Speedo Repair

Anthony Rhodes has produced a very useful document 'Repairing Jaeger and Smiths Speedometers' as typically used in MGBs. This is the 2002 version (there is also a 2000), one source of which indicates there is a 2004 version in the pipeline. However as I only came across that reference in 2014 it seems that it was never produced for some reason. It also does not go into much detail on the odometers, hence this page, which should be read in conjunction with his document.

Speedy Cables is often mentioned as a source of speedo repairs, but there have been complaints of these taking several weeks.
JDO Instruments offers a 48 hour turn round which has been verified by members of the MGOC MGB Technical forum.

Removing the mechanism from the case
How the odometers work
Changing the mileage reading
Reassembly to the case
What was wrong with Bee's trip?

Removing the mechanism from the case

The bezel has six (on the 4" instruments) tabs that fit over a flange around the case. There are six corresponding cut-outs (arrowed) in the flange, so the bezel needs to be twisted round until the tabs line-up with the cut-outs. If it's never been off before it will probably be stuck fast, but try first. Lever each of the tabs back a little at a time then try twisting, until it comes off. Only bend back the tabs the minimum amount, or it will distort the side of the bezel and be visible when fitted. The only seal is between the bezel and the glass, so it shouldn't be that making it difficult to remove. The glass should be stuck to the bezel seal so come away with the bezel, the edge reflector may also come away or be left inside the case. In the latter case (ho ho) it should hook out easily enough.

Getting the mechanism out of the case in two speedos I have is nothing like as simple as Rhodes describes. He says:

"Remove the two screws on the back and, if necessary, the tiny screw holding the reset cable (TR2-4). Then press the reset shaft (TR2-4) into the case and then push the threaded end (where the speedometer cable attaches) inward." For a start on mine there is a roll-pin connecting the reset knob to the spindle that has to be pressed out, but more importantly the spindle still projects through the case by about 3/4", which is a lot more than he shows for Triumph speedos. It doesn't push up into the case more than a fraction, and that holds the lower part of the mechanism back in the case, and only the top tilts forwards. This means the threaded boss for the cable tilts, which jams in the hole in the back of the case, even if the rubber gasket is pushed out of the way. The dial also tilts, and the lower edge of that fouls the lower half of the case. You need to remove the pointer then the dial in order to get the guts out, but as Rhodes says MG pointers are tighter on the spindles than Triumph speedos! He gives a method that works easily enough once the mechanism is out, but that's no help at this stage. The only way I have been able to do it - being nervous about levering on the pointer too much in case it damaged it or the spindle - is by removing the screws from the face, inverting the speedo so that the dial drops down onto the back of the pointer, then as the mechanism starts tilting in the case, ease the lower edge of the dial over the edge of the case. There is no clearance, so even this method requires some force, raising the possibility of damage to dial, pointer or spindle. The first one I did came out that way, but when trying to reverse the process to put it back in the case, the pointer pinged off! I was sure I had broken the spindle, but fortunately it had only become detached, and refitting it in the correct place was easy enough. You will probably need to push the gasket out of the screw holes so that it comes out with the mechanism, as getting that off the threaded input boss while the mechanism is half way out isn't easy.

The second time I opened up Bee's I didn't want to force it, so extended the hole that the reset spindle protrudes through into a slot, and that gave just enough forwards movement of the mechanism so the dial cleared the case while still attached to the frame. I cut it roughly with a pair of tin-snips with the mechanism still fitted, then cleaned it up after removal. This allows just enough additional forward movement for the dial to clear the edge of the case while still screwed to the mechanism. The felt washer still covers the extended hole, but you could always fit a bigger piece.

Once out of the case his description of removing the pointer: "Remove the pointer by gently turning the drag cup until the pointer is at 60 mph. Then gently hold the drag cup in place. Hold the pointer by the hub and pull and twist the pointer until it pops free of the spindle" did work for me. Then removing the dial exposes all the works.

Almost immediately I could see exactly what was wrong - the pawl and gear that advances the mileage one step at a time weren't engaged. The gear (A) was way out of line with the pawl (B), because the shaft on which it sits was partially withdrawn. I pushed it back which got everything working again, but very little pressure was needed to push it out again, so I had no confidence they would stay engaged. There are two split plastic locking clips (C and E) sitting in grooves in the shaft, each side of the frame. The outer one was very firm, but the inner one came out of its groove (D) all too easily. Could be the clip, or the shaft, but I didn't feel like sending off for bits yet. I reckoned if I could fashion a clip that hooked onto the frame and pressed lightly on the end of the errant shaft, that would hold it in place.

At first I thought about a springy strip, but settled for spring wire, and half an hour or so of careful tweaking seemed to do the job, and driving with a drill sent the wheels merrily round.

How the odometers work

All Rhodes describes is how motion gets from the input shaft to the odo. The input shaft has a worm gear, with a cog engaged with it. He says the cog can have 22, 25 or 32 teeth but doesn't explain why. It will determine how many movements of the pawl occur for each turn of the speedo cable, but it's only part of how many turns of the input cable are required to advance the odometers one mile. In order for the worm and cog to mesh correctly either the worm tooth pitch will have to vary with the cog pitch if the cog diameter stays the same i.e. both components change, or if the pitches stay the same the cog will be of different diameters and even though the worm will be the same the cog's shaft will have to be closer to or further away from the worm which means a major difference in the frame.

The cog is on a shaft with an offset pin at the other end, to which is attached the pawl (A). This offset pin and other features means that the pawl is pulled back and pushed forwards once for each full turn of the cog, and it is the push that advances the blue gear (B) on the odo one tooth. No matter which way the input shaft turns the pawl goes back and fore just the same, so irrespective of whether the car is going backwards or forwards, this mechanism means that the odometer will only go forwards, contrary to what Stephen King would have us believe in that a car going backwards will make the mileage reduce (let alone reverse the ravages of time and miles ...). There is no spring pulling the pawl down onto the blue gear as you might expect, indeed there seems to be a finger sticking out which looks like it is exactly for that. But even though the pawl is below the blue gear, and if you pull the pawl away from the blue gear, as soon as the pawl starts moving forwards it moves smartly onto the blue gear and engages with it. However, the forces that cause the pawl to move down onto the blue gear when the car is travelling forwards, also cause to it lift away from the blue gear when the car is travelling backwards, so in fact the odo won't move at all when the car is going backwards. The blue gear is keyed onto the main odo shaft, which goes freely through all the wheels, and is keyed into the black gear (D) at the opposite end. A three-legged spring presses the blue gear against a yellow component (C) that is keyed to the frame. The blue and yellow parts have serrations on their adjacent sides which acts as a positive friction grip. This is easily defeated by the action of the pawl on the blue gear, but prevents vibration from moving it independently of the pawl.

The number of teeth on the blue gear determines (for a given worm and cog) how many movements of the pawl are required to advance the odos one mile. That causes a variation in the size of the blue gear, but because the pawl can take up any position it can cope with any size of gear, so the blue gear is potentially the only component that needs to change to get different rates of odometer movement i.e. different numbers of turns per mile, as indicated by the coding on the dial. This speedo for example takes 1280 turns to advance the odos one mile.

For each turn of the blue gear, the black gear at the other end of the main odo turns once, and this advances the units wheel of the main odo by one digit. At the same time movement of the black gear goes via an intermediate gear and turns the tenths wheel of the trip odo via another black gear, which moves that one full turn for each turn of the main odo blue and black gears. The black gear on the trip odo has a similar friction arrangement to the tenths wheel as that between the blue and yellow parts of the main odo, which means in the normal course of events movement is transmitted from the gear to the tenths wheel of the trip odo.

As to how movement is transmitted from one wheel to the next, it's much easier to look at the trip odo as the gears on that are external. Incidentally, it is because the gears are internal on the main odo that makes it so hard to alter the reading, far harder than modern electronic odometers! Each wheel, on the side adjacent to the succeeding wheel (the 'up' side), has a single slot in its edge (A). This slot passes under the gear when the wheel changes from 9 back to 0 and advances the next wheel. At all other positions the only way the gear and the wheel fit together, is when one of the missing teeth (C) on a gear is facing the edge of the preceding wheel. The other side of the gear has teeth all the way round (D), as does the mating face of the next wheel up in the sequence (B). This means that each wheel turns independently of the next gear up except when it is changing from 9 back to 0, and it always moves in unison with the next wheel down when that is changing from 9 back to 0.

As the lowest wheel turns, a spring is pushing them all together, and friction is trying to turn all the wheels together, and each wheel is trying to turn its preceding gear. But because the 'up' side of each wheel only has one tooth slot in its circumference, the next gear up cannot turn, hence the next wheel up cannot turn. It is only when the digits are changing from 9 back to 0 that the single tooth slot on the 'up' side of a wheel comes round to the next gear up. The friction that is trying to turn all the wheels and gears pushes the next full gear tooth into the slot, and as the slot moves under the gear the gear turns two teeth, which advances the next wheel up by one digit. By that time the slot in the preceding wheel has passed, so the next missing tooth on the gear wheel rides on the outside of the wheel, until 9 comes round again.

Both main and trip odos operate in the same way, but the trip has the additional complication of being resettable. This also has a keyed shaft, turned by the reset shaft that protrudes from the bottom of the dash. Each odo wheel consists of two parts - the outer with the digits, and an inner. It is the inner that is keyed to the shaft, and that rotates freely inside the outer.

The inner has a finger (A) that lies inside the outer, which has a protrusion (B) at one point, and these two act as a ratchet. As the outers turn (anti-clockwise in this image) with increasing miles, this protrusion flicks past the end of the finger. But when the reset spindle is turned, all the inners turn together, again anti-clockwise relative to the outers. So now as the finger on each inner reaches the protrusion on its outer, it locks against the protrusion, and turns its outer at the same rate. Eventually all the fingers have locked against the protrusions, and all the outers are turning together. This goes against what is described above for normal movement, but under reset conditions the wheels are forced round which pushes the gears out of the way, lifting the pin up against the lever springs. At this time all the digits should be in line, and one keeps turning the reset spindle until all zeros are shown in the slot in the dial.

When zeros are displayed in the dial aperture a pawl (A) on the frame drops into a slot (B) on the disc (shown here above its normal position) adjacent to the hundreds wheel which is also keyed to the shaft and rotates during reset. This makes it harder (but not impossible) to turn the wheels any further, which helps the user stop at zero and not wind it past and have to go all the way round again. Note that this disc has the boss protruding more on one side, and that side goes against the frame.

While the reset shaft is turning the tenths wheel it's driving gear cannot turn as that is locked to the main odo, and the normal friction created by the serrations between the driving gear (A) and the ribs on the tenths wheel (B) is overcome to allow the tenths wheel to turn independently of the black gear.

Changing the mileage reading

With the mechanism out of the case I wanted to change the mileage reading of the main odo, in order to retain that from Bee's original. Very little info on this part in Rhodes, except to say all the bits have to be withdrawn from the frame and reassembled by hand, as short of driving the mechanism for the required 30k or so that is the only way to do it. Warnings in Rhodes of a fine coil spring that pings across the room at the slightest provocation in mind, I eased out the shaft which still leaves everything else in place between the side-walls of the frame, then gradually eased those parts towards the edge of the frame. When the centre hole just came into view, I used a thin stiff wire to push through all the components, so I could withdraw them all keeping them in the correct order, and more importantly avoid losing any parts.

Now I can separate all the components - 25! - laying them out in the correct order ready for reassembly to the required mileage, which is an extremely fiddly process. Each decade consists of the numbered wheel (A), a gear (C) which fits between adjacent wheels, and a carrier (B) to hold the gear in position. Incidentally the carrier can be either metal on earlier odos e.g. 68 to 73 MGB A suffix, or plastic on later e.g. 74 MGB BS suffix. Each wheel has a cavity both sides, the one adjacent to the preceding decade has teeth all the way round, and the other is smooth all the way round except for one position where it will accept a tooth, similar to the trip odo. The carrier has a slot to hold the gear, and a finger which engages in a slot in the frame, which holds the gear in the correct position between the wheels. As in the trip odo the gear has to be fitted the right way round so that the 'all tooth' side goes into the succeeding wheel, but in such a position that when any digit is square in the aperture in the dial, one of the missing teeth is facing the edge of the preceding wheel. It is only like this that the preceding wheel will fit over the gear. On the 'all teeth' side there are two teeth per digit, so it's also possible to get a wheel showing half one digit and half the next through the aperture in the dial - obviously incorrect.

So it is 'simply' a matter of sliding each wheel and each gear carrier onto a shaft one at a time, with the gear slot uppermost to prevent the gear falling out, installing the gear in the carrier the right way round, with a missing tooth facing upwards, the digit in the correct position i.e. not half and half, and the fingers on the carriers all pointing in the same direction!

It took me several goes getting each wheel, carrier and gear correctly aligned. I'd work along the row, only to find that by the end one of the preceding ones had moved out of line somehow. Finally I added the spring locator (H), spring (G), stepped washer (F - with the step facing the spring) and locking clip (E) over the end of the wire, and replaced the shaft with a thin stiff wire, keeping the wheels and black gear pressed together so nothing could slip out of place. I compressed the spring while inserting the whole assembly between the sides of the frame as far as the wire would allow - which is further than the shaft would allow, then withdrew the wire, and carefully pushed the parts further between the frame sides until I could just see the hole through the middle, and reinserted the wire. Between withdrawing and reinserting the wire is the tricky time, as everything could ping out. With the wire back in everything is safely retained, can be eased fully into position ready to withdraw the wire and reinsert the shaft. The shaft has a keyway but it only engages with the blue gear outside the frame, and the black gear at the opposite end. I'll leave it up to you as to whether you fit the locking clip (shown already fitted to shaft A), 3-legged spring (B), blue gear (C) and yellow component (D) to the shaft first then insert that from the pawl end as I did, or the other way round fitting them last. The final step is to get the fingers in the gear carriers in the slot in the frame. The later plastic ones will probably be easier in this respect as they are flexible, with the metal ones you will have to locate them before the shaft is fully in position. Note that the spring locator (H) fits round behind the 10k wheel and over the finger on that gear carrier. This centralises that end of the spring, with the stepped washer centralising the other end.

It was only when I'd done that, that I realised I'd set the mileage exactly, when really I needed to set it back a few miles, and turn it manually to check all was well before putting it back in the case and in the car. A pal had the lateral idea of driving the car with the old speedo for a few more miles, then turning the new one on manually to match. But given the salt on the roads impatience got the better of me, so I took it apart again! I wanted to achieve 60515, which was relatively convenient as I could set it to 60498 or so, then I'd only have to wind it forwards 17 miles, and could at least check the first three wheels. As it can only be put back together with the gears in the right place, and as long as I ended up with all digits in line, the others should be fine. So that was done, and by turning the blue gear by hand rather than relying on the pawl I was able to advance it pretty quickly, and all was well with both main and trip odos.

Reassembly to the case

When you are finished refit the gasket to the case, and put the mechanism back in the case without dial or pointer. Fit the dial, then the pointer. There is a mark on the dial (line on MGB, spot on Triumph apparently) just below the zero, and this is the zero mph datum mark for the pointer. Fit the pointer over this mark, then depress the stop (it will go straight into the dial against spring pressure) and move the pointer past it. How accurate this datum is I didn't know, particularly as the small amount of friction between the pointer spindle and its bearings means that the pointer is stable either side of the mark to a certain extent. Rhodes describes a very complicated way of checking the calibration, but all I did when I accidentally pinged the pointer off when refitting the mechanism to the case with dial and pointer already fitted, was to raise one rear wheel, run the engine at 1667rpm (for my 4-synch) in straight top, and popped the pointer back on directly over the 30mph mark! I subsequently checked it against road-side speed indicators, and it was spot-on, maybe I was just lucky. The tachometer is easy to check against an external device, of course. A slightly lower rpm - say 1620 - will give about a 1mph over-read at 30mph for both 3-synch and 4-synch boxes. A sat-nav will probably also give you a true speed reading to check the speedo against. This time having already fitted the pointer using the datum, I checked it the same way as previously. It came out either spot-on, or maybe under-reading by a fraction, but I'll leave it as it is until I get the chance to check it against a roadside sign or someone else's sat-nav.

There is some kind of seal between the glass and the bezel (bezel seal), completely hardened and cracked on both speedos, as well as a square-section O-ring between the case and the dashboard (dash seal). The only two places I have found that stock seals such as these in the UK are Holden and AES.

As far as the bezel seals go Holden don't list them. AES list a flat seal for an 88mm gauge but not for other sizes. They do list a sponge cord seal which you cut to length, in 3mm and 4mm thicknesses, 4mm being recommended for 'larger' gauges. However MGB 'half Vee' bezels have a triangular profile facing the glass. The inner edge of the bezel (B) rests right on the glass, but at the outer edge (C) there is about 7mm of depth available. The diameter of the glass is such that it fits inside the case, except that it rests on the 'reflector ring' (a pressed metal ring that is black on the front and white on the black, which reflects light from the internal bulb onto the dial). This means that the edge of the glass (A) is less than half-way between the inner and outer edges of the bezel. Even the 4mm sponge cord would just sit loosely in the gap between the edge of the glass and the outer edge of the bezel, I can't see how it would be compressed to form a seal. If it were glued close to the inner edge of the bezel, such that it was under the glass, then it would form a seal, but because of the slope of the bezel from the inner to the outer edges there would be significant pressure trying to push the seal towards the outer edge, and out from under the glass, and I can't see it staying in place.

AES also have a 4" under bezel seal, which I'm assuming goes between the glass and the bezel. This looks to be round profile like the cord seal, but without knowing the diameter or thickness this may be no better than the cord seal. If it sat just inside the outer edge of the bezel it would be useless. If it sat just outside the inner edge, then initially it would be far too thick to allow the tabs on the bezel to fit over the case rim, unless in the process of fitting it was squeezed down the slope of the bezel towards the outer edge of the glass. I've seen a statement that the original bezel seal is a dum-dum type of substance i.e. a kind of non-setting mastic. There are many mastic/butyl strip/bead products around, but most end up being 15 for 8 metres which is a daft prospect for a couple of gauges. A possibility is this Black Tack at 5 although at 19mm wide it would need cutting into narrower strips.

For O-ring dash seals Holden list 100mm (4"), 80mm and 52mm (2"). AES sell 4" square section and 2" square section but not the 80mm as far as I can see.

What was wrong with Bee's trip?

Flushed with success, and with no prospect of taking Bee out for an extended test of the odometers now we have salt on the roads, curiosity got the better of me in terms of wanting to find out what was wrong with Bee's trip. Using the slot I had cut previously I got the mechanism out quite easily, with pointer and dial in place. That allowed me to remove the pointer by turning the cup to 60 mph and twisting and pulling as with the 'new' speedo, and remove the dial.

Remove the felt washer (F), copper washer (E), steel washer (D) and spring (C) from the reset spindle. Remove the circlip (B) from the spindle just above where it goes through the lower part of the frame without losing it. Slide the shaft down and out of the yellow plastic gear (A), which will fall free.

Remove the gears (C) and pin (A) they turn on by lifting the pin up against pressure from the lever springs (B) above it and pulling forwards. Note that the left-hand spring sits in a thinner section of the pin to locate it. Take the gears off and refit the pin to hold the springs away from the wheels. Withdraw the shaft from the reset spindle end. Ease the wheels forwards together with spring (D) at the tenths end, and the white disc (E) from the hundreds end, and out of the frame, being careful not to damage the pointer hair spring - you may have to lift the pin up a little to do this - or lose the coil spring.

With the wheels separated you will see the inner part with the finger inside each of them. Insert the shaft into a wheel, lining up the keyway. Turn the shaft and the inner part should rotate very easily inside the outer part, at least until the reset finger locks against the protrusion. Test each of the wheels this way, and this is when I think I discovered the problem with Bee's trip.

The tenths wheel was very free, the hundreds a little stiff, but the units and tens were both quite stiff. As the problem usually occurred when changing a units or tens, I reckon the stiffness was preventing those wheels from trying to turn with the normal friction between the wheels, which was preventing the gear from going into the single slot in the tenths wheel and so moving the succeeding wheel. As to why it was stiff, I had noticed oil around the hair spring, and between the two parts of each trip wheel where one rotates inside the other. So is it possible the oil has caused the plastic to swell in some way? The white digits particularly on the trip odo are also discoloured to varying degrees, could that be from oil?

As to why oil should be so visible, many years ago I had a flicking pointer, but rather than the rapid flicking that indicates a problem with the cable it was only occurring as the tenths wheel on the trip odo moved. Taking the speedo out of the dash and testing the input shaft it did feel stiff, so I put a drop of oil on it, which freed it up, and after that the pointer was steady. Now people do say that too much grease in the cable can work its way up into the speedo and clog the works. But I did only use a drop of light machine oil, and looking at the back of the spinning magnet and cup there is no oil there, or on the frame round it, so I don't really understand how it could have appeared on the trip wheels from that source. And to be fair the odo digits have always been stained to some extent.

Cleaning the oil from the two halves of each wheel made no difference - not even making them stiffer. I pondered how I might free them up a bit, and eventually settled on putting each inner in turn on the shaft, where it is keyed, then putting the shaft in my drill-driver. A strip of fine wet-and-dry folded round the boss that fits inside the outer, and a few moments running the drill was making them freer. After that it was just a matter of reassembling and testing. Rather than running the input shaft from a drill it's much easier to use one of the black gears on either the main or trip odos as a thumbwheel which advances things much faster, slowing down as the tens (in my case) was about to advance to see if there is any tendency to stick, and after a couple of hundred 'miles' all seems well.