This Old Machine: Re-purposing an Engraver as a 3D Printer

I recently acquired an interlab robotics engraver from the mid 1990s. It consists of a Cartesian gantry style robot, an electronics enclosure and a Monorail 7245 computer. It has been in storage for almost 20 years and while it all still powers up ok, the engraving bits and the floppies the program was on have disappeared, and would cost too much to replace and update relative to the expected lifespan and utility of the unit at this point.
I have decided to re-purpose this old machine as a 3D printer, starting sometime after this year's nationals.



I took the covers off and cleaned the accumulated dust and grime off the guides and lead screws.



You can see from the photo the X and Y axis are relatively sturdy. Not sturdy enough for conversion to a mill IMO though. The X and Y leadscrews are multi-start with anti-backlash nuts and will not reverse drive. X and Y motors are NEMA23 1.8 degree 6 wire steppers with 5.4 Volts /1.5A rating stamped on the end, Z motor is 6V / 1.2A rating.

Removing the two aluminum rails and the black beams from the center of the machine will increase the room available for Z axis travel from about 1.2 inches to a more useful 5.4 inches of clearance. The Z axis will need a longer lead screw and rails to make this work. There is enough travel available and room to mount and reach all parts of a standard sized heated plate with a dual extruder, so the final work envelope will be around 200 mm x 200 mm x 100 mm. Big enough for battleship work, but not huge.

The stepper power supplies, control boards and the monorail computer will have to be replaced, of course.

So what parts do you printer guys recommend for this rebuild?

 

Very Cool!

 

Nice gantry! The steppers are the right type to be driven by a RAMPS 1.4 board, so I'll second that recommendation. You may be able to salvage the power supply too, what voltage is it?

 

Thanks for the compliment. I'm going to pass on salvaging any of the original drive electronics, most of it looks like it came out of the original apple computer. One good reason to replace the power supply is that while I might be able to identify the voltage, figuring out a safe amperage the power supply could deliver might be a little tricky and most likely isn't enough to power the heated bed. I think spending $30-50 for a supply is worth it in the long run.

CONSIDERATIONS ON SELECTING A POWER SUPPLY

A good starting question is what voltage supply to get for the steppers? Steppers don't really care what voltage they are run at, as long as it isn't more than about 20X the rated voltage and as long as the driver uses current limiting to keep the power level down. I run the steppers on my mill at 68V. The main advantages of running the steppers at a high voltage is increased low end speed, the RPM at which the torque starts dropping off is raised. The main disadvantage is motor heating, they can get hot enough to burn you. These old style round steppers probably need as much help as they can get, ideally I would like to run them around 48V if possible. (8-9X the nameplate voltage)

Another consideration is a standard RAMPS board is set up for 12V. Going higher on the voltage means making changes to a basic RAMPS board like: 1.) Powering the Arduino through a separate 5V line and cutting the RAMPS supply, 2.) possibly changing the electrolytic capacitors to ones rated for a higher voltage (They should be rated at least twice the operating voltage) 3.) Replacing the PTC fuses with real fuses. The stepper driver board rating is also a consideration, suitable Pololu drivers with adjustable current limiting are rated for 35V (A4988 black version) or 45V (DRV8825). A "normal" power supply for a RAMPS board is a dual 12V/5A+12V/11A, the 11A is for powering a heated bed.

I prefer an enclosed switching power supply for safety and ease of connection, which limits the available voltages somewhat.

Unless I stumble across a readily available control board manufactured for a different voltage, or get a case of stupid and decide to modify one to run on a higher voltage, it looks like using a RAMPS will limit the power supply voltage to 12V.

Until I dig into it more and identify different requirements, a 12V 20A enclosed switching power supply is my initial selection. These can be had for under $40

 

modifying a computer PSU gives you enough power as well if you have an extra around. not having enough power seems to be many peoples problems

 

Looks like you have an awesome starting point!! There are a few boards out there that are 24+ volt compatible (I think Smoothieboard is good for 24v, I think Duet is 36v tolerant). I have a Duet and am not very thrilled with it - firmware is a bit of a pain and support is not readily available. Ramps probably has the best support from the community as there are so many out there, but it may only be 12v capable. 24v is really plenty for a printer. Axis move speed isn't super important because you are limited on how fast you can extrude accurately, so additional voltage only gets you faster rapids, which is nice, but probably not nice enough to be worth a bunch of effort. Yes, the torque curve is better at higher voltages, but again, on a 3d printer, you are really just moving the carriage and it's associated weight around, not like on your mill where machining forces require extra torque on top of what it takes to move the axis.

 

I see I left the hot ends out of the power requirement. +30 Amps
Inertia should not be a problem if the "RAMPS" actually ramps; some sort of adjustable acceleration ramping should be included in the motion planner. I assume set acceleration is an axis-specific value in the Arduino that needs to be updated depending on machine performance?
Regular searches don't seem to turn up the best information on 3d printers. There is way too much promotional hype and clutter and re-sellers concealing any good useful content that may be out there. Any good pages you guys like that explain how this stuff works?

 

^^ +1. One of the pages I found useful at first is the List Of Electronics

 

I agree. I bought a 24v 13 amp power supply for about $50 from Circuit Specialists. It has overcurrent, over temperature and short circuit protection. So very hard to break with a stupid connection.

I use 24 volts and recommend it. You will find that a lot of Reprappers are scared of 24 volts and will recommend 12 volts, particularly with RAMPS. 48 volts may be a bit much. First, your stepper drivers are not rated for it. The most common stepper driver chips are the A4988 and the DRV8825 which have maximum voltages of 35 and 45 volts respectively. Second, you will have a hard time finding a heated bed rated for 48 volts, while beds with dual 12v/24v circuits are common.

Components on cheap RAMPS boards are very suspect at 24 volts so I took an unconventional approach on mine. I bought the board bare from Ultimachine.com for $10. So I know I started with a quality board. Then I ordered all the components from Mouser for <$20. I purchased 50 volt capacitors, better FETs, etc. I left the PTC fuses off completely and protected the board with an external circuit breaker. For the power to the Arduino MEGA2560 board I bought small 3 Amp adjustable regulator.

If you don't want to go to that much effort you can use a regular RAMPS board by replacing the PTC fuses, removing diode D1, and verifying that the capacitors are rated for more than 24 Volts before you buy.

 

And you can run the MEGA2560 board off USB as long as you don't have an LCD screen installed. With the screen y0u would need a separate power supply.

 

24 volts gives you twice the power, so they are faster and have more torque. For lightweight machines you may not notice a big difference. On this heavier machine I think you will, I sure noticed the speed difference on my mill. 24 volts on the heat bed also cuts the amps in 1/2... I can put my finger on the bed FET and it is barely warm.

The chart is from Geckodrives: http://www.geckodrive.com/support.html

 

To understand why I recommend a higher power setup, it might also help to understand my perspective on this kind of stuff. On my mill I started with 125 oz-in steppers and 12 volts. It was slowwwwww. I upgraded to 24 volts and saw an improvement. Then I upgraded to 300 oz-in steppers and saw an improvement. Then I upgraded to 48 volts, Geckodrives, and servos. I could have saved a lot of money it I hadn't gone cheap at the beginning.

 

I saw the graph from geckodrives and read the Stepper Motor Basics guide on the same site back when I was converting my mill to CNC around 2004. It was very informative. I ended up at 68 Volts (on 4 volt 259 oz-in steppers) and Geckodrives after progressing from a 24 volt Xylotec board that fried after a month (I learned the hard way about optical isolation from the parallel port). I agree that a much higher voltage is desirable on steppers. I never did find any servos that matched well with what the geckodrive servo drivers could do so I never converted to servos. My last upgrade was to ballscrews and linear rails, which are so smooth I haven't figured out how fast I can actually rapid the machine, I just arbitrarily set a 200 inch per minute limit.

I am planning to install an LCD screen, I like the temperature feedback and I don't want to tether this printer to a computer in order to run.

I'm slightly concerned about the 16Mhz speed of the processor on the ramps board. It seems (in my mind) to be barely powerful enough to do high quality trajectory planning with look ahead and actually handle everything real time. Throw in communication with a peripheral device (Like an LCD or a thermocouple reader) and ??? All it takes is a small hiccup of the wrong size in the output pulse train for a stepper and you can stall it.

After poking around a bit I think I like the way the smoothie software is set up better than the RAMPS. Plus it handles 24 volts without mods

 

I like the Smoothieboard too. 24v capability, 32 bit w/ 120 mhz processing, easy to connect larger stepper drivers (all necessary pins are broken out), and reasonably easy to mod firmware are some of the major draws. it is also supposed to provide much smoother processing/ movements.

I sort of sucked myself into buying the Duet for a few reasons, one major one was that I, like you, wanted stand alone capability, and the PanelDue is pretty much plug and play for Duet, adding pretty nice touch screen capability, but I am still considering going for a Smoothieboard instead - Duet is kind of a pain....

 

Looking at rebuilding the Z axis carriage first. The lead-screws seem to be 4 start 3/8-8 Acme screw, which means the lead is .5 inch. @400 steps per inch this gives a resolution of .0025 inch per step (.0635 mm per step). My experience is that I can barely see a feature that is .0025 inches proud at a 18" distance.

Is this good enough resolution for decent layer height control or do I need to buy something with a finer lead?

Changing the lead would mean I need to replace the $40 anti-backlash nut in addition to spending $20 for the screw and $10 for the guide rods. Forget about including micro-stepping in the calculations for now.

For a change, my itty-bitty lathe can actually handle machining the ends of this size lead screw.

 

I have also had some thoughts about improving the single hot ends - making it act more like a real plastic extruder instead of a just pushing a filament into the extruder and hoping. The idea I am kicking around in my head is to use an additional motor to rotate a carbide two flute down-cut router bit inside the melt chamber, and feed the filament into the bit from the side in multiple locations (Probably 6-8) at different speeds. This would allow mixing different colored filament inside the hot end and theoretically allow almost any color to be printed from a single nozzle, and would give a more consistent melt due to the rotational shear from the bit. Reversing the router bit for a second would stop any nozzle drool almost instantly. Keeping the feed area of the bit cool would be critical to reduce clogging. Color changes would not be instant though and would likely require extruding several cc of waste plastic over to the side to be consistent in color, every time the color changed. I've done a lot of injection molding of different colored parts so I'm pretty sure this approach could work well.

I will build a regular hot end first anyways so I can print additional extruders.