i could also test the needle holder here for you with my 0.6mm wire...Ab. So. Lutely. Sharing designs is what got me to here too so I'll definitely share all of bits I've created. This is the first cutting version in 30 tries though and I still haven't located the bearing assembly that flew off last night so I'm just going to get some cuts under my belt before I think it isn't dangerously flawed. I was wearing eye protection when the bit flew off though, it may only be 5g but it has a sharp needle going at 10krpm, of course I am going to wear eye protection testing it!
In just a few short tests I am wondering if it needs more mass though. I've seen that bit called the "flywheel" and I wondered why a needle poking through paper would need more momentum than the motor would impart, but I see some places where the needle didn't go all the way through the bottom paper on some cuts so I'm wondering if making the needle attachment lighter is a bad thing instead of a vibration thing. I still have a lot of trial to do to know for sure, but my needle is 0.8mm instead of 0.6mm so maybe it would be better if I got some more suitable material.
ERC TimSav - Cheap DIY CNC Foamboard Cutter
- Thread starter Edwardchew
- Start date
dkj4linux
Elite member
Thanks for the kind words.Thanks for all the great detail, David, you've always been someone I follow here for your great technical expertise and excellent taste in operating systems. You always seem to be on the cutting edgep) of putting software and hardware together for our purposes, and this CNCjs on my desktop inspired by your photo last week is just one memento from that.
You've given me a lot to consider in your post. You're right about the patience though, I've been 3D printing for coming up on 8 years now and know very well the product of pushing for fast over quality and accuracy. I'm pleased to hear that the weight of the flywheel is not an going to be an issue, as I spent last night working out a way to add 1g to it, and comparing that to the weight of the 3d printed flywheels (4-5g total) and the motor bell at ~25g. I have definitely considered some sort of vacuum to de-crown the foam board, and I am also trying to figure out where I'm going to put this thing to use it, since my workbenches aren't big enough. hahah Now the real fun begins: Thing built, now what?
SInce you are an accomplished 3d-printer, I'd suggest the following as a possible plan of attack:
1) Take a look at adapting and printing a "slip-on flywheel" for your motor... maybe this one, or this one, etc. It requires a pretty dial-in printer and some careful measurement but is a quick print, once you figure out all the "fits and clearances". It more properly uses the motor bell for a "flywheeel", seems to run smoother at higher rpm, and seems less fiddly to balance out.
2) Consider printing Moebeast's impressive vacuum fitting from here... and cut the 3 sheets of DTFB for a vacuum pad, same as in the video. I posted the DXFs here.
Also, in your thermal images... and as you've already noted... the needle guide (printer nozzle, Mig tip, sports ball inflation needle, etc) heats up to a greater and faster degree than other bits of the system. And running higher rpms only aggravates the problem. PLA can work (that's all I've ever used) for your cutter but you'll need to take care -- anywhere metal is in direct contact with plastic -- to insure that the plastic isn't softened/melted/deformed by the friction heat. Most of our cutters use a wooden "plate" (or, heatsink fins and/or external fan) to screw the guide tip into... and you might consider reworking your guide "boom" to incorporate an insulator, or cooling, of some sort for the guide. Some also incorporate a "cotton wad" oiler surrounding the needle, just above the needle guide, to lightly lubricate the needle and guide and help control friction heating.
BTW I'm thinking I may "borrow", or lift from, your servo-driven Z-axis design for a project I'm considering. Thanks for sharing!
-- David
CapnBry
Elite member
Oh yeah I am definitely going to make one of those suction things once I get to full size. I had also seen your slip on flywheel concept and I would be all in on that, but I don't want to cut the shaft off this motor, it really is only standing in since I didn't want to buy anything. That changed today when I bought a IFLYRC F2212 920KV motor to replace this one off my motor glider. It has a tiny little shaft so I won't even have to cut it off and it will make the nozzle arm 10mm shorter at least.
Another day another design though. I was not happy with the ~11.2mm stroke which also meant I had a pretty big left-to-right movement in the needle on every stroke. My bearing was already pressed right up against the motor shaft so it couldn't get any smaller. To get around this I decided to place the bearing on top of the motor shaft with the screw inside, so add ~6mm of distance on top of the motor. That way I can target any stroke depth of 6mm or more. It was easier to do just throw this together with a traditional cylinder flywheel shape.
6mm stroke seemed nice while cutting it close to the minimum amount of stroke to cut DTFB / MakerFoam, but the motor shaft grub screw was too heavy for it to balance properly. I had to drop to a nylon screw with a few washers on it to get close, so I increased the stroke depth to 7mm and it is much closer now. It probably could go to 7.5mm and then be adjusted with a washer for weight, but this works pretty well. You can see the two ponies cut-- the dark one is with the old flywheel and the light is with the new one, same needle. The kerf (tearout?) is much less with a lower stroke. The difference is also noticeable on the nozzle.
After running 20 minutes and cutting a pony, the cut was actually registering in the same ballpark of temperature as the nozzle. Super cool and not even PLA troublesome. The bearing ends up a little hotter for some reason in this, but it is actually another bearing from the same broken motor so it may be damaged slightly? In any case it is fine temperature-wise still. The base of the motor is the hottest part at 50C, so maybe some tiny fins could be added to the design to add a little airflow to the motor.
I also found some 22ga nichrome wire (0.025" / 0.635mm) in one of my drawers and it works acceptably too. That's what's in the first photo. I definitely want to use the stiffer music wire, but I don't have that size on hand, and it is nice to see that nichrome can work in a pinch.
I think I'm zeroing in on the last bits of the design. I just need to integrate some wire management into the carriage, but I will wait until I have the 20mm extrusion to see where the wires lay.
Another day another design though. I was not happy with the ~11.2mm stroke which also meant I had a pretty big left-to-right movement in the needle on every stroke. My bearing was already pressed right up against the motor shaft so it couldn't get any smaller. To get around this I decided to place the bearing on top of the motor shaft with the screw inside, so add ~6mm of distance on top of the motor. That way I can target any stroke depth of 6mm or more. It was easier to do just throw this together with a traditional cylinder flywheel shape.
6mm stroke seemed nice while cutting it close to the minimum amount of stroke to cut DTFB / MakerFoam, but the motor shaft grub screw was too heavy for it to balance properly. I had to drop to a nylon screw with a few washers on it to get close, so I increased the stroke depth to 7mm and it is much closer now. It probably could go to 7.5mm and then be adjusted with a washer for weight, but this works pretty well. You can see the two ponies cut-- the dark one is with the old flywheel and the light is with the new one, same needle. The kerf (tearout?) is much less with a lower stroke. The difference is also noticeable on the nozzle.
After running 20 minutes and cutting a pony, the cut was actually registering in the same ballpark of temperature as the nozzle. Super cool and not even PLA troublesome. The bearing ends up a little hotter for some reason in this, but it is actually another bearing from the same broken motor so it may be damaged slightly? In any case it is fine temperature-wise still. The base of the motor is the hottest part at 50C, so maybe some tiny fins could be added to the design to add a little airflow to the motor.
I also found some 22ga nichrome wire (0.025" / 0.635mm) in one of my drawers and it works acceptably too. That's what's in the first photo. I definitely want to use the stiffer music wire, but I don't have that size on hand, and it is nice to see that nichrome can work in a pinch.
I think I'm zeroing in on the last bits of the design. I just need to integrate some wire management into the carriage, but I will wait until I have the 20mm extrusion to see where the wires lay.
epsilon
Active member
Well... just read through the whole thread. I think I'm sold. Only reason I haven't hit buy yet is because I'm leaving town on the 26th and doubt shipping will make it in time. Been wanting to do a CNC for a couple years now, finally got my 3d printer a while back and slowly learning the ins and outs of that. Now it's time to dive down this rabbit hole too! I love how affordable and how portable this thing is! With traveling for work I'm rarely confined to my own house and workbench like I am now. Soon as my filament guide finishes think i'll start working on the parts. Oh, and I LOVE that hot wire cutter... This could be a very deep rabbit hole...
Lance Nordby
Member
I think that the shipping will take quite a while, epsilon. It seemed like it was a few weeks from order to arrival for most people. I'm still waiting. You might as well order now on the 7th batch if there are any left.
An FYI for everyone. I ordered the extrusion from PD Tech. It was arrived quickly and was at a good price. Best of all they claimed it was 6061-T6 and judging by how clean it tapped it's believable. It didn't booger up like 5 series aluminum is wont to do and as they claim it should be stiffer than 5 series as well.
I had a bit of trouble sourcing a 5mm tap at a good price and a decent delivery time. I had to order it online and pick it up but strangely enough O'Reilly's was the quickest and cheapest option ($3.99, 2 days). Also strange is that it was made in Japan (I was expecting cheap Chinese but it's good quality).
An FYI for everyone. I ordered the extrusion from PD Tech. It was arrived quickly and was at a good price. Best of all they claimed it was 6061-T6 and judging by how clean it tapped it's believable. It didn't booger up like 5 series aluminum is wont to do and as they claim it should be stiffer than 5 series as well.
I had a bit of trouble sourcing a 5mm tap at a good price and a decent delivery time. I had to order it online and pick it up but strangely enough O'Reilly's was the quickest and cheapest option ($3.99, 2 days). Also strange is that it was made in Japan (I was expecting cheap Chinese but it's good quality).
epsilon
Active member
dehager, curious if you got this up and running and how the reading is output and/or factored in? Considering going this route versus buying a standalone tach. Though, i suppose having a handheld unit could be handy... idk anyway, just curious.
epsilon
Active member
Yeah, that's what i was afraid of... problem is i don't know where i'll be yet. Don't suppose you live in St. Louis? lol
Lance Nordby
Member
Nope, sorry.
Can someone tell me if there is extra belt length if I want to make the axis' a bit longer?
Can someone tell me if there is extra belt length if I want to make the axis' a bit longer?
oh yeah, theres definitely lots of extra belt slack.
CapnBry
Elite member
I've used their stuff as well and agree it feels stiffer than the openbuilds stuff. Two other major differences though. The corners of their profile is round instead of square. This isn't an issue in this incarnation, but if you're building a frame where the pieces butt against each other, the square edges are far more "self-aligning" since there's more length in contact. The other thing is that their v-slot slot is slightly different geometry. The pieces I made to fit tightly into the v-slot grooves didn't fit the PD Tech stuff. I'm not sure if the difference is in the grooves or the face, but something about it is very slightly different. V-wheels still roll in it just fine though, do the TimSav pieces fit tightly onto it?
CapnBry
Elite member
I'm getting close! Got my extrusion and did a new print of the carriage for the full sized v-wheels. Because the needle toolplate extended so far down, I also extended the carriage to support it through more of the length which also provides more lateral stability.
For the wiring, I used 4 conductor 22AWG LED wire which is nice because it all holds together and strapping it down sideways as it leaves the carriage and at the stepper end means it folds up nicely and stays out of the way as the carriage moves without any additional structure.
But wait, you're saying. Four wires go in, eight come out? What a mathematical failure. Really all we need down here is 12V/GND and two servo signals. The ESC has a 5V BEC on it, so simply power the ESC with 12V/GND, the 5V is fed back from the ESC signal connector into the servo pinheader, and it is jumpered (along with GND) over to the servo. Here's what it looks like before I covered it in hot glue and heatshrink, which is crazy secure. In retrospect I probably could have just wired ground directly to the JST going to the ESC and used two actual jumpers to join the 5Vs and GNDs on the 2x3 pinheader.
The wiring is also quite tidy at the CNC shield now. The servo / ESC connectors are 3 pin shrouds but the only wire coming out is the PWM signal. Single pin dupont connectors don't get a lot of grip on just one pin and tend to pull off easily in my experience, so having all 3 pins filled with the female sockets but no wire makes them really hold nicely. Also note my X and Z stepper drivers are populated, not X and Y. This is because the Y-step pin is used for the ESC signal.
My Y stepper is less than 1.5mm from the table, plenty of clearance, except the extrusion bows down a bit in the middle (either that or my table is slightly proud) so the stepper was just scratching the table when it moved through the middle. I tried flipping the extrusion over just in case it was bent, but same problem. I cut off 10cm from the 2040 which took up 0.2 or 0.3mm, but it wasn't enough to make me not worry about scratching the table over time. To combat this I added a V-wheel like the X end support that is just the right height so it doesn't touch at Y=0 or Y=760, but does make contact through the middle. It should give me a more consistent depth of cut. You probably can't see it in the overview here but it screws in through the loop in the Y belt going to the stepper. All the wiring is then bound up in some spiral wrap and after running gcode to move it back and forth 100x, it never once even came close to moving toward the work surface so I think we're good there.
I am still waiting on my motor and ESC (been in Orlando for 6 days now) so the wiring on top of the carriage is a bit of a mess. I don't want to cut my nice ESC wires to put the right length connectors on so they'll just be wrapped up and strapped to the back. The ESP8266 running esp-link has been working great for running GRBL Controller to send G-Code from my desktop, or using the web-based terminal or telnet to type gcode commands to test.
Just need to figure out a way to strap the thing down and then we'll get on to making some cuts.
For the wiring, I used 4 conductor 22AWG LED wire which is nice because it all holds together and strapping it down sideways as it leaves the carriage and at the stepper end means it folds up nicely and stays out of the way as the carriage moves without any additional structure.
But wait, you're saying. Four wires go in, eight come out? What a mathematical failure. Really all we need down here is 12V/GND and two servo signals. The ESC has a 5V BEC on it, so simply power the ESC with 12V/GND, the 5V is fed back from the ESC signal connector into the servo pinheader, and it is jumpered (along with GND) over to the servo. Here's what it looks like before I covered it in hot glue and heatshrink, which is crazy secure. In retrospect I probably could have just wired ground directly to the JST going to the ESC and used two actual jumpers to join the 5Vs and GNDs on the 2x3 pinheader.
The wiring is also quite tidy at the CNC shield now. The servo / ESC connectors are 3 pin shrouds but the only wire coming out is the PWM signal. Single pin dupont connectors don't get a lot of grip on just one pin and tend to pull off easily in my experience, so having all 3 pins filled with the female sockets but no wire makes them really hold nicely. Also note my X and Z stepper drivers are populated, not X and Y. This is because the Y-step pin is used for the ESC signal.
My Y stepper is less than 1.5mm from the table, plenty of clearance, except the extrusion bows down a bit in the middle (either that or my table is slightly proud) so the stepper was just scratching the table when it moved through the middle. I tried flipping the extrusion over just in case it was bent, but same problem. I cut off 10cm from the 2040 which took up 0.2 or 0.3mm, but it wasn't enough to make me not worry about scratching the table over time. To combat this I added a V-wheel like the X end support that is just the right height so it doesn't touch at Y=0 or Y=760, but does make contact through the middle. It should give me a more consistent depth of cut. You probably can't see it in the overview here but it screws in through the loop in the Y belt going to the stepper. All the wiring is then bound up in some spiral wrap and after running gcode to move it back and forth 100x, it never once even came close to moving toward the work surface so I think we're good there.
I am still waiting on my motor and ESC (been in Orlando for 6 days now) so the wiring on top of the carriage is a bit of a mess. I don't want to cut my nice ESC wires to put the right length connectors on so they'll just be wrapped up and strapped to the back. The ESP8266 running esp-link has been working great for running GRBL Controller to send G-Code from my desktop, or using the web-based terminal or telnet to type gcode commands to test.
Just need to figure out a way to strap the thing down and then we'll get on to making some cuts.
Does anyone know if Edward still sells the kit?
delli1221
New member
Yes he's still putting these kits together. Looks like they have been selling well for him lately. Here's a link to his site. ERC TimSav Lite Kitshaike said:
Hi, I stumbled upon this design on Thingiverse for Z-axis movements. I wanted to hear your thoughts on it. Looks like a more stable solution?
@CapnBry your z-axis design looks more stable than the original one. And this might be an improvement to it?
@CapnBry your z-axis design looks more stable than the original one. And this might be an improvement to it?
CapnBry
Elite member
Oh wow that is neat. ERC's servo mechanism works really well but an issue is that the Z height is proportional to sin(angle) so you don't get linear response through the whole travel range. A gear-based system would solve that. I also want to try to wrap PTFE tape around my wheels now too but I'd have to take them off to do that so that's going to have to wait. I seem to not be having any problems with binding currently but it couldn't hurt.
My mechanism appears to be pretty robust too because I was doing a dry run without the motor spinning and had my M3 set too low so I plowed into the side of the foamboard and pushed my TimSav around on the floor even though it was held down by 2x concrete blocks. No damage or loosening to the Z-axis.
I've posted my OpenSCAD file to Thingiverse. To get all the parts I used, comment out explode(); and uncomment out parts(); The old spinner is spinner() and the one I am currently using is spinner2().
EDIT: And after an hour of test cuts the spinner bearing worked its way out and threw the needle, despite needing a mallet to get the bearing in originally. Back to the drawing board. I am going to try to add a grub screw to secure the bearing which, looking at where the screw will end up, will throw off the balance so I'm going to have to up the stroke to 8mm or possibly replace the aluminum standoff with a brass one to even it out.
Last edited:
CapnBry
Elite member
Well the grub screw seems to be working but it is a lot harder to balance now. I'll have to work with it some more.
Here's a tip for people without a tachometer who want to know what the RPM is, just download "Noise Meter" by JINASYS from the Android App store. It says "Contains ads" but I think the only ad is in the menus where a line of text suggests you donate. Go into settings and set Digital Filter -> Low Pass -> 400Hz. Then fire up your motor and look for the peak.
The red lines are the motor noise and their harmonics, so we're only interested in the bottom one. On the bottom right, under the waterfall, you see the current peak is 160Hz. 160Hz * 60 seconds in a minute = 9600RPM. I try to shoot for 150Hz or 9000RPM. Make sure you also try it while the cutter is running, since the cutting action slows down the motor. For me M4 S40 was 9600RPM with no load, but it took M4 S60 to get it up to 9000RPM again under load.
Note that if the number is much higher, like 400Hz, then likely the first (second?) harmonic is being picked out as the loudest and your actual Hz is half that. So if it says 360Hz, you're at 180Hz/10800RPM. You can use the + icon to zoom in and switch to logarithmic scale with the "linear" button to visually verify you're getting the right peak.
Here's a tip for people without a tachometer who want to know what the RPM is, just download "Noise Meter" by JINASYS from the Android App store. It says "Contains ads" but I think the only ad is in the menus where a line of text suggests you donate. Go into settings and set Digital Filter -> Low Pass -> 400Hz. Then fire up your motor and look for the peak.
The red lines are the motor noise and their harmonics, so we're only interested in the bottom one. On the bottom right, under the waterfall, you see the current peak is 160Hz. 160Hz * 60 seconds in a minute = 9600RPM. I try to shoot for 150Hz or 9000RPM. Make sure you also try it while the cutter is running, since the cutting action slows down the motor. For me M4 S40 was 9600RPM with no load, but it took M4 S60 to get it up to 9000RPM again under load.
Note that if the number is much higher, like 400Hz, then likely the first (second?) harmonic is being picked out as the loudest and your actual Hz is half that. So if it says 360Hz, you're at 180Hz/10800RPM. You can use the + icon to zoom in and switch to logarithmic scale with the "linear" button to visually verify you're getting the right peak.
Hi...could you please send me the stl of the cylinder flywheel ? i am struggeling with my needles getting hot and jumping off the bearing...i would love to try this one...thank you
dutchmonkey
Well-known member
You should change out the bearing to a flanged model. I did that back towards the front of this thread. It helped fix my problem of the wire slipping behind the bearing and causing a ton of issues.
3 x 6 x 2.5 would be fine ?
CapnBry
Elite member
The source files are on Thingiverse and it is likely you'll need to tweak some of the sizes to make it fit just right on your printer, but here is the spinner I have on my TimSav right now.
You'll need:
Needle Assembly (in assembly order)
M3x7 button head
3 x 8 x 4mm bearing
washer of some sort that prevents the standoff from scraping, I used the one that was in the broken motor the bearing was in
M3x8 standoff with a hole drilled in the middle
M3x7 button head (another)
Fixings
3mm grub screw
M3x10 socket cap screw
M3 nut
0-4x M3 washer to balance it out, stuck on the socket cap screw
Last edited:
Similar threads
