Calculator blog

As most of you know, HP will market the new HP Prime in late september 2013. I have been playing with one of the prototypes. The color on its back is textured white. We have been told that the final model will be textured black. It will make it more “professional” in aspect - on the back, since the front is perfect already. There is a plastic cover that is a departure from old leather-like calculator cases of yore. However, it protects well the calculator screen against direct hits and accidental drops.

This machine is thin! Much thinner than the 50g it supersedes. part of the reason is that it uses a cellular phone battery - itself much thinner than even AAA-type batteries. It is impressive in a professional desk, and it was the chat of all engineers in the office.

The unit has been designed to be upgradeable through firmware versions. It was a breeze to upgrade the firmware to the latest version, and I used the virtual Windows XP session in a Mac, so no need even for Vista to be able to upgrade. The newer version has better fonts and probably less bugs - not that I am able to find any significant bug with my scarce experience with it.

One of my main concerns about a calculator is the keyboard feel. HP units of yore always excelled in that. Lately, HP units have been a little bit of hit and miss: the HP35, HP15c and silver HP17bII+ are excellent, while the HP50g, 39gs, 40gs are barely enough. Prime is a little bit like the HP30b: small distance to get the click, but once gotten, the key press is clear and there are no missed or doubled keys. In this respect, it is even better than the HP17bII+.

It is as well a discovery to be able to navigate around the menus with your finger - it gets you there much faster. Also, in some graphical apps, is nice to change scale by pinching in or out with your fingers, or moving upwards or sieways to the graph areas of interest. Remember the HP50g scale screen, and how many times you needed to go there to get a good view of what you were doing? No more!

Enough for today. Tomorrow, more! 

If your HP41CL seems to lose the configuration after a low battery sign, or a change of batteries, and you are fearing to go through the whole installation procedure, do not fear! You don't need to do that. If you have assigned some keys using functions from a plugged in port, and you see XROM xx, do not worry: the memory loss has not been so big or so long as to lose the keys assignments, and therefore most likely the whole setting can be reinstated.

You just need to reenable the memory management: XEQ MMUEN. This will make the calculator again aware of the whole memory area. You can now use re-assigned keys and see that they keep the right function name on it.

This method does not work when you receive the dreaded "MEMORY LOST" message (below that of the HP41CL when it was born!)

Sad news...

I have received a mail from the designer of the circuit board of the HP-41CL (Monte  Dalrymple), where he tells me that he has run out of units in his last production batch. He will only restart production in case there is enough interest on it. So, if you were planning to buy one, you're out of luck for the time being.

However, you can try to make a free reservation, just by sending an email to jose@thecalcuatorstore.com, so that I can forward to monte together with other requests ( I have already three of them)

• 5 graphing apps (blue) to explore graphs –including the new Advanced Graphing App!
• 2 Special apps (red): the Geometry app and the Spreadsheet app
• 4 Statistics Apps (purple) for descriptive and inferential statistics and data collection
• 4 Solver Apps (orange) for solving specific types of problems (triangles, finance, etc.)
• 3 Explorer Apps (green) for investigating a function’s equation and its graph

You may have bought an HP-41 in EBay, just to find that the battery bay is utterly rusted. There goes your investment! Maybe you wanted it for CL conversion - maybe it was your first 41c and you wanted to see what’s all the fuss about it!

Whatever the reason, letting the rust inside your calculator is a recipe for later disaster. It will corrode the circuit lines, communicate to other parts of the circuit, destroying your investment (and, in the case of the CL, it IS an investment!)

You can easily buy port covers, even battery cover; but it is much more difficult to find this piece in internet. And with a good reason: it is the first thing that fails in the HP41c, and therefore you can only salvage the other parts.

Well, there is an option: you can buy in our shop the adhesive circuit that adapts to the plastic body of the battery connector. It has an extremely strong bonding, and the shape is perfect for the task. I have to say that it is a little difficult to put it properly: you need a firm hand and some experience to do it. I would not recommend to buy one to repair your only calculator, since it will not work properly the first time. At least it did not work for me the first time, but I got the experience of what failed. The second time went well already. So, my advice is: if you want to repair just one calculator, you’ll be better off buying two. If you’re repairing more, then n+1 - so that you can afford one error. Or you may be more skilled than me - in than case buy just as many units as repairs you need to do.

The circuit comes with clear repair instructions, but I would stress the fact that you need to bend it properly to shape beforehand, using the guidelines, and that you should start in the cavity between both ports.

I have used it to repair a calculator that would not start, and that I bought in internet as “for parts”. Now it is my CL. Today I have just repaired a non-working CV - this time keeping the original board in. This will end up in another CL as soon as I receive the new boards, with Sandmath 2x2, among other new features.

Please take a look at the beautiful finish of the battery once repaired.

You can find it here.

Programming the HP41CL - some comments.

While looking at the many programs found in the Solution Manuals or the modules, in many cases I find that the data entry to the program is too formatted and not flexible enough. In my opinion, what is really needed is a clear register usage, so that the user can put the correct data into the right registers beforehand - no need to go through all of the data in order to start the program again.

One typical case is the cashflow-related programs. If we agree that the 0 register is now, and register 1 = period 1, and so on, things can get very easy for data entry. You just need to define how many years you’re using, store that in a register outside of the possible year length scope, and then enter all the year data in the order you want, correcting it as many times you want, checking the contents whenever you want, etc. No need for cumbersome data entry procedures.

I would like to be able to apply the same logic to matrix entry, but here the knowledge of how a matrix is stored is much more involved. There, the example to follow is the HP15c user mode matrix data entry for serial entry, and the register 0 and 1 use to fix the pointer in the right cell.

Let’s show it with a couple of examples that I much prefer to the options found in the financial pac or in the financial solutions book: Net Present Value (NPV) and Internal Return Rate (IRR)

Lets agree that the maximum cash-flow length is 20 years (periods). This is OK for most investment projects. This assumption is important, since it is a “hard” limit in the program. If you feel that 10 years is enough, you can modify the program accordingly and be able to run it with a lower “SIZE” arrangement. However, this is not a concern anymore for us CL users.

We will focus on the NPV function, to make it as simple as we can, and forget about IRR. We will use the SOLVE function found in the Advantage pac to find it.

We use register 21 to store the number of periods to use, in order to make the loop shorter; we will use register 22 for the countdown, as the indirect register used in HP15c and others. And the registers 0 to 20 are self explained with regard to periods.

Now, the IRR is extremely easy, based on the SOLVE capability of the Advantage Pac:

In a past blog issued we gave Angel Martín’s indications on how to change the behavior of the %T function in the excellent ROM module Sandmath Lib.4. The method intended to be used with the HP41CL calculator.

In short, the %T function does not appear in the basic function set of either the HP41c or the newest HP41cl. There are a number of ways to implement it in either machine, one of them being the Sandmath module. However, its behavior there is different from that found in HP financial calculators. In them, the “total value”, stored in register Y, is not “consumed” by the function, as is the case with most binary functions. This way, you can clear the x register, disabling stack lift, enter then a new “part” number and continue calculation. I can attest that this behavior is very practical when doing financial analysis.

Angel Martin was extremely helpful in finding and offering a solution for my specific case, Sandmath Lib4 ROM module.

But if you want to perform the same change on the newest version of Sandmath (currently called 2x2), you need to change the memory address. The procedure would be as follows:

Copy the module to RAM. The location of the 4SMT module was 126 and 127 (is a two block module)

"148>840", YMCPY

"149>841", YMCPY

"14C>844”, YMCPY

"14D>845”, YMCPY

The byte to change is 0xpF44, relative to where you put it. So:

Step 1. introduce in Alpha the text: "840F44-1111"

Step 2. execute YPEEK. You should get "840F44-0369”. This is to check that the previous step did work, and that you’re operating on the right byte.

Step 3. modify the Alpha value to "840F44-0331"

Step 4. execute YPOKE.

Step 5. Plug it to the port of your choice (mine was 2):

"840-MAX”, PLUG4

And that should be all.

Interesting adventures with the HP41cl

As you may have found in this blog, despite being engineer, I have a financial job. Therefore, I have used all HP financial calculators and I am a little bit accustomed to how they work.

In the routine costs analysis, you find yourself trying to see which % of a total are several quantities that form part of that total. HP developed several percentage functions, one of them, %CH, is included in the standard set of HP41 functions (as well as in the HP15c, for that matter). It is logical, since in science you usually try to find how much a variable has increased. However, the “percentage of a total”, %T, is absent in both calculators.

One key feature in all % functions in HP calculators is that they leave the y register untouched (the number from which the x percentage is taken, in the % function; the origin for change, in the %CH function, and the total, in the %T function). Most other functions “consume” the operands, which “collapse” into the result. Why?

Typically the y value needs to be used again. For %, you typically add or subtract the result of % to the base number. For example, if you want to add VAT to a price, you just enter the number, enter, the VAT rate, %, + and you’re done.

In the case of %CH, you may want to try with a different “new” value. You just CLX, which disables stack lift, enter the new value, and you’re ready for another %CH calculation.

With the %T function is the same. Once you have entered the total, you may want to calculate the percentage as part of the total of a number of quantities. Again, for each one, you press CLX, enter the new part, press %T and you’re there.

As I said, there’s no %T function in the plain HP41c. The HP41c manual, in its programming part, shows a supremely elegant way of performing this function:

LBL “%T”

1/X

%

1/X

As % keeps Y, so does %T defined as above.

But when I started working with the HP41CL, and having the Sandmath module as one of the plugged in, I decided to use the %T function that resides there. But when I started using it, I discovered that it did not keep the Y register! It rendered unusable the function for my purposes. What a pity.

I then wrote to Angel Martín to inform him of the function behavior. He had modeled it as the typical binary function, that consumes its parameters, but he came up with the solution: you just need to change a byte in a specific position in the ROM and that’s it. He did it from the top of his head - that’s how well he knows his modules!

As I am not too much versed in the inner working of the HP41cl, I asked him for more directions. Actually, he told me three different way of doing it, from the easiest to the most difficult and dangerous.

The first involves downloading the latest version of the module with the serial cable. A no-no in my case, since I decided early on not to have it. It is ugly, and in the long term I can live with the HP-IL.

The second was to copy the module from its flash location to RAM, then modify it there with YPOKE, and the redirect the PLUG to a port of my choice.

The third involved the initial part of the second, but then writing it back to flash - I did not dare to do it by fear of destroying something.

Here’s the procedure for number 2:

Copy the module to RAM. The location of the 4SMT module was 126 and 127 (is a two block module)

"126>840", YMCPY

"127>841", YMCPY

The byte to change is 0xp8EC, relative to where you put it. So:

Step 1. introduce in Alpha the text: "8408EC-1111"

Step 2. execute YPEEK. You should get "8408EC-0369”. This si to check that the previous step did work, and that you’re operating on the right byte.

Step 3. modify the Alpha value to "8408EC-0331"

Step 4. execute YPOKE.

Step 5. Plug it to the port of your choice (mine was 2):

"840-DBL", PLUG4

I have found a time module for the HP41CL and I am now enjoying the facilities it provides.

The most important part, in reality, were the time functions that were already available from the beginning, thanks to the architecture devised by Monte; but it is nice to use your calculator as a machine to keep your appointments. Of course, an smartphone is much better for that purpose, even as a chronometer; but together with the HP IL interface, the HP41CL is able to control many things, like lighting, multimetres, even oscilloscopes.

You may have read my messages about the HP3468A. This is an HP-IL enabled machine. I have two units (one for me, one for the shop), that I have converted to 230v/50hz use. They work flawlessly with the HP41CL.

We have an apartment in a mountain area. We know that the voltage there is not too stable, and the drops have damaged in the past some appliances. But, of course, the local utilities company denies everything (although they have started working to improve the supply with a new transformer house nearby). To prove the point, I have set up the voltage meter of the HP3468A so that it is triggered and read by the HP41cl each 5 minutes. This way I am able to map the whole performance daily. It is a pity that this method does not allow me to see the real peaks and valleys, but at least it gives a good proxy to the real situation. Then you can make some statistical calculations, apart from knowing the waveform (since I am also storing all the points - it is a pleasure to have such a big memory). I have seen voltages as low as 193, which is 16% lower than the European standard at 230.

(By the way, do you know the reason for the 230v? I have been told that the European bureaucrats, faced with different voltages across the English channel (240 and 220v), decided to set the standard at 230 and just enlarge the tolerances. That way, the UK would continue at 240 and the rest of Europe at 220, and everyone would be happy - except the appliances, working with much wider voltage tolerances)