Slowest UK MPCNC build, South Manchester, in the Shed on Lavington Avenue

Thanks Jeffe, re read those docs and yes will orient z to be pos = up, and then reapply right hand rule for the other axes.

printing a new xyz piece which has just been posted here yesterday. mine was cracking and i tried to superglue and anneal in an oven. Lets say that experiment wasn’t perfect.
The reprint of these bits is less time than going full burly but eventually i may as i would like to mill metals eventually.

Everything is always a million jobs just to get something done. I moved my 3d printer into the garage last night to appease the wife. waited till this morning to calibrate it. That literally took 1hr45mins because im using a modded anet a8 with a inductive sensor and skynet v2 by scou firmware which probably marlin can now do but I’ve never upgraded as it worked well enough. It has a 1 mm nozzle. I am hoping that strengthens these bits i’m reprinting, it certainly helps with print times. I lose a little resolution but i can clean the holes with a drill so wont affect function. anyway shall post some photos once i’ve got the new centre assembly back in action.

Here are a few snaps of todays progress. I reprinted the top and bottom xyz J piece and reassembled the centre assembly. In doing this i unearthed more stress broken old blue PLA at the xy stepper mounts so i have attempt to fix the cracks with loctite superglue and applied some downward force over night using a trusty orange cable tie. You can see now how many parts have failed due to poor quality printing and maybe me using sledge hammers to crack nuts. I am motoring along nicely now and can reprint the legs to burly if and when they fail. The top mount of the z failed a while back due to eccentric wobble caused by using a couple which didnt centre to the correct diameter of the threaded rod. That is why they are white. Anyway im approaching a point where i feel i could draw something which will be a massive milestone. Also sorry for the mess, I have too many projects and too little focus and organisation.

2 Likes

So looks like the overnight glue up with the woodfile compression helped. had alook at all the other falling side rollers and this was the worst in that it was in pieces not just on its way out. problem is it is easier to glue a properly broken piece as you have more surface area to get a good bond. Anywho ive done the small bit of work on cutting some thin spacers to help close the gap between the top and bottom of the rollers. After all the analysis of the failure here i think ryans design employs good use of triangles. For example the belt once tightened on the stepper provided a third central downward force on the pipe which complements the two bolts at the front. I think over tightening without spacers was not helping the situation where the lower part was cracking.
To help this i cut some shims out of hdpe (milk bottle) and this may help a little to spread the force so that it is not all at the plastic around the bolts. Will post more pics later from my mobile as it takes me a while to resize them for web viewing.

Just an fyi, those pieces should have a gap between the top and bottom part. 4 or 5 mm I believe.

@barry99705 I’m just trying to prevent cracking again and think that filling that gap should help spread load around the hole but i could be wrong. tbh i am just trying this as the original design didnt work for me but this might not either. A reprint may be in order but i am trying to rescue my old prints with superglue, perseverance and patience. They take a considerable amount of time to print so I’d rather try remedial fixes for now.

Don’t crank so hard just firm not power tool or Hulk tight

1 Like

@timonjkl yep, I don’t use power tools on this now, I also try to temper my hulk mode.

1 Like

I’m still running the design before this one. Used hot wood screws to fix it when it finally cracked. :rofl:

1 Like

Hey @jeffeb3 the magic number for my steps is 4872 but 4500 ballpark not too bad.

How did i test this, well…

at first i considered cable tying a vernier to the z but could get a nice orthogonal angle to the base and didn’t want to have any error so…

I found a something to use as a spacer (a roll of brown packing tape) measured its width approx 51mm

I then jogged the z axis without a bit in the router to touch the base and then jogged it up to where i could slip the masking tape roll between the collet and the base. took the measurement did the math and came out with 4925.5

I used the M92 to set z to 4925.5 and tried again. then i noticed that it wasn’t right and i was pretty sure i was right with the math. Anyway when i was trying to lift the Z the stepper motor was clearly losing steps. It occurred to me that the could be the source of the error. i disassemble the sliding part from the main xyz section and looked at my threaded rod. it wasn’t anywhere near central in the hole causing a lot of friction. with a quick tappy tap tap tap on the two plastic spacers (one of which holds the hex nut to trap the thread) i managed to centre it nicely in the hole.

I reassembled it, did a similar attempt and found that the stepper still loses step when retracting from right off the base with no bit in the collet. I disassembled again took the thread out and greased with some oil to remove more friction, reassembled. I still am losing steps.

The i realised i am quite far down and with a bit in the collet i will never be plunging the z that low so i changed tack and when to a two tape system. Armed with masking and packing tape rolls which have a difference of 23mm in height I found the readout on the screen when jogging the Z to go from -236 to -96 and so 140m change. Out of the box the software(firmware) default is 800 steps per mm when you do a M503 for the Z axis.

so 140/23 = 6.09 approx and 800 * 6.09 = 4872. put the new steps per mill in via
M92 Z4872
on pronterface serial via my pc. (as a temporary measure since the eeprom is disabled so i will reflash with this number now i am sure)

i think this is a livable setup as i think the resistance causing the skipping is when the Z axis is lower than it should ever be when there is a bit in the collet. Time will tell and it wont be long before i probably upgrade it to the lead screw since my new prints have the recess for the brass fitting.

anyway ive set up those axes to respect the +z is up and the others to follow the RHR. seems like the next step is clear the cutting area of all the crap and do a crown print.

did some tidy up and cable tied a pen to the side of the router. downloaded estlcam and repetier host and began drawing crowns. havent done a good one yet, what seems to be bad is that between each travel the Z is getting lower and lower meaning the line is getting darker and darker on each engrave section. Either i am losing steps on the z up or my gcode is wrong and its doing some wierd relative movement on the Z but i think its most likely losing steps due to friction. Very much considering doing the lead screw upgrade.

Is it possible that i just require more voltage. I’ve read that you can run the drv8255 steppers i have at 36v max

reading the wiki it is possible to run the ramps at 24v by removing d1 diode and powering the mega separately via the 2.1mm barrel jack. checking you dont have a cheaper ramps with 16v caps. Dont power the heated bed (dont need that for cutting) as it has a fuse which isnt rated above 16v.

https://reprap.org/wiki/RAMPS_24v

this all seems overkill like maybe i need to go back and test the drv boards are calibrated right it was a long time ago i did that (if indeed i ever did).

Also i may be able to push the voltage a little higher on the 12v supply and it be enough.

EDIT before i try all that i used the search in the forum and found VmaxZ could be too high as there was a thread called something like “Z axis losing steps” which sounded just like my issue in that i don’t lose steps when manually jogging but as soon as i run the output of the estlcam gcode of the crown it does. Anyway i am flashing the mega again, changed my Z steps to default to 4872 and the VmaxZ from 30 down to 4. Might be a big jump down but does the Z axis really have to move that quickly?

24V isn’t needed, unless you are using very high speeds.

You do need to adjust the potentiometers on the drivers. The default current will be wildly wrong.

Have you tried the test crown? That gcode is very conservative and works on any (working) machine.

Success, i now have a couple of nice looking crowns.

so the VmaxZ feedrate drop from 30 to 4 flash didnt seem to help entirely but it led me to the problem. I looked at the Gcode in Repetier host and it was doing the plunges at F180 i changed all these to F30 using search and replace and this gave me the first reasonable crown. Then i went back to Estlcam to see why my feedrates were high. It said i had 3mm/s as the F(z) like in Ryan’s tutorial but this is producing plunges of 180 so i did some math and turned it down to 0.5mm/s

1 Like

did some cutting, made a useful thing but not what i intended.

First thing i tried to make was an enclosure for my new hestia pi opensource thermostat. (annoyingly i have a boiler in the uk that has two wires with control a single heater where as the hestia pi is design to use a single wire and the live feeding the boiler so i have to do some more hackery to get it to work with the hestia pi)

ANYWAY. i had my Mckeller router cutting at full 230v and it cut fine into some pretty hard wood but my god it is noisy. i am about 10 metres from the house in a concrete building with a tin roof and i can hear it from within the house through the double glazing. Anyway I had abort the full cut due to time and lateness of night out of an abundance of compassion for my neighbours. Also it would have been keeping our 4 and 2 year old up.

The mckeller at full Volts give a no load speed of 31000rpm which was more than capable of cutting the wood at the speeds set out in ryans estlcam tutorial.

In the end with the hestia pi i 3d printed the case as it has vents on the side which presumably increase the longevity of the raspberry pi zero within.

The useful print i did was for a spacer which was to mount my guitar back the the wall. Its hanger bracket had fallen out due to use of tiny wall plugs which were barely screwed into the plaster.

I used fusion 360 to make the sketch dxf and then estlcam to make the gcode mostly using the automatic cut and holding tab options but double checking and ensuring that the depth of cut was deeper than my wood. I elevated the wood using a couple of similar thickness offcuts and clamped at the edges under the belts.

I ran the router using an SCR (dimmer switch) to drop the voltage and hence the speed so that i couldnt hear it in the house and then performed the cut. It did the three 120 degree screw holes perfectly but when it came to cut the 80mm diameter outer cut it lost steps and what started as a circle quickly became and ellipse due to my reduction of cutting speed to far with no complementary reduction of feed. My guess is that as it cut across the grain it lost more step which is why i didn’t just get a spiral.

It is becoming clear to me i need to find the sweet spot for pinewood feeds and speeds and i need a more sound damping enclosure.

So anyway what was my useful object. well i took that damaged spacer, cut it out on my bandsaw and used it a drilling template which protected my plaster work which id repaired recently and ensured the holes into the wall were orthogonal.

Pretty sure this is a useful machine already but its going to take a bit more work before i can use it for long periods.

the nagware of estlcam (stick) and the bloody usefulness of it (carrot) has got me looking at my deep pockets for the 49 euros (whatever that is in GBP)

big thanks to everyone so far who’ve helped me.

1 Like

I have a homemade thermostat. It was the first project in my “Evil House” things. My furnace needs two wires to control it, but it is either open or closed. I just wired it to a relay. It has some kind of 24VAC on it, but I couldn’t figure out how the other thermostats get power off of it. I ended up just pulling through some more cables and powering the thermostat with 5VDC from a wall wart.

I need to redo it, actually. The oled screeb I used has some burn in, so it is too hard to read (and generally doesn’t look as cool anymore).

Oddly enough thermostats run off 24vac. :wink:

Yeah, but…with a two wire system, drawing any current requires closing the circuit…which turns the furnace on.

I don’t know how the new thermostats that “run” off old 2 wire circuits work either. Perhaps they “leech” enough power when the furnace is on to charge a battery, and run off that the rest of the time? Or perhaps it’s possible to use a current limiting resistor to limit the current draw to a value too low to actuate the solenoid in the furnace?

I have a homebrew thermostat too. Mine’s just a ESP8266, without a display. The UI is web based. I’m using a “wall wart” for power too.

I do have the wall wart, raspberry pi, and wifi router all on a UPS so short power surges / outages don’t affect it.

I wired mine in parallel with the old school bi-metallic thermostat, so I have a back up. I keep the old thermostat set to ~45*f so I don’t freeze pipes if the homebrew job fails. It’s been surprisingly reliable though.

I did have a problem with it freezing or something, but it didn’t take too long for me to correlate the failures with the use of my ham radio. Eventually determined that transmitting more than ~500 mW on the 40 meter ham band would induce failure every time. 8^) Twisting the ~6ft long wires from the wall wart and changing my antenna configuration a little solved the problem.

The old thermostats were just a bimetallic coil with a mercury switch. Then they went to battery powered ones. Now they’re at least 4 wire, so you have constant power.

I installed a programmable thermostat to replace the bimetal/mercury one. It didn’t need a battery. I am guessing it has enough resistance when the furnace is off to avoid the detection of closed by the thermostat. I also have some weird idea about them using the fact that it is AC to create a voltage from a single wire, but I can’t quite figure out if that is possible.

My house has these 3 other people in it that will tell me if the thermostat is broken. It has only failed once, and it took me a few mins to put the programmable one back on while I replaced the original arduino yun with a esp8266. Wiring another one is serial, especially in the basement is a good idea.

I have the bimetal one in my shed, hanging by a nail. I like to adjust it when people visit and see how long it takes them to realize it doesn’t do anything. It does also act as a thermometer, between like 40F and 90F.

1 Like

I bet you could fancy it up and make it run an exhaust fan. Just don’t run 110 through a mercury switch, mercury vapor is kinda harmful.