Advanced Search

Ups, I have just seen the correct IO standard can be configured /edit in the plugin of the blink module in the riocore software. Hence, it is not necessary to do it manually in the pins.cst file.

have tried to get linuxcnc from buildbot 
... any such change will see no buildbot to help, or that help is delayed ... 
so I have to build LinuxCNC from source

   


qtvcp_panel_preferences already work perfectly here 

 

I did not know about the LED on the FPGA board. Now it makes sense.

However, as the bank 3 of the FPGA is powered by 1.8V, the IO_TYPE should not be LVCMO33, which is in conflict with the power of bank 3 causing an error during place & route.

I have not tried the toolchain of the riocore, as Compile does not work with my Windows setup. I may look into it later as I can live with the Gowin IDE.

File pins.cst:

// ### blink3 ###
IO_LOC "PINOUT_BLINK3_LED" 10;
IO_PORT "PINOUT_BLINK3_LED" IO_TYPE=LVCMOS33 DRIVE=8;

IO_TYPE=LVCMOS18 solves the problem.

Pins of bank 3 are 3 to 11, 13 to 16 and 79 - 88 (some of the pins are used for the right sight of the board)

Just for the case that somebody else uses the IDE....

If you create a component (called comp, with a float pin named length) in your script, you can use this for the box:

Box(comp, 60, -30, -30, 'length', 30, 30)

or with no component in the script this should work:
Box(60, -30, -30, 'comp_name.length', 30, 30)

All the other entries can use a string to force evaluation of a HAL pin too.

Chris

Spammer and their post removed.  

I have decided to switch to LinuxCNC from EdingCNC. This weekend I got a Mesa 7i76s (and 7i73 for the pendant); I already had the inductive proximity sensors and Teknic Clearpath motors from before, but I have a lot of reading up to do!

I asked ChatGPT to help me figure out a new grounding scheme, because all of a sudden I need two separate power supplies for the sensors and board, and boy did it come through for me! Where would I be without AI?

greg23_78 wrote:

i don't have the electrical cabinet with the robot
i'd like to start with a second-hand 400 to 240V high-power three-phase autotransformer because it's “quite easy to find”. i'd like to share the phases and neutral to supply the drivers with 110V single-phase power.
i have 3xbig motor and 3xsmall motor so 1 phase for 1 big and 1 small.
after some research i managed to find some data on robot or control electrical power.
robot KR16(more recent than my robot but quite similar) = Installed motor capacity 8.8 KW
cabinet : Rated power input depending on the drive class: 7.3 -- 13.5 kVA.
I don't know which transformer power I should look for 12 KVA minimum? 400V/18.1A - 230V/31.4A

assuming that large motors will be at a maximum of 50%-60% of their maxspeed (110V instead of 180V)

KR16 powerKW
 

KRC2 power KW
 

 

Note that with Pico Systems servo amps, the supply MUST be isolated from the mains, an autotransformer must not be used.  This is because there is no isolation barrier between the control logic and the main transistors to keep cost down.
Jon

I'm running Ubuntu 10.04 on my Hardinge CHNC lathe with no problems on an old PC. Been running it for years and don't need to upgrade what is not broken.

JT

tommylight wrote:

In Europe you can get toroidal transformes for any secondary voltage, that is the easy part, the 5.4A rating is most probably the long term current the motor can withstand, peak current will be much higher so the drives are usually 20A or 30A rated, hence the power supply must be rated accordingly, in your case at a guess at least 60A transformer and rectifier. Or have 2 of the big motors on one power supply and 4 smaller on another power supply, at 30A to 40A each are easier to manage and find and might even be cheaper. You might easily make do with much less as it is a small weight, and use moderate accelerations.
Does the control box have power info on it? Should be in KW,
 

Yes, two toroid supplies might be the best solution.  Note that these transformers can be overloaded for short intervals, and that the PWM servo amps are switching power supplies, so that power input is just a little over power out, so they can deliver large output currents when the motor is moving slowly, while drawing MUCH smaller current from the DC supply.
Jon

greg23_78 wrote:

ok, so the 160 V on your website corresponds to a peak voltage?
For the input voltage, for reasons of economy, a DC power supply to manage the 6 motors is likely to be expensive.
What's more, being in Europe, I have to manage with 240/400 volts. It's easier to find an auto-transformer to output 110V and add a 3-phase bridge rectifier with capacitor.
is it possible to supply your amps with RMS voltage?

 

Sorry, I really ought to change that.  The amps won't blow up at that voltage, but you have a high probability of having nuisance trips of the overcurrent fault logic.  So, I really recommend 122 V as the highest DC supply voltage.
for resolvers, is there a way to manage robot position data with batteries (like kuka) without continuously powering the computer with linuxcnc open? because manually homing robot axes could take a long time ..

For the input voltage, for reasons of economy, a DC power supply to manage the 6 motors is likely to be expensive.

A simple power supply can be built with a transformer, rectifier and capacitor bank.  In the US, there are commonly-used "step down transformers" that have split primary and secondary windings.  These can be strapped to give various voltages, and are used in large machine tool controls to provide 120 and 240 V from 480 V mains. These can often be found at scrapyards for the salvage metal price. To get 122 V DC with a capacitor-input filter, you would want about 86 VAC from the transformer.  Possibly this can be arranged with a buck-boost autotransformer before the isolation transformer.  If you can find suitable tranformer(s) surplus, this can be done quite cheaply.  The amps definitely need isolation from the mains, and DC input.

for resolvers, is there a way to manage robot position data with batteries (like kuka) without continuously powering the computer with linuxcnc open? because manually homing robot axes could take a long time ...

Our resolver converters have no way to provide absolute position via quadrature output.  That's one of the reasons why some vendors use exotic absolute encoders.  Also, most robots have gearing on the joints, so the motors make many turns for the full motion range.  Fanuc serial encoders, for instance, report # of full turns from the home position, but need a backup battery..
Homing should not be a huge issue, especially if the robot is placed in a good pre-home stance when LinuxCNC is shut down. LinuxCNC can do a sequenced home procedure where each joint is homed in a specified order to be safe.

Jon
 

Hi Mark,
Nice to see that the bearings look okay.
In the 70's we had a 4 foot diameter spherical roller thrust bearing for a 150 tonne rotary regenerative pre-heater. It arrived with false Brinelling. SKF took it back, reground the races and fitted new rollers. It might be possible to do that with these bespoke spindles if there were damaged.
I re-read the docs for my spindle and realized I didn't include the attached page. It's not all that applicable to your machine except for the end float I think. I heated my bearing to 50 C at they explain but I made a silly mistake. I used my paint stripper gun on low and almost immediately the grease started to smoke. It seems that the grease heats up much faster than the steel. So I repacked it and put it in my Siemens domestic oven. It has a nice thermostat that goes low enough to proof dough. It seems the key thing is to make sure that the races are all hard against the end float shoulders. And yes I put it into may Emmert pattern makers vise and hit it repeatedly with a rubber mallet. I've fitted it with wooden face plates.
It seems they set them up with a few micrometers float and are fussy it isn't larger. It certainly ran hot on first start even at the lowest speed. So I went through the whole stop, cool, start procedure at every speed. And it still got a bit hot at full speed until it bedded in later.
So I think the idea is the spindle runs hotter than the surrounding parts as the heat has to get to the fins behind the cover plate. That means the end float is eliminated and the bearings are pre-loaded. In you case you have  shims. I think I would make a similar tiny float when the races are seated and remember to rotate the shaft to settle the balls and cage. When these taper rollers or angular contact balls operate they nutate like small gyroscopes. The races must drive that nutation or at high speed the balls will scuff themselves and the races. I think possibly the tapered rollers in mine have too much gyroscopic moment to run at your higher speed, so the use balls for that?
If you go back you will see I had some issues with the cooling fan being loose. It very important that it blows down the fin slots. I made a bigger copy of you ring lamp. They work great, but be careful it doesn't block the cooling air coming down those slots
The labyrinth seal on my quill was full of muck from the swarf. So I agree that flood cooling is probably better.
Cheers,
David

Hi everyone

I'm new but I've been reading every scrap of info on this forum, raspberrypi.org, raspian.org and youtube on getting my Pi 5/Mesa 7i95T setup working.

I'm a bit rusty on Linux but it's coming back.

I ran though the full install a few times with the RPI5 download from the download page and got most things working - screen, network, apt updates, etc.

But I had trouble getting my 52PI m.2 NVME hat working with my drive (which is not a SATA)

I started looking through the journalctl boot files and saw quite a few bcm2835 and bcm2708 entries, so my first question shouldn't the drivers in question be bcm2712?

I see a few bcm2712 along with the others but I began to worry that I may have 'updated' my system to some older drivers for pi0/3/4 etc.

So I just downloaded:

rpi-5-debian-bookworm-6.6.54-rt39-arm64-ext4-2024-11-06-2037_qtpyvcp_stable_networkmanager.img.xz

I tried to burn it with the Raspberry image burner and it gives me an error:

Input file is not a valid disk image.
File size 6700000000 bytes is not a multiple of 512.

Did I download it incorrectly from Google Drive?
Is my 32Gb eNVME drive to small or an off brand?

Having read hundreds of posts on this WIKI over the last six weeks I'm blown away by time and effort everyone has put into this project.

Thanks for everything,

Richard

PCW wrote:

SPI should work though a Efinix specific top level file would need to be
made from the Xilinx source.  Efinix tools don't automatically
translate "Z" types into tristate pins so you have to export the signal
_and_ enable pins from the top level file. Note for the current SPI source
(GCSPI) The SPI clock must be routed to a FPGA clock pin.
 

Thanks for the quick response, This should get me going in the right direction. Currently I have only gone so far as to generate new .bin files when required for my 7i92t/7i96s cards.

PCW wrote:

Any package is OK
 

Awesome, glad to hear it.