There are a few problems with using adaptor plate and flywheel kits to join a Subaru engine to a manual VW transaxle. Some of them are inherent to using adaptor plates for this application, and tend to be to do with the dimensions of VW and Subaru parts which you cannot change. Others are to do with the way certain adaptor plate kits are designed – they’re certainly not all equal. For manual VW applications, adaptor plate / flywheel kits are never an ideal way to join Subaru engines to VW gearboxes, even though for some applications (type 1 gearboxes), they’re your only option. Adaptor plate kits for automatic VW transaxle applications generally don’t have the same problems, as you’ll see below:

Anyone who has been involved with old rear engined VW ‘s for a while is likely to be familiar with the fact that the starter motors in manual models can fail in an unusual way. They can work perfectly when cold, but consistently lose most of their torque when hot. The result is that they still turn, but are not able to crank the engine fast enough to start it when hot. Very much like if the battery is flat, but it isn’t. This seems to be more prevalent in VW buses (at least the T2a, T2b and petrol T25 / T3 / Vanagon models) than Beetles. This fault is not fixed by many who recondition starter motors, so can happen on newly reconditioned ones as well as old used ones. (If anyone has definitive information on the exact cause of this failure, it’d be interesting to know what it is – as a guess, maybe the lacquer on the armature windings breaking down, partially shorting them only when hot?).
Anything which increases the loading on these starter motors tends to exacerbate the symptoms, and that includes having to crank a larger capacity and / or higher compression engine, such as all of the typical Subaru engines which get fitted into VW’s. You’re far more likely to get this problem with a harder to crank Subaru engine joined to a VW gearbox with an adaptor plate kit, which forces you to still use a VW type starter motor than you would if you still had a VW engine.
Some adaptor plate kit manufacturers will try to sell you a high torque aftermarket starter motor. These are often based on the Denso spur gear starter motor design which has the solenoid inline with the pinion, and the motor offset (as used in all MY90-97 Subaru models). This is because the Denso design is very modular, and was used by many manufacturers – they can be made to fit just about any engine by designing one or two new parts for them. If they were all made using all genuine Denso parts (apart from the one or two special parts), they’d probably all be good products. The problem with them is that the design has become generic – so generic that every part for them has been copied by multiple aftermarket companies, to various quality standards. So take your chances with what quality they’ll be. Most probably not as good a the Denso real thing. – it’s not uncommon to hear of them failing.

If you want a real, standard OEM quality high torque starter rather than aftermarket, there are two options. You can use VW part number 003 911 023 A from an automatic bus, semi-automatic beetle, Porsche 914 and some ’69 – ’71 Porsche 911 applications. They’re 0.8hp (590W). The new Hella aftermarket version of this starter (8EA 011 611-181) is very reasonably priced, but they are made in China, so may or may not be the same quality as original German Hella. Or you can use Porsche part number 911 604 101 03 – they’re used in some ’73 – ’89 911’s (the 915 gearbox models) and are either 1.5hp (1100W) or 1500W depending on where you look – either way, they’re considerably more powerful than the standard VW or the auto / semi auto VW / 914 or earlt 911 starters. You may want to check that your starter positive terminal, earth strap and battery negative lead are at least 35mm2 though. The original VW Beetle / bus manual starters were 0.7hp (515W).

With the VW automatic models (003 and 090 transaxles), an adaptor plate and drive plate kit is the only practical way to join any alternative engine to them, as they don’t have removable bell housings / torque converter housings. However, they also avoid the problems mentioned above, as they don’t have a clutch, and already use a more powerful starter. This makes adaptor plate and drive plates to join alternative engines a perfectly acceptable, as well as being the only option, if designed well.
Most competitors’ adaptor plate kits for this application require modifications the the VW final drive / torque converter housing, which is cast iron. This usually involves drilling out threads in an existing hole, or similar, and is only required due to bad design, and the fact that most who make them never truly designed them at all – they just bought someone else’s product and attempted to copy it. There is no need to modify any VW auto gearbox to join it to a Subaru engine using an adaptor plate / drive plate kit if the parts have been better designed. Our auto adaptor plate kit requires no modification to the gearbox – everything bolts on, and this is one of the reasons we designed it back in 2011.

This is an obscure one, but it can, and sometimes does happen. Some adaptor plates for Subaru – VW applications are made from steel, other from aluminium. The latter are always machined from rolled aluminium plate, and this can introduce a potential problem. Unlike steel or cast aluminium, rolled aluminium plate can be susceptible to a little-known type of corrosion called exfoliation corrosion (also known as lamellar corrosion). During the rolling process, and the aluminium grains (or crystals) get flattened into thin layers, relative to their more even shape in the cast ingot before rolling. It is a type of intergranular corrosion – where the corrosion spread is between the grains, but their very large surface area flattened shape makes the corrosion far more severe. The corrosion takes up to 6 times as much space as the metal it used to be, so it effectively wedges the layer shaped grains apart, leaving them peeling or flaking away. We only know of two instances of this happening to Subaru engine adaptor plates, both from the same supplier. They’re a prominent one, not someone who those involved with Subaru engine conversions will have not heard of.

The alloy grade used is apparently a 6000 series grade. They’re generally not the most susceptible to exfoliation corrosion (5000 or 7000 series grades are generally worse apparently), but the it is likely that it’s one of the grades which contains copper, as it along with magnesium – one of the main 6000 series alloying elements, are known to induce exfoliation corrosion susceptibility. The heat treatment spec also has an effect on exfoliation corrosion susceptibility. Our guess is that the affected adaptor plate shown (another example can be found on line with a bit of Googling) is probably 6061 T6. The example shown was a UK customer, and the manufacturer blamed the “massive corrosion environment you live in with it raining 90 percent of the time” for the corrosion. They recommended trying anodising or sacrificial anodes (presumably they think this is ‘normal’ galvanic corrosion – it clearly isn’t if you compare with online pictures of exfoliation corrosion), but made no mention of exfoliation corrosion. Maybe they’re unaware of it?

OK, so this isn’t anything to do with the concept of using an adaptor plate, but it’ll still be of interest if you’re shopping for one. Some of them are exceptionally bad quality. We were once asked to help a friend / gearbox rebuilder whose customer had had 3 or 4 gearbox input shaft failures. He took the ‘box back, stripped it and proved that there was nothing wrong with it that could be the cause. He also knew the customer was using a Subaru engine adaptor plate, and knew it must be the cause of the problems, but didn’t know how to prove that one way or another. He asked if we could help, and sent the adpator plate to us. It was a pretty crudely machined steel adaptor plate, made seemingly on manual machines via a combination of turning and milling or drilling. They’re a very common product sold in the UK for many years by a business who seem to claim they’re their product. They’re not – they’re made (very badly) in Germany, and ate apparently always painted silver.
Investigating this proved to be very interesting. The first thing of interest was that the spigot diameter which locates into the gearbox had been turned onto the plate rather then milled, but was significantly out of round! It resembled an ellipse, with an obvious maximum and minimum about 90 degrees apart. How an ellipse is turned on a lathe is a mystery to us. Next we measured the engine to gearbox alignment dowel positions on a coordinate measuring machine to the centre of the apparent ellipse, and found that the closest the plate could hold the engine and gearbox to correct alignment was way out. Out by 14 times the maximum engine to gearbox alignment tolerance we use on out bell housings! There is no way that the oil seal or the clutch disc can last with that much misalignment. We welded up the dowel holes and re-machined them far more accurately as a favour to the gearbox builder friend (as shown below), and his customer used it for a while, but eventually switched to one of our bell housings. The nowhere near horizontal machining of the flat at the top of this plate is a good indication of the level of attention to detail put into the rest of it:

Overall, an extremely amateur product which does not live up to the standard you’d expect with it being made in Germany. Engine to gearbox alignment is something which very few building adaptor plates for one off applications realise the criticality of. Some businesses seemingly don’t either. A bit of Googling and you’ll be able to find other example of the dowels not lining up on the same product, and also how the manufacturer refused to accept the product back, and the customer had to correct it himself. Nice! These are extremely cheap adaptor plate and flywheel kits, and you get what you paid for. Don’t assume that because a product is available, it’s been engineered to any kind of acceptable standard at all. Some, such as these, have clearly not had an engineer or a competent machinist anywhere near them!

A well known brand of Subaru – VW engine conversion adaptor plates makes their plates from aluminium, and used to rely on the gearbox studs just being screwed into tapped holes in the aluminium. As the strength of material decreases, the amount of thread engagement needed when threading fasteners into that material increases, so you need more thread engagement when screwing into aluminium than you do with steel to give the same strength. There are some common rules of thumb for this with the common engineering materials which any design engineer working on products made from them should know. Customers having problems with the studs pulling out of adaptor plates was all over the VW engine conversion forums back when we came up with the Subaru engine conversion bell housing back in 2002, and the cause was obvious to us – the design had nowhere near enough thread engagement for aluminium, so the standard VW torque was pulling the threads out of the plates. This was one of a few reasons we made our parts from scratch, rather than just buying an adaptor plate kit. Some of the products out there at the time were clearly being designed by folks with no idea what they were doing.
Most, maybe all adaptor plate kits which use aluminium plates learned from this (they all copy each other), and now rely on either a nut or a bolt head on the other side of the plate, but it’s something to look out for if you’re shopping for one.

Adaptor plate / flywheel kits for joining Subaru engines to VW gearboxes are made to suit VW clutch sizes 200, 210. 215 and 228mm. The flywheels are all unique to the clutch size. Only two of these really make any sense. 200mm because a large range of uprated clutches are available in that size (intended for the type 1 tuning world), and 228mm because it’s the biggest VW size (and also the biggest size that will fit inside the VW bell housing). Both were used in quite a wide range of VW applications over many years. Uprated clutches used to be available for the 228mm size, but no longer seem to be.

The 210 and 215mm sizes are far more obscure options, only used in 1700 and 1800cc bay window buses and certain 1.9 wasserboxer T25 / T3 / Vanagon models. Nobody is ever likely to make uprated versions of these, as the demand for them is so much smaller.

For reasons best known to those who sell them, most adaptor plate kits in the UK use 215mm clutches. This as absurd, and the main seller of them (they don’t manufacture them, despite seeming to make out that they do – they buy particularly horrible quality kits (as pictured above in this FAQ) from Germany, and they seemingly always supply, or always used to supply the 215mm version for all applications, including turbo engines.

To work as they should, clutches always have a ‘safety factor’ of at least 1.2 – i.e. they are designed to be able to hold 20% more torque than the engine they were designed to be used with has. Rear engined VW clutches tend to have a slightly higher safety factor. This is so they cad stand an acceptable ampnt of abuse, like towing, or the vehicle being overloaded, or the driver pulling away very hard and still have a reasonable life. If you use up that safety factor by using the clutch on an engine with more torque, you’re reducing amount of abuse it can stand.

The VW engine with the highest torque which used a 215mm clutch had 142 Nm. Assuming a safety factor of 1.5 was used for the design of the 215mm clutch, you have no safety factor on the clutch at all when you put 1.5 x 142 = 213 Nm through it. Alternatively, the safety factor gives you a torque overhead of 213 – 142 = 71 Nm. Compare that with what the users of these 215mm VW clutches in adaptor plate kits are asking their clutches to do:

Phase I n-a EJ20, 172 Nm, a safety factor of 213 / 172 = 1.23, or a torque overhead of 41 Nm
Phase II n-a EJ201, 184 Nm, a safety factor of 213 / 184 = 1.16, or a torque overhead of 29 Nm
Phase II n-a EJ251, 223 Nm, a safety factor of 213 / 223 = 0.95, or a torque overhead of -10 Nm
Phase I turbo EJ20G, 290 Nm, a safety factor of 213 / 290 = 0.73, or a torque overhead of -77 Nm

To put this into context, no VW clutch which we have the data for has a safety factor anywhere near as low as 1.23. No VW or Subaru clutches which we have the data for have a safety factor anywhere near as low as 1.16, so you’re overloading the 215mm clutch significantly relative to any VW or Subaru OEM application with just a normally aspirated EJ201.
Or with even a low powered turbo engine, you’re trying to operate the clutch with a safety factor of 0.73, or 36% more torque than it could handle with a safety factor of zero, or 0.48. That’s 204% more torque than it could handle with a safety factor that VW intended for a normal working life. I.e. the 215mm VW clutch is destined to fail very quickly in this application.