8 replies to this topic

[alpder]

Thought this might be useful. There's probably a pictorial guide out there already - quite possibly on this forum - but I didn't find it. Here's one way, though surely not the correct way, to disassemble and reassemble a Smiths mechanical odometer, using only rudimentary tools and without the [re]assembly jig which the factory would have used. Corrections and additions most welcome.

 

Warning: the odo is a pretty robust item. But the speedo is not. In particular it contains a very delicate spiral spring which returns the pointer to the stop. Excessively rotating the needle - in either direction - is likely to distort the spring and so affect the calibration of the speedo. So unless you're also planning to work on the speedo - perhaps to recalibrate it - avoid touching the needle.

 

These odo's frequently jam, usually when turning a thousands or a tens-of-thousands with all the zeroes to the right of the turning number lined-up. This is when the mechanism has to do the most work. Potentially, dismantling cleaning and reassembling the wheels will reduce the friction and fix the problem. If you've virtually given up on your speedo, and are about to replace it... well, this might be worth a try instead. As Spider mentions below: work in a clean space - anything magnetic on your tools or workbench will risk being deposited on the speedo's magnet and risk ruining it. A sheet of clean cardboard on the workbench. And wipe off any iron filings which may be on the tools you're going to use. AFAIK the odo is not oiled. Adding oil - and especially WD40 - will likely not help and can contaminate the speedometer itself, causing the needle to stick.

 

As with any dismantling job: take lots of pictures as you go for reference during reassembly.

 

The pictures show an odo being assembled to read "00000.0". I've since realised the holes manufactured in the parts enable them to be built on a jig in the factory to read "99999.9" because the first test would then usefully roll over all the wheels. A note in the text shows how to adapt this method to get "99999.9", still without the need for a jig.

 

A Smiths mechanical odometer:

 00a.SmithsOdometer.jpg   28.88K   1 downloads

 

First task is to get into the case, by removing the crimped-on bezel. Before unbending the crimps, make a note of how they're only bent over at no more than about 45 degrees:

 00b.UnbendCrimps.jpg   41.87K   0 downloads

 

Remove the face. If the glass is stuck to the bezel, best option is to leave it there: the gasket between will likely be perished and will be destroyed completely by dismantling. If you prefer to take it apart (for cleaning, say) then a length of soft "4mm sponge cord gasket" (ebay) will be useful to replace the perished rubber gasket between glass and bezel:

 00c.GlassParts.jpg   20.75K   0 downloads

 

If the face is in two parts (inner and outer) tape them together temporarily:

 00d.TapeSplitFace.jpg   30.39K   0 downloads

 

Undo the two screws on the back, and lift out the mechanism. Try to handle it only by the casting. Try not to get fingerprints on the face. Don't take the needle off: it is non-trivial to get it back in the right place.

 

This cogwheel moves one tooth for every rotation of the cable:

 00e.Cog.jpg   28.09K   0 downloads

 

And this white pawl moves the black toothed wheel one tooth for every one turn of the cogwheel. Consequently the TPM for the odometer is (cogwheel teeth) x (toothed wheel teeth):

 00f.ToothedWheelAndPawl.jpg   17.15K   0 downloads

 

Before proceeding, use a square screwdriver-bit or a small screwdriver to rotate the input shaft about 30 or 40 rotations (anti-clock, as looked at from the back). Observe the pawl's movement. It will probably try to engage a tooth on the toothed wheel, then slip. Have a good look at that one tooth the pawl is slipping on: it's probably seen thousands of these slips and if the result is that it's worn so badly that the pawl doesn't have a hope of catching it, you'll need a new toothed wheel. Also check that the pawl is free to move on its mounting - it should exhibit minimal friction. Check also its straight metal spring, which should spring the pawl down onto the toothed wheel - if that doesn't do its job then the pawl can't pull the wheel round.

 

Protect the input stem of the casting and lightly fit into a vice. It'll need to be mounted at different angles for different operations. Start by easing the pawl's straight metal spring off the end of the shaft and then slide it out of the pawl and put it aside:

 01.HoldInVice.JPG   46.92K   0 downloads

 

Ease the outer nylon circlip out of its groove. Incautious prising (like shown) can tear it. Better to remove by pulling it outwards with tweezers. Then slide it off the end of the shaft.

 02.RemoveOuterNylonCirclip.JPG   24.29K   0 downloads

 

Remove the three-pointed spring and the toothed wheel and the white ratchet behind it:

 03.RemoveSpringToothedWheelRatchet.JPG   39.39K   0 downloads

 

Now the bit that stumped me for a while: there's another nylon circlip inside the mechanism. To remove it from its groove, gently use water-pump pliers to press the tapered end of the shaft inwards:

 04.UnclipInnerNylonCirclip.JPG   43.37K   0 downloads

 

Partly remove the shaft, without letting the number wheels tilt (if they do tilt, the spring at the 10,000's end may ping away - that'll be game over):

 05.PartlyRemoveShaft.JPG   36.06K   0 downloads

 

Slide out the heavy metal washer then tease the spring out - the number wheels can be tilted to assist this but get a good hold of the spring first:

 06.RemoveWasherAndSpring.JPG   44.31K   0 downloads

 

The wheel parts, in the order they fit:

 07.ThePartsInOrder.JPG   34.26K   0 downloads

 

Disassemble the stack of wheels. Some types have a brass axle-tube and metal spacers. Others have plastic spacers with the axle moulded into them in short sections. But there's little difference to how the stack is assembled. Odo's which have plastic plates (this example has metal plates) seem more likely to jam because I can feel considerably more friction between the small cogwheels and the plastic plates than between the small cogwheels and the metal plates. Chuck all plastic parts into a plastic pot with some soapy water and wash. Strain through a sieve (not a colander) to recover them:

 08.DisassembleAndWash.JPG   37.78K   0 downloads

 

Note the two teeth - with a gap between - on one side of each number wheel:

 09.TwoTeethNearSeven.JPG   42K   0 downloads

 

Lay the first (tenths) wheel down in this orientation. If there is a brass tube, fit it:

 10.TenthsOntoShaft.JPG   37.65K   0 downloads

 

Assemble one plate and cog exactly in these orientations. The cog has three big teeth and three small teeth. A big tooth must be pointing outwards:

 11.AssemblePlate.JPG   34.73K   0 downloads

 

Flip the plate and drop it onto the number wheel with the prong to the right and the cogwheel towards you. The big tooth must go into the gap between the two teeth on the number wheel.

 12.OneBigToothIntoTwoTeeth.JPG   32.95K   0 downloads

 

Rotate the prong so that it lines up with the middle of the '5'. The small cog will rotate 1/6th of a turn as it moves slightly away from the 7:

 13aLineUpWithFive.JPG   33.93K   0 downloads

(This pic is of a different odometer with plastic plates - note how the main grooved shaft is temporarily fitted to steady the assembly, and that the first plate is different to all the others.)

 

Add another wheel, with the numbers exactly in-line:

 13.NextWheelExactlyInLine.JPG   34.61K   0 downloads

 

Repeat to make a stack, always inserting the big prong of each small cog between the two teeth adjacent to the '7' and then rotating each plate to keep all the prongs in line with all the '5's. (Edit: I guess in the factory the prongs would be alligned with the '4's, so the assembled odo reads 99999.9 - so that when the odo is first tested all the numbers should roll over, confirming correct assembly.)

 14.RepeatToMakeAStack.JPG   23.01K   0 downloads

 

It's really important that the wheels don't separate, even slightly: if they do then the small plastic cog may turn 1/6th of a rotation into the wrong orientation and, when the odo tries to turn that cog's wheel (maybe in 10,000 miles time) it'll jam. So cut a thin (1cm) strip of masking tape, and stick it across the 7s to create a solid cylinder:

 15.TapeAcrossSevens.JPG   32.61K   0 downloads

 

The metal disk would be fitted in the factory with its hole matching the hole in the tenths wheel - the jig would give the assembler no alternative. Edit: As far as I can see, though, there's no reason to use this particular alignment: instead, aligning it 180degrees around has a potential advantage, which is that the two or three teeth on the black toothed wheel which have been doing the most work (turning over the zeroes together at 10's/100's/1000's) are relieved of that work, which is transferred to potentially less-worn teeth on the opposite side. But I may have missed an important feature which requires the plate to be where it is... I leave you to experiment:

 16.MetalDiskHolesLineUp.JPG   29.87K   0 downloads

 

Rotate the speedo so that the tenths end of the fixing is slightly higher than the 10,000s end - this stops the metal disk falling out of the stack when fitting. Pass the pointed end of the shaft into the tenths end of the body, then push the inner nylon circlip onto it - just onto the very end:

 17.Rotate.StartShaftWithCirclip.JPG   26.16K   0 downloads

 

Drop in the stack of wheels with the 9s uppermost:

 18.DropIn.JPG   41.91K   0 downloads

 

Using a screwdriver from below, tease the prongs into their slot on the body. Little force should be needed to do this, although the prongs close to the tenths end may need modest persuasion so work on those first and the others will follow:

 19.RotateToLineUpProngs.JPG   45.88K   0 downloads

 

Press the shaft part way in:

 20.ShaftPartWayIn.JPG   32.69K   0 downloads

 

Rotate the body in the vice, then carefully tilt the wheels a little so that the spring can be placed (uncompressed) onto the end of the stack:

 21.DropSpringOn.JPG   26.75K   0 downloads

 

Tilt the stack as far back into place as it will go - so that the spring slots into the bracket on the body:

 22.SlidePartlyIntoPlace.JPG   33.9K   0 downloads

 

Use needle-nose pliers to gently grip the wire of the exposed part of the spring, and rotate it to wind it in towards the wheels. This will 'screw' the spring into place. (Trying instead to compress the spring and fit it in one go into its space would likely result in the spring pinging away to be lost):

 23.RotateSpringClockwise.JPG   25.65K   1 downloads

 

Holding the stack firmly in-line, slide the heavy metal washer between spring and body. Then push the shaft in until the inner nylon circlip (which didn't fall off earlier, or did it?) clicks into place in its slot on the shaft.

 

At the other end, fit the ratchet, noting how the two pegs fit either side of the bracket:

 24.FitRatchetBack.JPG   32.92K   0 downloads

 

Fit the toothed wheel, the three-leged spring, and the outer nylon circlip. Then slide the straight spring into the white pawl and engage it under the shaft:

 25.FitOtherParts.ReplacePawlSpring.JPG   22.56K   0 downloads

 

Remove the tape from the stack. Pull it slowly starting from the 10,000s end and do not allow the wheels to spread apart:

 26.RemoveTape.JPG   37.01K   0 downloads

 

Chances are the split face will have wobbled a bit, and allowed the needle past the stop. If so, remove the tape from the face, lower it down, and gently reposition the needle. Then refit the whole lot back into the case and fasten with the two screws on the back:

 27.Oops.JPG   33.87K   0 downloads

 

Test, by winding the odo on at least 1.1 miles, If you have a square driver bit for a drill then it's easy. But a small flat bit can also work:

 28.SquareBit.JPG   20.4K   0 downloads

 29.TestToOnePointOne.JPG   41.7K   0 downloads

 

To re-crimp the bezel, there'll be some proper ways to do it. Bashing it with a hammer and punch surely isn't one of them, even if tempting. This photo shows one way of doing it without the correct tools, while still giving good control over the amount of crimp. Press the speedo body downwards firmly into the bezel and onto the bench with one hand. Simultaneously engage the serrations in the uppermost jaw of the water-pump pliers onto the edge of the metal then - without trying to close up the pliers - pull the plier's handles towards you to tilt the top jaw towards the speedo. This movement - if the serrations grip well - will bend the edge of the bezel in a way almost indistinguishable from the original crimps. A very modest crimp is all that's required: and will be easier to dismantle in future than a full 90-degree crimp.

 999.Recrimp.jpg   39.93K   0 downloads

Edited by alpder, 22 June 2025 - 02:28 PM.

[Spider]

Great post, thank you for taking the time not only to do this sort of work, but for your well detailed photos and tips.
 

 Don't take the needle off: it is virtually impossible to get it back in the right place.

 

If you look just near the HB lamp, you'll see a dot (or on some faces, it's 2 dots close together);-

 

 

When re-fitting the Needle, in it's free state, align it to the dot, press it back on, then carefully move it back up to the right side of the Neddle stop pin

 

The only other tip I'll add is to be sure you are working in a very very clean environment, be mindful that there's a magnet in the Speedo too, so any metal filings etc will find there was on to the magnet and easily jam the unit.

[KTS]

i did the same a while back, but just put the details in my project posts

 

i removed the speedo pointer as it made it much easier to get to the odometer (..to be honest I didn't think it would be possible to do it otherwise, so kudos to you for proving otherwise)

 

a couple of things i noted when doing it was that the small (transfer?) cogs were asymmetrical; not immediately obvious, but will mean the tumblers don't sit square if put in the wrong way round.  the other was that there's what appears to be an assembly hole in all the tumblers and transfer cogs that makes putting everything back together again much simpler

 

https://www.theminif...ckup/?p=3655300

[alpder]

Great post, thank you for taking the time not only to do this sort of work, but for your well detailed photos and tips.
 

 Don't take the needle off: it is virtually impossible to get it back in the right place.

 

If you look just near the HB lamp, you'll see a dot (or on some faces, it's 2 dots close together);-

 

 

When re-fitting the Needle, in it's free state, align it to the dot, press it back on, then carefully move it back up to the right side of the Neddle stop pin

 

I have seen this trick, and thought about using it. The mark is, clearly, the notional zero of the meter though the needle can't get there because of the stop. However, I first experimented by removing the stop and seeing where the needle would come to rest. Even with the meter exactly level, the needle would stop at different places quite far either side of the mark after being disturbed - it doesn't really settle at the mark every time. That's presumably why there Smiths don't try to display speeds between zero and the stop: they're too unreliably reproduced. So I figured that removing and replacing the needle was likely to be a bit of a lottery because of this. I imagine it might be possible to spin the speedo up to a known speed, and fit the needle at that point. That way, there are two strong forces (the spring on one hand, the magnet on the other) opposing each other so that the spindle is forced to a specific position (as indeed it must be: at a given speed from ~10mph upwards, the needle always points to the same place). That would require - say - a stepper motor driven by an Arduino... pretty simple but still not in most people's shed.

[Spider]

There is a 10 Photo limit in each individual posting, but as other said, just do a Pt 2, Pt 3 etc.

 

I have seen this trick, and thought about using it. The mark is, clearly, the notional zero of the meter though the needle can't get there because of the stop. However, I first experimented by removing the stop and seeing where the needle would come to rest. Even with the meter exactly level, the needle would stop at different places quite far either side of the mark after being disturbed - it doesn't really settle at the mark every time. That's presumably why there Smiths don't try to display speeds between zero and the stop: they're too unreliably reproduced. So I figured that removing and replacing the needle was likely to be a bit of a lottery because of this. I imagine it might be possible to spin the speedo up to a known speed, and fit the needle at that point. That way, there are two strong forces (the spring on one hand, the magnet on the other) opposing each other so that the spindle is forced to a specific position (as indeed it must be: at a given speed from ~10mph upwards, the needle always points to the same place). That would require - say - a stepper motor driven by an Arduino... pretty simple but still not in most people's shed.

 

If the magnet and bearings of the speedo are clean, it will consistently settle back to the same place. The Bearings are a 'Jewel' style (that may or may not have a Jewel actually fitted) These are designed to run on points are are bone dry of lube as any type of lube has 'stiction' that needs force to overcome. I clean this style of bearing with Shellite as it dries without residue. Most Brake Cleaners will do the same. I loosely fit the Needle, zero it to the Dot, then check that it's consistent by flicking the Needle a couple of times, then press it on firm. I than gently bend over the Stop a little and at the same time, lift the needle a wee bit to get it over the Stop.

 

Spinning it up would work, but you need to be sure it's in calibration in the first instance,,,,,
 

[alpder]

If the magnet and bearings of the speedo are clean, it will consistently settle back to the same place. The Bearings are a 'Jewel' style (that may or may not have a Jewel actually fitted) These are designed to run on points are are bone dry of lube as any type of lube has 'stiction' that needs force to overcome. I clean this style of bearing with Shellite as it dries without residue. Most Brake Cleaners will do the same. I loosely fit the Needle, zero it to the Dot, then check that it's consistent by flicking the Needle a couple of times, then press it on firm. I than gently bend over the Stop a little and at the same time, lift the needle a wee bit to get it over the Stop.

 

Spinning it up would work, but you need to be sure it's in calibration in the first instance,,,,,
 

That makes sense. The speedo I was playing with, when I tested the return-to-zero, had previously been opened. On dismantling I found it very dirty. Possibly even someone years ago oiled it to "fix" it.

[Spider]

I see you've managed to edit your opening post and get it all in there.

 

Good work and well done on both fronts !!

[alpder]

The assembly method can be slightly adapted to rebuild an odo with a non-zero reading, but it is a bit trickier. Also, building with the odo reading 99999.9 allows a proper test, as the numbers roll over to 00000.0 . If you'd still like to build with a non-zero number...

 

Take the number you want. Say, 05921.5. Round down to the nearest 10. This example gives 05920.0 .

 

Assemble as far as having two wheels and two plates fitted, with the prongs lined up with the 5's. This'll read "0.0" when fitted.

 02PairOfFives.JPG   19.02K   0 downloads

 

Add the third wheel, with the desired number adjacent to the zeroes. In this case '2' to read "20.0". Optionally, tape together (starting at the 7's) as you go:

 03ThirdWheel.JPG   19.28K   0 downloads

 

Then fit the plate with the cog and, just as before, insert the large tooth of the cog between the two teeth adjacent to the third wheel's '7'. Tip: a small pen-mark on the cog indicates where the big tooth is hidden underneath, giving confidence that it has been correctly positioned between the teeth on the wheel below:

 04FitCog.JPG   27.16K   0 downloads

 

Then rotate that plate so that its prong exactly lines up with the other two prongs. Then continue to build up with the remaining wheels ('9' then '5' then '0') in a similar way:

 05Stack.JPG   28.66K   0 downloads

 

When fitted:

 06Fitted.JPG   45.58K   0 downloads

 

After fitting into the instrument case, wind on the odo to the desired reading as a test.

Edited by alpder, 22 June 2025 - 07:56 AM.

[Spider]

Great work.

 

I'll move this over to FAQ in due course.

 

Many thanks for taking the time to document and post here.