Re: WR90-SMA transition update

Neil Smith G4DBN

I can't get a traditional flycutter to produce a really good mirror finish though. What happens is that when the cutter hits the edge of the workpiece, even with the table locked, there is an audible thump and some signs of resonance, so the surface has some weird waves from the harmonics created. You can just about feel them with a finger, but they are down at the 1  to 4 micrometre level when measured with a Mahr Millimess on my big granite surface plate.  More cosmetic than anything, but annoying.

I found that laying some 2mm thick lead sheets on either side of the vice and putting big lumps of iron and/or buckets of sand on top really kills vibration.  Also I cast and turned up a 2 kg lead flywheel collar to fit the flycutter spindle, and that helped by increasing the inertia of the tool.  However, I had a cheap and rather nasty Chinese 80mm mill and arbour, which was almost unusable because the insert pockets were at varying heights, with up to 15 micrometres of variation. After seeing some advice on a machinists' forum, I tried removing all but one insert and runing a slower traverse speed and it was a revelation.  The mass of the head gives a reasonable flywheel effect to reduce the harmonics and wibble in the Bridgeport spindle and drive (Mostly in the varispeed drive I am guessing).  I might try try making a lead collar for the arbour, but I think the real answer is to make a copy of the Suburban Tool flycutters with a heavier head, maybe even machining pockets on the back of it to fill with lead, and using a 25mm lathe tool with a large radius insert.  The 2HP motor might struggle with really wide cuts, but the biggest things I need to flycut are heatsink faces, and 70mm radius should be sufficient for those. I would need to find a way to counterbalance the off-centre mass of the cutter though, as there is a limit to how long a 25mm insert tool can be, and I'm NOT getting into machining insert pockets, for there lies madness.

I am limited by the spindle speed of the BP, so I have made up a bracket to fit a 14,000 RPM spindle motor to the quill for doing really tiny milling with end mills of 0.5 to 1.5 mm, but even that isn't really fast enough.  What I should do is make a coaxial spindle with offset motor so the single bearings are in-line with the BP quill, but finding precision bearings rated to 35k rpm that would fit inside an R8 collet is pretty much impossible.  The bearings I used for the toolpost spindle on my lathe to cut the 0.5 mm (or less) choke grooves on the ends of the 122 GHz horns were 35 mm OD and cost around £90 a pair, but have sub-micron runout and can go to 48,000 rpm, which is almost fast enough for the 0.5 mm end mills.  Almost.

I'd certainly trade surface finish for the capability to do complex curves and 3D tapering forms.  Next move though is to get one of the new large Anycubic Photon Mono X 3D UV resin printers to make pattern pieces for sandcasting. The I'll be able make arbitrary shapes for brackets, mounts, feedhorns, enclosures and suchlike in aluminium or bronze without being dependent on my pattern-making skills.

Neil G4DBN

On 10/09/2020 07:59, Robin Szemeti - G1YFG wrote:
Thanks for the detailed info, I am a big fan of WD40 for aluminium. It is useless as a lubricant in most situations, but as a cutting fluid for aluminium, it works quite well!

You clearly have the old Bridgeport running well. My CNC router does not have the stability to achieve anything like that sort of surface finish, sadly, even though it cuts aluminium quite nicely otherwise.

On Thu, 10 Sep 2020 at 07:54, Neil Smith G4DBN <neil@...> wrote:
I ran a load of simulations in OpenEMS, but I’ll file the probe to length and measure the actual length, then see how repeatable it is. Without a trimming screw, I’ll have to file them all to get them on frequency I’m sure.
I’ll publish the OpenEMS model when I publish the drawings
Neil G4DBN

On 10 Sep 2020, at 06:27, Alex May M1BSX <alexmay@...> wrote:

A thing of beauty! :-)

Very intrigued as to how it was milled and whether you're using an empirical approach to determine the length of the "probe" ? 

73s Alex (M1BSX)

Alexander R. May

On Thu, 10 Sep 2020, 00:01 Neil Smith G4DBN, <neil@...> wrote:

This pic is my first try at machining an SMA coax to WR90 waveguide transition.  Milled it from a lump of 1 3/4 inch aluminium square bar on my ancient Bridgeport.    It needs a bit of trimming to get it bang on frequency as I left the pin a little long. Now I'm waiting for a load of Radiall SMAs to arrive from Mouser.

Neil G4DBN

I modelled it in OpenEMS with a three-inch length of WR90 mated with the oval cavity. Return loss (black) and through loss (red) look promising, I need to check the real thing once I have the probe filed to resonance.

Fusion 360 CAD image


Robin Szemeti - G1YFG

Join to automatically receive all group messages.