This car has long had bouts of overheating. In the glove box when I got the car were receipts for a new head, belt, and radiator. Spendy!
On a trip to Jackson Hole, the car overheated quite a bit on the way over the pass. This was with a full load of passengers, mid-summer with the AC on. We had to stop a lot on the way up the hill to let the car cool off. Embarassing. In Jackson, I determined that the main fan wasn't engaging, and though the aux fan was flailing away it was insufficient to cool the engine. I wired the fan to the clutch using a coat hanger, and was able to get home that way.
At home, I shotgunned the problem. The car got a new clutch and mating fan (upgraded parts), water pump, thermostat, and fan belt. With the high mileage on the car, I figured it was time for these items anyway. With these repairs, I really didn't notice any more problem, though it did seem as though the car still got a little hotter than it should at times.
Fast-forward two years, and the car overheated again in high summer, but not a lot. It was also rarely a problem, and we made it through summer OK. In the winter, no problem.
Add another year, and a hot summer, and the car is again misbehaving. Temperatures are sometimes getting dangerously high, in 100 °F weather (indicated) with the AC on. It's time to do something. I look, and find again that the main fan isn't engaging. So I order another fan clutch, reluctantly.
Except that it doesn't fix the problem! Braniac then actually does some diagnosis, and finds that the radiator behind the fan is cold. So of course the fan clutch won't engage. So I installed a new thermostat, but the car still has the problem, intermittently!
Oh no, that's really pointing to the radiator, and those are expensive. (Or worse: leaking head gasket or cracked head!) Plus, it's not the original radiator, this one is only ten years old. I do some more research, and find that MB recommends a citric acid flush of the cooling system when this happens. The radiator is apparently easily clogged with deposits from corrosion. (They also recommend a new radiator, but I'm not going to go that far yet. I have not been maintaining the cooling system as it should be, mea culpa.) Also, I noted some corrosion on the bypass seat of the thermostat housing. This could help keep the bypass channel partially open, reducing flow through the radiator. Some have reported that corrosion there can be a significant cause of overheating.
Next step citric acid flush. I picked up 5# of citric acid powder at a baking supply house. (You know the kind, located in an industrial area, no marked front door, wholesale only kind of place, except that they don't care how much you get, so long as it's in a quantity they stock it in. I had my choice of 5# or 50# containers.) Cost was $10.79 out the door. You need 1kg (2.2#) to do one car, so I have another treatment in reserve. I took the old thermostat, probably still good, and converted it into a blocked-open thermostat like the special tool you can order from MB for flushing. A slit and cut-down .357 shell was employed to hold my extra thermostat open. (I find that I use shell casing brass for lots of things.) Also very handy was an old flathead valve-spring compressor, which pulled the thermostat open for me. Much easier than trying to he-man the thing open while my wife tried to wedge a stop into it!
I went to the boneyard and nabbed some radiator hosing and some fittings ($2), and I rigged up a flushing adapter (for the upper radiator hose) to go with my forced-open thermostat. With tools and materials in hand, and a free morning, I went at it.
Before I started, I had a better look at the corrosion in the thermostat (waterpump) housing. It was minor, and I don't think it requires replacement of the housing. I took a 1×1.5" stone (a chunk of marble flooring, flat but not polished on the back) and rubbed it around the thermostat bypass seat. This took off the high points of the corrosion, and should help ensure a better seal. Then I installed the open thermostat. I found quite a few bits of chalky orange crud in this housing. Looked like deposits to me! BTW, access to the thermostat is made infinitely easier by removing the air filter box and the inlet boot to the turbo, along with the EGR vacuum hoses.
I followed the manual's procedure that I had been given, which involved a 5-minute water flush, followed by a 15-minute (sealed) acid bath, followed by another 5-minute water flush. All of this with the engine running at various RPM's.
When I restored the normal thermostat, I found no sign of the chalky deposits from before. The acid bath had turned from clear to a yellowy-orangey color.
The only incident occurred during the re-fill. I poured in a gallon of new coolant, and as much water as I could get in. I had a serious air pocket in there though, and coolant wasn't circulating. The temperature gauge got all the way to the red line. Scary. While this was occurring, I shot the engine at various points with the IR thermometer. The head where the sender screws in measured 115 °C, closely correlated with the dashboard gauge. But the aux fan didn't come on, and its sender, in the radiator hose fitting, only measured 75 °C or so. The location of the dash gauge sender is a good one, it is able to monitor head temperature even with a severe coolant flow problem. I pulled the top hose, and managed to pour 1/2 gallon of water into there, slowly. Sounded like a steam iron. Ugh. After this, the temps were rock-solid at just over 80 °C on the gauge, and the various points of the engine measured good. For what it's worth, I think I measured 76 °C at the top radiator hose fitting, and about ten degrees less at the thermostat housing.
The lesson here is to back-fill the head from the upper radiator hose before you start. That'll save you much trouble.
Then I buttoned the car back up, belly pan and all, and threw the IR thermometer and a gallon of water in the trunk, for topping off and checking later.
However, it's cooled off a lot here, so I won't know if the problem is fixed for awhile. It ran at the right temperature coming in to work, but that's not too significant.
At noon the same day, indicated temperature was about 85 °F. Driving around with the AC on, coolant temperatures indicated about 90 °C on the freeway: roughly normal for this car. But, in stop-and-go traffic the temperature crept over 100 °C readily. The aux fan cycles when parked and running. (Though due to AC or coolant I can't really say.)
The IR thermometer showed the fan clutch at about 80 °C at these times, but it still doesn't roar when you rev the engine. The radiator is plenty hot behind it, too! (Burned myself...) At shutdown with the hood already open I could get back to the front of the car to find the blade just spinning to a stop. That seems pretty wrong to me.
When I got home that day it was still 85 °F outside, so I played with the car some more. The AC was on, and the coolant temp had risen to about 100 °C at this time. With the hood up, it was apparent that the main fan wasn't engaged. The aux fan cycled on and off. Radiator face temperature at the outlet ranged from about 89 °C when the aux fan kicked in, down to about 69 °C (explained later).
With the aux fan off, if I stuck my gloved fingers into the main fan to stop it, the exit temperature would go up about 1 °C per second. Letting the fan go again (still uncoupled) slows this rate considerably. With the aux fan running the exit temperature dropped at about the 1 °C/second rate.
All this was pretty consistent. The car seems to cool well enough if it can just get air. I'm sure that the presence of both belly pans doesn't help, but they sure quiet the car down so they're staying.
Exactly once when I wandered into the house on an errand and came back out I found the main fan coupled. There was significantly more air flowing even at idle. I grabbed the throttle linkage and revved the engine, and the air really started flying. Then the fan uncoupled again (I may have over-revved, or it just cooled itself enough) and when I looked, the indicated coolant temperature was 90 °C and the exit temperature was at its lowest, 69 °C. I was never able to get it to couple again. It was cooling off some outside by then. Of note is that the asymmetric (new-style) plastic fan doesn't make that howling roar that our older SL does. So it might be pretty hard to tell if the fan is coupled from inside the car.
So, it appears that the new clutch can couple, at least under some circumstances, but it's also pretty apparent that the indicated coolant temperature has to be well above 100 °C to make it happen. This is not right, this new clutch is probably in marginal condition.
I dug around in my crap piles (I have way too many car parts) trying to find the 'original' metal fan and clutch. The only likely candidate I found had a 116 part number on the fan. Is this what early SDL's would have had? (We don't own any 116 cars.) Of note is that the clutch it was attached to had a 95 date code on it. So, if this was the fan I took off the car, it's not the original clutch. So, that makes at least four clutches this car has had: The original, the '95 one I took off, the new one I bought at that time (with new fan), and now this one. And I don't think any of the three in my possession work right. I've got fan clutches out the..., well, I have a lot of fan clutches. Including several for older cars that I bought at the boneyard, for potential spares.
I think the lesser coupling of the newer clutches (when compared to the old ones) may have been done as a fuel-saving measure. But I think that they may have gotten a little too close to the edge, the clutches don't seem to hold up. Putting more silicone oil into them is starting to get mighty attractive.
There are several web sites out there about repairing viscous clutches, though most aren't for MB's. It appears that the most common problem is gradual loss of the silicone oil, with a concomitant loss of fan function. One such site, with pictures, is this. It shows the internal construction of the newer style of fan clutch.
I have seen threads talking about the fact that some Toyota clutches shipped without oil, which was separately available. Toyota reputedly offers three viscosities of oil: 3,000, 4,000, and 10,000 centistokes, of which the first two are the recommended viscosities for various applications. Toyota supposedly wants $6–7 per small 'dose', of which two can be required to recharge a MB clutch. A quick search for other sources of oil turned up The Chemistry Store, which charged $25 per gallon. Quite a bit cheaper, though with shipping and handling it's more like $40 per gallon. However, they only offer viscosities of 350, 1,000, and 12,500 centistokes; only the latter is likely to be useful in a clutch. I ordered a gallon of it, and hope to defray the cost by parceling out some of it to other interested parties.
I still have to figure out how to get the stuff into a clutch!
My 300E has been running too hot in summertime, especially in slow stop & go conditions. I verified that the electric fan was working properly, and yes, it kicks on high speed when coolant temps exceed 105 °C, as it should (thanks to a new switch—the old one wasn't kicking the fan on.) But, the electric fan isn't enough to cool the engine by itself, it's only there to help the main fan. With the engine temps at 110 °C, I killed the engine, and the fan kept spinning for several seconds... it was definitely not coupled. Radiator fins are clean and there was no cold spot in front of the clutch. Looked like a good candidate for the silicone refill theory.My thoughts on this:The Surgery:
I pulled out that half-liter bottle of 12,000 cst silicone fluid which I purchased a while back (from someone who bought a gallon of the stuff, and was re-selling smaller quantities.) The bi-metallic strip (BMS) is simple to pop out on the M103 clutch, it's more glued in place than mechanically locked in, believe it or not (unlike the OM60x clutches I've seen.) With that out, I used the previously mentioned method of heating the clutch, and slowly dripping in the fluid as the clutch cooled. I'm not sure how much I ended up putting in, as I didn't measure the quantity. I lightly clamped the back side of the bearing in a vise, so I could spin the fan. Before I started it spun easily with no resistance. I stopped filling when I noticed a decent increase in resistance. I glued the BMS back in place with a dab of RTV at each end, same as it was originally.
Results:
Well... good & bad, I suppose. First the bad. When cold, the clutch is fully coupled, and doesn't appear to de-couple after a few seconds (as a normal clutch would). The good news is, when warmed up (after a 20-minute test drive), the clutch does de-couple, and appears to operate normally. I let the car idle (in 100 °F ambients, with the AC on max) long enough to get the engine temp to about 100 °C, and around that point, the clutch coupled as you would expect (due to the BMS expanding). Now, I couldn't find anything in the M103 service manual about temperatures or engine RPM's, but when hot & coupled, the clutch stayed coupled up to at least 4000 RPM... it finally appeared to back off above that speed (I think the OM603 clutch is supposed to stay coupled up to 3500 RPM when the BMS is triggered.) Hot or cold, the fan now stops very quickly when the engine is shut off. It's not instant, but I'd say it stops within a half-turn. Oh yeah—so how does it cool? MUCH better... so far, I haven't seen the temps over 100 °C yet. I'll keep an eye on that though. Remember, this was compared to an old, shot clutch that was probably just floating along at idle speed all the time. I was trying to avoid plunking down ~$175 for a new fan & clutch.
Side note—the M103 clutch uses an 11-blade plastic fan. I may be doing this same procedure to a friend's M103 clutch, if so, I'll take some photos next time. I'd also prefer to locate some thinner fluid (6,000 cst might be good), at least for the clutches with plastic fans.
The source I used only had 12,500 cst oil, or very much lighter stuff (1,000 cst). I think that one could get a gallon of each and mix them to get the target viscosity you wanted, with a fair amount of leftover thin stuff, but that was more involved than I cared to do. The 12,500 cst stuff was a gamble.
I have a vacuum pot I made out of an old pressure cooker. That is what I'll use if I ever try this experiment, it should go a lot faster than the thermal method, since I have a nice shop vacuum pump. That and a clay dam to puddle the stuff in. Just puddle a known amount of oil and alternately apply and release the vacuum to drive it all in. One could incrementally add oil until there was an improvement, with less risk of overfilling. (Tedious to pull, doctor, replace, and test the fan a bunch of times though.)
I wonder about the quantity he put in, it seems like it may be somewhat critical to proper operation of the fan. Too much and the internal reservoir can't hold it all and the fan'll couple too much of the time, which might explain his problem. Or not.
