Restoration of 1952 MG TD 2

Author: Bob McCluskey
First posted: 1 Sept 2000
Last amended: Dec 2015
Please email Bob McCluskey
Car No TD/11935
Engine No XPAG/TD2/12333
Body Type 22381
Body No 11301/78948

Oaking it Work This section deals with the problems I had with making it run reliably. It is written as though it was solved linearly. But it wasn't, it was backwards and forwards with many false starts and hopeful signs and disappointmrnts.

At first, after the resurrection, all went well. But after the car had done a few hundred miles, when I started to have more confidence and took it for longer runs, a problem developed with overheating. This car had always been prone to overheating, but it seemed to be more severe after the resurrection. This may be because previously it had usually been used only for relatively short trips - to work and back, or to the shops, and when it got there, it had a long wait during which it could cool down. In contrast, now it had frequent trips of an hour or so, followed often by a relatively short recovery time.

The first attempt at a fix, long ago, had been to remove the thermostat. Well, no thermostat at all must be better than a fully open thermostat, because it must have even less resistance to flow, right? I could never work out why it made the problem worse. Of course, it was due to the radiator bypass. As everyone else knew, the bypass was designed to allow the engine to warm up quickly, an absolute necessity in cold climates, but a bit of a luxury here in Sydney. Originally the thermostat had a sleeve to cover the bypass: when the thermostat was closed (ie when the engine was cold) the sleeve would leave the bypass open, so coolant would be pumped back around the block without going through the radiator; but when the engine was warm, the thermostat would open, and the sleeve was supposed to blank off the bypass, so coolant would be forced to circulate through the radiator. By simply removing the thermostat there was nothing to prevent coolant from simply recirculating through the block, avoiding the radiator altogether, even when hot. The fix was pretty simple: blank off the bypass circuit altogether, so the coolant had to be pumped through the radiator, giving both a chance to fulfil their destiny. Most people make a little blanking plate to blank off the bypass; I was not so professional, and - just as a trial, intending to do the professional fix later - I used a cork to block it off. Luckily we had a champagne cork, left over from celebrating the car's first run, which was just the right size, and which, regrettably, is still there...

Anyway, on a run up into the mountains, it ran well, and because there was a long lunch at the top before returning it had time to cool, and I thought the problem was fixed.

But it wasn't.

At the next club run there was only time for coffee at the rendezvous, before setting off. And here either the problem, or a new one, resurrected itself. The engine simply wouldn't start. The carburettors flooded, so it couldn't have been a fuel starvation problem. When it did finally start, I drove home, the car spluttering and stuttering all the way: as long as it was in first or second gear, and with no load, it ran OK, but as soon as I opened the throttle, or changed up a gear, it missed and threatened to stall. Well, I had seen that problem before in one of the mistakes I described earlier: while there was no load, so the cylinder pressure was low, the spark could fire the mixture OK, but when the load came on, and the cylinder pressure was higher, the spark failed. The problem was that the condenser had failed, giving an open circuit, and wasn't fulfilling its critical role in the ignition circuit.

FIRST FIX: THE CONDENSER (or capacitor, to give it its proper name). I couldn't find a replacement to the original specification, but I bought one designed for high performance engines, and made it fit; I thought it would fix that problem, but after another failure I thought a bit harder.

The capacitor was originally constructed of two layers of aluminium foil, separated by impregnated paper. Nowadays the paper is replaced by a thin film of mylar, giving significant improvement in reliability (but, as everyone who has suffered a failure of the capacitor knows, not enough improvement).

When the contact breaker points close, current builds up relatively slowly in the primary of the coil - relatively slowly because it is resisted by the inductance of the coil, but in fact the charge time can be no more than about 1 millisecond for a four cylinder engine running at 6000 rpm.

When the contact breaker points open, there is no path for d.c. current. In the absence of the capacitor, the primary current has a problem. It cannot continue to flow, but, because of the inductance of the coil, neither can it stop instantaneously. It solves its problem by making a spark at the contact breaker points. Almost all the energy in the coil is used up in this spark; there is very little left to make enough spark at the secondary to fire the mixture. It might fire the mixture at low throttle settings, or in low gears, but will not fire it at higher cylinder pressures when the throttle is opened, or gears are changed up. You can try this by simply disconnecting the capacitor at the distributor, so you will recognise the symptoms when it happens in reality.

As it is, the current in the coil discharges rapidly into the capacitor. Consequently the magnetic field collapses rapidly, intersecting both coils of wire as it does. As it intersects the primary coil it generates a back-emf of about 300V.

Steve Maas has shown that the capacitors usually fail internally due to poor construction techniques giving a poor internal contact, and it is easy to see from his results that the internal contact could be intermittent or fail completely when hot.

When the capacitor fails it almost always gives an open circuit (if it failed to a short circuit, it would short out the points, and the current would never get established, far less interrupted). In that case, as discussed, the energy in the primary is dissipated in a spark at the points, instead of in the secondary.

Following Maas, I cut one open. Exactly as he says, the capacitor capsule itself consists of two layers of aluminium foil, separated by a very thin layer of mylar. They overlap the mylar: one layer of foil extends beyond the mylar at one end, and the other layer at the other end. The capsule goes into the case, where it is located not by any positive contact mechanism, but by the case assembly itself. In the picture, you can see the three indentations on one end of the capsule, which match the three spotweld points on the inside of the case, where the fixing bracket is spotwelded to the case. Although you can't see it in the picture, there is a similar deformation at the other end. This is all that locates the capsule, which is otherwise free to move inside the case, and it is all that provides electrical contact. I believe the brown colouration which you can see on the end of the capsule and inside the case is evidence of arcing and malfunction. I think you will see that it is less than reliable, and you would not bet your house on it.

According to Maas the capacitor should be close to 0.2 µF, and although the voltage at the primary is only about 300V, he suggests a replacement should have working voltage of 600V dc, and he suggests a possible manufacturer of a suitable device. I sourced a couple of MilStd 0.2 µF capacitors with 1000V dc working. Although the capacitor originally made a neat fitting inside the distributor, in fact physically it can go anywhere as long as electrically it goes between the contact breaker points and ground. I fixed one to a bit of printed circuit board and fixed it to the firewall, where it will be away from the hot engine. I did try it between CB and supply, which has the same ac equivalent circuit: it worked, as any electronic engineer should have expected, and I could have made a neater installation if I had left it, but I thought one change at a time was enough.

And that should be the last capacitor failure I suffer, and I expected that my stalling-when-hot problem would likewise be corrected. But it wasn't.


Next club run we had to take a ferry, and naturally had to turn off the engine while aboard. And the engine wouldn't start when it was time to disembark. There was a group of cyclists aboard, who helped push the car off the ferry, and enjoyed making clever jokes about the reliability of mechanical versus organic transport. Very embarrassing! The first thought was a vapour lock, which of course is known to be a problem with these engines. The fuel pump appeared to work properly: it was quiescent until I pressed the float chamber tickler, when it ticked away as expected. And when I slackened the carburettor union, it pumped petrol. So there didn't appear to be a vapour lock, and there wasn't a blockage in the fuel delivery line either, which I have seen. And after about 20 minutes, the car started normally and ran properly. And we passed the cyclists labouring up the hill on the other side of the river, but we didn't stop to discuss who was entitled to last laughs.

The problem - whatever it was - was clearly related both to overheating in the first place, and to getting a little rest after overheating. Paul Ireland has described a problem with almost identical symptoms, and in an impressively comprehensive series of tests, reckons the higher volatility of modern fuels accounts for this problem. Well, of course I couldn't get fuel to 1952 specs, so I began a systematic effort to eliminate any possible cause of running hot.

Well, one known cause is lean running. I couldn't set the idle speed, and in the past I have found that this can be due to a leak in the inlet system somewhere, probably in the inlet manifold, which of course leads to lean running. Using a piece of plastic tubing as an ear trumpet, I couldn't hear any leak around the manifold itself, but there was a faint hiss around the carburettor linkage. So off to the carbie shop to get the butterflies renewed. And that fixed the idling problem, and must have helped with cooler running. And the carburettor man suggested that I would get even better performance by advancing the ignition to a point just before the point where it started to "pink" under load.

SECOND FIX: THE DISTRIBUTOR The spark was timed perfectly against the timing mark on the pulley. But advancing the timing made a huge difference to power and to petrol consumption: now I could drive up several hills in top, whereas previously on those hills I'd had to change down to third. And it's getting well over 30 mpg, even around town. So now less fuel is being burnt, and it's being used to make the car go, whereas before at least some of it was being used to heat the water jacket and exhaust manifold. This was explained in another article by Paul Ireland, but it wasn't published till well after I needed this knowledge, and in any case I wouldn't have read it in time!

The distributor has three jobs to do: to make the spark by opening the points, to make sure it comes at the right time, and to make sure that it gets routed to the correct spark plug. For a four cylinder engine running at 3000 rpm, the spark must come reliably, with accurate timing, 200 times p sec, a cycle time of 5 mSec.

For maximum power, the mixture needs to be pretty well burnt, generating maximum pressure in the cylinder, when the piston is about 15º after top dead centre (ATDC). It takes a finite (and pretty well constant) time of about 3mSec for the spark to initiate maximum burn rate. (Not all accounts agree with these figures: some say less, some say more). But using my figures, because it makes the arithmetic easy and because anyway there is so much variability in the literature, it's a simple matter of arithmetic to calculate that at 1000rpm the crankshaft rotates at 6 degrees p mSec.

So at 1000rpm, it takes 2.5 mSec for the crankshaft to rotate from TDC to 15 deg ATDC. It follows that to get optimum timing, the points should open 0.5 mSec BTDC, or 3º BTDC. And at 2000rpm, the crankshaft rotates at 12º p mSec, so it takes 1.25 mSec to reach 15º ATDC, and the spark must be initiated 1.75 mSec or 21º BTDC. And at 3000 rpm, the spark must be initiated 69º BTDC. But to make it a bit more complicated, the burn rate isn't a constant function of engine rpm: fuel burns faster at higher concentrations, which correspond to wider throttle settings, which of course correspond pretty well to higher speeds. So the rate of advance isn't constant: although it needs to be advanced further betwen say 2000 rpm and 3000 rpm, it doesn't need to be advanced as much as between 1000rpm and 2000 rpm. In fact, the advance at 3000rpm isn't much more than at 2000rpm. All of this means there must be some way to make the contact breaker points open earlier at higher engine rpm, and the advance curve isn't linear, but tends to flatten off at higher revs.

But to make it even more complicated, if the engine is under maximum load at at 3000 rpm, the fuel density will be higher than if it is cruising at 3000 rpm, so the burn rate will be faster and the engine won't need so much advance. With our simple mechanical advance/retard, there isn't much we can do about that, but more sophisticated distributors also monitor manifold pressure through a pressure transducer and adjust the advance curve accordingly (vacuum advance/retard). (Actually, this explains one of the problems which have been troubling me on the Lotus: back-firing in the exhaust when running with a trailing throttle at about 4000 rpm. The Lotus has a similar mechanical advance/retard which is set for optimum with the throttle open, where it needs less advance; but when the throttle is closed, the mixture has lower concentration and it needs longer to burn, but doesn't get it because ignition advance is a simple function of engine rpm. So the mixture isn't fully burnt when the exhaust valve opens, and unburnt mixture ignites in the exhaust.)

To achieve this, our cars rely on a simple mechanical device driven by centrifugal force.

Two Lucas distributors fit my TD: the first is so-called DKY4A, fitted to TCs and early TDs, and the other is D2A4, fitted to late TDs and TFs. My car initially had the earlier distributor, but following a critical incident it now has the later. (What happened, one of the springs came off and wedged under the advance plate, and stopped the distributor from rotating; the pin locating the driving gear sheared, and the gear fell into the sump, where it will remain until I have to take the engine out again. This was yet another show-stopper.) The distributors are not identical, and parts from one will not fit the other. The weights are taller and heavier in the second system, and (presumably) the springs stronger. To accomodate the taller weights, the body of the distributor is taller. The shaft length is not identical, so the driving gear does not mesh identically with the driving gear on the camshaft; but close enough. Apert from that, they work the same.Bob Grunau has discussed the differences in more detail. A centrifugal advance control plate is attached to the vertical shaft. The bob-weights are pivoted on this plate, and engage with the plate by way of studs which engage with holes drilled into the plate. This limits the movement of the bob-weights and so limits the amount of advance the distributor can achieve. The plate is marked (underneath) with a number indicating the maximum advance. My initial distributor plate was marked 15, which meant 15º at the distributor; and since the distributor rotates at half engine speed, that corresponds to 30º at the crankshaft. If you're brave, you can increase the advance available by drilling out these holes. But don't do it.

As the distributor spins faster, the centrifugal force makes the bob-weights move outwards on the pivots. The movement is constrained by a pair of springs, one lighter and one heavier, which is loose on the weight. Initially the light spring controls the advance, and when it is spinning faster, the heavier spring takes over and the weights move less, so the advance curve flattens out, and becomes completely flat with no further advance at the limit set by the holes in the advance plate.

As the centrifugal weights move outwards, the cam is rotated anticlockwise, corresponding to advancing the cam compared to the distributor body and ignition points.

It really is pretty foolproof, and can go for long periods without maintenance.

The contact breaker points are the final critical component in the ignition system. Although simple in concept, yet like everything else MG, they can find many ways to cause frustration, and they have tried them all out on me (for example, early systems used a fibre cam follower in the points, which rarely needed adjustment; later systems used plastic, which looked identical, but which wore out quickly and needed frequent djustment).

Engine Timing

While it lived with me, this car had never had a working tachometer, and it had never had the generator gearbox to driive it. I had had the tacho itself rebuilt by the instrument maker when I had the speedo rebuilt; but I had never connected it. Suddenly it seemed important to me to have one.

When I first started the car after the long rest, the generator didn't work. I took it to the auto electrician, who told me the armature was unserviceable. He sourced a new armature for me, but of course it didn't have the all-important extension for the gearbox drive. So it lived without until now. The generator case was original, with the extended back plate and threaded extension ready to take the gearbox; so it was simply a matter of modifying the armature. The clever Dutch worked out how to make this modification and wrote it up many years ago in SQUARE FRONT - the magazine of the T-Type owners of Holland. I gave the problem to my mate Steve. Not to trivialise the problem but Steve made it look easy. He made up the drive extension (in Imperial units, not metric as the Dutch had), and sorted a couple of other problems with the generator body - the frontbearing, the threads on the rear extension, and the generator's mounting brackets. He made a bracket to hold the gearbox, to stop it from swinging down and shorting out the points - a well known problem.

Once I had a tachometer, I was able to measure the entire advance curve. This entailed marking the front pulley in ten degree increments. It wasn't easy to mark them off accurately, and next time I take the engine out, I'll do it more accurately. But in the mean time, the table below shows current best knowledge. The static timing is currently set to spark At Top Dead Centre, so I could perhaps advance the timing by a few degrees. In any case, I had expected a more steeply rising curve:

RPM 20060010001500200025003000
º advance 51517202530


And the next to be addressed was the cooling system itself, comprising water jacket and radiator. The first element here was a two-part rust and scale remover from Liquid Intelligence. Judging by what came out, there had certainly been a lot of rust in the system, but I cannot be sure that it got it all. The second element was the waterless coolant, also from Liquid Intelligence. They won't tell me what's in it, but they claim it has better heat transfer because it is more thermally efficient. And they claim it has a corrosion inhibitor, apart from having no water to cause scale and rust. But most importantly, it has a high boiling point - 190 deg C. They claim that with normal coolants, as they approach boiling point, bubbles start to nucleate inside the block, and these bubbles impede the heat transfer, leading rapidly to a thermal runaway. The high boiling point of this coolant means that these bubbles never get to form and the heat transfer is not compromised. Well, I have no way of verifying all these claims. But subjectively, judging from the temperature in the cockpit, it does seem to run cooler.

Horst Schach (p96) identifies another issue. He says that some engines have a small hole drilled through the intermediate water jacket, behind the core plugs at no 1 and no4 cylinders. Some engines have the hole drilled for no1, but not no4, some v.v, some both, and some not at all. These are claimed to improve water circulation and lead to better cooling. Neil Cairns reckons these holes, where present, were to simplify removal of sand from the sand casting, and they're so small compared to the size of the water jacket that they can have only minimal effect on the flow of coolant. I incline to believe Cairns, as I believe that not even MG with their casual approach to drawing numbers would have made such a change to some engines but not others wthout some documentation. But I still had the bloody problem.

Now I began to get desperate. I had previously rebuilt the petrol pump, with a new diaphragm, new valves and gasket, and of course new points. And even though I knew the problem lay somewhere else, I took it all apart again, and even reset the diaphragm return spring. And then I put it on the shelf, and bought a new pump, with electronic control. (It looks and sounds identical, and, although I haven't taken it apart, I'd be pretty sure the working parts - the diaphragm, the valves, the points, and possibly even the escapement - are the same, but with electronic contact breaker to avoid the old problem of burnt out points. It is out of warranty now, so I might take it apart to see).

And I bought a new coil - a Bosch GT40 Sports Coil. I found there was a lot of confusion over modern coils, which are intended for -ve earth systems and are marked simply + and - , and how they should be connected in a positive earth vehicle.

I resolved the issue to my satisfaction by connecting the + on the coil to the negative (live) battery terminal, via the ignition switch of course.

But it still did not fix the starting-when-hot problem, even though, shamefully (believing it to be wrong), I did try with the coil the other way.

Finally I thought to use a timing strobe to make sure the timing hadn't slipped, or worse. And there was no spark at all. And after about 20 minutes, as before, while I was still cranking the engine and trying to find the fault with the strobe, the spark suddenly reappeared and the engine ran as though there had never been a problem.

I am still a long way from resolving this problem. What could make the spark disappear and then magically reappear? I am currently leaning towards the contact breaker points as the problem: I think when it gets hot enough, the points bind on the bearing spindle. Of course they cannot jam closed, because the rotating cam would force them open; so they must jam in the open position. In that case, of course, no current would be established in the primary, and therefore no spark. I believe I have measured an open circuit when the points should have been closed, but it is not easy because firstly, it only happens after the car has been running for about an hour, and secondly, merely putting the probes of a multimeter onto them is enough to move them and make contact again.

Now, several thousand miles later, with the pivot on the contact breaker having been addressed, no more trouble, so I am convinced that that, not modern fuel, was the source of my problem.


It recurred one hot day at the busiest intersection in Chatswood. Everyone sounded their horns and waived as they edged past, and I was grateful for their sympathy and understanding. And it happened again one coolish Saturday afternoon after only about 10 km. It seemed to be happening more often and sooner. Everyone told me it sounded like the coil, so, even though my existing coil was only a few thousand miles old, I sourced a Bosch Blue coil - made in Brazil, standard equipment on Porsches, epoxy filled and supposed to be more tolerant to heat and vibration. It worked OK, and so on the first club run - about 200 miles - I set out with a light heart. On the way home, we drove up the motorway, perhaps a little faster than I normally did. And in the first traffic jam, around the airport, it first started to miss and then cut out completely.

We pushed it around the corner, out of the intersection, and I began to look for the problem.

Nowadays I carry my strobe and multimeter, so I was well equipped. As I expected, there was no spark at all. I measured voltages and resistances at all points in the circuit, and all was as it should be. I checked the little contact inside the distributor. I went over and over the whole system in the hope that it would have changed since the last time I looked. And one strange thing: if I took the HT lead from the coil out of the distributor, I could draw out a 1cm spark to the block. But when I put it back in the distributor, nothing.

Finally, I gave up, and let Margaret call the road service people.

And while we waited, a pedestrian came by, and stopped to chat. Her husband enjoyed tinkering with old cars; perhaps he could help? I could not dissuade her, and sure enough her husband came with a full bag of tools, and got stuck in. He did everything I had done, with the same result. But he did one thing more. We pushed the car until the heel of the points was on the highest point of the distributor cam. And the points weren't opening.

A quick application of his feeler gauges, and we were on our way, and I am extremely grateful, Elaine and Leigh.

It certainly gives me one more point of diagnosis. But as always with things like this, as one problem is solved it leaves more questions:
Petrol consumption had fallen off a bit, and possibly power. Those are both consistent with ignition being retarded, and that is consistent with the points opening later. But not enough to have made me suspicious.
Why didn't the points open? the bearing plate hadn't moved, so the heel of the cam follower must have worn. (And in fact in this composite picture, which compares a new cam follower with the existing points, you can see that the heel is very substantially worn).Why?
Why did it stop when it got hot, and recover when it cooled down? Something must have expanded and contracted marginally, allowing the points to open again. What?
How is it that I was able to draw out a spark from the coil if the points weren't opening, and why did it disappear when I connected the HT back to the distributor?
Next day when it was cold, I measured the gap again: it was tight at 15 thou. Leigh had set it to 20thou. Why had it changed? Had it worn further? Did it reflect hot v cold? But that would be wrong - if the gap got smaller when it got hot, until the points stopped opening, then it should have got wider when cold. Perhaps it was simply that the 20thou had been set in the dark when everyone had had enough.
The timing was significantly advanced by Leigh's adjustment - about 12 - 15degrees BTDC at about 700 rpm, compared to 5degrees at 600 rpm I had just measured. Well, that's just the result of increasing the gap, so the points open earlier. And it goes to the fact that I had thought the advance curve had been too slow.

I'm not sure this is the end of this saga, but at least I'll be better prepared next time.

More to come...

send me an email

Top of Page