Advanced Search

Why don't you use the VFDMOD user component instead? forum.linuxcnc.org/24-hal-components/387...modbus-rtu?start=150

I have had two commercial P&P machines over the years. First I got a fully functional Philips CSM84 (made by Yamaha) in 2007. It came from an operational contract manufacturing line who was highly motivated to sell as he had new machines coming in 2 weeks. It was VERY easy to program, and the manuals were quite good. But, it used mechanical centering jaws and was not quite accurate for fine-pitch parts.

In 2020 I got a Quad QSA30A from an auction in Texas. This was made by Samsung, but fitted with Quad electronic feeders and Quad alignment cameras. It is a LOT more complicated to program, but works amazingly better! The manuals were TERRIBLE, but I got hooked up with the guy who did training at the factory, and he was a HUGE help!

Anyway, one of the biggest parameters to consider is the number of feeders a machine can hold. And, remember that feeders for wider tape will take up several slots on the feeder rails. Second is availability of, and durability of, nozzles. Cheap plastic nozzles will wear pretty fast or break when there is a mis-pick. My first machine did not have a nozzle changer, I really didn't know what I was missing!

My first machine had spring-powered mechanical feeders that were too weak to peel some cover tapes. Feeder jams were extremely frequent. I tended to put C-clamps on the empty tapes to provide more pull to help the tapes advance and peel.
The electronic feeders on the new machine have plenty of pull to keep those tapes working.

Another think to look at is error recovery. My old machine would just stop after a few mis-picks of one part, and wait for you to fix it. The new machine detects a few mispicks and marks that part as skipped, then when it gets to the end of the program, it beeps, waits for you to fix the issue, and you press start and it places all the skipped parts.

If you go for a used, commercial machine, make sure it is a commonly-made model, and parts will be available for some time. My new machine was made around 2000 but parts are still VERY easily available for it.
Jon

Due to my inexpert testing I may still have a bit of a problem.

I am using a simple brushed motor spindle controlled by a relay.
When I hit pause the spindle-cw pin is going low but the spindle.0.on pin is remaining high so that the spindle is actually still running.

Should the spindle actually turn off.

I have included the appropriate files. (about line 148)

Alex

PS I have connected the spindle relay to the spindle-cw pin for now. That seems to work, though I don't know if it will generate other problems later.

Yup, most are still linked to 2.9.2.
Thank you.

More info....?

Try www.modbusdriver.com/diagslave.html (30days free trial period) to verify that your computer can actually communicate with VFD. That is the only package that runs under Linux,that I know of, may be other members know of better Modbus testing package for Linux.

If you use windows there a lot more choices. Some commercial software packages are easy to use but are not free, however, they all offer free trial period. (google is your friend)

Are the buttons connected to the appropriate QtPlasmaC external HAL pins?

linuxcnc.org/docs/2.9/html/plasma/qtplas...l#plasma:ext-hal-pin
 

Hi Endian,

Sorry I wasn't clear...what I meant is the LinuxCNC dies because lcec errors out on not being able to map the PDO (for example 0x6060) even though it was configured with SDOs but for some reason not visible until doing the "ethercat rescan".

Perhaps these issues are not related but would be strange coincidence that EoE is involved in both. I'm not sure anyone is actually using it with LinuxCNC.

Glad we got you fixed up.
A HAL pin for dialogs requiresa bit of a re thought of dialogs.
I will look intro this when life slows down a bit - not to long.

I don't have a patrion page, but a mars bar and espresso is a nice thought - Thank you for the offer. Maybe I will look into that in the future.

Chris

Could you post a sample long program to test with?

Hello, all I have suddenly been getting numerous errors when trying to run my table. I have tried all the fixes on radius to end of arc errors. I have a suspicion that the recent fusion update may be causing this. I reprocessed an old file I had, and it now also won't cut. is anyone else having issues. 

Hi all!

I am currently trying to get a Folinn BD600 vfd to work via Modbus (rs485) to work using MB2HAL (as VFDmod seems to be offline)

I integrated the MB2HAL component but when using the HAL Configuration Monitor, I only get 0s.
My computer has a RS485 IF included but I also tried using a RS485 USB converter, no differences.
My User is already in the dialout group.

Is there a way to test the Modbus connection to the VFD on this computer without using linuxcnc? 

I also added my Hal and ini files. Any major findings?


 

linuxcnc.org/downloads/
under the heading "LinuxCNC Packages"
most of the links for v2.9.3 actually link to v2.9.2

don't know whom this concerns...

@Rda,

I agree with you. And that was the reason to use Thermal Dynamics high def for me.

But how do you send the commands? I would assume that json would be pretty good at sending commands too? So you read states/parameters from the received json and you could also send json messages to change speeds, set states etc.?
Yes, the transfer protocol for recieve and send is in json format. It's just a text string, but then uses a json layout.

EDIT: Sorry if this is confusing but on the server side you would have all the commands etc. mapped.
Yes indeed. You need at least a map or a list. In json we can use an array of Id and Value.
For example you send id="load" value="/home/gcode.ngc"
Or just send the id="run" value=""
Or send multiple id's and value's in one message, i have to test if we can send 30.000 lines of gcode in one message.

Then the client could send json messages to ask the server to do what it can.
Yes. The client can send id="stop value="" or the client can ask to load a mdi command like id="mdi" value="G01 X0 Y0 Z0"
And so on.
But everything from and to the server has to be mapped.

I am now trying out webassembly. This is c++ code compiled into html and java to run on a webbrowser.

Edit, my previous post : i could connect the websocket server to the linuxcnc nml's server in c++.
This is likely not the case, as we can better create a linuxcnc nml client that connects to the linuxcnc nml service or server.
Then we make a bridge from the linuxcnc nml client to a websocket server. This then should work excellent.
We already have the nml mapped, so we know all the command and status channels. It just a matter of coding it to the finish line.







 

Hello,
Something is wrong. If Qtdragon is running for a while, the retract function does not work for Z measurement or tool measurement. Z simply remains on the button in the triggered state. If Linuxcnc is restarted, it works again for a short while. Maybe there is a connection to the program length of the program being processed, approx. >50,000 steps. Qtdragon also feels a bit slow. BS and Linux CNC Version: 2.10.0-pre0-3685-gcbfe160071 up to date. I noticed the effect with the last version, but I suspected my wiring, but that was not confirmed.

Best regards, Frank.