NP-41 Emulator (may be)

[Jlouis]

(10-11-2015 11:23 PM)Chris Chung Wrote:  Got the keys working, will need more timing adjustment (scanning) as they are sometimes not very responsive.

I am running this on a partially depleted coin cell at around 2.85V. A 3V battery will produce too much ghosting on the LCD.





Will be designing a new PCB to move on. The charge pump still fails after 1/2 second. It drops to 1.66V instead of staying at 2.8V. I had tried different values and types of capacitors and still cannot get it to work. I had also ripped the capacitor pad off after excessive re-flows. Will include more options in a new PCB design for charge pump trials (also allows custom biasing to see if contrast can be improved.

Also found out that the current design w/ the use of a port 9, I cannot assign the "ON" key the role of h/w interrupt after getting into deep sleep. So the current unit does not have a good power saving sleep scheme.

I am just showing the video here as there is a lot of interest for it. My next PCB will just contain the MCU and LCD, plus a 9-10 pin connector to a keyboard. This will save cost at the current stage.

The final unit (for this PCB) is a bit large for my liking. I was indecisive when doing the LCD and had chose to match the HP-41C LCD size. It appears that based on that the product is more likely a desk unit (a slim one though) instead of a pocket unit.

Wow, very nice indeed!

By the way , nice sound track too!

Can't wait tod see your project done!

Keep on!

Cheers

[rprosperi]

(10-11-2015 11:23 PM)Chris Chung Wrote:  Got the keys working...

The final unit (for this PCB) is a bit large for my liking. I was indecisive when doing the LCD and had chose to match the HP-41C LCD size. It appears that based on that the product is more likely a desk unit (a slim one though) instead of a pocket unit.

Amazing progress Chris! Awesome indeed!

Are you going for a 41C, 41CV, or 41CX?

Oh, and can we upgrade this with a 41CL board?

Cool! Looking forward to the next stages...

[Massimo Gnerucci]

Thanks for your update.

Keep up the good work!

[Jlouis]

(10-12-2015 02:12 AM)rprosperi Wrote:  

(10-11-2015 11:23 PM)Chris Chung Wrote:  Got the keys working...

The final unit (for this PCB) is a bit large for my liking. I was indecisive when doing the LCD and had chose to match the HP-41C LCD size. It appears that based on that the product is more likely a desk unit (a slim one though) instead of a pocket unit.

Amazing progress Chris! Awesome indeed!

Are you going for a 41C, 41CV, or 41CX?

Oh, and can we upgrade this with a 41CL board?

Cool! Looking forward to the next stages...

May we dream all modules in it?

Now I can't sleep anymore!

[brouhaha]

(10-12-2015 10:56 AM)Jlouis Wrote:  May we dream all modules in it?

There's this one, which supports almost all modules, has USB and MicroSD, and has an optional four-level stack display (turned on by flag 56).

The consensus at HHC conferences was that everyone wanted a 42 rather than a 41, so there hasn't been much further development on the 41 code. The more recent hardware is packaged a little better than that prototype was, but is shown running Free42 rather than 41.

[Harald]

(10-12-2015 06:04 PM)brouhaha Wrote:  The consensus at HHC conferences was that everyone wanted a 42 rather than a 41, so there hasn't been much further development on the 41 code. The more recent hardware is packaged a little better than that prototype was, but is shown running Free42 rather than 41.

Yes, and that makes a lot of sense. Your calculator is supposed to be better than what was already available from HP. And better than the WP34s because of a more capable hardware.

Chris is aiming at something completely different here. I think somewhere he said it is supposed to be a "retro toy". But it will be a very useful toy as well. And I am sure I will want one too

[Chris Chung]

I agree w/ Eric and the NP-41 (and NP-25) are a very different class of projects. The DIY series is targeting usability, takes serious potential users input, with enhanced functionality and extension capabilities. It will be the calculator you can rely on for serious work.

The NP-xx is quite different. I create them just for fun. They are hobby grade w/ focus on building electronics than creating an everyday calculator. The keys are not comfortable, the LED are tiny and hard to read. And I actually enjoy building / designing them more that I use them. The project is satisfying to me on what I achieve technically, like the clever "1-wire" shift register usage on the NP-25 and the customized 14 segment LCD glass I am trying on the NP-41. Having said that I still want them to be workable as a show piece, although they are not meant to be your "go-to" calculator.

I need to learn more on the HP-41 peripheral systems to see if it is feasible to add them to the NP-41. Currently it does what Nonpareil does as the core engine is that. The MSP430FR6989 chip I used has 128K FRAM (program and volatile memory both uses it) and 2K SRAM.

The emulator size is around 26K right now, and I think I am emulating the 41CV w/ 336+176 registers. That takes up 532*14 bytes, or less than 8K, so there are potentially 80K+ FRAM memory for other use.

The currently design has a 9600 UART port which can talk to a PC via USB-Serial. The processor itself supports way more peripherals like SPI, I2C among others, So IR can be added, etc.

Right now I am working on the next PCB which would be display + MCU only (somewhat like DIY-4).

[Waon Shinyoe]

Update from Zweb's STM32 version:

-The display has been driven successfully.
-But there is an error on PCB design. The COM1~6 pin were reversed... so the signals have to be reversed by software. The reversion process is slow, and cause the low refresh rate of the screen. Trying to use LUT.

-Still working on keyboard input. Register for GetKey doesn't work properly. -When switched back from alpha mode or nomenclature mode the screen will be filled with grabble.

[brouhaha]

(10-12-2015 10:33 PM)Harald Wrote:  Chris is aiming at something completely different here. I think somewhere he said it is supposed to be a "retro toy". But it will be a very useful toy as well. And I am sure I will want one too

I fully agree with you. I only brought up what Richard Ottosen and I built as an example of something that can be done, albeit at greater expense.

I actually considered trying to develop a calculator variant with a segmented LCD, but retaining the four-level stack display. The problem is that it needs 831 display segments, more than any of the common microcontrollers with LCD drive can handle, or even any common standalone LCD driver chips. At one time I thought doing that would almost force going to a custom chip-on-glass display module, which is unfortunately much more expensive to have made than the purely segmented LCD of the single-line NP-41. However, when I researched it more, I found that Chinese masked-ROM microcontrollers such as the GeneralPlus GPL191C offer sufficiently high segment counts. They don't have enough RAM for a sophisticated calculator, but could be used as a smart LCD controller. Compared to US and European IC vendors, the Chinese vendors have low mask NRE charges and low minimum volumes, so that's not completely unreasonable for a "high-end" hobby project.

Another way to get the four-line stack display would be to actually use four LCDs, with four LCD-driving microcontrollers, where three are just used as slave display drivers.

When I hacked the 41 to do the multiline display, I only implemented that for the stack, because that was fairly easy. It would also be possible to display all 24 characters of the ALPHA register on two lines, rather than scrolling, and to display multiple program lines in PRGM mode, but that would take a lot more hacking. It's been somewhat easier adding these features to the Free42 code base.

[brouhaha]

(10-13-2015 01:41 PM)Chris Chung Wrote:  Right now I am working on the next PCB which would be display + MCU only (somewhat like DIY-4).

The DIY4 and DIY5 boards were designed as single-board calculators, but with the keyboard and battery partitioned from the CPU and display, so that they could be separated and connected by a ribbon or flex cable. However, we were not able to identify any sufficiently low-cost flex cable and connectors. It's amazing how expensive SMT flex cable connectors are, as well as the flex cables themselves.

If we did have suitable cable and connectors, we would have wanted to put the keyboard on the "wrong" side of the PCB, in order to assemble the PCA using single-sided reflow, and have the keyboard section scored. After PCA assembly and test, the keyboard would be snapped apart from the display and CPU, flipped over, and installed in the case.

I've considered using a separate microcontroller to scan the keyboard, so that the wiring between the keyboard PCA and display/CPU PCA might be as low as three to six pins, depending on the interface used and whether the battery and speaker connect to the keyboard PCA. With only a small number of wires needed, it becomes plausible to simply hand solder a few wires between the boards, trading off component cost for labor cost. The trick is to find a low-cost microcontroller with *extremely* low static power dissipation, well under one uA. The Energy Micro Zero Gecko appears to fit the bill. In the course of researching that, I found several dedicated keyboard scanner chips which claim to be low power, but their idea of low power is much different than mine, typically on the order of 5 to 10 uA. They're almost as expensive as the Zero Gecko, and not as flexible.

For the upcoming XC-40 run, we're sticking with basically the DIY5 design with relatively minor improvements to the electronics, but the same overall layout and packaging. We may revisit the keyboard-on-the-bottom idea and/or the separate keyboard microcontroller idea in a future revision.

[Chris Chung]

A mini update.

PCB fab house had sent my new design and I expect to receive it in 2 weeks time in slow mail. I am saving cost now with standard shipping as I burnt too much getting the 1st prototype to work. Last PCB cost $20 to make, I use DHL that cost $25 and the custom clearing is another $16 (turns out for value over $20 they will add clearing and other fees). I am sticking to snail mail this time.

In the meantime I had played w/ 3d prints and keycap buttons and I can keep you entertained w/ my failure.

   
I designed it w/ a online tool and get it made for $16 from a hub via 3dhubs. They are keycaps and I am trying to make them fit into some buttons.

   
I was suppose to "break" individual keycaps out from the waffle and sand them down (that was what I thought). But they are too thin and resolution is not defined enough.

   
Here are the keys that I intend to use. I was trying to match up the "nibs". The keycap should have a 2.4x2.4mm square hole to match the button's nib.

Unfortunately I over-estimate the 3D printer's resolution, the squares on the keycaps are not precise enough for a good fit.

It looks like I need to invest in a 3D printer to further my trials, one button + keycap at a time.

[jebem]

This is great news, Chris!
Excellent idea and effort.

[Chris Chung]

Apologies for the lack of update.

Been distracted w/ other little projects.

So I had the new boards for 3 weeks now. Had assembled one, failed again. The super fine trace does not agree with me and I always manage to 1) over do the solder paste, 2) short a trace or two, 3) trying to correct them w/ soldering iron, load of flux, solder wick,e tc. 4) after a few tries the tiny thin trace can stand the heat and pressure and give up, copper isolating from PCB.

I kind of feel tired w/ it and stayed away from the project until a few days ago when I assembled another board w/ my last MCU (these are costly, almost $20 ea). Again can't really control the solder paste perfectly (I think I need a solder mask made), and this time there were less shorted traces. I was able to correct them w/o destroying them. Took me two evenings though.

And the result is spectacular! With the 1st version "fat boy" PCBs, there were no contrast control due to charge pump failure, the contrast was not desirable and there were a lot of ghosting. To a point where I don't think I will want to use it. They are all gone in this version w/ the charge pump working, for now I had set the LCD voltage to 2.8V and there is very good contrast and almost no ghosting (unless you look really hard / angle it).

   

This version is a "headless" design where there is no keyboard and peripherals. Only the "brain" and the LCD. On the back side there are two 8 pin female headers, they are going to connect to a "base" board which is mainly 1) keyboard, use 9 pins, 2) power, 2 pins, 3) uart, 2 pins. There are 3 pins left for other purpose like may be a buzzer, and / or IR.

The idea is that I can try different keyboard designs and peripherals while saving my dearest parts (MCU + LCD). I got the idea from one of the DIYs Eric was showing (he had very expensive sharp low power LCDs, I think).

Show back of PCB, above is the failed unit.

   

To be honest, I am losing steam on this project. I've got the most difficult part working and the rest would be really boring. Like trying different keyboard schemes etc. So this will be put into a slower pace. It is also really difficult to build this, w/ all the SMD parts and 0.5mm trace.

I originally want to make this into another pocket calculator so that one of my children can take it and "really" use it in high school. The NP-25 looks more like a toy than calculator and she's reluctant to use it in class (although it works fine). Now the NP-41 is getting too big, and too delicate to carry around. When me and Waon did the LCD, I just decide that it's about the same size of a HP-41, which now looks enormous. Anyway, I am still very happy to be able to work out this custom LCD thing.

OTOH, there was this original "plan B", in case the 14 segment LCD does not work out. My DogM display module came and it's the right size for a pocket calculator. Actually it came more than a month ago. I worked on and off on a smaller, easier to make version.

The DogM display being narrower will make into a pocket size, I had order a 100x60mm board for it. Here is the prototype.

   

w/o the back light, the contrast is worse than it's bigger brother, as it's a graphic display. I am using a lesser chip (I like to challenge myself). This is not a FRAM chip, but a MSP430G2955 "value line", just got 56K flash and 4K ram. It is not a DIP device, but in TSSOP packaging, it will be easier to solder than the MSP430FR6989's LQFP packaging.

I think this one can address a broader audience of builders (as kits, etc) if I do it right. With a 132 x 32 graphic display, there are potentials to add bonus (like addn registers contents) features.

So that's it for now. Next move is the DogM PCB (when they arrive), and I am on and off designing a "base" keyboard PCB for the 14 Segment design.

[quantalume]

I'm very pleased to find out about this project. I was just about to embark on a similar one myself. I've always wanted a desktop version of the 41C with 14-segment LEDs and real keyswitches like the Cherry MX. I had in mind using a Raspberry Pi Zero as the brains, which would permit all sorts of connectivity options.

Any updates from our Chinese friends? What is the status of LCD production and cost?

Many thanks for your efforts!

David

[Kinma]

I must say; I love the picture of the spectacular LCD!

I would like to see you use that.

However; I understand the size issues.

[Massimo Gnerucci]

(12-17-2015 07:28 PM)Kinma Wrote:  I must say; I love the picture of the spectacular LCD!

I would like to see you use that.

However; I understand the size issues.

Ditto! Fantastic contrast.

Thank you Chris, however this turns out it will be definitely interesting (and I want more than one!)

[Chris Chung]

(12-17-2015 04:05 PM)quantalume Wrote:  I had in mind using a Raspberry Pi Zero as the brains, which would permit all sorts of connectivity options.

Any updates from our Chinese friends? What is the status of LCD production and cost?

Using a PI would be very different, as it would run on top a OS. It may not the the instant-on instant-off attribute of a direct MCU build. But definitely will work for a desktop calculator. Although not very common, there are 6 digits x 14 segments LED module that you can use.

We have not initiate batch LCD production. I have about 10 samples. The combined cost for a batch of 300 units would likely be a bit less than $1000. Which is very attractive if we have the scale. Consider the graphic option (plan B), the display alone would cost $30 (if acquired in US, from China it's $12, but not reliable supply).

(12-17-2015 09:18 PM)Massimo Gnerucci Wrote:  Ditto! Fantastic contrast.

Thank you Chris, however this turns out it will be definitely interesting (and I want more than one!)

I am also very impressed w/ the contrast outcome! I was discourage w/ the last build.

I was thinking to create plan B for a kit, it will be easier to assemble and more reliable (fewer traces, fewer problems). There are many things that can go wrong w/ the custom LCD version (which did). I was thinking to just leave it there after I finish my build, and if there are a few here would like the LCD they can get them from me.

But l will let the project lead itself, will work on plan B, get a keyboard for plan A and see how it goes. Plan A will be slower though.

[Helix]

It's an amazing project Chris, and I hope you will succeed in it.
Personally, I know I will be pleased with your Plan B. I've never used a real HP 41 so I don't feel nostalgia for the 14-segment display, and I will certainly buy the small graphic LCD version if it becomes available.

[Chris Chung]

Okay, one question for HP-41C owners.

I tried the NQueen problem. It took me 15 minutes on the current NP-41.

The result is 876.0. Can anyone confirm this is correct?

I got the program somewhere in this forum.

[smp]

(12-17-2015 07:28 PM)Kinma Wrote:  I must say; I love the picture of the spectacular LCD!

I would like to see you use that.

I absolutely LOVE this!

I was ambivalent before, but I want one of these now.

Please keep up your effort.

smp