"Smooth Scaled Deciduous Trees of the Southern Nordic Countries"



#2

Hello all and a special thanks to the attendees that sat through my 10 pm presentation.

The first day (Thursday) involved driving with Wlodek from Los Angeles to Guy Ball's house to see his collection and discuss book publishing. Wlodek also had to pick up copies of his book which has been revised to include the 12 Anniversary and 15C LE which were available to the attendees.

Guy is one interesting guy! Unfortunately we were three weeks to late to see the collection as it was sold in its entirety!

Then on to San Diego arriving at midnight thirty!

Friday stuffing proceedings envelopes with Günter Schink. Günter introduced me to "Weizen" beer during our stuffing break. Later in the evening there was more beer call as others arrived.

Hello Namir, Gene, Richard, Wlodek, Joe and his brother Jim and, well check out the list at:

attendees list

As for Saturday, excluding NDA topics, the WP34S with both Niel and Marcus stole the show. Then Erics demo and discussion on the overlay. Now for me, the more complex and CLUTTERED the keyboard the more I like it. The WP34S is a work of art and for me, the clutter discourages any one from borrowing it in the cockpit.

Andreas Möller talk HP 49G+ / 50G - O.S. Extension Pack is simply amazing. What Andreas has single handedly created with help menus and language modules on the 49G... is incredible.

For everyone out there, Joe Horns presentations are as informative as they are entertaining. No wonder he won the best speaker award, and no, it was not because he tallied the votes ;-)

Patrice again wowed us with his prowess in programming with his talk on "HP-15C Benchmark: The Devil is in the Details".

Eric Smiths DIY was simply amazing. I got to hold it and was allowed to turn it on. Of course being a 41C I felt right at home but was totally blown away by the LCD display. It looked so sharp you would have thought it was printed on a clear vinyl overlay and applied to the LCD panel!

Richard entertained us with his time keeping abilities and the discussion on the PPC ROM which turns out to relevant after 30 years. Appearing in the CL ROM library and the matrix routines form the base of the WP34S matrix solutions.

Jake Schwartz filled us in on the matrix/PPC ROM aspect of the WP34S mentioned above!

Yes DAVE, you are one lucky guy, and I don't mean your TI! HOWEVER the TI is fantastic, what a display to cart around from Reno! Yes bring the others, I may bring the 9825A with the chicklet keys if I can. By the way, I am looking into those glasses, who cares if you look like a muggle (Potter).

Wlodek and Dennis harms discussed the 12C and Wlodek brought and displayed his multiple variations of the 12C since its inception.

As far as the HP crew goes, Cyrille and Tim, many many thanks and I did not think I would learn so much MORE about trees, especially since my background is Botany/Forestry! Who knew.

There were more talkers but I have to refer to my notes which are still packed. Download some photos and etc.

Here is a youtube video I just uploaded. It was part of my talk and is relevant to a thread below about benchmarks and speeds vis a vis the 41CL.

Short explanation of what you are seeing:

There is a requirement to convert metric altitudes in various countries to foot altitudes for use in the cockpit. The standard slope has been arbitrarily altered to maintain aircraft separation at altitude. The two slopes can be regressed with a test as to which slope the metric altitude input belongs. Also a subroutine to round correctly so that three significant digits are used.

ex.

13,100 meters should convert to 43,000 feet. You can see simply multiplying by the conversion "1 meter = 3.2808399 feet" does not give the correct result. An alternative approach as a thinking exercise is to create a data base where the metric is paired with the feet measure. Search for the metric and display the associated feet.

Yes I know it is cumbersome but this is just for demo purposes and it works perfectly in the CL due to its speed.

CL to CX speed comparison on data base search in extended memory.

More posting to follow!

Cheers, Geoff


Edited: 27 Sept 2011, 8:04 p.m.


#3

Quote:
Eric Smiths DIY was simply amazing. I got to hold it and was allowed to turn it on. Of course being a 41C I felt right at home but was totally blown away by the LCD display. It looked so sharp you would have thought it was printed on a clear vinyl overlay and applied to the LCD panel!

Now this is extremely exciting! I think I'm very much out of the loop, maybe Eric has posted about this already, but I don't remember seeing it. Has he tooled up a 41C-style case for his DIY calculator? Does it look like a 41C?

Do tell more!


#4

It's effectively a 41Cx with a display that can (optionally) show all four stack levels, in a nice high-resolution font. That makes it longer than a normal 41CX, though slightly shorter than a 41CX with a card reader.

I haven't had time to write up anything or take photos or video. I'll post something once I find the time.


#5

Very much looking forward to that! Do you have plans to produce it (however scale) or will it remain a DIY project for your own enjoyment?


#6

We're trying to come up with a way that we can manufacture and sell the DIY5 without losing our shirts. The problem is that they will likely be fairly expensive, because we expect that we can only amortize the NRE over a relatively small number of units. How many people today would want to buy an improved -41 if it cost more than a new 50g? Not many, I would think.

On the other hand, Monte is apparently able to sell some 41CL boards for $235, and that is an upgrade rather than a complete calculator. And the 15C LE sells for almost as much as a 50g.

On the gripping hand, we can leverage the same design as a hardware platform for a derivative of Free42, or other community-developed firmware (similar to WP-34S project), so it isn't necessarily just an improved -41.


#7

Quote:
How many people today would want to buy an improved -41 if it cost more than a new 50g?

I of course don't know the answer to that question, but sure enough the 50g price isn't that high to me for such a wonderful project. How much more is too much is the tricky part, and even that is relative.

But sure you didn't mean that price would be for a complete DIY5 calculator, but only the enhanced display? Not having seen the unit at HHC I'm clueless about your implementation, sorry.

On reflection I guess this won't be an upgrade - which would be hard if not impossible. I read your other post where you indicate 41CX+HEPAX capability into the unit, will it also be "upgradable" so that you could install the CL board to it?


#8

I only meant that the selling price of a DIY5 calculator is going to be more than the price of an HP 50g. Most likely a *lot* more, possibly more than double, but I don't really know at this point.

I expect that there is only a tiny market for calculators that cost more than the most expensive HP, TI, and Casio models. If there was actually a non-trivial market for $250 calculators, HP, TI, and Casio would be selling some.

If we could sell 10,000 units, it might be possible to sell them at a retail price under $130 each.

If we only sold direct, and not through wholesalers/distributors/retailers, we could price it lower at the same production quantity (or the same price at a lower production quantity) since the price only would need to include a margin for us, and not for the middlemen. That would work fine to sell to people here in the forum, but would we get sales from anywhere else?

There's somewhere between $25K and $100K of NRE for industrial design, mold tooling, and FCC testing that has to be amortized over the production run. If the production run is small, that's a *lot* of cost per unit.

And of course that NRE doesn't include paying us *anything* for the ten years of (part-time) effort we've already invested into the development of the electronics and software. If we had to amortize that over a small production run also, the calculator would be far too expensive even for the hard-core enthusiasts here. Fortunately that is a sunk cost that we don't expect to recoup.

I'm looking into CNC machining aluminum molds myself. That has the potential to reduce the NRE substantially, but since I don't have experience doing that it will take a long time with many attempts, and could ultimately end up costing more than paying a mold designer.

Edited: 28 Sept 2011, 2:31 p.m.


#9

I hope you respond to my email requesting more information. Since you have substantial time and money invested I will be more than happy to sign a NDA if necessary, but I can definitely help. For example: I know a bit about mold making and can get fabricate CNC aluminum for the cost of machine time ($35/hr.) The market for expensive calculators is out there, but build quality must be top notch.


#10

Machined aluminum cases? :-)

- Pauli

Edited: 1 Oct 2011, 8:21 p.m. after one or more responses were posted


#11

I think some machined cases would be fun as an option. Not to mention, once the tooling is set for CNC the cost is materials plus machine time. Want an extra nice case? I don't think I'm the only one around here who would throw in an extra $50 for it.

#12

Without experience in injection molding you should not try to create your own mold. The plastic needs not only to be injected into the form at very high pressure to reach all corners, straps and edges, it must also be cooled down in a controlled manner and later removed from the mold by mechanical means. Then you need to prepare for shrinking which depends on the temperatures and thickness of the material. The results wouldn't be very appealing.

Maybe a mold for vacuum forming is a better option but the available shapes are restricted compared to injection molding.

Pauli's idea sounds interesting, though. :-)

#13

Quote:
How many people today would want to buy an improved -41 if it cost more than a new 50g? Not many, I would think.

Probably, but count me in!

Massimo

#14

Count me in (2 probably)

#15

You may want to look into doing a kickstarter. There is no risk. You set a funding goal and if the goal is not reached, no money is exchanged.


A few examples:


Coffee Joulies


Cosmonaut Stylus

#16

How many forum members know what Eric was talking about? I did.


#17

So did I. Unfortunately, one of the only books ever that I haven't finished..
while The Mote in God's Eye is still one of the best I ever read, and I've re-read it at least three times..

Werner

#18

I just reread both "Mote" and "Gripping Hand" last month. Great stuff.


#19

I enjoy listening to the unabridged audio of Mote while driving to/from work. Takes a couple of weeks. :-)

#20

Quote:
I just reread both "Mote" and "Gripping Hand" last month. Great stuff.

I was once using the allegory of the King and the Thief without realizing at that moment that I had it from the Mote, and of course, the others had no clue what I was talking about. That must have been 10 years after I read it and I still think of it when the talk is about diplomats.

#21

I missed it but then again hadn't read a niven book since 81 or so. At one time i had every paperback in current print. Lesson learned......whatever you leave behind when heading out into the world is instantly trash! Lol

#22

I came within an inch of including that in my post!

My favourite part of the book, too.

If anyone could teach a horse to sing hymns, it would be a Motie mediator.
;-)

Werner

#23

The "other hand" and "gripping hand" construct was frequently used by Jerry Pournelle in his "User's Column" in Byte Magazine, the column was renamed as "Computing at Chaos Manor" at some point in time, IIRC.

BTW, during the 90's I worked at a PC hardware supplier in Buenos Aires and was invited to write a short, technical feature in the local Novell Users monthly newsletter. I decided to use "Vendor's Row" as a title, intending a low-quality wordplay with "User's Column" and similar titles. Of course, some knowledge of matrices was needed for this pun to acquire the desired meaning.

While Pournelle used to mention Larry Niven's books and also their joint projects, the only one I read was "Inferno". Those were "pre-Amazon" and "pre-Internet for all" times (Pournelle made us envy with his frequent references to his ARPANet account and so); and that was the only one of their books I managed to find during one of my trips to the USA (BTW, HHC 2011 was my 32th visit to the USA).

I once mentined the "gripping hand" expression to my English instructor here. As knowledgeable as he indeed is, this particular expression was not familiar for him.

#24

Hi Eric,
I LOVE the large display, the only exception being the typeface used for the zero with the dot in the middle which brings back painful memories of slaving away punching cards for the IBM mainframe on campus. How about prepending X: Y: Z: and T: in front of the stack register displays, and can you display LastX as well?

With the large display, would it be possible to emulate the printer with the upper part of the display, perhaps based on flags 21 and 55 status? The 41 app for the iPhone might serve as a model.

I would certainly consider buying your 41-like machine, if it's comfortable in the hand. It has to be somewhat affordable, too.

Take your time and get it right.


#25

The font I used is "DejaVu Sans Mono" font, which is based on the Bitstream Vera fonts. I used this only because it was close at hand, and I had few other monospaced fonts available. I wrote a program to render bitmaps of the characters, and added a few HP-41 characters by manually creating editing the bitmaps.

I wanted some obvious distinction between the letter O and the numeral 0. The dot in the middle is just what that font happened to have; I probably would have preferred a slashed zero, which is what the actual -41 used.

There is no reason why a different font couldn't be used, and the source font doesn't have to be a monospaced font as long as you bear in mind that it will wind up monospaced on the calculator display. I plan to make it possible for the user to install different fonts. The current requirement for the bitmap font is that the character cell is exactly 32 pixels wide, which has to include the character, punctuation (period, comma, or colon) and inter-character spacing. I could change the font display code to allow for character cells less than 32 pixels wide.

I'll probably provide a 14-segment font for users that want the actual -41 display appearance.

I could prepend the stack level identifiers. I'll probably use another flag to enable that. The identifier will be in a much smaller font.

I didn't expect anyone to want to also display lastx, but certainly I could do it. I'll add a flag 57 for that as well. Do you want that shown above T, or below X?

I think I might like to use a smaller font for the Y, Z, T, and L displays than is used for X. Maybe another selectable option.

Printer emulation is certainly possible, but at the moment it's a low priority.

Affordable is the biggest challenge. What's your upper bound on affordable?


#26

I thought the digit spacing was a bit wide, too, but I assume that's easily adjustable.

What's my upper bound on affordable? Quite high, actually, but I'm not a good demographic. Probably about what I'll pay you to fix my 9830 tape drive if I can't.


#27

The character spacing has to allow for the period/comma/colon to fit between the characters.

#28

I would think L above T would be less intrusive and allow for changes to the display behavior without having the X value jump around. My 2 cents, although I would pay more. ;-)

#29

Quote:
The font I used is "DejaVu Sans Mono" font, which is based on the Bitstream Vera fonts. I used this only because it was close at hand, and I had few other monospaced fonts available. I wrote a program to render bitmaps of the characters, and added a few HP-41 characters by manually creating editing the bitmaps.

I wanted some obvious distinction between the letter O and the numeral 0. The dot in the middle is just what that font happened to have; I probably would have preferred a slashed zero, which is what the actual -41 used.

There is no reason why a different font couldn't be used, and the source font doesn't have to be a monospaced font as long as you bear in mind that it will wind up monospaced on the calculator display. I plan to make it possible for the user to install different fonts. The current requirement for the bitmap font is that the character cell is exactly 32 pixels wide, which has to include the character, punctuation (period, comma, or colon) and inter-character spacing. I could change the font display code to allow for character cells less than 32 pixels wide.


Please stick with monospaced fonts. My quibble with the zero having a dot in the middle probably isn't worth worrying about. It just brings back the pain of re-typing punch cards ad infinitum for my freshman Fortran 66 class. Having survived that, it's been all downhill from there :-)
Quote:
I'll probably provide a 14-segment font for users that want the actual -41 display appearance.

That would be interesting, though not a requirement.
Quote:
I could prepend the stack level identifiers. I'll probably use another flag to enable that. The identifier will be in a much smaller font.

I didn't expect anyone to want to also display lastx, but certainly I could do it. I'll add a flag 57 for that as well. Do you want that shown above T, or below X?


I agree with Gene: please put LastX above T.
Quote:
I think I might like to use a smaller font for the Y, Z, T, and L displays than is used for X. Maybe another selectable option.

The smaller font is a great idea. I don't care if the size is fixed or user selectable, but others may have a different opinion.
Quote:
Printer emulation is certainly possible, but at the moment it's a low priority.

Yes, please keep it on the to-do list.
Quote:
Affordable is the biggest challenge. What's your upper bound on affordable?

Hmm. Less than the cost of a 10" tablet?

#30

The display will definitely be monospaced. My point was that the font used doesn't have to be designed to be monospaced, though it will always be rendered into a fixed-width cell. In other words, if you want to use Times Roman, you could do it, but the character size should be chosen such that the widest character (probably an upper-case W) would fit in the character cell.

A font designed for monospaced use will usually end up looking better, but there are relatively few monospaced fonts to choose from. I looked through the entire Adobe font library and didn't really find any monospaced font that I like.

If there are good font designers hanging around here, maybe they would care to try their hand a creating or modifying a font.

The calculator uses bitmap fonts in a custom format. I can convert them from BDF format. My preference is to use Opentype, Truetype, or Adobe Type 1 fonts, which I can render into the right size in BDF format. I've assigned code points for the special 41 characters from one of the "private use" regions of Unicode.

#31

Quote:
Less than the cost of a 10" tablet?

According to Google, those start at $139 for an off-brand Chinese tablet. The iPad 2 has a 9.7 inch display, and pricing starts at $499.

Making a few assumptions:

  • NRE of $50K
  • CoGS of $70
  • direct sales at $140 (same as cheap Chinese 10-inch tablet)
the minimum sales to break even would be 714 units.


At the opposite end of the spectrum, high price and low volume,
the assumptions are:

  • NRE of $50K
  • CoGS of $100 (higher due to low volume)
  • direct sales at $500 (same as iPad 2)
the minimum sales to break even would be 125 units.

Neither scenario includes making any profit or recouping any of the investment already made in development. I don't think I can actually sell 125 $500 calculators. The big question is whether there is somewhere between those scenarios that is viable.


#32

My 2 pennies is keep it south of $250. For me that is pretty close to the break point. At $250 or less I would buy two. Between that and $500 I would buy one. Any higher and I'm priced out of the market.

Again, just my 2 cents.

Cheers,

-Marwan

#33

I'd think you should watch 15C LE sales as close as you can.

I'm dying to know how many of those sales are satisfying demand from people like us and how many are being bought by technical users who might not be enthusiasts to the degree common around here. Demand from the latter group ought to drop off sharply as the price goes up. Demand by the former group might be less elastic.

#34

I think at $500 you will have very few sales. At $140 or so you will sell a fair number, but would you sell over 700 of them? Some quick calculations indicate you would need to sell about 330 machines at $250 to break even, 250 machines at $300, or 500 machines at $200. The figures are at zero profit and no safety margin, which are unrealistic.

You have to be a little conservative and make a modest profit as well. I know you would be fair and don't think anyone expects you to work for free, especially considering the effort and expense you have already gone to.

#35

Here, here! The conference was amazing, as always.

Thanks to all the organizers, conference committee, HP geniuses and great speakers!

Bruce


Possibly Related Threads...
Thread Author Replies Views Last Post
  Any Southern California HP Calculator clubs left? Jedidiah Smith 7 680 02-19-2013, 01:40 PM
Last Post: Jedidiah Smith
  Free42S - Android 3.2 Scaled to full screen Bill (Smithville, NJ) 0 253 09-27-2011, 01:06 PM
Last Post: Bill (Smithville, NJ)
  HP Codes for other countries Bill Lafferty 0 204 11-26-2005, 12:13 AM
Last Post: Bill Lafferty
  Same model calculators manufactured in different countries Nick Nicholas 3 472 01-28-2003, 10:56 PM
Last Post: Randy Sloyer
  Quality of HP calcs. made in different countries Iqbal 9 841 12-05-2001, 12:47 AM
Last Post: db(martinez,california)

Forum Jump: