ColorCNC Colorlight 5A-75E/5A-75B as FPGA controller board

The board has some unused space in flash memory and it's available through etherbone access or at compile time by modification of building script.

The answer is that mesa boards use self-designed CPU that is programmed in assembler and its own toolchain to write such a programs. Also mesa boards integrate a hardware ethernet MAC. In contrary to mesa, litex implements UDP/IP stack right in the FPGA. Such a concept is dictated by schematics of existing colorlight boards that connect ethernet physics directly to FPGA chip. Therefore, we __have__ to have a MAC and UDP/IP level in FPGA design. Making litexcnc project compatible with mesa protocol is possible, because out-of-the-box litex offers several risc-v based cpus with some examples. That lowers entry threshold in world of system-on-chips. But except of implementation of flexible computer-independent safety featurs I don't see any reason to go in this direction

I see the logical continuation is to create own FPGA board and use ethercat protocol instead of udp/ip

Some previews of a break-out board for HUB-75.

FRONT:

BACK

I would love to put the LinuxCNC logo on it, but I'm wondering whether that would be allowed...

TOLP2 wrote:

I am interested. I saw in th downloads that the PDF with the schematics something went wrong with the export. Each page is shifted a little bit downwards in the margins, so the last pages are mostly blank.

Would be wonderful to have a sort of HAT for the 5A-75E with for example:
- 6 stepgens (with DIR, STEP, ENA and ALARM (input)) = >24 pins
- 6 PWM with 0 - 10 Volt ouput
- some GPIO - in (for limit switches, button encoders, etc). How many?
- some GPIO - out - how many do we need?
- 1 or two expansion slots for future developments (like MPG, LCD or sensors?

I finished the split between stepgen and pos2vel-converter. I will push that tonight, as I need to add the example how to connect is in the HAL. Runs smoother then ever before.

 

Sorry for the late reply.
The project is not yet finished. I need to make more improvements on the board.

The board has 5 analog outputs (+/-10V) or 5 steps / direction (by changing the jumper) for this purpose I use AM26LV32.

There are 24 isolated inputs (Multiplexer input)
and 16 isolated outputs with LED indication.

One RS485 input/output

besriworld wrote:

No need for apologies, because a late reply is always better then no reply. Most of us, including me, do this in their free time if they have time to spare. So we have to appreciate the willingness of people spending their valuable time on these projects.

Your board looks very promising. Though I was trying to source some of the components here (NL), which turn out to be either expensive or difficult to get. Another possibility would be that I'm not looking at the right places. 

I hope with a board of this size the proto-typing phase will go effortless (i.e. not too many bugs in the design). How are you going to do that and keep the costs acceptable for a hobby project?

If allowed, I would take some inspiration from your schematics for my family of BOBs for HUB75. The license on these boards will be CC BY-NC-SA, which means that users:​​​​​​

• Share — copy and redistribute the material in any medium or format
• Adapt — remix, transform, and build upon the material

As long as:

• Attribution — You must give  appropriate credit , provide a link to the license, and  indicate if changes were made . You may do so in any reasonable manner, but not in any way that suggests the licensor endorses you or your use
• NonCommercial — You may not use the material for  commercial purposes . Of course you can use the machine you created or modified using these boards for commercial purposes. 
• ShareAlike — If you remix, transform, or build upon the material, you must distribute your contributions under the  same license  as the original.

The repositories for the HUB-75 boards is placed separately from the LitexCNC, as it may benefit other developments of the Colorlight boards as well. The location of the repository is:  github.com/Peter-van-Tol/HUB-75-boards

I bought the components before designing the board I order from Farnell and some parts from AliExpress. I spent many hours finding the right components that would be available and would fit into the project.
I was also inspired by various projects .
I don't have any experience with making PCB boards . I hope there are no mistakes and it works the first time The board is very big ... I don't know how it will work in real life I don't like the 5v power backup how it's designed... need to fix it. 
I am uploading the project from KiCad 

Thanks for the great work you do. This is a great project. My card has arrived 5A-75E . It would be great if I could run it on a virtual machine.


 

romanetz wrote:

I see the logical continuation is to create own FPGA board and use ethercat protocol instead of udp/ip
 

Hi
other than the compatibility with existing hardware, what are the technical avantages of ethercat vs etherbone?
Just out of curiosity, I've never studied ethercat internals. I suppose that it could be implemented in the same hw boards we are using now.

From my understanding ethercat has even lower latency than etherbone, but comes at the expense of not being routable. Imho latency is a non-issue with GBit anyway, one roundtrip should be around 1us for this application.

EtherCat has another advantage as well, consider what would happen if you would have two cards? With our implementation it is not guaranteed that the commands are processed at exactly the same time, as each card will have a different wall-clock. There might be small (micro-seconds) delay between the two cards. EtherCat on the other hand synchronizes each card, but this does require specialized hardware. As far as I understand, EtherCat cannot run just off every networkcard.

The advantage above makes the design of EtherCat very different then our approach. Multiple EtherCat-devices can be easily chained together and each will pick their packages from the stream. Each component is relatively simple, as they have only one function and maybe some GPIO. In our case all functionality is located at a single card which does all functions and interface with drivers.

Maybe a poor man's solution, but at least something which can be created by an average hobbyist.