Rebuilding an old surface grinder's spindle.

[Flynth]

This grinder is a polish 1960s make Jotes Spc20a. It's spindle is somewhat unusual so I decided to document my effort to rebuild it here. Hopefully it will prove useful to anyone trying to do the same in future and interesting for others

I've had this surface grinder for years and I can achieve very good finishes on it, but I always felt it took way too long for the surface finish to cleanup. Also to get good finish it requires a long warmup (likewise for truing a new grinding wheel).

So recently I started investigating and I found runout exceeds the limit for this machine (twice when hot, 4 times when cold). I disassembled the spindle and I found many issues with it.

Also the wheel hub thread was bent when I bought it. I always wanted to straighten it. A spindle rebuild is a good opportunity to do that.

I was hoping to remove the spindle entirely(it's a tube-like unit). Unfortunately it wasn't possible. I suspect because of corrosion.

Here we have few photos of the disassembled spindle and a drawing showing it's layout. It runs at 2700rpm and it uses very light oil (iso 4).

The outer taper and spindle housing tube:

The shaft with two roller bearings. Note the oil groove cut in the shaft, not the bronze bushing.


The bronze bushing, note unusual shape and the fact it is not split. That is very weird. It will certainly get deformed and it will contact only on three points! I guess that's fine?


Finally the drawing:

It is a nice design overall. It uses rubber clutch in the back and the pulley has its own bearings separate from the spindle this way the pulling force is not transferred to the spindle.

Many people modify this spindle for modern roller/ball bearings, but I enjoy my machines in their original configuration so I'll do my best to restore it leaving bearing swap as the final option (if everything else fails).

I started by indicating the shaft. It isn't easy (it took me a while to figure it out) as both centres are mangled. The front is on the bent thread, the rear has been drilled and tapped for a screw. In the process it got 20 thou (half a mm) off :-(

This is why I thought the shaft was bent, but thankfully it isn't.

Eventually I found a way using a lathe, a 4 jaw chuck and a steady rest.

I've managed to indicate it to 5micron runout (a bit over a tenth). Then I confirmed the shaft is straight. The taper is 0.01mm bent (4 tenths) which is sort of acceptable, but the thread was bent by almost 40 thou (a mm).

This is the setup I ended up using on the press. Sorry for a busy picture. The silver pucks are lead ingots used as padding to prevent the shaft from getting damaged.



Despite the indicators it was extremely difficult to measure how much the shaft deflects and how much it moves sinking into lead pucks. However I found a great method. I put the nut on the thread and I used gauge blocks tightening it so a block is grabbed. Then I could determine the movement by trying to insert the block between the nut's back and the front of the taper.

I managed to get it down to 0.2mm (a bit under a thou) which I decided is acceptable. Also I'm very happy I managed not to bend the shaft in the process

 

[Flynth]

So today I'm making a bearing spacer, but yesterday I managed to do some lathe toolpost grinding with a dremel-like tool. Long time ago I bought this good quality (rather expensive) dremel with a stand made by proxxon to drill precise holes in pcbs. It has very low runout as I successfully drilled 40+ 0.4mm (under 2 thou) holes with it in a pcb and the drill is still usable (a carbide pcb drill).

So I was expecting it to work well as a tool post grinder and it did. I mounted a cluster diamond dresser in the tailstock and I used green wheels from a cheap dremel set.

That's my dressing setup:

Here it is grinding a blued (with drying ink) shaft:


I was taking tiny passes just bumping the lathe wheel. Overall I took only about a thou off the shaft (radially). By seeing how much blue remains you can see how out of shape it was.

The end result:

And the wheel thereafter. Still usable

The shaft is not ideal. It is 0.01 mm out of round (much better than it was) and the finish has very slight wheel hop. You can see it in a reflection if you look very closely (it can be seen on the photo above). So I might lap the shaft with aluminum split laps.

This is the first time I tried tool post grinding with this dremel tool and dremel wheels(I don't have a proper tool post grinder). And I have to say, I'm very pleased with the results.

 

[Flynth]

The rebuild is done! I'm very pleased with the results. But first let's talk about the final stuff I did. I'm pleased to say after straightening the threaded piece in front of the nose the nut actually threads all the way I could've straightened it more. I left 0.2mm - almost a thou of a thread bend in fear of bending it the other way.

I've spent some time making a very precise bearing spacer (I managed to hit a tolerance of about 7 microns, further would be possible by stoning). However... Having taken measurements from the old spacer I failed to notice it is a wrong spacer for the bearings... Ahhh

It looks like someone replaced the rear bearings and left the original spacers with the outer spacer hitting the roller basket when tightened.

I spent quite a while measuring tolerances of the current bearings, but I haven't noticed the spacer is to thick radially. The correct spacer would be less than half the thickness.

So I gave up on the spacer for the time being and I decided to set the rear bearings preload with a nut and loctite it in place. The spacer I made will come handy if I ever decide to replace rear bearings with the correct type (unlikely to happen soon as they cost $140 for 2).

Additionally I made a comically large pin wrench (photo here https://www.hobby-machinist.com/threads/a-comically-large-pin-wrench.105873/) For adjusting the spindle's main bearing clearance.

I fixed the threads on the bearing clearance nut. You can see on my photos someone has beaten the front of the nut so hard it deformed threads a quarter inch deep. I chased the thread on the lathe and it became possible to thread the nut onto the bronze bushing (out of the spindle) all the way till the end of the thread making use of the entire adjustment range.

Then I used diesel, gas, an old toothbrush and compressed air to clean the inside of the spindle.

Finally I decided to press the old bearings back onto the shaft and reassemble the spindle. Here I hit the first snag... I planned to use 50~100N of rear bearing preload. When I removed all clearance and I set the lightest preload there was a spot on the rotation of the spindle that felt tight... I backed it off and gently approached it many times while measuring how far the shaft moves. In the end, not only I had to forego any preload... I had to put in axial clearance (while cold) of half a thou... This gets squeezed a little once a back cap is installed. I wasn't happy, but I could do nothing else...

Then I warmed up the spindle for 30 minutes which removed the axial clearance of the back of the bearings set. I kept increasing the front nut tightness. Here it is worth noting how difficult it is to measure a runout of a hydrodynamic bearing. While it runs it uses oil at very high pressure to provide stiffness. When off, there is no oil film... It seemed I hit my target runout way too soon, but I had no sure way of knowing the requirement is met already.

Having found a scientific article that talked about a very similar spindle (similar bearing shape, same top speed, same lubricating oil, only much bigger diameter) I scaled down the oil thickness they gave and I got a result of 10~15microns of desired clearance(I measured it by how much the shaft lifts at full speed vs stationary position). I slowly approached it while running the spindle and measuring the temperature every few minutes.

Eventually my indicators showed me under 10 microns of TIR (6~7), but I didn't really believe them. I couldn't measure it at speed, just coasting down which is not ideal. So I decided to do a grind test and it is a lot better.

I got this very good finish after 2 passes on mild steel with a 46K wheel. You can see in the reflections there is not even a hint of a wheel hop



Previously I would have to run 10 sparkout passes to get this. This was with a basic, not very open dress. I took a 1 thou "roughing" pass (5mm step over, 470ipm). Which left rhythmic scratches, then this cleaned up beautifully with the next single 4 tenths inch pass.

The spindle gets to 33C (in a rather cold 15C shop) after 45min of running.

So overall I'm very happy with the results I could probably make that clearance even less, but I'm not going to for the fear of ruining all my work by seizing the spindle.

Also I did some reading on hydrodynamic bearings and it turns out having the bronze bushing "crushed" non circular and deformed is actually preferred for high speed spindle because it prevents "whirl" which may even destroy it when it causes uncontrollable heat. Whirl is when the shaft starts "orbiting around thd geometric centre of the bushing with half the speed of rotation".

So here we are, my spindle rebuilding log. Sorry for long winded text.

 

[Firstram]

Very nice!

 

[Flynth]

Very nice!

Thank you

BTW, during this rebuild I learned one interesting fact that is very surprising to me.

Specifically, when I discovered the bronze bushing is not circular but what I could call pinched star shape I was expecting I'll have to scrape it to bring it to circular. However, that would be a very bad idea. I read approximately above 2k rpm the bushing is on purpose designed not to be circular. Elliptical shape, made by crushing the bushing is often used as well as various ways to pinch it resulting in varying number of outward and inward lobes. This is done because a circular shape exacerbates whirl. Whirl is a situation when the shaft doesn't stay put, but it starts to "orbit" the geometric centre of the bushing at half the rotation speed. Allegedly it causes energy loss and consequently temperature raise as well as complete loss of stiffness.

Also the design that pumps oil into the bearing gap through relative motion of its components is the very definition of a hydrodynamic bearing. It is similar, but also somewhat different that the usual slow speed "plain" bearing. Your typical slow speed plain bearing with gravity fed oil supply (as found in old lathe's for example) is often working in so called "boundary mode". This is a mode when 10% of the shaft/bushing surfaces still make some contact. In such situation you definitely do want it to be circular.

However, as speed and oil pressure increase this is replaced by a condition where there is zero surface contact. This is how hydrodynamic bearings work and for them various non circular shapes are used. See this drawing as example :


Then there comes another interesting bit of info. Hydrodynamic bearings are still today used in some high performance grinding spindles. For example I read about a large 90mm diameter spindle (that uses 35kg grinding wheel) bought new by some researchers that still uses a hydrodynamic bearing today. It runs at up to 2500rpm and it has up to 1.5 micron of runout... That's how good those spindles get.

So people that replace those "plain" bearings with rolling bearings have to use class 4 bearings for comparable results...I think I understand why the designers chose "a plain bearing" instead of a rolling element one. A rolling element bearing of comparable parameters would probably be a lot more expensive. Of course this machine was made long time ago in back then communist Poland so money wasn't that much of a consideration, but availability of high precision bearings certainly was. There were shortages of everything (famously, and somewhat hilariously toilet paper was in short supply too). Quite likely it was a lot easier to source bronze than class 4 roller bearings...