Early 1.9 to 2.1 conversion - stop me if I'm being stupid!

[bigherb]

I guess Bigherb's point about the total mechanical advance on the DG distributor is an important one. I don't know what they'll be able to tell me about pinking/knocking from their kit. .

You use a microphone or a piezo sensor to detect pinking/knock.

[CJH]

I haven't checked my own one, but I guess I have a C or J variant on my DJ, since the advance didn't pick up until about 1500rpm. I do have a spare Q variant which I could try.

I've checked - mine is indeed a 'C' variant, which explains the late appearance of the centrifugal advance (~1500rpm). I think it's at least encouraging that this 'homebrew' method of measuring is accurate enough to distinguish.

I think it would make sense to refurb my 'Q' variant and fit that, as the early part of the centrifugal curve matches the DJ variant better. That would mean I would set the idle advance to 10º, instead of 12º, since there'd be no need to try and compensate for the lag in the C variant.

[adie89]

Brass nut (like the ones on the exhaust studs in the earlier photo) on the left, copper lock nut on the right.



Except the copper nuts aren't copper after all. They must be copper plated steel.



When you say that copper nuts don't last, what happens to them. Is copper plated steel ok?

Brass is used because if it corrodes it doesn't bond to the material it's attached/screwed into. That's why a lot of old timer cars use brass and copper fittings for coolant/water. The brass will never stick, it will also fail before damaging the harder item it is usually attached too

[CJH]

Brass is used because if it corrodes it doesn't bond to the material it's attached/screwed into. That's why a lot of old timer cars use brass and copper fittings for coolant/water. The brass will never stick, it will also fail before damaging the harder item it is usually attached too

Thanks Adie. Beats me how exhaust studs corrode anyway - the only time they can possibly get wet is when the van's on the road, at which time they'll be far too hot for moisture to remain. Anyway, I went with stainless studs and brass nuts and a stainless exhaust - so I reckon I'll be fine.

[CJH]

I've prepped a 'Q' suffix distributor, with the aim of fitting it this evening and measuring the centrifugal and vacuum advance curves. I bought it for a song off ebay a while ago as it was described as faulty. As a bonus it came with a cap that seems unused.



The vacuum unit didn't work at all, and I'm pretty sure that was the only problem. I stripped it down, cleaned it thoroughly inside and out in a paraffin bath, blasted it dry with an air line, lubed it up and reassembled it with a new vacuum unit, new dust cap and a new oil seal. It all seemed ok inside - centrifugal weights move freely and the spring is in tact. The timing test this evening should show if there are any other faults. Hopefully the centrifugal advance will begin at a much lower rpm, so that it better matches the lower end of the DJ curve and doesn't need the extra couple of degrees to compensate for the slow start.

[itchyfeet]

is the new vacuum unit a powerspark?
do you have a link please I need one too.

[CJH]

Yes, it's the Powerspark unit - their model number IA_Bosch_vac. They list it several times as variously for Audi, Porsche, VW. I bought one from them at a show a few years ago, and asked whether it was exactly the same as an original unit - they said it may be a few degrees different, but that many people had used it successfully. So I ran my DG on one of these for a year or two and it was fine. I've since bought a couple more to put into 'stock' as spares, and it's one of those that I fitted to the Q suffix unit last night.

I initially assumed that the critical part was the advance curve, as a function of vacuum, and that the shape might be a bit different from an original unit. And I figured that measuring and comparing that curve would be difficult. But with a bit more understanding of how they work I think it's the total advance that's more important, since that's present from idle - the curve plays a part as the load goes up and the advance reduces, and I suspect the transition, i.e. the length and shape of that curve, is not the critical factor - I could be wrong, but from those earlier measurements I feel that a difference of a few degrees isn't critical. So with that in mind I've measured the amount the arm moves between zero vacuum and full vacuum on the Powerspark unit, and when I swap the distributors over this evening I'll do the same with the genuine vacuum unit that's currently in the van, so that I have an idea how different the total advance is. And of course the timing measurements will tell me also.

[CJH]

I fitted the 'Q' suffix distributor - it works fine. So then I measured it, and was quite surprised by what it shows. I had it set to give 10º at 850rpm without vacuum.



The chart needs a bit of explanation. The solid lines are the DG and DJ charts from the plots in the wiki. The DG plot has been shifted to match the 12º of centrifugal advance that I had inadvertently set on my engine. And it now seems that the DG plot in the wiki is for the 'C' or 'J' suffix version, since the centrifugal advance doesn't kick in until about 1500rpm.

There are two measured DG 'C' plots (dashed lines with dots) - one with just mechanical advance (grey) and one with vacuum advance added (green).

There are also now two measured DG 'Q' plots (yellow and light blue).

There's also a plot with three yellow dots, representing the data points in Bigherb's earlier table (upper values of the quoted range). These three dots ought to lie on the yellow DG 'Q' line. The first one does, the second one does if you squint a bit, and so maybe the third one would be where I'd have been if I'd gone to 4200rpm.

So it seems the 'Q' suffix distributor looks like it adds centrifugal advance rather quicker than the 'C' suffix version, but actually the rate is similar at lower rpm, it just starts earlier. According to Bigherb's table all the different variants should give the same maximum centrifugal advance at 4200rpm.

The Powerspark vacuum unit seems to add about the same amount of advance as the genuine unit (compare light blue-minus-yellow with green-minus-grey).

So where does this leave me? No idea really. Is the earlier accumulation of advance by the 'Q' version better? VW must have thought so or they would have stuck with the 'C' and 'J' variants. But maybe it's only better for the DG engine, since the 'N' suffix (for the DJ version) seems to accumulate advance uniformly over the whole range. And there's still this question of *total* advance to answer, since the 'N' suffix version doesn't seem to specify any more than 16º of advance beyond 2400rpm.

Having had this 'Q' suffix version apart, I suspect it might be possible to limit the total centrifugal advance by putting some packing between the weights and the end stops - that would be quite an exercise though. I think I'll wait and see what the folks on the rolling road say.

[waltraud]

Marco Mansi uses coated studs I believe, strong but protected.


Sent from my iPhone using Tapatalk

[CJH]

I've borrowed Andrew's 123Tune distributor and had a play this afternoon - it all seems to work fine and I can change the advance curve using their software. So with that, Itchyfeet's DJ 'N' suffix and my two DG versions (C and Q suffix) I should be good to go for next week's rolling road session.

I found the data points for the two 123Tune curve definitions that they supply for VW distributors. They don't say which engines they're for so it would be wrong to try and draw any conclusions from them. They supply one that's described as 'One-fits-all' and one that's described as 'Performance'. So I thought it would be interesting to compare them, and also look at how they compare to the measurements that Itchyfeet and I have made on the 'C', 'Q' and 'N' suffix versions. This plot is for the mechanical advance only - vacuum advance confuses the issue as it's dependent on the throttle opening.



The two 123Tune plots are the dotted lines. Firstly their 'performance' curve accumulates advance more slowly than the one-fits-all, and it maxes out at a lower total advance. I don't know enough to explain why that would be. But the low-to-mid rev range of the 'performance' curve is more like my DG 'C' spec, while the limited maximum advance makes it more like the DJ 'N' spec.

I'm hoping the folks at the rolling road will be able to talk me through the subtleties of these curves and help me decide which one I should use. I'm really hoping that it's one of the DG versions, because the DJ 'Q' spec is apparently quite rare and I can't afford a 123Tune unit!

[itchyfeet]

Proper geeky....love it

[CJH]

Long post alert!

Well I had a very enjoyable and instructive morning on the rolling road this morning. Got through quite a few setups, and was beginning to struggle to keep everything in my head towards the end. The main thing for me was that it was full of surprises! Quite a few things came out contrary to what I was expecting.

I've spent some time digitising the printouts that I was given after the session, so that I can plot different combinations to try and draw some comparisons. By adding in the plots from the previous session when I had the DG in the van, I've now got three 'variables': Engine (DG or DJ), carb (DG or LT) and distributor (suffix 'C' and 'Q' are DG versions, suffix 'N' is Itchyfeet's DJ version, 123 is Ajsimmo's 123tune unit).

So to try and make sense of the plots I've labeled each curve according to engine/carb/distributor, so DG/LT/'C' means engine=DG, carb=LT, distributor='C' suffix.

A few caveats about the plots - the dyno computer plots 'smoothed' curves, and odd things happen to the smoothing algorithm when there's an abrupt change, such as if the rev limiter cuts in, or if the driver tries to stop the test before it cuts in but is a bit hasty or a bit late pressing the button. So this means that the curves at the very top end should be taken with a pinch of salt. And the other thing is that no two runs are ever the same - so small differences between plots are within the noise of the technique.

So to start with, I've got the plots from the earlier session with the DG, that showed there was no point in fitting the LT carb to the 1.9.



Next up is the comparison of the 'standard' DG versus a representative curve from the DJ. The DJ setup was the one that I arrived in this morning: LT carb and Itchyfeet's 'N' spec distributor. Wow, what a difference. Max power up from 76.7HP to 102.7HP (34% up) and max torque up from 106.4lb-ft to 147.6lb-ft (39% up). I was delighted with the 102.7HP, bearing in mind the injected models are rated at 112HP. But then came surprise number 1: he told me that he'd measured the drive train losses (during the slow down phase of the test) to be 18.3HP, so my power at the flywheel is actually 121HP. I was skeptical so I asked him how it could be possible for a carbed engine to make more power than the injected version, given all the previous discussion about the relative size of the throttle body versus the venturis. He said it's not uncommon, and is apparently due to the better atomisation of the fuel through the carb versus a spray mist from the injectors (there's more evidence that the bigger throttle body may not be important in one of the later comparisons). We also discussed my non-standard Speedshop exhaust, and my non-standard (Silverbullet) camshaft, both of which could help boost the performance compared to the standard injected DJ.



With this DJ/LT/'N' setup we did some small tests with different timing, and saw very tiny differences at both ends of the rev range, from which he concluded that the timing curve was already spot on - normally changing the base timing will pivot the power curve around the mid point, since it will improve one end at the expense of the other. That didn't happen with the 'N' suffix distributor, meaning the curve is good everywhere already - no real surprise there, since it's the distributor that VW specified for this engine. But surprise number 2 was that the LT carb was just about right everywhere too. On the DG it was far too rich, and given that the LT is a 2.4 litre engine I expected that it would still be too rich on the DJ. In fact, the main reason I went today was to get it re-jetted. In fact it was only ever-so-slightly rich across the board, but apparently not enough to warrant reducing the jet sizes.

So next we tried the DG carb, to prove what a clever boy I'd been in fitting the bigger LT carb, and also to see what could be achieved by re-jetting the DG carb for the DJ. Surprise number 3: there isn't a huge difference between the two carbs. Power is lower than with the LT carb from about 3500rpm, and reaches a maximum of only about 10HP off at around 4000-4500rpm. Torque is maybe 10lb-ft lower across the board. I was expecting the bigger carb to have a bigger effect than that, based on venturi areas. The mixture was a touch lean across the board though. So he opened up the stage 2 main jet. On the DG it's a 110 (1.1mm), and on the LT it's a 135 (1.35mm). He opened the 110 up to 120 (1.2mm). That improved the mixture everywhere (keep in mind that these tests are done in 4th gear with the throttle wide open from start to finish, so the secondary throttle is probably open most of the time). But surprise number 4: the power and torque curves are now at least as good as those from the LT carb - the torque is perhaps even better below about 4000rpm. How can the smaller carb better the larger carb? Who knows - faster airflow leading to better atomisation maybe? Anyway, that's the carb I've kept on - it's my 'test' carb with my 3D printed spare parts so I'm quite pleased that it still has a life to live on my DJ. And the fact that the increased venturi size in the LT carb doesn't help compared with a re-jetted DG carb could, I think, show that the airflow is already sufficient, so might this be the reason that the much bigger throttle body in the injection setup doesn't leave the carbed version behind?



Finally we did some tests with different distributors (actually, it wasn't finally - we did this earlier, but it's more logical to explain in this order. So these tests were done with the LT carb fitted).

We put the DG's 'Q' spec distributor on. The general consensus was that the engine sounded less smooth. The suggestion was that this may in fact be pinking. He said pinking would not be audible at higher revs, but the slightly rough edge could well be the result of pinking/knocking. I didn't get a printout of the curves, but they were noticeably down - maybe not quite as much as the difference between the stock LT and DG carbs, but not far off. From the discussion I think this was due to the ever increasing centrifugal advance that the 'Q' suffix distributor gives - he said it was better to have a lower maximum advance if possible, but I didn't get why. I think my options are to tinker with my 'Q' suffix version to limit the maximum advance, to look for one of the 'rare' 'N' suffix versions, or just keep Itchyfeet's

Or to get a 123Tune distributor.



If it wasn't for the cost I'd be tempted. We replicated the published curve of the 'N' suffix version in the software of the 123Tune unit, and it turned out to be even better than the genuine 'N' suffix unit. Not much, but a bit. No idea why - maybe the initial advance wasn't quite the same, meaning the maximum advance was different. Or maybe the vacuum advance on the 'N' suffix unit had a small residual effect - the throttle was wide open as mentioned, but the rolling road only provides a light load, so maybe the vacuum advance wasn't completely gone. In the 123Tune unit the vacuum advance was set to zero across the board.

So all in all a very enjoyable morning. I'm pleased that my engine doesn't give anything away to an injected version after all. And I'm pleased that my 'test' carb lives on, and that the change to a 1.2mm stage 2 main jet has made it so good - I suspect others with carbed DJs could get a 120 main jet from a carb spares place rather than messing abut with a 1.2mm drill. But I'm disappointed that my DG distributors aren't the best for the engine, as it means I'll need to hunt around for one of the rare 'N' spec versions.

[itchyfeet]

Excellent work Chris

I'm going to havevto print those and have a good look.

Great to know a carb can be almost as good as injection, of course without comparison of a FI DJ you don'f know if the injection curve at lower revs is the same but it probably doesn't matter if you can drop a geaf and get the power higher up.

[CJH]

Great to know a carb can be almost as good as injection, of course without comparison of a FI DJ you don'f know if the injection curve at lower revs is the same but it probably doesn't matter if you can drop a geaf and get the power higher up.

Yes, hadn't thought of that - the curve could be a different shape I guess.

Bigherb - any chance you've got the specs for the 025 905 205 AF ? This apparently replaced the 'N' suffix. Might help widen my search for a suitable distributor if the advance specs look similar.

[bigherb]

Bigherb - any chance you've got the specs for the 025 905 205 AF ? This apparently replaced the 'N' suffix. Might help widen my search for a suitable distributor if the advance specs look similar.

Not off hand, They where only fitted to 1991 model year. UK models had MV engines fitted from 90 MY.
VW parts sold 205AF as a replacement for the 205N so probably very similar.

I'll see if I can find which market used the DJ after 1989 but I'm away for a week.