Saturday, 14 November 2020

Bantam CNC - sequence of events

 There's a right way to do this:

Some of the critical machining work will need to be done on the lathe itself - stuff like modifying the existing X axis (cross slide) leadscrew and perhaps machining up the housings for the ballnut and thrust bearing for the cross slide. It's a question of "physician heal thyself", something that can't be done while said physician is out of action.

What to do first? Seems to me I should get the cross slide assembly done first, then move on to the saddle. Rather like doing the Z axis on the Bridgeport conversion before the more straightforward X and Y drives, as that's where the tricky stuff lay.

Sequence of events:

  1. Clean up the cross slide and saddle bodies and machine the cavities for the encoder scale, limit switches etc. Ideally but not necessarily including the ballnut yoke fixing holes, if I can finalise them in time. This will be done on the mill.
  2. Reassemble the lathe (apron, cross slide, power feed shaft etc), so I have a functioning machine again. Obviously I will need to mount the modified motor and recommission the VFD at this point.
  3. Machine up the cross slide thrust/drive housing and the cross slide ballnut yoke. These will likely require boring in the lathe but otherwise is mostly milling.
  4. Bore out the Rotex coupling for the cross slide ballscrew drive and the driven pulley, then machine up the cross slide ballscrew. This needs to be done on the lathe.
  5. At this point, the old leadscrew can be chopped down and the new ballscrew, ballnut, yoke, thrust bearing and pulley fitted.
  6. The drive servo could be fitted at this stage, although it would mainly serve to show it assembles correctly.

If this goes well, I should have a functional manual lathe with a ballscrew cross slide (X axis) and a servo drive. I can then move to the more straightforward saddle drive (Z axis), glass encoder scale etc.

So first, before I can proceed much further, I need to make my mind up about the saddle and cross slide mods I will finally go for. Although I'm not shy when it comes to machining the original parts, I'd rather not end up making a pig's ear of it.

The cross slide itself is fairly simple - machine a shallow slot on the hidden underside of the body to receive the magnetic trip. Then drill and counterbore the fixing holes for the ballnut yoke. I may do that later, as removing the cross slide is fairly easy.

The saddle is a bit more involved. Requires cleaning up the rough cast leadscrew cavity and making some slots for the encoder body and the limit switches. Let's get those features finalised next....

Tuesday, 10 November 2020

Latest progress on the Bantam CNC conversion - CAD assembly

So how's the CAD work going then, Fatty?

I've put lots of effort put into this recently. At least I seem to be making some sort of progress. I'm not the kind of guy who just holds a part up against the machine, screws it in place and then figures out how to add the next bit. There are a lot of moving parts here and not a lot of space.

I've modelled up most of the critical parts to a appropriate degree of accuracy, so I can design stuff to fit together with a fair degree of confidence that it will assemble and work.

Here's the section view through the cross slide ballscrew. I've got a simple rigid sleeve shown here but that wouldn't allow any float, which would make setup and adjustment rather tricky. In fact I plan to have a 14mm Rotex "spider" coupling joining the new ballscrew and the original leadscrew stub. 

The back of the machine is looking somewhere between "busy" and "tricky". I hadn't realised the mounts for the large ballscrew would be so large. Fitting them and the saddle and the ballscrew and the encoder scale in that space will take some juggling to end up with a tidy solution. Looks workable but isn't your slam dunk.

From another angle:

And stepping back a bit:

The "missing component" that houses the cross slide ballscrew thrust bearing, connects the saddle to the saddle ballnut and picks up the encoder read head will be a critical part. It also holds the servo motor, houses the drive belt and pulley and provides some adjustment to adjust the belt tension and align the saddle ballnut.

Next - figure out a sensible solution for the "missing part" that I can make without needing a PhD in machining....

Saddle ballscrew assembly - getting some bits together

Preparing to get on with the Bantam CNC conversion

Here's what I've got in the pipeline, heading its way from various emporia in Korea and the UK. Spending beer tokens is the easy bit, although getting them in the door past the Domestic Manager can be fraught. But that's no reason to hold back.


I've gone for parts from the DY Global scrapyard in Seoul, stripped from some bygone machine or other. Rather than take my chances with some shark on AliExpress flogging bent (literally) rolled ballscrews, I thought I'd go for second hand decent ground jobbies. In some cases, they offer new old stock parts from the likes of HSK, Hiwin, TBI etc. In this instance, there was nothing suitable for my needs. However, I finally settled on the following:

1205 ground ballscrew for the cross slide. I've convinced myself the ballnut will (just) fit in the cavity in the saddle under the cross slide. I'll clean up the rough cast cavity to give it a bit of margin but it The ballscrew itself is longer than I need but I'll machine it down when I have actually designed something for it to go in.

And a 2005 ground ballscrew for the saddle. It's longer than I need but as I'm not yet certain what that is, I'd rather cut down a longer ballscrew than curse a shorter one.

From DY Global's description:

Main Size (Metric) 
  • Shaft Diameter = 20mm
  • Lead  (Pitch - distance by 1 rotation) = 5mm
  • Nut Diameter = 44mm
  • Flange Size = 60mm X 44mm
  • Nut PCD = 55mm
  • NUT Length = 41mm
  • Overall Length : 1090mm
  • Helix parts Length : 980mm
  • Motor Side  (Fixture Part) diameter for coupling : 12mm
  • Motor Side  (Fixture Part) diameter : 15mm (applicable to AK15, EK15, BK15, FK15)
  • Support Side diameter :  15mm (applicable to AF15, EF15, BF15, FF15)
  • Ball Screw Support Unit : BK15+BF15
Ballscrew supports:

The big one seems to come with a set of supports. These are the generic BK15 and BF15 supports that you see on all the Chinesium sites. There's a reasonable drawing here ballscrew supports - Hiwin.

And for the ballscrew / ballnut themselves, the Issoku ballscrews home page is a good start. Issoku HTR2005C7 ballscrew (page 68) and Issoku HTR2005C7 ballnut (page 12) look about right.

As for the cross slide ballscrew, I simply modelled it myself, along with a ballnut created from the dimensions on the ebay page. I'll check and correct that when I have something to measure up. I suspect it's so old there's no data on the internet.

Linear encoders:

I've also received an "EHM" miniature magnetic linear encoder from M-DRO that should fit underneath the cross slide table itself. The scale is a magnetic self adhesive strip of about 1mm thickness and the head is just 10x12mm (40 long), so can be buried out of the way. Resolution is 5um - not much point going for any better. Wasn't cheap but hopefully it will at least work.

For the saddle scale, I have a full sized HXX glass scale that I bought in China, also 5um resolution. It was sized for this machine so is 700mm long, with 600mm measurement range. I can cut it down if required but ideally it will be OK as it is. I've modelled it up in Fusion.

Drive pulleys:

For the cross slide drive, I plan to use an HTD 3mm pitch, 15mm width timing belt. The largest pulley I can fit in under the cross slide in its extended position would be 32 teeth (32-3M-15) and for the motor I should be able to get something like the 10t (10-3M-15), assuming that's not too small for the motor shaft. With a 150mm long belt (150-3M-15) on these pulleys, the centre distance will be around 42mm. I want to keep it fairly compact.

Cross slide limit switches:

ZF seem to do some sealed miniature microswitches. These should just about be able to hide themselves under the cross slide, within cavities in the saddle - I don't mind machining bits off here and there. Got some Omron ones too, in case they don't work out. These are not sealed but appear to be half decent quality (hopefully) and appear to be intended for industrial use.


Finally (for now), I ordered some oilers. The Bantam is rather light on lube points, as John Stevenson pointed out. I will press a few of these into the cross slide etc to facilitate oil getting where it needs to be.

Thrust bearing:

I want something half decent here, so I've gone for a Nachi 6200-2NSE9C3 Sealed Ball Bearing (C3 Clearance) 10mm x 30mm x 9mm. I'm not planning on any fancy paired back-to-back assemblies here. This is a single row deep groove bearing and it should be up to the requirements without having to step up to a double row bearing. I can't see it collapsing under the load it will see somehow.

Rotex coupling:

14mm Rotex coupling and orange spider will join the stub of the original leadscrew with the new ballscrew, removing any axial load. This is only allowing the handwheel to operate the ballscrew, while the servo and encoder are all on the ballscrew side of the coupling.

Sunday, 1 November 2020

CAD assembly for the Colchester Bantam in Fusion 360 - bed, saddle and cross slide

Modelling the Bantam:

Yes, I'm still talking to Fusion, despite our recent misunderstanding. I'm not the kind of guy to attempt a CNC conversion without spending weeks designing everything in CAD. Rightly or wrongly, I prefer to model up the base machine, then design and import the new components to make it work. Ideally I'd be able to import stuff like the fasteners, pulleys, bearings, leadscrews etc from places like Misumi, Traceparts, McMaster Carr etc but I won't find a Bantam CAD model anywhere, so I need to model it up myself.

This bit takes forever and doesn't really achieve much in itself but with a bit of focus, a half decent model can be crafted up in a few days, around all the other stuff that needs to be done in this world. It doesn't need to be photo realistic but I want to be able to create parts that can be CAMed up and made without having to redesign them - or find they don't actually fit once I've made them.

This is basically an imperial machine, albeit with metric cross slide and top slide leadscrews (the gearbox and leadscrew are imperial, natch), so the dimensions are all made in inches and the fasteners are all UNC / UNF threads etc. Luckily I have the manual in PDF for reference, showing all the "standard" parts and the exploded views.

Bed casting:

Rear view:


Rear view:

Cross slide:

And a quick idea of where the 10t and 32t pulleys might fit:

Still quite a bit to do to both the model and the concept itself but it's a start...

Friday, 30 October 2020

CNC the Bantam?

Where have you been, fatty? 

After sulking for the last couple of months over the cunning stunt Autodesk have pulled, resigning my shiny new 4th axis to the back shelf, I've been planning how to pull myself out of this pit of self pity and get on with something in the workshop.

It's not just been the sulking that's kept me away from the machines. I've been under the Domestic Manager's cosh, making some progress on the main bathroom refit. So far I've:

  • Completely stripped out 2-3 layers of old tiles.
  • Replastered all the walls
  • Repainted all the walls
  • Completely replaced the ancient tongue and groove flooring
  • Totally replumbed the hot, cold, heating, waste and soil pipes
  • Fitted a new spa bath with Mira Bluetooth electronic thermostatic filler
  • Fitted a fancy Roca karzi and wall mounted basin unit
  • Installed a 2m x 0.7m walk-in shower with Grohe thermostatic control
  • Installed shower panels at both ends of the room (alongside the bath and inside the shower)
  • Fitted 2m x 1.4m glass shower panel
  • (Almost) fitted tongue and groove PVC flooring
  • Replaced the door and replaced much of the door frame and architraves
  • Fitted new LED downlights in place of the halogen things that need replacing every few months.
That leaves the skirting boards and a couple of pieces of architrave, once the final pieces of flooring are in place. Then I can make a start on the master ensuite. Can't wait.

Sorry - CNC the Bantam??
Been thinking about the worn out old dog that is my Colchester Bantam lathe. I've made a few parts for it so far, such as the screwcutting clutch that I made but never quite got round to fitting. And the DRO scales that arguably would allow me to make accurate stuff on the thing. Currently I do my best using trial and error but having seen how the DRO transformed my manual Bridgeport mill, I can see how much more accurate I could get this thing to be.

I still have a Mesa 5i25 / 7i76 controller setup from 4 years ago, when I was first thinking of converting the Bridgeport clone to CNC. I took fright in the end, due to the almost (to me) impenetrable process of installing and configuring LinuxCNC and ended up using a Chinesium Newker controller for it instead - and a Centroid Acorn for The Shiz. So rather than simply fit the DRO to the lathe, why not go the whole hog and CNC the fucker? It would seem rude not to. 

I looked at a Denford lathe recently but the guy was asking to steal urine on the price so that went nowhere. Saved a major domestic, which I suppose is a blessing but left me still wondering WTF to do in the workshop next. 

My plan is looking like the following. Of course, plans are liable to change but it seems like a starter for ten:
  • Swap the cross slide leadscrew for a 12 or 16mm(?) ballscrew, with a timing pulley on the far (back) end, driven by a servo motor. By dropping the motor through a belt, it will clear the cross slide and allow a 3:1 reduction.
  • Fit a DRO encoder scale on the side of the cross slide. I'm always rather dismissive of people who believe closing the loop at the machine slide level can magically eliminate backlash and any other mechanical shortcomings. You need to get that right before you try to close any loops electronically.
  • Fit a ballscrew behind the bed to operate the X axis (carriage), as there's no room left at the front given all the leadscrews etc and the apron cluttering up the space. The only place you could fit one at the front would be so far down, away from the saddle that there would be all manner of issues with the saddle rocking as the ballscrew changed direction.
  • Fit a DRO scale alongside the X axis ballscrew. Again, this isn't some magic panacea but with a well assembled carriage assembly, there shouldn't be a lot of backlash to try to compensate.
  • I may try to allow manual operation by some means, such as a disconnector on the X axis ballscrew but if my experience is anything like The Shiz and the Bridgeport, the wireless MPG makes the handwheels pretty much obsolete unless you plan to make something quick and dirty like a toilet roll holder.
You said something about fans and motors?
First of all, I thought I'd start by uprating the puny 0.75kW motor to something a bit more becoming. For one thing, if I plan to CNC this machine, I'll need to be able to control the speed electronically rather than by endlessly changing the gear selection. Luckily I have a 3kW 4 pole motor from the anals of time (Wavedriver days in fact ie the 90s). I think this was bought to run a compressor to supply the air brakes on an EV bus but proved not man enough for the task. This is just a bit too big to fit into the available space in the Bantam. Bollocks.

Replace that with this:

But wait, surely I can chop it down so that it fits? With the pulley snug up against the motor at one end and the fan / shroud shortened, it looks as if it should just about fit. Let's do this. The main effort will centre around reducing the size of the fan and the shroud. 

Make a stub to hold the plastic (nylon?) fan:

Cut those vanes back a bit. Can't imagine it's going to compromise the thermal behaviour much - not on a Bantam at any rate.

And trim the register back a bit on the other size, to reduce the gap between the fan and the motor housing:

Now cut the cowling back, otherwise all that fan trimming will be in vain (vane?). I think this is the very first time I have ever used this Mitutoyo height gauge, to scribe a line at the desired height. They look nice but like a shaper machine they seem to have very little application in my workshop.

There. I need to hack the shaft back a bit, otherwise it's looking reasonable

A new, larger bore Taperlock insert arrived from Bearingboys (£6), as the shaft is 28mm instead of 19mm. I can mount the pulley right up against the motor, although the shaft sticking out doesn't appear to foul anything, so I shouldn't need to cut it back.

Now for some manual work. This is why machine tools were invented but there's no point messing about with the bandsaw here.


Finished. Used some bearing retainer to hold the fan to the shaft.

Ready for action. That seems to measure up against the space available.

Now to modify the mounting plate for the larger motor. Or more accurately, to replace the plate, as the motor foot mounting holes coincide with the existing slots in the original plate. Luckily I seem to have some 3/8" loominum plate that is almost exactly the right size.

Tuesday, 18 August 2020

Thermal Arc 220 AC/DC TIG welder!!

Yet another welder?
Oh no. Not another accident involving a mouse? Yep. But this time it's a non-functional TIG welder, spotted on a machine tool trader group on Facebook.

An AC/DC TIG welder like this is usually about £1k new (plus vat). You might get a hooky one for half that price off the back of a Chinese lorry / back of the factory / bootleg copy if you want to take your chances but even if I did, the Domestic Manager would crucify me if she ever found out.

This one cost me £100 plus carriage from Colchester to Lancashire. Despite weighing 19kg, that bit only cost me £7. I used parcel2go to compare the options and found that Parcel Farce 24h service was the cheapest. Bizarrely, Herpes consider this to be in the "Light and Large" category, thus warranting a £25 (+vat) charge. Well they can go whistle.

Sure enough, it turned up this lunchtime. Lots of cardboard. Hope the delivery guy didn't throw it out of the van.

Aha. There it is. 

Some ape has been in there previously.

The side covers slide out the bottom once you have removed the screws.

Thermal Arc was acquired by ESAB recently. One of the things they wasted no time over was erasing any previous technical data, service manuals, parts lists etc. In fact, you'd hardly know they'd even existed. The centre in Chorley where they possibly designed these things seems to have been closed down, so there's no possibility of ringing up and getting a copy of the manuals. Ringing ESAB technical dept resulted in a phone number. Ringing that led to a recorded message from a lady, giving the following numbers. Naturally both of those went to voicemail. So that trail petered out quickly enough.

I didn't bother turning it on, as I know it's DOA and there's no point charging up all the caps if I'm going to be digging around inside it. So let's do just that...

This is clearly the output stage / secondary board:

  • Date code Feb 2006.
  • An array of ISOTOP power devices. These are a mixture of diodes and IGBTs. They rectify the transformer output and switch the output to provide the AC or DC voltage. The copper thing works as a busbar but also as a current shunt for measuring the output current.
  • There are 2 optoisolators at the top. These drive the IGBTs on the output side.
  • 3 transformers provide isolated supplies for the controls - and gate driver circuits presumably.
  • Various connectors for the torch control, chiller control, gas solenoid, display, HF circuit and power supplies.
  • The big fuck off copper tabs lead to a pair of heavy windings in the output choke.
  • The feed from the primary board connects up to the underside of this secondary board, so you can't see it here.
  • No obvious damage or signs of magic smoke.

I need to note the connections, as I'll be taking this plate off to get to the ISOTOP devices. Note the red and black current sense wires. The big lug is the -ve output connection to the torch.

Here's the primary board. 
  • The rectified AC comes in from the input board as the heavy-ish red and black wires. 
  • There's an IMS board underneath this PCB, visible through the holes. It's a half bridge circuit, so presumably some discrete IGBTs and diodes.
  • The main power transformer is visible at the bottom of the photo. The primary winding comes onto the board as 2 heavy-ish black wires.
  • There's a current transformer that monitors the transformer primary current.
  • The secondary is centre-tapped and connects to the -ve (torch) output via the HF transformer. This injects a high frequency, high voltage signal onto the torch to enable arc striking. The HF transformer is at the bottom left, with the primary side wires in a yellow sleeve.
  • Control is done by an SG3525A (PWM controller with complementary outputs) on the piggy backed daughterboard. Fairly low tech stuff, as you'd hope.

The plate's out of the way so I can see the ISOTOP screws. Checking with the DVM, one of the IGBTs seems to be short circuit. Let's remove the board to have a closer look. 

I need to note these screw positions for later. It should be fairly obvious. Yes, it should be. So I'll take a photo then.

Noting the myriad connectors which will all have to be disconnected to get the board out:

Nearly there


Nothing particularly obvious

Some slight darkening but no real overheating when you look closely

This is the culprit. 
GA200SA60S ie 600V 100A IGBT from International Rectumfriers. Short between collector and emitter and also between gate and emitter. Nothing else damaged - the current transformer must have done its job.

The gate drive circuit is pretty simple. HCPL-3031 optoisolated gate driver IC with just a 22R resistor to each of the 2 IGBTs. Both resistors appear to be intact and both gate driver optos measure identically which suggests the damage is limited to the IGBT. Looks as if I should be able to simply replace the IGBT and be on my way.

Fortunately, Farnell stock an equivalent ST part, so I can get a couple delivered next day. I guess that will be Friday, given how late in the day it is now. But that's better than "20-40 days" via AliExpress.

Apart from cleaning it up, I can't do much more than just sit back and wait...

Friday, 7 August 2020

4th axis - finally completed

Finally all assembled and all machine screws fitted. No cockups or fettling required.

36mm through bore:

Fixing slots spaced to suit the table of The Shiz

Chuck at top front of box, to minimise intrusion into the machine space

4th axis - done.

TFFT. Now I guess I'll have to mount it on the machine and try it out.

Bantam CNC - sequence of events

 There's a right way to do this: Some of the critical machining work will need to be done on the lathe itself - stuff like modifying the...