Prajna
Fixing things for the love of it
- Messages
- 622
- Location
- Castelo Branco, Portugal
They weren't yesterday but are today.
- Thread starter Misterg
- Start date
Munkul
Jack of some trades, Master of none
- Messages
- 7,520
- Location
- Cumbria
How will you balance the rotor? Watching with great interest.
The turbine blades would have been the one thing I'd have been tempted to rob off an appropriately sized turbo... even with the wrong geometry they'd probably work ok...
The turbine blades would have been the one thing I'd have been tempted to rob off an appropriately sized turbo... even with the wrong geometry they'd probably work ok...
Hmmm why does that sound so familiarActually making glass was quite good fun - it was the management b******t that spoiled it
Misterg
Member
- Messages
- 1,062
- Location
- North Wales
Thanks for letting me know - Not sure what happened there. The SSL certificate for my site was renewed yesterday, not sure if that affected anything. No errors or bandwidth problems showing on the server...
Please let me know if it happens again.
slim_boy_fat
Forum Supporter
- Messages
- 29,540
- Location
- Scottish Highlands
Misterg
Member
- Messages
- 1,062
- Location
- North Wales
Thanks.
I'm only intending to do a static balance - the [ready made] compressor wheel should be balanced already; if I'm careful making the shaft it should be in balance; so by statically balancing the turbine, the whole assembly *should* be in reasonable dynamic balance.
The static balance is nothing fancy - just spin the turbine & find the heavy spot. Grind a little bit off it and try again. (I haven't finished this yet, btw!). There are designs for electronic balancers around, but I hope not to have to go down that particular rabbit hole for a ~100k RPM turbine (newer designs like the Wren mentioned above can run to 250k RPM+)
I believe there are issues trying to re-use the turbine from a turbocharger - AFAIK they are welded to the rotor and you can't drill a hole in them or they burst. They also use radial inflow turbines which have a completely different shape to the axial one here (the NGV is a ring around the outside of the turbine directing the flow inwards, rather than backwards). There are designs that use this type of turbine (there's a Russian guy on Youtube who builds them, and I think the GTBA are also working on a design) but I don't have the skill or knowledge to try and adapt them to this design.
Misterg
Member
- Messages
- 1,062
- Location
- North Wales
After procrastinating for a long time, I slowly came to the view that twisting and grinding the blades was more about consistency than precision.
I made a holder to index the disc for the 19 blades and got stuck in with a piece of bar slotted to fit over the blades. I had it in my drill press to stop it slipping off the end of the blade while I swung on it with an adjustable spanner.
The bending was done cold, as I found out that the ductility of Hastelloy X decreases with temperature unless you can get it to over ~900°C. I don't believe that is achievable with a blowlamp, and I don't have oxy-fuel kit, so cold it was,
Next, the meat between the blades needed to be ground out. The book doesn't supply any details. After a bit of thought, I made a holder to mount the turbine blank in the vertical milling slide from my lathe and mounted my angle grinder on a skid to keep the grinding disc at a fixed height above, and parallel to the bench. This meant that I could set the angle and height of the blank and make a controlled cut across the turbine blades. The diameter of the holder acts as a guide to getting a consistent depth.
2.5mm thick disc first to open up the slots:
Then a series of facets using a 1mm disc
I put the disc back on the arbour and skimmed the tips of the blades to true them up before starting to finesse the shape.
The front face of the blades needs to be ground to a concave shape. I'd convinced myself that I could do this with a hollow roller on a belt sander, so I spent an afternoon adapting my Aldi belt sander to take a narrow sanding belt instead of the sanding disc.
As it happens, while I can grind the front of the blades with this, the contact angles are all wrong.
After trying various small grinding wheels in the Dremel to hollow out the concave face of the blades, I found that a carbide burr was more effective (but the term is relative!).
The narrow belt sander attachment worked well to smooth the convex faces of the blades though.
After another afternoon's work, the upshot was this:
The bevel gauge is at 34° which is the angle specified for the chord of the blade at the exit - close enough?
The trailing edge of the blade is getting quite thin, so I hope I haven't overdone it. The convex faces aren't perfect, but I am hoping are good enough - there's maybe a few thicker spots that could be evened out.
Another few hours of careful grinding tidied it up a lot. An oval carbide burr worked well to blend in the concave surfaces:
Afterwards I went over it with a small Scotchbrite wheel in my Dremel-alike.
There are small areas near the root on the convex face of each blade which I've so far not been able to sand out. At the moment I'm sort of ignoring them - I don't want to take much more off the blades as it's difficult to get pressure at the root without also sanding the outer trailing edges (I've already lost the very corner of a few blades where the grinding burr snagged them).
This short video gives a better idea of what it's like:
I'm leaving it alone now until I mount it on the shaft and try and balance it.
I made a holder to index the disc for the 19 blades and got stuck in with a piece of bar slotted to fit over the blades. I had it in my drill press to stop it slipping off the end of the blade while I swung on it with an adjustable spanner.
The bending was done cold, as I found out that the ductility of Hastelloy X decreases with temperature unless you can get it to over ~900°C. I don't believe that is achievable with a blowlamp, and I don't have oxy-fuel kit, so cold it was,
Next, the meat between the blades needed to be ground out. The book doesn't supply any details. After a bit of thought, I made a holder to mount the turbine blank in the vertical milling slide from my lathe and mounted my angle grinder on a skid to keep the grinding disc at a fixed height above, and parallel to the bench. This meant that I could set the angle and height of the blank and make a controlled cut across the turbine blades. The diameter of the holder acts as a guide to getting a consistent depth.
2.5mm thick disc first to open up the slots:
Then a series of facets using a 1mm disc
I put the disc back on the arbour and skimmed the tips of the blades to true them up before starting to finesse the shape.
The front face of the blades needs to be ground to a concave shape. I'd convinced myself that I could do this with a hollow roller on a belt sander, so I spent an afternoon adapting my Aldi belt sander to take a narrow sanding belt instead of the sanding disc.
As it happens, while I can grind the front of the blades with this, the contact angles are all wrong.
After trying various small grinding wheels in the Dremel to hollow out the concave face of the blades, I found that a carbide burr was more effective (but the term is relative!).
The narrow belt sander attachment worked well to smooth the convex faces of the blades though.
After another afternoon's work, the upshot was this:
The bevel gauge is at 34° which is the angle specified for the chord of the blade at the exit - close enough?
The trailing edge of the blade is getting quite thin, so I hope I haven't overdone it. The convex faces aren't perfect, but I am hoping are good enough - there's maybe a few thicker spots that could be evened out.
Another few hours of careful grinding tidied it up a lot. An oval carbide burr worked well to blend in the concave surfaces:
Afterwards I went over it with a small Scotchbrite wheel in my Dremel-alike.
There are small areas near the root on the convex face of each blade which I've so far not been able to sand out. At the moment I'm sort of ignoring them - I don't want to take much more off the blades as it's difficult to get pressure at the root without also sanding the outer trailing edges (I've already lost the very corner of a few blades where the grinding burr snagged them).
This short video gives a better idea of what it's like:
I'm leaving it alone now until I mount it on the shaft and try and balance it.
One the one hand this thread is utterly fantastic. I'm in awe of the talent, skill and application.
On the other hand it drives it home that I'm a talentless oaf who can barely glue 2 bits of metal to each other
That engine would be wonderful just sat on the sideboard as a piece of art.
On the other hand it drives it home that I'm a talentless oaf who can barely glue 2 bits of metal to each other
That engine would be wonderful just sat on the sideboard as a piece of art.
Maker
Most folk just call me; Orange Joe
- Messages
- 10,487
- Location
- Don't ask questions
This is incredible, who needs five axis!After procrastinating for a long time, I slowly came to the view that twisting and grinding the blades was more about consistency than precision.
I made a holder to index the disc for the 19 blades and got stuck in with a piece of bar slotted to fit over the blades. I had it in my drill press to stop it slipping off the end of the blade while I swung on it with an adjustable spanner.
The bending was done cold, as I found out that the ductility of Hastelloy X decreases with temperature unless you can get it to over ~900°C. I don't believe that is achievable with a blowlamp, and I don't have oxy-fuel kit, so cold it was,
Next, the meat between the blades needed to be ground out. The book doesn't supply any details. After a bit of thought, I made a holder to mount the turbine blank in the vertical milling slide from my lathe and mounted my angle grinder on a skid to keep the grinding disc at a fixed height above, and parallel to the bench. This meant that I could set the angle and height of the blank and make a controlled cut across the turbine blades. The diameter of the holder acts as a guide to getting a consistent depth.
2.5mm thick disc first to open up the slots:
Then a series of facets using a 1mm disc
I put the disc back on the arbour and skimmed the tips of the blades to true them up before starting to finesse the shape.
The front face of the blades needs to be ground to a concave shape. I'd convinced myself that I could do this with a hollow roller on a belt sander, so I spent an afternoon adapting my Aldi belt sander to take a narrow sanding belt instead of the sanding disc.
As it happens, while I can grind the front of the blades with this, the contact angles are all wrong.
After trying various small grinding wheels in the Dremel to hollow out the concave face of the blades, I found that a carbide burr was more effective (but the term is relative!).
The narrow belt sander attachment worked well to smooth the convex faces of the blades though.
After another afternoon's work, the upshot was this:
The bevel gauge is at 34° which is the angle specified for the chord of the blade at the exit - close enough?
The trailing edge of the blade is getting quite thin, so I hope I haven't overdone it. The convex faces aren't perfect, but I am hoping are good enough - there's maybe a few thicker spots that could be evened out.
Another few hours of careful grinding tidied it up a lot. An oval carbide burr worked well to blend in the concave surfaces:
Afterwards I went over it with a small Scotchbrite wheel in my Dremel-alike.
There are small areas near the root on the convex face of each blade which I've so far not been able to sand out. At the moment I'm sort of ignoring them - I don't want to take much more off the blades as it's difficult to get pressure at the root without also sanding the outer trailing edges (I've already lost the very corner of a few blades where the grinding burr snagged them).
This short video gives a better idea of what it's like:
I'm leaving it alone now until I mount it on the shaft and try and balance it.
@Misterg you should have reminded me about this!! I’d totally forgotten and have basically been on-site since I messaged saying I was struggling to process that material…. I’ve still got the material which as said your welcome to, just haven’t had chance to slice it up. Excellent work so far with this though, I’ll keep hold just in case your likely to need a new turbine wheel. I actually want to prove I can cut it up more than anything else… may have to invest in some cbn parting tips…
Misterg
Member
- Messages
- 1,062
- Location
- North Wales
Thanks for all the very kind comments.
@Rrobson - no worries, I thought we'd decided to call it a day. Thanks for trying !
The turbine wheel fits on a shaft that passes through the middle of the engine to drive the compressor wheel at the front. It's supported on a couple of ball races and both wheels are clamped onto the shaft with left hand threaded nuts. The whole thing needs to be as true as possible.
The shaft is chunkier than I was expecting and made of heat treated steel: 817M40T (EN24T). It needs to be stiff enough for its resonant frequency to be higher than the running speed of the assembly (to avoid the shaft 'whipping') and strong enough not to take a permanent set if it doesn't run perfectly true (I learned from the book...)
It also has to transmit ~20 horsepower to the compressor wheel at full chat!
I made the clamping nuts first, trying to keep the tapped bore true to the clamping face.
An aluminium spinner nut to secure the compressor wheel:
And a steel nut to secure the turbine wheel:
I turned down the ends of the shaft blank, leaving a shoulder, centre drilled, and screw-cut the thread for the appropriate nut. (I won't tell you how long it took me to figure out that a left hand thread cutting tool isn't the same as a right hand one upside down
)
While it was all running true, I could tighten the nut up to the shoulder and finish the outside true to the shaft thread.
I hogged the rest of the shaft out 0.5mm over size then set it up between centres for finish turning (the first time I've turned anything between centres). There was about 0.05mm of runout in the shaft after roughing. When finished, it was ~0.004mm TIR which I was very pleased with.
I did cock it up though - I got both bearing journals to be a very good fit by lapping them to size, but when I was trying to finish the 7mm shaft through the compressor I ended up with it undersized and tapered.
I'd been hoping to achieve a slight shrink fit, so was looking to finish at somewhere around 7.005mm / 0.2758" but ended up at 6.989mm / 0.2752" at the outboard end and 6.972mm / 0.2745" at the inboard end. I didn't know how this had happened at the time (I'd noticed that the tool seemed to be producing more swarf than expected, but didn't have the presence of mind to stop and check anything!
).
In hindsight, I believe that the shaft heated up and became too tight between the centres and started bowing - I had something similar happen when finish turning one of the tapers - I was checking the tailstock quite often, but not often enough, obviously! I think the long, slender section of shaft probably heated up more quickly than the thicker parts I had been working on at first and it caught me out.
The compressor wheel slid onto the shaft quite nicely, and I'd almost convinced myself it would be OK, but after sleeping on it (the problem, not the shaft!) decided that it wasn't good enough.
Not having either enough EN24T stock or the stomach to make another shaft, I thought I'd try and rescue this one by lightly knurling the undersized section and machining it back - if it didn't work, I'd be ordering some more stock and making another shaft anyway.
This time, I managed to hit 7.007mm / 0.2758" for the full length after lapping. As this isn't a bearing surface, I'm reasonably happy with the result.
I lapped a bit of a lead onto the compressor shaft and the compressor wheel slides on about 6mm before meeting resistance. If I pour boiling water onto the wheel, it slides smoothly into position. When the wheel cools, it is absolutely rock solid on the shaft. More boiling water and it will slide right off again - very, very pleased with that!
I didn't get a picture with it assembled though!
There are a couple of spacers that need making, too. These need to be as parallel as possible. I ended up finishing one face, the bore and OD then parting off and turning an arbour to mount them on (with superglue) to face them to length. Seemed to work OK - I made a couple of wonky ones, but on the good ones, I can't measure any taper with a vernier mic.
Thanks for reading
@Rrobson - no worries, I thought we'd decided to call it a day. Thanks for trying !
The turbine wheel fits on a shaft that passes through the middle of the engine to drive the compressor wheel at the front. It's supported on a couple of ball races and both wheels are clamped onto the shaft with left hand threaded nuts. The whole thing needs to be as true as possible.
The shaft is chunkier than I was expecting and made of heat treated steel: 817M40T (EN24T). It needs to be stiff enough for its resonant frequency to be higher than the running speed of the assembly (to avoid the shaft 'whipping') and strong enough not to take a permanent set if it doesn't run perfectly true (I learned from the book...)
It also has to transmit ~20 horsepower to the compressor wheel at full chat!
I made the clamping nuts first, trying to keep the tapped bore true to the clamping face.
An aluminium spinner nut to secure the compressor wheel:
And a steel nut to secure the turbine wheel:
I turned down the ends of the shaft blank, leaving a shoulder, centre drilled, and screw-cut the thread for the appropriate nut. (I won't tell you how long it took me to figure out that a left hand thread cutting tool isn't the same as a right hand one upside down
)While it was all running true, I could tighten the nut up to the shoulder and finish the outside true to the shaft thread.
I hogged the rest of the shaft out 0.5mm over size then set it up between centres for finish turning (the first time I've turned anything between centres). There was about 0.05mm of runout in the shaft after roughing. When finished, it was ~0.004mm TIR which I was very pleased with.
I did cock it up though - I got both bearing journals to be a very good fit by lapping them to size, but when I was trying to finish the 7mm shaft through the compressor I ended up with it undersized and tapered.
I'd been hoping to achieve a slight shrink fit, so was looking to finish at somewhere around 7.005mm / 0.2758" but ended up at 6.989mm / 0.2752" at the outboard end and 6.972mm / 0.2745" at the inboard end. I didn't know how this had happened at the time (I'd noticed that the tool seemed to be producing more swarf than expected, but didn't have the presence of mind to stop and check anything!
).In hindsight, I believe that the shaft heated up and became too tight between the centres and started bowing - I had something similar happen when finish turning one of the tapers - I was checking the tailstock quite often, but not often enough, obviously! I think the long, slender section of shaft probably heated up more quickly than the thicker parts I had been working on at first and it caught me out.
The compressor wheel slid onto the shaft quite nicely, and I'd almost convinced myself it would be OK, but after sleeping on it (the problem, not the shaft!) decided that it wasn't good enough.
Not having either enough EN24T stock or the stomach to make another shaft, I thought I'd try and rescue this one by lightly knurling the undersized section and machining it back - if it didn't work, I'd be ordering some more stock and making another shaft anyway.
This time, I managed to hit 7.007mm / 0.2758" for the full length after lapping. As this isn't a bearing surface, I'm reasonably happy with the result.
I lapped a bit of a lead onto the compressor shaft and the compressor wheel slides on about 6mm before meeting resistance. If I pour boiling water onto the wheel, it slides smoothly into position. When the wheel cools, it is absolutely rock solid on the shaft. More boiling water and it will slide right off again - very, very pleased with that!
I didn't get a picture with it assembled though!
There are a couple of spacers that need making, too. These need to be as parallel as possible. I ended up finishing one face, the bore and OD then parting off and turning an arbour to mount them on (with superglue
) to face them to length. Seemed to work OK - I made a couple of wonky ones, but on the good ones, I can't measure any taper with a vernier mic.Thanks for reading
Misterg
Member
- Messages
- 1,062
- Location
- North Wales
Ebay
The wheel specified in the book is the KKK 5326 123 2037 which I believe was originally numbered 5626 123 2018.
The ebay listing says it's OE for an Audi 200.
(It was discovering the ebay listing that turned this from wondering about it to trying to do it.
)
Misterg
Member
- Messages
- 1,062
- Location
- North Wales
Bearings are a bit of an issue for these engines.
The ones in the previous post are bog-standard steel 608 bearings. Various people in the GTBA flung their hands up in horror at the prospect of 'going back' to using steel on steel bearings (even though the book gives them as an option) - they used to be OK in the good old days when you couldn't get anything else, but inevitably failed (unless replaced regularly as a preventative measure), often taking the engine with them.
At the speeds involved, cages get torn apart by the centrifugal forces causing the bearing to collapse and whirly bits to hit non-whirly bits. There's also a problem with steel balls in a steel race picking-up and seizing the bearing if the balls skate on the race for any reason (oil mist lubrication and preload springs help with the skating). The firm consensus was that steel bearings have been overtaken by full complement hybrid ceramic bearings.
The bearings are available from Boca in the US (list price $125 each!
)
However the GTBA also provided a workaround...
If you can get hold of steel races and a supply of suitable silicon nitride balls, one can assemble one's own bearings.
You can measure the balls in a steel bearing and order identically sized Si3N4 balls (Aliexpress, etc.) or, as I did, buy four (fairly cheap) caged hybrid bearings, dismantle them and re-arrange them into two full complement bearings.
The key to getting it to work is to carefully machine down one side of the inner race at a slight taper without touching the bearing track itself.
If you get it just right, the inner race will click into position inside a fully populated outer race:
Obviously these can't support any load on the side of the race that has been machined away, so they need to be treated as angular contact bearings, but as they're assembled opposing each other with a preload spring, this (apparently!) isn't a problem.
It sounds pretty straightforward as I write it here, but there was a heck of a lot more faffing than the description suggests!
Thanks for reading
The ones in the previous post are bog-standard steel 608 bearings. Various people in the GTBA flung their hands up in horror at the prospect of 'going back' to using steel on steel bearings (even though the book gives them as an option) - they used to be OK in the good old days when you couldn't get anything else, but inevitably failed (unless replaced regularly as a preventative measure), often taking the engine with them.
At the speeds involved, cages get torn apart by the centrifugal forces causing the bearing to collapse and whirly bits to hit non-whirly bits. There's also a problem with steel balls in a steel race picking-up and seizing the bearing if the balls skate on the race for any reason (oil mist lubrication and preload springs help with the skating). The firm consensus was that steel bearings have been overtaken by full complement hybrid ceramic bearings.
The bearings are available from Boca in the US (list price $125 each!
)However the GTBA also provided a workaround...
If you can get hold of steel races and a supply of suitable silicon nitride balls, one can assemble one's own bearings.
You can measure the balls in a steel bearing and order identically sized Si3N4 balls (Aliexpress, etc.) or, as I did, buy four (fairly cheap) caged hybrid bearings, dismantle them and re-arrange them into two full complement bearings.
The key to getting it to work is to carefully machine down one side of the inner race at a slight taper without touching the bearing track itself.
If you get it just right, the inner race will click into position inside a fully populated outer race:
Obviously these can't support any load on the side of the race that has been machined away, so they need to be treated as angular contact bearings, but as they're assembled opposing each other with a preload spring, this (apparently!) isn't a problem.
It sounds pretty straightforward as I write it here, but there was a heck of a lot more faffing than the description suggests!
Thanks for reading
Barking Mat
Cuddler of hedgehogs.
- Messages
- 12,917
- Location
- France, Brittany.
I'm in awe at this thread, and am just utterly impressed by your skills.
Misterg
Member
- Messages
- 1,062
- Location
- North Wales
Very, very kind of you to say so - I surprise myself most of the time!
