Some time ago I gave up on using a Dremel mini drilling machine. It’s one of those that take 3.18mm (1/8″) tools and has collets to hold the bits. I should have returned it when it was bought it because it was really quite bad right from the start. I do not know how many drill bits I broke with this machine. Finally, I opened it up to replace the bearings and guess what; I could not find any! As far as I could disassemble it, I really did not see any. Add to that the drill stand which is mostly plastic on a metal column, would easily move a millimeter, probably two sideways. But what could one expect at that price? Well maybe a bit more….. So out it went.
Time for something better
Now I got something-something that looks a bit more robust, a Proxxon IB/E. It has the front end of the house in some metal alloy while the back end is plastic. Not quite built like a machine spindle but it is not intended to be, and not priced as one either. I knew it had a proper ball bearing near the spindle end so there was hope. The Proxxon is priced about twice the price of a Dremel, while machine spindles start from at least twice more again and go as high as you can afford.
Next, I slowly turned the drill around with my fingers, taking readings here and there. The dial was swinging about 40µm end to end, that is plus-minus 20µm peak to peak.
The Proxxon is definitely much more robust than the Dremel was. Much lower noise, vibration, and definitely lower runout than the Dremel even if I never did any measurements to compare. I did, however, make measurements on the Proxxon, after it had a few hours of use.
I measured with a 1µm (one micrometer) resolution micrometer dial and the machine fixed to the table, with a 3.18mm drill bit and the meter right outside the collet.
First I measured the radial lag. by pushing the drill with my fingers to one side, letting go, taking a reading, pushing it the one way, taking another reading. There was a lag of around 15µm end to end.
Next the axial lag; placing the meter along the axis of the drill. Pulling the drill bit out with my fingers and letting go. To my surprise, there was a 300µm axial lag, just by the spring force in the micrometer, which is really not much. Actually, one can easily feel this axial lag it with the fingers.
Summary of spindle measurements:
|Radial lag||15µm end to end|
|Radial runout||40µm end to end|
|Axial lag||300µm end to end|
Radial movement is quite acceptable to me while the axial movement is not.
Can we improve on that?
I was quite convinced could be better than the measured lags and run-out. After all, I am a tool geek!
Looking at the drawings of the machine, (attached) one can see that there is one bearing on the spindle shaft end and other bearings are inside the motor, probably not accessible. The front bearing size is 8x14x4mm, a fairly available size, quite easy to get so I ordered a new ceramic precision ball bearing to replace the original one.
One nice ting with quality tools, and basically all older machines and tools, is that you can repair them, refurbish, fix them. Usually, the newer and cheaper things are, the less possibility do you have to take them apart and repair.
So now we disassemble. Void the warranty!
The whole unit is held together by four long screws.
They are Torx size 10
type, so do not try with hex / Allen key. The machine comes apart nicely; a back end with the electronics, a center body with the motor, and the front end metal house with the spindle.
A dead end
I assumed the hub was threaded on the shaft. However, it would require a special tool to remove it. So I greased the hub, filled it half full with epoxy and stuck a 6mm hex wrench in there. I let it cure for a day and then tried to unscrew the hub. It did not move even he slightest, despite heating the hub with hot air and applying as more force than I should. The conclusion was that the hub was not removable and the fan must come off.
Removing the spindle
The spindle is mounted with a circlip holding the bearing in place. The problem is that the fan a shaft coupling is blocking access to the circlip. There is no way one can get a tool in there to reach the circlip, without first removing the coupling and fan. Well removing the fan should be enough.
Looking in the parts list at Proxxon, the fan and spindle are available as a spare part if something breaks. So I tried as gently as possible to turn the fan on the shaft and again, with some force, yes, it moved.
So I the hob oiled it with a thin penetrating oil left it for a while and tried again. It would turn but was obviously not threaded, not coming out. So time to try to lift and pry it out. Hmmm, what tool? Finally I found that two teaspoons, the thinnest ones I had, could slide in between two of the cooling slots, just under the fan.. I had to bend the spoons a bit to curve them around the hub. Then, gently prying, the fan came of nicely! Now the circlip was accessible with a narrow long nose pliers or, in my case, a narrow medical needle holder, and the circlip could be removed!
After that, with a light tap on the spindle end, spindle and bearings come out.
Fitting the new bearing
Now it was becoming obvious that the problem with the lag was probably not the bearing itself, but the mounting of the bearing. In the parts list there is a spring washer that was not there on my spindle. That would explain the axial lag. The bearing itself was also neither tightly pressed into the housing or onto the shaft, which would probably explain the 10 mil or so lag in radial movement. So what to do? The proper way would be to get a ball bearing with a tighter fit or to add some plating to the shaft and housing, but both are really out of the question here.
So I went for a simpler option; locking the bearing to the shaft and housing by some epoxy. I could not use thread locking fluid or superglue, as I would need some minutes to assemble the whole thing before the liquid cured.
I used a semi-fast epoxy which allows 30 minutes of aligning, and I used about half of the time assembling. Most of the time was used to get the circlip back in place. The little epoxy used to fill the axial movement was probably a little too much. And when pushing the spindle in, that excess epoxy really has no place to go. It took some quite hard pushing with a thin screw driver to get that locking ring into place.
It was at once obvious that the axial movement was gone. The original 0.3mm movement was easily felt with the fingers. After the modification, there was no obvious axial movement felt.
After assembly, the front-end housing and spindle were warmed with a hair dryer to speed up the epoxy curing, while at the same turning the machine on for short periods to make sure the spindle shaft was centered.
Already now, the machine was obviously running quieter than it had been. Most noise is from the motors carbon brushes.
The machine is quieter and has fewer vibrations. Axial lag has gone down from 300µm to about 10µm, radial lag has been reduced while run-out is virtually unchanged, so it could be due to movement in the upper spindle coupling.
But now, overall movement is within about two 100ts of a millimeter which is good enough for me. And just knowing the bearing will withstand hundreds of hours is good.
I will now give it some hour of use before doing measurements again.
Any comments appreciated!