Did the HP-65 record data?

[Matt Agajanian]

Hello all.

As easy as it is to record programs on the 65, was there a means to record a data card on the 65? Or, when a card is recorded, were the memory registers recorded as well?

On that note, if a data card was created, how was a data card loaded?

Thanks

[Don Shepherd]

(03-21-2017 11:59 PM)Matt Agajanian Wrote:  Hello all.

As easy as it is to record programs on the 65, was there a means to record a data card on the 65? Or, when a card is recorded, were the memory registers recorded as well?

On that note, if a data card was created, how was a data card loaded?

Thanks

Matt, the 65 records program steps only, no registers.

[Matt Agajanian]

As I found out. But, even as a workaround, writing a program to enter constants and store them in registers, then to record that program on a mag card might just be a suitable compromise.

[Geoff Quickfall]

That sounds like a good idea.

How many registers per digit are used. I have to dig mine up and find out. Something like:

6
.
0
2
E
2
3
STO
0
(Next constant)
STO
1
(Next constant)
STO
2
.
.
.
Load data program card; hit R/S

Then insert second program card and load the program referring to the previously loaded data. Does a new card erase the data registers? Again, got to get mine running.

Geoff

[Dieter]

(03-22-2017 03:13 AM)Geoff Quickfall Wrote:  That sounds like a good idea.

This method was even used for some HP67/97 programs although these calculators support data cards. For instance take a look at the Stat Pac's Normal Distribution program. Card 1 stores the constants, card 2 does the calculations.

BTW, some time ago I posted a 67/97 program with a more accurate approximation where the constants are intended to be stored on a data card.

(03-22-2017 03:13 AM)Geoff Quickfall Wrote:  How many registers per digit are used.

"Registers"?!?
The HP65 can store up to 100 program steps. One step per digit, decimal comma/point, "CHS" or "EEX", so a number can require anything between 1 and 15 steps.

(03-22-2017 03:13 AM)Geoff Quickfall Wrote:  6
.
0
2
E
2
3
STO
0
(Next constant)
STO
1
(Next constant)
STO
2
.
.
.
Load data program card; hit R/S

You got the idea. ;-)

Since "STO n" is merged in a single step, each constant requires between 2 and 16 steps to store, so you can handle between 6 and 49 values (a final RTN is required). With one card, that is. ;-) But there are only 9 data registers, so in most cases the complete register set should fit on a single card.

(Edited this section since I originally made the same wrong assumption as Geoff: STO is merged and does not require two steps).

(03-22-2017 03:13 AM)Geoff Quickfall Wrote:  Then insert second program card and load the program referring to the previously loaded data. Does a new card erase the data registers? Again, got to get mine running.

I have never even touched a HP65 but I would be very, very surprised if reading a program would clear the data registers. Why should it?

Dieter

[SlideRule]

I wrote numerous 65 programs which relied on the retention of the contents of the data registers while altering the contents of the program registers.
According to the 65 manual (page 63); "Reading or writing a card records all 100 steps of the program memory. However, it does not change the contents of the registers, which enables you to utilize data developed by a prior program."
BEST!
SlideRule

[Willy R. Kunz]

When I created RPN-65, the HP-65 simulator for iPhone and iPad, I decided to add the capability to read and write data cards, just like the HP-67 does. In fact, you can easily exchange data cards between RPN-65, RPN-67 and RPN-97.

Alternatively, you can store and load registers along with the program on the same card by simply flipping a switch. This actually eliminates the need for data cards or data-loading programs. Touch-ID-enabled devices even let you protect your register data.

BTW, the "67/97 program with a more accurate approximation where the constants are intended to be stored on a data card" Dieter mentioned above may now be downloaded for RPN-67/97 here. (Scroll down to the Math section.)

[Geoff Quickfall]

Haven't played with my few 65s lately so all my info from memory.

Each program line can only contain one digit. So entry of a multi digit constant takes up as many program lines as the constant contains digits. Yet one memory register will hold the complete digit as the display will hold. So right, one memory register holds more data then one line of programming.

I didn't think the introduction of a second program card would erase or overwrite the memory registers; that sentence should have been more of a question; read unfamiliarity with the 65. It would be the only reason that a data card in the form of a program loading data would not work.

Forgot that the STO n command was a merge command in the 65 as it is in all other programmable HPs.

Cheers, Geoff

So much to know, so little memory to retain it, or as the 41CX would say: 0 STO C

[Dieter]

(03-22-2017 04:49 PM)Geoff Quickfall Wrote:  Each program line can only contain one digit. So entry of a multi digit constant takes up as many program lines as the constant contains digits.

Plus any signs, commas and EEXes.

(03-22-2017 04:49 PM)Geoff Quickfall Wrote:  So much to know, so little memory to retain it, or as the 41CX would say: 0 STO C

I'd rather say 0 STO c. ;-)

Dieter

[Dieter]

(03-22-2017 04:49 PM)Geoff Quickfall Wrote:  Each program line can only contain one digit. So entry of a multi digit constant takes up as many program lines as the constant contains digits.

Plus any signs, commas and EEXes.

(03-22-2017 04:49 PM)Geoff Quickfall Wrote:  So much to know, so little memory to retain it, or as the 41CX would say: 0 STO C

I'd rather say 0 STO c. ;-)

Dieter

[Gene]

I've been told that memory is the fourth thing to go.

No idea what the first three were.

[teenix]

(03-22-2017 01:36 PM)Dieter Wrote:  Since "STO n" is merged in a single step, each constant requires between 2 and 16 steps to store, so you can handle between 6 and 49 values (a final RTN is required). With one card, that is. ;-) But there are only 9 data registers, so in most cases the complete register set should fit on a single card.

A gotcha for the HP-65 is that R9 does not have a merged code, and this register is also used for various trig and comparison tests.

cheers

Tony

[Matt Agajanian]

(03-22-2017 04:34 PM)Willy R. Kunz Wrote:  When I created RPN-65, the HP-65 simulator for iPhone and iPad, I decided to add the capability to read and write data cards, just like the HP-67 does. In fact, you can easily exchange data cards between RPN-65, RPN-67 and RPN-97.

Alternatively, you can store and load registers along with the program on the same card by simply flipping a switch. This actually eliminates the need for data cards or data-loading programs. Touch-ID-enabled devices even let you protect your register data.

BTW, the "67/97 program with a more accurate approximation where the constants are intended to be stored on a data card" Dieter mentioned above may now be downloaded for RPN-67/97 here. (Scroll down to the Math section.)

Yes, Willy. Thanks for making the vintage modes an option. That way, a true 65, 67, 97 experience can be had and, when you want to take it up a notch or two, you can.

[Dieter]

(03-22-2017 04:34 PM)Willy R. Kunz Wrote:  BTW, the "67/97 program with a more accurate approximation where the constants are intended to be stored on a data card" Dieter mentioned above may now be downloaded for RPN-67/97 here. (Scroll down to the Math section.)

As already mentioned in the thread with that program: the RPN-67 version contains two data sets for two slightly different rational approximations. However, the provided program version will work correctly only with the first, original data set that is stored in the primary registers! Do not use the alternate set with that program!

Willy, please update your RPN-67 program and data cards. There are two options:
(1) Leave the program as it is and remove the second coefficient set from the data card.
(2) Preferred method: Only use the second, slightly improved coefficient set and (!) apply two changes to the program:
(2a) Change the constant 1,38 in line 033 ff to 1,42
(2b) Change the constant 5 in line 027 to 4,68.

Users that already have downloaded the RPN-67 version simply should not use the coefficient set in the secondary registers. If you want to use the optimized approximation, read the data card, do a CLREG and P<>S to delete the first and activate the second coefficient set, and finally write a new data card. Then apply the mentioned changes to the program and save the corrected result to a new program card. At least unless Willy provides a corrected version.

Dieter

[SlideRule]

(03-22-2017 10:45 PM)teenix Wrote:  ... HP-65 is that R9 does not have a merged code, and this register is also used for various trig and comparison tests...

From the 65 Manual;

pg 59) Also notice the STO 9 and RCL 9 are not mergd. This serves as a helpful reminder that the HP-65 uses R9 to store intermediate results when using trigonometric functions, rectangular/polar conversions, or numerical comparison tests."
&
pg 84) The DSZ (decrement and skip on zero) key subtracts 1 from the contents of R8 and then tests for a non-zero value.

Although it is included in the merged key codes, R8 is also semi-reserved or serves a dual-role.

BEST!
SlideRule

edit: R8 vs R1 - thanks Don, one of those all to often senior moments.

[Don Shepherd]

(03-23-2017 02:22 PM)SlideRule Wrote:  

(03-22-2017 10:45 PM)teenix Wrote:  ... HP-65 is that R9 does not have a merged code, and this register is also used for various trig and comparison tests...

From the 65 Manual;

pg 59) Also notice the STO 9 and RCL 9 are not mergd. This serves as a helpful reminder that the HP-65 uses R9 to store intermediate results when using trigonometric functions, rectangular/polar conversions, or numerical comparison tests."
&
pg 84) The DSZ (decrement and skip on zero) key subtracts 1 from the contents of R1 and then tests for a non-zero value.

Although it is included in the merged key codes, R1 is also semi-reserved or serves a dual-role.

BEST!
SlideRule

SlideRule, I think R8 is used with DSZ, not R1.

[SlideRule]

(03-23-2017 02:37 PM)Don Shepherd Wrote:  .. R8 is used with DSZ, not R1...

Gosh darn senior moments, Thanks for the gentle proof!
{revision made}

SlideRule

[Matt Agajanian]

(03-23-2017 01:58 PM)Dieter Wrote:  

(03-22-2017 04:34 PM)Willy R. Kunz Wrote:  BTW, the "67/97 program with a more accurate approximation where the constants are intended to be stored on a data card" Dieter mentioned above may now be downloaded for RPN-67/97 here. (Scroll down to the Math section.)

As already mentioned in the thread with that program: the RPN-67 version contains two data sets for two slightly different rational approximations. However, the provided program version will work correctly only with the first, original data set that is stored in the primary registers! Do not use the alternate set with that program!

Willy, please update your RPN-67 program and data cards. There are two options:
(1) Leave the program as it is and remove the second coefficient set from the data card.
(2) Preferred method: Only use the second, slightly improved coefficient set and (!) apply two changes to the program:
(2a) Change the constant 1,38 in line 033 ff to 1,42
(2b) Change the constant 5 in line 027 to 4,68.

Users that already have downloaded the RPN-67 version simply should not use the coefficient set in the secondary registers. If you want to use the optimized approximation, read the data card, do a CLREG and P<>S to delete the first and activate the second coefficient set, and finally write a new data card. Then apply the mentioned changes to the program and save the corrected result to a new program card. At least unless Willy provides a corrected version.

Dieter

I'm a wee bit confused here.

I have updated to the most recent version of RPN-67 and the Normal & Inverse Normal Distribution card, in the Stat library, labeled ST1-09A, has the coefficients and storage commands in the program, I don't see a separate Data card. What Normal & Inverse Normal Distribution card is this discussion referring to?

Thank you.

[Dieter]

(03-23-2017 06:04 PM)Matt Agajanian Wrote:  I have updated to the most recent version of RPN-67 and the Normal & Inverse Normal Distribution card, in the Stat library, labeled ST1-09A, has the coefficients and storage commands in the program, I don't see a separate Data card.

Sure, that's the 40 year old HP program with ..."mediocre" accuracy, especially in the distribution tails.

(03-23-2017 06:04 PM)Matt Agajanian Wrote:  What Normal & Inverse Normal Distribution card is this discussion referring to?

The program is the one linked to in this thread. Even in the message you quoted:
"As already mentioned in the thread with that program: ..."

This program is now available as a RPN-67 card. Willy has posted the respective link in his message a few posts above yours.

I posted this program last summer in the museum's HP65/67/97 software library. This program uses a higher order rational approximation with much better accuracy than the good old Hastings version, as quoted in Abramovitz & Stegun and used in several HP software pacs. The required nine coefficients for this improved program are supposed to be stored on a data card. This program and a matching data card are now available as a RPN-67 card on the RPN-67 webpage Willy already linked to (section "Math").

In the meantime Willy seems to have addressed the mentioned problems and the current download should contain a fixed version.

Dieter