41CL/Time Module "Correct" doesn't?

[BobVA]

Hi:

I'm probably just being thick here, but I cannot get the "CORRECT" function to work on my HP-41CL. If I enter the correct time and execute "CORRECT" the time is set appropriately (just as if I'd used "SETIME"), but the accuracy factor isn't stored ("RCLAF" always returns 0.000000 afterwards).

I've tried this twice but unfortunately didn't check the time error beforehand, thinking I'd just made a mistake, so I'm not sure what the AF should have been. Since I'd waited a couple of weeks in both cases, zero wasn't expected.

I tried a quick experiment over the course of a couple of minutes and did get the AF to set to 0.1000 by entering a new time that was a minute or so off and executing "CORRECT".

So, if there is a problem, it seems there needs to be a longer interval between the "SETIME" and the "CORRECT" to cause it to surface.

If that's the case, it may point to some kind of issue with where or how the time span interval is stored?

I'm curious if anybody can either see what I'm doing wrong with "CORRECT", or has observed the same issue?

I'll try again in a couple of weeks, but before trying "CORRECT" I'll note the error so I can manually figure the AF.

Regards,
Bob

[ColinJDenman]

Not sure if it helps, but over several years of correction, my AF settled down at around 12.7 (but possibly minus! it was a long while ago).

I don't think you should expect much change over such a short period of time -- indeed frequent correction, with its inherrent residual error can result in poor AF and timekeeping. In my experience and recollection, the interval for correcting was never more than once a month and often several months. It might even recommend such a frequency in the manual.

If you chose to keep DST, you'd need to remember not to allow that to slip into the equation, either use local time consistently or use UT consistently, making any systematic hour correction with T+X (again my recollection is hazy).

[BobVA]

Thanks for the reply - very interesting. The "CORRECT" function seemed to work this time - not sure what, if anything, I did incorrectly before though. ("Applied superstition" would suggest it was because I manually calculated the AF before trying "CORRECT"!). It may have been something as dumb as typing "SETAF" instead of "RCLAF" when I checked the result, but I hope didn't do that *twice*.

CORRECT gave an AF of +55.4, which was pretty close to what I manually calculated. I suspect the difference was caused by the calculator having the exact date/time of the last SETTIME operation stored, and me guessing based on the date of my post

I'll give it a try again in another few weeks and see what it looks like.

Regards,
Bob

[Joe Horn]

(09-25-2014 09:43 PM)ColinJDenman Wrote:  In my experience and recollection, the interval for correcting was never more than once a month and often several months. It might even recommend such a frequency in the manual.

In the 41CX Owner's Manual (August 1983 English edition), Volume 2, page 238, it says "the time span between uses of CORRECT should be 1 week or more."

Also, the entire Appendix F (pages 374-378) is about the clock's accuracy. The last footnote at the bottom of page 375 says, "The longer you wait to execute CORRECT, the smaller the error due to keystroke variation becomes in proportion to any error resulting from a combination of all error factors. A practical time span for many applications is 1 week." On the next page, it says "Remember that increasing the time span between execution of SETIME and CORRECT and execution of the next CORRECT will result in a more precise accuracy factor."

[BobVA]

(09-26-2014 09:27 PM)Joe Horn Wrote:  ..."The longer you wait to execute CORRECT, the smaller the error due to keystroke variation becomes ...

The time module manual also has an interesting footnote:

Quote: Approximately +/- 0.1 second is the maximum keystroke precision for most users. You can reduce precision error by executing CORRECT as a function assigned to a key instead of by XEQ ALPHA CORRECT ALPHA. This is because the calculator takes less time to internally locate and execute a function assigned to a single key.

Which is what I did.

I'm not sure what effect the TURBO setting on the CL might have on this, but I believe the speedup is disabled for actions related to the keyboard, so I'm guessing none, regarding that particular delay.

Bob

[ColinJDenman]

(09-26-2014 09:27 PM)Joe Horn Wrote:  In the 41CX Owner's Manual (August 1983 English edition), Volume 2, page 238, it says "the time span between uses of CORRECT should be 1 week or more."

Also, the entire Appendix F (pages 374-378) is about the clock's accuracy. The last footnote at the bottom of page 375 says, "The longer you wait to execute CORRECT, the smaller the error due to keystroke variation becomes in proportion to any error resulting from a combination of all error factors. A practical time span for many applications is 1 week." On the next page, it says "Remember that increasing the time span between execution of SETIME and CORRECT and execution of the next CORRECT will result in a more precise accuracy factor."

Ah... only a week. I do recall the keystroke delay being the issue with frequent CORRECTs. Perhaps my months where just my natural caution :-). Trying to synchro a keypress a fraction of a second before the due time to counteract the delay in processing was always fun, especially if you held the key down too long and Nulled your CORRECT.