Ok I’m still getting some problems with interference
while using the CNC machine. The last thing that happen while making my super chisel was
the spindle moving up in the z axis. In fact it moved in that direction, and luckily because
I had hard limits in place when that crashed into the top it just stopped automatically.
I also had another problem while cutting where the job just stopped mid cut… That could
have been from a loose connection somewhere along the wiring. Something else that happen, and I don’t know
if this is related or not but if I stopped a job mid cut the coordinates changed on my
software – and in particular the z axis always seemed to change and it’s never consistent.
I’ve obviously done something wrong with how I’ve wired this up so I think I need to investigate
this and possibly redesign how this is put together. The first thing I should mention is that…
I don’t want this video to become over-complicated – and I can already admit I hadn’t filmed
everything, so what I’m going to do is just talk about where I think I’ve made mistakes,
the things I’ve changed and how I think these changes will improve the machines performance. The first is that I used braided earthing
strap to shield cheaper cables to the from the VFD. The problem with this was that the
metal braided shield is not insulated from other metal contacts, and it would be with
SY cable which is what I used in the end. So I’m going to have to replace all of these. I think I’ve also created problem areas
where I’ve tried to expand the use of the controller – so I could connect it to a
laser engraver. And to have the option to switch between this (the laser) and the VFD.
In reality it’s meant I’ve had to run wires a lot further than I would otherwise had done. I also don’t feel like I did a great job
with the aviation panel mounts, which I used between the controller and both the spindle
and steppers. Not only was my soldering skills poor, but the panel mounts made the continuity
of the cable shield more complicated to maintain. So this one here, really what I should have
done was use a gland and kept the wire intact passing through this point and then to my
VFD. I also think I may not have earthed the shielding
correctly – all of which should led back to a single star connection, which I had instead
branched. Everything is connected but in reality something may be looping in ways which may
not seem apartment. I’ve also over-used power supply units – and
cluttered the enclosure with wires. In retrospect, my hard limits switches, and homing proximity
sensor, steppers motors and controller could be run from the 24v PSU – which would have
been a lot easier to wire up. The VFD I used was poorly documented and made
wiring more difficult. That said I worked out how to programme it – but basic instructions
were not well translated in the English manual. I doubt the VFD had an internal EMC or EMI
filter, and I may have wired the external one I bought incorrectly. I also didn’t place
ferrite cores on the wire to the spindle and earth terminal on the spindle hadn’t been
connected to its housing. Moving on from the VFD there’s other areas
that I need to look at. One thing is the voltage dividers that I’ve used… I use voltage dividers
to convert the signal from the homing proximity sensors from 12v down to 5v, however this
method doesn’t isolate noise returning to the control board. Ok so, what I’m going to start doing is slowly
dismantling the cables and taking things apart so I can have better access to the core wiring,
and then I’m going to look at the power management first, and the go through the various stages
of re-fitting the VFD and re-wiring the homing switches and so on – and hopefully by the
end of it this will look a little bit neater and also work. On a manual for a different VFD whic his available
in the UK from RS tools there is one section here about wiring up to the motor. EMC guidelines.
So this is the bit I don’t get. To prevent radiated noise, motor cables should be put
in a metal conduit. Alternatively an armoured or shielded type motor cable should be used.
For effective suppression of noise emission the cable armour or shield must be grounded
on both ends to the motor and the inverter ground. I was told you should only ground
on one end so that’s confusing me. These connectors should be as short as possible. Motor cables
and signal lines of other control equipment should be at least 30cm apart. So there’s
lots of information in this actual manual, that I can interpret which is illustrating
lots of things I should think about. So this is the enclosure since I’ve done all
the re-wiring it looks a lot simpler and there’s a few new things inside. The first is I’ve
taken all the power supply units out and replaced them with a single 24v 400w unit. I’ve also
added a opto-isolator board which is connected between the controller and the terminal block
with runs off to the sensors. I’ve also done a lot of tidying up of the wiring. I think the first thing I’ll do is talk about
the VFD. I’ve taken off any shielding which I additionally added for example onto this
ribbon cable because I was worried that would might contact something and create some kind
of ground loop. Everything is pretty much the same in here although I ended up changing
the wires to thinner gauge ones for some of the signals. The main thing I did was re-wired
the power coming in and the ground I took it to the EMI filter terminal and then run
it to my terminal block. And the section here is the start connection. Here you can see the ferrite core which the
cable loops though. I also put ferrite cores on the main plug and the USB cable. I’ve removed the aviation panel mount here
and replaced it with a gland, so the cable and shielding is not interpreted . This runs
all the way to the spindle. The spindle is a little bit different now.
There’s two things that I’ve done. The the armour or shielding – this braided stuff in
the cable. I’ve actually let it extend out a little bit and clamped the aviation socket
to it. But it actually isn’t connected to the chassis of the spindle. Just compare it
to Nathans machine – which is connected. So clamping the braid onto this isn’t doing
what I though it was going to do. That said I have used the machine a couple times since
finishing it and it seems to be performing ok but only time will tell. The other thing
I did was, most spindles I’ve seen have four pins. One is written down as the earth but
there’s never anything connected to it. So what I’ve done in this case is actually wired
to a crimp which I’m putting through a mounting machine screw. And this connects the housing
to ground. If you do ground the housing of the spindle it will mean that the terminal
you use to do your probing have to be placed in a specific way so the ground is then connected
to the bit and the signalling is essentially your touch plate. I think it is important
to connect the spindle because in my current set-up it’s obviously isolated because of
the mounting method and material. Obviously having a permanent connection makes
it a little more hard to move around. At the front of the machine you can see I’ve replaced
all the aviation panel mount with glands. So they are all through cables and nothing
interrupting that. Soldered at one end and physically connected to the controller or
the opto-isolator at the other end. You can see where I got rid of the switches for the
laser here. And this is where I’ll plug in an extractor. The only thing I’ve kept the
same is the probe wire. The reason I’ve done this is because I’m planning to use a couple
different probing methods and tools which I’ll talk about in another video. I also changed the wire which runs to the
hard limit switches around the machine, to a continuous wire and there’s a gland there
just to stop it from being pulled out. I really wanted to keep the display panel
at the front of the machine plus this selector switch or toggle switch to go from the PWM
signal on the controller to the potentiometer. I’ve created solder connection for a lot of
things instead of screw terminals where I can, to create better connections. I did swap
a lot of the cables to thinner stuff. I realised I didn’t need such chunky wiring as I was
using before. I just twisted it and made sure it was clear from everything else. And here
we got the isolation board. 24v comes in. I think there must be some kind of resistors
in there as when I take a reading from this side the voltage drops to 8v. So on it’s un-triggered
state it’s 0.45v and when the sensor is triggered is jumps to 5v. I also 3d printed this bit
at the bottom here to hold this in place. It’s not great but at least it won’t flip
over or get pulled. And I made a similar thing for the bucks voltage regulator which I’ve
connected to the power supply unit, and it’s drops the voltage down to 12v so I can turn
the LED’s on – which I put in a while ago. This is the 24v relay. One side is going to
the hard-limit switches while the other goes to the socket for the coolant although it
would be a lot better if it could function as a switch for an extractor to be honest
– and that could turn on and off, or I could just wire this up instead. So this is the main terminal block where the
power comes in, and then goes to the contractor and fuse and allow the power to go to the
rest of the system. While filming I did a bit more reading about this particular cable.
I’m not sure if it can be deemed as shielding but I think it will work if I do create a
physical connection between the metal and the metal underneath. The other thing which I should have done was
used galvanised steal as the backing plate here and not aluminium. Coz this can oxidase
and effects how electricity flows through it. The is the manual for the VFD I was considering
buying but I didn’t particularly want to spend £150 on it just yet but maybe I will still.
There’s a lot of really interesting things and worth a read. Here it says if using an
external EMC input filter it must be mounted under the drive and connected directly to
the line supply via an un-shielded cable. You can see how they’ve shielded the cables
– literally screwed directly to the backing plate by the looks of it. The a cable must
be connected to earth on both ends, the shielding must be continuous. So that’s essentially what I’ve learnt,
and the changes I’ve made to the electronics of the CNC machine and it’s controller. I’ve
been cutting stuff out, on and off now for several weeks, and I’ve not noticed any
unusual behaviour. So it seems that the changes I’ve made to the controller have made a
difference – but I still need to fine-tune the steps, acceleration and nut-block tensioning
for the x, y a z axis, which is what I’ll be looking at in the next video.
Thanks again for watching and you’ll catch me then.

#33 CNC Controller Rebuild #33 / All the things I wish I knew
Tagged on:                                                                                                                                                                 

21 thoughts on “#33 CNC Controller Rebuild #33 / All the things I wish I knew

  • January 5, 2019 at 11:25 pm
    Permalink

    @Educating Savvas, Love the vids and have watched 90% of them! Keep up the great work! I am still waiting for your detailed wiring diagram that you were going to send me for the relay switch that I have commented about on the other video (https://youtu.be/ZUavvKD3tzE). 😉

    Reply
  • January 5, 2019 at 11:55 pm
    Permalink

    Interesting stuff…

    Reply
  • January 6, 2019 at 12:00 am
    Permalink

    That ground the cable at both ends is new to me too. Maybe they expect that the motor is isolated like your is? I'd like to read what an expert has to say on this.

    It sure seems like you did a ton of work before knowing what exactly was causing the problem, I hope you got it. The wire cabinet looks a lot better now with less in it but sometimes you want separate power supplies to reduce electrical noise between devices. That optocouple board might be better suited to being mounted on the CNC and not in the cabinet if you think that's where the noise is being picked up, I don't think it is but… I always shuttered when you referred to those connectors as "aviation", what they really are is amateur radio microphone connectors. Aviation connectors, also known as bulkhead connectors or firewall connectors have potted pins and are sealed to reduce the chance of air moving through them, you know like keeping smoke out of the cabin, that's why real ones are really expensive. Anyway looks much better now, and your wiring skills have improved greatly. Keep it up!

    Reply
  • January 6, 2019 at 12:43 am
    Permalink

    star point shielding on your signal level stuff would be the most important as that would almost nullify a nukes emp. The spindle driver and output cable would be better out of the box, as that's where most emi would be coming from , grounding the spindle case would only be good for safety reasons, you can make a battery powered probe for manual z . Glad you put optos in, apart from supply noise isolation, if for some reason the divider losses ground, the output of it will go to the voltage of the input at whatever ma the top resistor will allow straight into the 5v system, blowing stuff up untill something fails open. Stepper cabling isn't that emissive if twisted, but can cause glitches if run paarraallleell with signals. Is that a double L or not? Haalloommii, can never spell that right either

    Reply
  • January 6, 2019 at 3:20 am
    Permalink

    How are the chinese linear rails doing?

    Reply
  • January 6, 2019 at 3:26 am
    Permalink

    I noticed the occasional issue with my Bcnc – GRBL setup as well. When it was first setup, I witnessed several instances where it plunged a drill bit into a material. This, it was supposed to. Then it moved sideways, with the drill bit fully submerged in the material. SNAP. Breaks the bit, not before loads of noise in complaint. I don't believe there was any issue with my G-Code, I believe it was the controller. It seemed to just forget to pull back up again.

    I've witnessed the other occasional Gaffe, though I can't say as I recall exactly what those issues were off the top of my head. I am wondering if GRBL does not check its queue against the list of G-code commands in GRBL. If I could, I would have it load up a buffer of the next 5-10 seconds worth of commands, and do a double check before doing those commands.

    Anyways. Probably not related to the problems you are experiencing. but, curious I am not the only one noticing limitations or issues….

    Reply
  • January 6, 2019 at 11:05 am
    Permalink

    I suspect just adding the optos to the inputs of your controller (opto board CLOSE to your controller board) from your limits etc would have fixed all your issues. Everything else is just working towards reducing the noise on those lines which goes away as soon as you opto isolate those lines. It's unfortunate that the use of optos isn't preached at a religious level all over every CNC forum. You can't go running high input impedance logic level lines any significant distance if there is high voltage switching etc nearby. Unfortunately the market is full of "amateur" designed controller PCB which kind of, sort of, often work until they don't! The little GRBL Arduino shield is typical of these, close to useless without external opto isolation on all inputs if you have any high power electronics around.

    Reply
  • January 7, 2019 at 2:16 am
    Permalink

    I'd move the VFD and power supply out. Put the VFD in a separate shielded metal enclosure (e.g. https://www.hammfg.com/part/CHKO12106) and add ferrite cores on spindle out wires. Optical isolators on all inputs. Ground both ends of the spindle cable braid. Star point ground everything in the enclosures.

    Reply
  • January 7, 2019 at 7:14 pm
    Permalink

    It's interesting you mention the aviation connectors and the Z axis control issues as I had EXACTLY the same issue…I had some really nice connectors with spring clamp connectors in them so it wasn't even a soldering issue it was just too many antennas for interference I believe. One note additionally is in automation controls we only ground the shield at the sensor end never both and while your setup is a bit different from a normal sensor setup it should follow similar rules. Also USB cables are a general source of issues in my experience as well.

    Reply
  • January 8, 2019 at 12:04 pm
    Permalink

    Man that high frequency spindle driver should be as far away from other electronics as possible, as well as shielded in a faraday box to be sure..

    Reply
  • January 8, 2019 at 5:27 pm
    Permalink

    I don't see any drain wires coming out of your cables going from the drives to the steppers. I see what looks like green ground cables all tied together (not the same thing). If you're not using shielded cables then I would say there's your problem.
    Also, you should be using shielded cables on your limit switches because stray electrical noise can false trigger these devices.

    Reply
  • January 8, 2019 at 9:00 pm
    Permalink

    My limit switches would get tripped by noise, until I added capacitors at each switch and at the controller for each switch (3×2 + 3 total)and resistors at the controller. I was getting false tripping in the middle of long jobs. I haven't had a problem since. I don't have grounded shields on the wires.

    Reply
  • January 9, 2019 at 2:51 pm
    Permalink

    Mechanical relays will trigger an EMF spike when deenergized. This be remedied with a small diode across the coil leads. You can look up online if your unsure how to install them, they are directional. When I use metal shielding I ground both ends. If that interferes with your metal touch plate simply insulate the bottom of it. I avoid contact type limit switches due to switch bounce but it's unlikely to cause other issues since their is very little current on them so no real arcing to worry about. I use those panel connectors on my cnc mill but I do use actual shielded wire,.

    Reply
  • January 17, 2019 at 11:06 pm
    Permalink

    Try a UPS for your electronics. I forgot, this is what I did, and the vast majority of my GRBL problems evaporated.

    Reply
  • January 19, 2019 at 6:30 am
    Permalink

    I change my controller and have no more issues. I had the same controller you have

    Reply
  • February 14, 2019 at 3:47 pm
    Permalink

    knowing the Steppers, 6wire unipolar, … 3A 400-600 oz/in. ??? … 12-24 v on a potientiometer, and a power supply rated for the sum of your motors, +2A…. so my 4axis TB6560 controller tech says… methinks a 24v 15A
    p/s is my next $$.

    Reply
  • March 1, 2019 at 2:54 pm
    Permalink

    Just recently found an interesting video about types of lasers

    https://www.youtube.com/watch?v=3JqLa4yRwuk

    Reply
  • March 6, 2019 at 8:13 am
    Permalink

    It seems I'm right behind you. From my research grounding is the main issue for me. Like grounding shielded cable only on one side. Not Daisy chaining them. I have basically I have same vfd same drivers and motors. My limit switches caused all my problems. I threw them away.

    Reply
  • March 7, 2019 at 2:03 pm
    Permalink

    Hey guys – wanted just to share some interesting thing.

    Do you know that Endurance make a diode laser attachment for any 3D printers and CNC machines.

    I would suggest you to check it out in google something like google )Endurance laser attachments)

    Reply
  • June 25, 2019 at 8:08 pm
    Permalink

    Your VFD is a cheap copy of another Chinese brand. Everything that could be removed to save money, have been removed.

    Reply
  • November 20, 2019 at 9:17 am
    Permalink

    I'm loving your videos but rule number 1 in root cause analysis is…Change one thing at a time and test. Maybe you did this but just compressed the video. It would be interesting to know what actually fixed the initial problem.

    Reply

Leave a Reply

Your email address will not be published. Required fields are marked *