Sunday, 30 September 2018

Moment of inertia calculations - think again, fat boy

Despite the fact that the inertia of the Blidgeport table and load masses, expressed as moments of inertia at the servo motor are pretty small relative to the MoI of the X and Y servo motors themselves, the best response I could get out of them wasn't exactly lightening fast. These motors are rated at 400W continuous, with something like 2-3 times that in the short term.

Of course, I made an early decision to retain the handwheels so that I could retain a lot of the manual features of the machine. And the handwheels on this particular model are pretty large in terms of MoI, when compared against the simple "ball" handles often seen on Bridgeports.

When you look at the contribution that the handwheels make to the total MoI of the combined system referred to the servo motor (through the 2.5:1 reduction stage), it is dominated by the MoIs of the handwheels. There is only 1 on the Y axis but there are 2 on the X axis.

I reran the calculations with and without the handwheels. The "inertia ratio" is the ratio of the final MoI to the MoI of the servo motor alone. Without the handwheels, the motor MoIs dominate. With the handwheels, the MoI is almost an order of magnitude greater than the motor. The servos are basically expending most of their effort twiddling the handwheels. 

As you can see, the net MoIs can be reduced by more than an order of magnitude and the effect on the P, I & D terms is significant. So I should be able to improve the response significantly by removing them.

Saturday, 29 September 2018

Laser engraver head for the Cetus

I've only just got the damned thing and only run 2 test parts off it but already I've spotted an enhancement. 

This whacky German guy has spotted that the Cetus has a spare header that is intended for driving a small thermal laser - and shown how to get it running. 

Looks good to me, so obviously I've wasted no time wasting my money on one.

I should get on with making up the bracket while I wait for the slow boat to bring mine over here. 

Looking forward to playing with this. Meanwhile I should also be looking into how to use g code with the Cetus. Doesn't look too difficult but still requires me to RTFM.


Cetus 3D - what it looks like and first tests

When I posted the pros and cons of the Cetus, I didn't have access to the photos I'd taken. Now I do.

Can't stand "unboxing", although I suppose that's what the first part is. 

The Jiffy bag contains the bonus extra set of 3 nozzles and the middle package is the bonus 2 x 500g reels of PLA filament:

Inside the main package are the std contents including the instructions (in something like 8 languages):

Under the top layer of packing is the main vertical columns (the Z axis slide):

Removing the separating layer exposes the base and the table mount (X axis):

Here it is, all pitched out onto the table:

First, assemble the column (3 screws) and the build plate (3 screws), then attached the preassembled hot end (head) unit (2 hex socket bolts). Plug the cables in, fit the filament guide and Bob's pretty much your aunty:

Off we go, printing out one of the test pieces from the library in the iPhone app:

Although there are no instructions or illustrations, this seems to be how the filament spool holder goes together. It's free standing.

Getting there slowly....

Bugger me, this is slow. Admittedly I chose a fairly complex test piece and went for a fine layer thickness (0.2mm).

Left it running overnight and then went to work. When I came back in the evening it was finished. TFFT!

The slicer begins with a raft for the part, extending out a fair way - for stability I suppose.

It's full of support structures....

Used the simple model generator in the PC app to make this,using faster, coarser (coarsest in facT0 settings. This took about 30 minutes:

Note again the foundation layer and the base layer. They both peel off but together they amount to a fair bit of filament.

Figuring out which parts to remove and which to leave was a challenge, not least as I wasn't sure what the finished part should look like. Looks as if this unfortunate being has contracted some form of giant smallpox - largepox perhaps? Don't know who she's been hanging out with but I can't see that ending well. Either way she needs to get down the clinic pronto. But the printed part looks pretty reasonable.

I don't have a macro function on my iPhone but you can see the finish isn't bad at all. I must do a more scientific test piece so I can check the dimensional accuracy.

Wednesday, 26 September 2018

Another rodent accident - Cetus 3D printer this time

Ooops - again:

Working away from home and workshop doesn't seem to play well with me - the resulting cabin fever seems to make me accident-prone and - in particular - susceptible to rodent-related outrages.

Lat time I had a serious accident with a mouse it resulted in 3 tonnes of Japanese cast iron turning up at the house, which the Domestic Manager wasn't entirely happy about. I think (hope) I've lived that one down now and TBH, the machine itself (The Shiz) has been a revelation. It's just a shame that it resides 200 miles away when I am at work during the week.

When I lived in Canada (2010-2014) I was Engineering Director at Delta-Q and bought an Ultimaker2 through / for work. We found it to be a really handy piece of kit. Having said that, it seemed to require endless buggerage to get it to work and keep it working, so at best you might just about describe it as "prosumer", rather than professional. Luckily we had an "intern" to keep it running. It also cost something like £4-5k by the time it had been shipped across from the Netherlands.

I've been keeping an eye on the budget end of the 3D printer market over the last couple of years. The price has come down and the performance up to the point where there are some almost decent choices available now. Products such as the Monoprice come to mind and occasionally there are similar offerings from the likes of Lidl and Aldi. 

Anyway, on Wednesday evening last week I was online, looking at the Cetus3D MkII which appears to be a pretty decent machine. Before I knew it, I'd suffered one of those finger spasms and I found I'd ordered one. In one night, all was lost!

Anyway, I've had the thing up and running now and here's what I think about it now. I'll look at the machine and its first output later....


  • The price is pretty good. The cost was £347 delivered. A Prusa i3 Mk3 fully assembled is £900 plus delivery (with a 1 month leadtime currently) and an Ultimaker 2 is about £2150 delivered - that's over 6 times the price and the reviews are pretty convincing....
  • Delivery was quick (and free). I ordered it on Wednesday evening and it arrived at work on Friday by 10am via Parcelfarce 24h service. 
  • Came with a bonus pack of 2 x 500g blue PLA filament.
  • Came with every possible fastener and tool required (apart from crosspoint screwdriver) and some filament samples. And a paint scraper. And a pair of wire cutters.
  • Came with 6 nozzles (2 each of 0.2mm, 0.4mm and 0.6mm) - 3 were part of the bonus pack.
  • Very nicely made - "minimalist but functional".
  • Uses genuine Hiwin ball sides.
  • Requires only screwing together 3 subassemblies and mating 3 electrical connectors.
  • The UP iPhone app is excellent and comes with a large library of sample models. You can load and slice models, set all the controls from your phone, start the job and follow progress.
  • The proprietary UP Studio PC app appears to support g-code now although I have not investigated this yet. Sounds as if this would potentially allow output from Fusion360 etc rather than .STL output, although I'm not sure about suitable post processors etc. You can also use slicers such as Cura.
  • Has wifi, so no need to bugger about with SD cards, USB cables etc, although there is USB too if preferred.

Not so good:

  • The build plate had some scratches and marks but nothing serious enough to be bothered about. But it has an interesting (heavily anodised?) surface finish that doesn't seem to require Kapton film, Pritt Stick etc - it remains to be seen how effective this is.
  • No instructions given for wifi setup. Took a while to figure it out.
  • No instructions for assembly of the parts for the (separate) spoolholder frame, or pictures on the website. Luckily somebody had posted about that on the forum.
  • The filament guide tube didn't fit into the hot end housing without some fettling. The 3D printed housing quality was very rough, which is ironic really.
  • The PC app doesn't seem to include the rather excellent sample models that come with the iOS app unless I missed something. The settings GUI is a bit unconventional and some of the screens were tricky to find. Otherwise not actually bad.
  • The Cetus user forum is spectacularly useless. It's difficult to imagine how such an outcome is possible. The search function barely works and the displayed font is so light, small and spidery it's almost invisible, just to add to the annoyance. I had to go back out to Google to search for relevant content within the forum. The sister forum for the UP printers looks a lot better but there is no Cetus content there.
  • Cable routing is a bit tricky to get right and there was only one cable tie to hold the cables in position. Also there is no strain relief on the signals connectors, so eventually we can expect wires to fatigue and fail there.
  • There is no integrated spoolholder which is a pity. Instead, you stand the supplied frame alongside the printer, once you've figured out how to assemble and use it. You could perhaps mount it onto a simple baseplate, along with the printer itself, if it becomes sufficiently annoying. 


  • Typically noisy stepper drives. Could replace with a silent (3rd party) drive board if bothered eg Azteeg X3  Some guy on Youtube did his own board but doesn't seem to have any plans to share it.
  • No enclosure (helpful for ABS).
  • Heated build plate is an optional extra at $40 (it's required for ABS) - but currently not in stock. Also sounds rather basic / crap.


Thursday, 6 September 2018

Moments of inertia calculations for DMM Tech servo tuning.

I buggered about with the DMM Tech servos on my CNC conversion previously and had a fiddle with the DMM Tech "DMMDRV" tuning software

The results were encouraging but without some super loving, self tuning function, there is clearly some work to do before I can claim the parameters are optimised.

DMM Tech published an app note showing how to calculate the MoIs of the system and how this can be translated to starting values for the Kp, Ki and Kd gains.

 Here's my spreadsheet that implements those calcs:

I ignored the trivial MoI of the small toothed pulleys and instead included the relatively large MoI of the 10" diameter handwheels. For a handwheel like this, the MoI is essentially Mr^2, whereas for a solid disk, the MoI is Mr^2/2.

The moving mass for the X axis is the table, vise and work. For the Y axis, it is the table, vise and work - plus the saddle. However, with a ballscrew pitch of 5mm, the effect of even this mass is relatively small in terms of total MoI, particularly with a 2.5:1 reduction.

The final outcome is the "Inertia Ratio" which is the ratio of the MoI of the system to the motor. From that it's possible to read off some starting values for the P, I & D gains, namely:

  • Inertia Ratio = ~4.9
  • Kp = ~78
  • Ki = ~24
  • Kd = ~28

Next time - try these out.....

Iscar Tang Grip TGFH parting tool tests

Stung into action by several months of forced separation from my workshop, I suffered another serious accident involving a mouse. On this occasion it resulted in the delivery of a 3mm version of the 32mm blade Iscar TGFH parting tool kit from Zoro and 5 of the uncoated IC20 inserts for loominum cutting. They were on offer, so actually cheaper than currently shown and I also got an additional 5% coupon code to apply at checkout.

Found myself back home and in the workshop this afternoon due to impending leave, so it would have seemed rude not to have done some initial tests.

Found a piece of 2" dia mystery steel and mounted the tool up in the holder. Being an industry standard parting blade size, I was able to swap out my existing Korloy "Saw Man" blade for this one quite simply.

The parameters for the (steel cutting) IC808 inserts are 4-10 thou per rev and 80-180 m/min (sorry for mixing units but my Bantam has an imperial feed gearbox). That translates to about 700rpm min surface speed.

Started out using girly settings ie 250rpm or so with tentative hand feed and the whole machine almost shook itself apart. Quite alarming and surely not good for anything.

Then upped the speed to 700rpm and took a more serious cut. Job done - no juddering and a nice curly ribbon of swarf. Juddering only started when the tool approached the centre, at which point the surface speed was obviously quite a bit lower. Engaged the feed (about 3 thou per rev) at 7000rpm and it cut like a good 'un.

Seems to confirm what has been said here before - the cure for juddering during parting with these indexable tools is to use the proper surface speed and feed per rev and use power feed. Trying to girl out simply makes it judder unmanageably.

I should probably find some mystery loominum and repeat the exercise now with the (uncoated) IC20 inserts. In the interests of science of course.

Rust eater?

What? I've had to do some derusting on a variety of components recently and it seems to be an expensive and/or hazardous and/or labour i...