That arrangement would also further reduce belt length which would be a plus w.r.t. potential stretching effects from the weight of your Z plate assembly. However, unless you implement it as a rack and pinion linear belt (belt-on-belt drive or tractor) you will still likely issues with belt stretching unless you employ some constant force offset springs to, in effect, make your Z axis weightless. However, in the arrangement you have above, you only need to use a belt to drive your lower rod shaft. If you used cables instead of belts for the actual bed it would eliminate any stretching problem - you would just need to find type of pulley with enough traction to drive a cable or find a suitable roller chain/sprocket.

 

I have a friend that designed large format (12" x 24" or bigger, can't remember) 3D printer with counter weights for his Z-axis. Haven't seen one here though... Seems like a set of constant force springs would be more compact/less complex to implement and able to more evenly distribute the force though.



Not doing that on my build though - gravity effects on my 400mm x 400mm Z axis actually assist in further reduction of potential backlash and gantry movement acceleration/deceleration effects (my gantry moves on Z - build plate/bed is fixed). I like Giuliano's belt drive idea though - though I don't anticipate it, if I encounter any unacceptable Z banding issues, a similar belt setup would be worth a look.

 

Now those would do the trick quite nicely! A Rostock-like setup would probably be ok (and definitely cheaper) - but I believe the performance of those belts would be superior. And with minimal torque losses, you could actually drive both sides with a single stepper via another (steel reinforced) belt if desired. Looks like you'd need to find a different supplier though (or just go the Rostock route).

 

I like it! Properly balanced, constant force springs would also reduce Z axis drop forces in the event of loss of power to the steppers. In fact it may still allow the bed to be suspended in absence of power to the steppers. Regardless it would certainly prevent the build plate from slamming to the bottom during such an event.

Depending on your pulley selection and stepping resolution, you should be able to obtain ~10 micron resolution with this setup as well. I think this is the max my lead screws can do so I don't think you aren't giving up anything going this route - just adding a slight amount of cost/complexity but not too much. Looking forward to seeing it's performance!

 

You already have a horizontal rod available for your upper pulleys - 8mm right? You could use 8x15x24 linear bearings ($0.80 from robodigg) - or for this application even a self lubricating bronze bearing ($0.60) would work. And if it were me I would probably go with the 9293K54 spring (same price but 4.12 lb.) for a total of 16.48 lbs force. When you add screws and other hardware, heating element, glass or Ultem be plate, etc., you are likely to be closer to 16 lbs. Also, a net positive force may not be a bad thing (the less total force you have in either direction will reduce potential belt stretching issues).

 

There are also these @ 3.5lbs. each:
http://www.amazon.com/dp/B0085ZXEDW/ref=biss_dp_t_asn

They are a little larger but are only $12.67 shipped for 5 of them...

 

During one print cycle, the average printer cycles up and down the Z axis twice. Once for homing/calibration and once for the actual print. So if you printed one object a day 365 days a year it would take you 5 1/2 years to cycle 4000 times. In addition, the (minimum guaranteed) cycle life is for the guaranteed load tolerance of the spring. It doesn't mean the spring will stop working after 4000 cycles. And since the use case is for a load offset (not necessarily for perfect balance) all it means is that, using the springs from Amazon noted above, if four of them offset 14 lbs when new, after say 6-8000 cycles they may only offset 11-12. Since the load is distributed amongst 4 belts than each belt may be in tension by a half lb. or so - still completely usable and unlikely to cause stretching issues but if you are worried about it you can spend $13 and replace the springs every 5 years...

 

I suspect that for greatest longevity it would be best for the plate the be left at the top (almost fully coiled springs) - not something it will default to unpowered without mechanical assistance unless your net force is positive (which if you did design it for, would also provide an intrinsic indication of weakened springs).

 

Vertical movement of your lower rod will provide tensioning adjustment for your red belt. And if you move your Z stepper and drive pulley closer to one end you would likely not need additional shaft support.

 

Seems kind of pricey for that thin of a piece! And I would be afraid of trying to mount that thin of a piece of material semi-permanently to my build plate/heat spreader. I've seen some decent results from a simple acrylic sheet (which also isn't cheap but certaily is compared to ultem) using lower bed temps.

https://lesagegp.wordpress.com/2013/10/06/acrylic-sheet-for-print-surface/

I think I am going to try acrylic first. I plan on buying a large sheet to enclose my printer anyway so I will have plenty left over - and since it is thicker I can probably sand and reuse as needed. And if it doesn't perform well enough I will probably get this:

http://www.ebay.com/itm/25-x-16-x-1...158?pt=LH_DefaultDomain_0&hash=item3a9577fa16

More than twice as much as the zoro option but imminently resurface-able through light sanding and less likely to damage (I saw a picture of am ultem sheet someone was using that cracked - suspicion was difference in thermal expansion of aluminum plate vs. the ultem - this would also be a concern if you are planning to glue it to your heat spreader).

 

I didn't realize you were increasing your build area to 20"x16" - I haven't seen any decent prices for anything over 16"x16". I have read that link before, as well as many others with mixed results so I'm not ready to jump just yet. And if I do, I plan on using a thick enough piece to be removable so I can use different print surfaces (e.g. bakelite for nylon, ultem for general printing, glass for some prints, etc.). Though I suppose I could tape a thin sheet of ultem to glass and have a reversible plate and just switch to the bakelite sheet when I want to print nylon...