brydling

I've now received advance payment for two boards, and one more guy has told me that he's going to pay for a third board but hasn't done it yet. After that, there's only one more order needed for me to place the order for the fifth batch of B256A13's. That will most likely be the final batch. I thought so about the previous batch too, and orders are coming in an even lower rate now. That will also allow me to shift focus to producing the DTS-board If you're interested, send me an e-mail as stated in the first post in this thread.

I've now received advance payment for two boards, and one more guy has told me that he's going to pay for a third board but hasn't done it yet. After that, there's only one more order needed for me to place the order for the fifth batch of B256A13's. That will most likely be the final batch. I thought so about the previous batch too, but orders are coming in an even lower rate now. That will also allow me to shift focus to producing the DTS-board :) If you're interested, send me an e-mail as stated in the first post in this thread.

Nice! Send me an e-mail when you're ready to do the magic :) Best regards, Niclas

You're welcome! The board uses the 18F4550. It does not have a programmer connector, so you'll need some type of programmer adapter board or a breadboard so that you can make your own. I use a breadboard. The PICkit2/3 are good programmers. Best regards, Niclas

More updates :) I debugged the kernel today on Ubuntu 12.04 with kernel 3.2.0-54 and found the issue. I am working on a solution for fun, but I don't know if it will make it upstream :) I have never done kernel development before, so I really don't know how to get the code into the mainline kernel, but when I have a fix I guess I'll try to find out :) After browsing the kernel code online and checking various newer versions, I noticed that in version 3.8 and later a bug was fixed that missed 16 buttons, so with a newer kernel you will get 80 buttons per logical joystick instead of 64. I tried to patch that change to my 3.2-kernel and tested it with good results. I am upgrading to Ubuntu 13.10 now to verify that it works with an unmodified newer kernel too. I will then make my fix to that kernel version instead of the one I am using now, to have a newer base to work from :)

Ah, I understand ;) A couple? You need more than one extra? If you have access to someone with a PIC-programmer, I can send you the bootloader file so that you may upgrade it yourself for free. If you want a new chip, it is 60 SEK each + shipping. The shipping is a bit cheaper than it is for a whole board since I don't think it is necessary to pay for insured shipping and they weigh almost nothing :) Regards, Niclas

Hi Bigfoot, In the first post in this thread you can read the ordering instructions. Thank you for your interest :)

Hi Peter! Firmware is upgradeable over USB. See the first post in the official thread over at the For Sale-section, under the section "News:". The bootloader is not upgradeable however, so if you have a board with a serial number of A0036 or lower, you can't fix the first problem listed without a new chip. I don't think that the first problem is critical, so I don't see a reason for anyone to buy a new chip because of it. Just shout if you have any questions regarding the upgrade procedure or anything else!

Thank you for your trust and those fine words Adrian :) I will respond to your e-mail. /Niclas

The applications are jstest-gtk, available as a package for Ubuntu, and the test application that is included with Bodnar's js_serial_preload. I am compiling a kernel right now, to see if I can understand the HID drivers enough to find out why they don't find all 124 buttons per logical device. I have a theory as to why there are 2 extra analog axes too. I think it is the POV hat, because I saw something in the kernel drivers that looked like it was code for exposing a POV hat as an X- and a Y-axis.

I have now tested several joystick applications under Ubuntu with my laptop. The results are the same as if you were to use two BU0836-boards. They both show up with the same name. Applications that rely only on the name will therefore have problems seeing both virtual joysticks on the board. However, a library similar to js_serial_preload.so will solve the problem. I have already tested that. I don't know if X-Plane uses the name to uniquely identify a joystick, because I couldn't get it running due to the GPU being too low-end and old. I was hoping to at least get it to boot and show the menu so I could test joystick mappings, but no. When I run some joystick testing tools they report 10 axes and 64 buttons on the first virtual joystick, and 7 axes and 64 buttons on the second virtual joystick. I don't know why Linux thinks that there are two extra axes on each virtual joystick, but I don't think it should be a problem either. The "64 buttons" are more worrying. Maybe the old Linux Joystick API doesn't support more than 64 buttons on a single joystick device.

1. The intention with the analog inputs on this board is not to make a control stick/yoke, but to have rotaries on panels and things like that. 1024 steps is more than enough for a panel knob. I haven't heard of anyone that have used the B256A13 for building primary flight controls. I have not got any feedback about the resolution being too low. 2. I will make a Debian installation on my old laptop and see if I can get the X-Plane demo installed.

1. The resolution is 10 bits or 1024 steps 2. It is a standard HID device, so it should absolutely work under Linux. I have not tested it though, so I can't give any guarantees. Do you know of any free Linux-application that I can use to test it? What distribution are you using?

After the last post I got e-mails from several people. Unfortunately not enough people to feel confident that I will get my investment back if I order another batch. I am now trying to make this new batch a reality anyway. If you are interested in this board, please see this post: http://forums.eagle.ru/showthread.php?p=1909972#post1909972

Current status I am now one order away from having sold 60 boards :) Apart from the two software bugs that was an issue for some people in the beginning, I have only heard of one board that has broken and the guy needed to get a new chip for it. I think that is a pretty good result, so I feel satisfied :) Right now the situation is this: I have one board left, and that will probably be sold in the next few days. After that, I have a confirmed order of two boards from one guy, plus some guys that have asked for quotes but haven't responded anymore after that. For me to order a new batch at the current rate of B256A13 orders, I need to be sure that I get four boards sold right away so that I get my investment returned. About half of the people that announce that they are interested doesn't follow through with the order, so that means I will need to have about eight orders to feel confident that I will get four boards sold when I have ordered the batch. In the beginning that was not a problem. Every time I announced a new batch I got response from about ten people. Now I think this community is starting to get a little bit saturated on joystick interfaces, so I don't think I will get eight orders in a short timeframe again. Those times have passed ;) The alternative would be to get the four orders pre-paid, so that I am sure that I will reach return-of-investment. To sum it up: To make a new batch reality, I will need payment in advance for four boards. After I have gotten that, it will take about three weeks for me to get the PCB's from the manufacturer and about another two weeks for the boards to reach your home. If you are interested, please type it in this thread so that others can see it. I know I wouldn't catch on something like this if I couldn't see that others are doing it as well. You may of course get your money back up until the time where I place the order to the manufacturer.

AJS 37 Viggen 3-axis ADI running with three of my DTS-boards:

Hi again, After the last post, I got more orders than I had expected. I now have more buyers than I have boards left. I am therefore writing here to check for more interest, to see if it would be possible to order another batch without economic loss. Please send me an e-mail as stated in this thread http://forums.eagle.ru/showthread.php?t=80201 if you are interested.

I have responded to your PM.

I am going to order a batch of DTS boards tomorrow. The question is, should I order some B256A13's as well? Right now I only have two boards left, and possibly got buyers for three boards, so I am one board short. I can't order one board however, because that will be a really expensive PCB. Anyone more interested?

All boards from the first batch are now soldered, programmed and calibrated. I will order the second batch tomorrow.

Need some ADI's up and running? ;-)

My advice is learn C++ while you still think learning a programming language is interesting. Once you have learned one language there is the risk that you will get comfortable and not as motivated to start learning another language, and C/C++ is the only language without any limitations as to what you can do with it (almost, you'll need assembly to write some OS kernel code). When you know C++, you will have no trouble writing C# or Java code. If you've only learned Java or C#, you'll have trouble writing C++ code. Among the people I have worked with and gone to school with, the only ones that are really good programmers are the ones that have learned a "real" language early on, like C/C++/Pascal. I have several theories as to why beginning with C/C++ makes a better programmer. One of them is that C/C++-compilers aren't often good at telling you what you have done wrong. While that is generally a bad thing, it forces you to develop a real understanding of the language. You can't just use trial and error, correcting the things the compiler tells you to correct, because the compiler error messages can't always be trusted. Another thing that makes C/C++ a good beginner language is that you'll develop an understanding of pointers and memory, and you're probably going to need it sooner or later. In fact, if you start programming for real you'll probably have to write some C/C++-code sooner or later, and that is where things are going really wrong for people that have only learned C# or Java. At the very least you'll have to interface C/C++-code some time, and it just get's so much easier if you know the language. Another reason as to why C++ is the best language to learn is that you'll maybe want to go to other hardware sometime. C# and all of .Net are intended as Windows only. You may say that you'll never start using Linux at home anyway. And that may be true for your desktop computer, but what if you would want to buy a Raspberry Pi or a Beaglebone Black to do some home cockpit-project? Java could of course be used instead. But let's say then that you want to use a tiny microcontroller, then you're restricted to pure C in many cases. To put it in a shorter way. If you want to start programming for real, I'd advise you to learn C/C++ first, beacuse chances are you'll never have the time or motivation to learn it if you learn another language first. If you know exactly what you want to do, and want to accomplish this project in the quickest way possible, chose whatever language is best suited for the task. In your particular case I'd suggest C#. Edit: Didn't notice this thread was over one year old! So, which road did you choose? :)

Unfortunately not. Maybe I'll build my own when (or if) I need to run an indicator like this at home. The museums that are going to buy my boards have got them from the air force.

Yes, the price is 1000 SEK so it's more than $100. The USD is very weak now compared to what I was used to when growing up. Back then it was about 10 SEK for 1 USD, now it's about 6.5 SEK for 1 USD. The boards have some expensive components and will take about 2-3 hrs each to assemble and calibrate.