So I have the mod finished, calibrated, and I think it's great results.

Great job!  I think the .3mm results look effectively perfect and probably about as good as one can ever expect to get. 

For the .1mm prints, software compensation or one of the slip-joint solutions might be able to finally rid the slight banding that is still there, but if not, going to a .5mm thread pitch will squeeze the bands 37% closer together, which IMHO looks even less noticeable than banding at .8mm.  The flexibility and tighter thread tolerances would probably help too.  I don't think anybody has done a side-by-side comparison of an M3 vs an M5 with the same printer--so I'm curious--but my gut reaction is still that the finer the thread, the better the print will likely be.

Comments are welcome:)

10mm x 10mm single wall cylinders.

As far as microstepping-based banding, I agree it can be one cause of banding-like symptoms, but I am skeptical that it is the *primary* on the solidoodle cause for two reasons:

1) Many people have tried changing the layer height to one that will result in no microstepping.  A few have claimed "some" improvement, but only one AFAIK on the solidoodle has found total success, and he may not have had the typical banding most of us have seen.  It did not work for me.  Yet, I think you'll agree that if banding were due to microstepping, and you set the layer height to one of these full-step values (1/200th thread pitch) that one should see *no* banding at all, since no microstepping is occurring at these values.

2) The banding most everyone else describes is sinusoidal in amplitude with a period exactly equal to the thread pitch, regardless of layer height.  I can't see how microstepping-based banding could exhibit this profile, as a graph of the attempted microstep values shows a repeated function with a higher frequency than the thread pitch that also varies depending on the layer height.  Instead, I'd expect to see any banding to have a more chaotic profile with a higher frequency than the thread pitch.   If the microstepping behavior were more random as you describe, then of course the profile would be even less sinusoidal than observed.

However, I think everybody who has changed z-rods has gone to an m8 or smaller metric zrod, which tells us very little about the actual cause of any improvement they see, in terms of metric versus imperial, coarse vs. fine, or low-quality vs. a high quality rod.  My suspicion is that changing to *any* other z-rod (including a higher quality 5/16-18 rod) would probably yield significant improvement.  The choice of how fine to go is probably much less important than making any change at all.

Often as you reduce cost in manufacturing, mfg processes, and materials at the compromise of quality control you increase your customer service, lost customer, returned and replacement costs which can easily out pace your savings in manufacturing cost reduction. I think especially for a company that is relatively new they're reputation is the most valuable asset they own.  And there is nothing like poor QC and corner cutting that can tank a reputation. Ya know. 

I actually have designed a few manufacturing processes for short run manufacturing in the realm of two to five hundred units a week per assembly line.  But not necessarily the know-all of it by any means.

Anyhow I think solidoodle has great opportunity to really compete. It's just odd to me that they don't pick a finer pitched lead screw when it would really help hide some major deficiencies in their design/quality control process.

Thanks for the excellent info. 

Ironically I meant to say a .8 pitch above... That is what I put on mine, and I am certain it is more then enough to make banding a non issue when properly aligned, backlash removed, etc. 

But I side with your views anyhow. 

I've always thought the errant plus or minus 1 micro step "controversy" that I have seen on the forums is a bit funny. at least for the solidoodle Z axis.  Even on a stock machine a tolerance of +/- .0004mm per layer is the least of your precision/accuracy concerns.  Even if you multiply that across all 200mm of z height that is at the worst  case scenario (where the errant step is always in the same direction and miss stepping at .1mm layer heights) something around .8mm tops... The chances of that happening are in my mind pretty slim.  Most likely its in the .1 to .2 range at full height.... Which to me... Well that is a non issue for these machines. But for the most part I think anyone hardly maxes out their z height on a regular basis. 

It's sort of like the numbers games in computers. Where most people look at two maybe three variables when buying a computer for speed. Processor ghz, ram, and maybe your video card.... Now of course I know we are all too savvy around here to make that mistake.  But the irony is of course pretty much any computer you purchase new these days will be blistering fast for the average user.  Sometimes seems like more of an ego game. "Guess the size of my computer's ram processing liquid!?!"

Right let me know when anyone cares kind of a thing.

Anyway - I'm no expert by any means - the above is based purely on empirical evidence and observations of the many other printers out there..

The imperial/metric argument is also misunderstood by many. The issue is if you don't consider the steps-per-mm on the Z-axis and your microstepping settings etc - *often* usual heights end up being 'partial steps' - this will result in the motor ending up returning to the previous detent position, or advancing to the next, based on whimsy (i.e, it will change its mind with no predictability). The key is to try and align your steps to *complete* steps - this can be as simple as avoiding a height, moving to another height that *is* a complete step; or looking at what you use for microstepping. For example, on my M8-2.0 ; 1 full step = 0.01mm - I can therefore pretty much run *any* height I want and will always have complete steps...  Now this issue is also present in Metric - there is various pitch-to-step combinations that end up as partial steps.  Anyway - the issue isn't Imperial is bad, Metric good. Its more about what imperial, what steps and what layer height. But the idea that imperial is 'at fault' purely because its imperial is actually a complete misnomer....

As for why drill a 5/16" rod and not run a coupler - they did used to run couplers - but switched to a center drilled rod as A: its less assembly intensive (press rods onto motors, put on shelf for assemblers...) B: Cheaper as they have a lower component cost - A lot of what happens on a solidoodle is about reducing assembly costs and steps - e.g, the change to the printrboard was arguably motivated by reducing the BoM on drivers and the need to install them and reduced the header count on the PCB by a large factor (and its headers that hurt - they are often one of the most expensive BoM costs on microcontrollers).  A pressed on central drilled rod requires no tightening of screws, no sub-assembly, and no alignment issues (in terms of to the motor - assuming the rod is drilled straight). If drilled and pressed to the same depth everything comes out 'identical' compared to tolerances that can be introduced through couplers (unless you employ a jig to insert everything etc - which is totally doable but again adds steps to the assembly and therefore - costs).

I love it, but sometimes certain upgrades make you wonder what is going on over there.

the m3 rod flexes instead of the bed carriage.  i believe im going the way of a metric leadscrew. its all a matter of preference, but anything is better than the off-center drilled rod they come with.

I have seen a few people remark about the rod being drilled off center.  Are you sure you're not seeing an illusion?  Because of how the thread tapers off at the end of any threaded rod, a center hole will always appear to be off center.
Just wondering..

I've never found a fine-threaded m5 or m3;  I'm not sure how available they are, so I have the .5 mm (standard) m3, which is the same pitch as a fine m5.  I limited the maximum speed of my z-motor because on long moves up and down I found the tight z-nut I made could bind up once in awhile, but I don't think the motor gets any near this fast when printing or retracting.  My guess is that a fine m5 would have a similar problem at the motor would be spinning at the same speed.

Lots of people have gotten great results with an m5, but I concur that the finer the thread of an m3 is better IMHO, for two reasons:

1) I finer pitch automatically has less possibility for backlash due to the geometry involved

2) The printed layers can only reproduce banding with a band-spacing equal to or greater than two layers in height (i.e. highest frequency banding equals alternating layers of narrow and wide).  If you get the thread pitch narrower than this (the nyquist cuoff), then it's mathematically impossible for the layers to show banding at this frequency, and the most I think you might see is a much dampened down secondary aliasing effect at a lower frequency and amplitude.  For .3mm printing, the nyquist cutoff is .6mm, so I believe going to a .5mm pitch should theoretically give you no primary banding effect.

I did however take lots of care to machine two 5/16" x 1/2" Allen head bolts so that I could leave the Aluminum bed nut in place, and reutilize my favorite backlash tool the trusty well nut/expansion nut. 

Anyhow with no hysteresis or z-wobble fix turned on my .3mm prints looks promising! My .1mm not so much, but I think that comes back to my very very horribly machined aluminum coupling.  When the new one arrives in the mail I'll post follow up pictures.

Successfully changed frequency, now if I could just machine a proper coupler...

adrian wrote:

2n2r5 wrote:

I see Z-lift as admitting defeat and yielding to poor calibration and sub-par equipment.

Hey, call me old fashion - but I like my linear motion to be symmetrical

Symmetrical motion is for those that can't divide by 0.

So... I take it you can

2n2r5 wrote:

I see Z-lift as admitting defeat and yielding to poor calibration and sub-par equipment.

Hey, call me old fashion - but I like my linear motion to be symmetrical

Symmetrical motion is for those that can't divide by 0.