.0 went out of production in 2016 and you need a programmer for it. As I have some units left I can still produce some boards. LTC6900 of version 1.0 is still available.