Growing a Grumman - Part 4

2007-12-17: Part3 was written towards the end of 2004 and a mere 3 years(!!!) have gone by to bring us to the point where Part 4 of this odyssey can be written!

Apart from some work on the airframe and other smaller components, the main effort of the past 18 months has been the engine and particularly the home-brewed PSRU. Incidentally the original plan was for a chain-drive unit using HY-Vo chains - but on investigation it was found that these in SA were difficult to source and very costly, in the sizes and lengths needed at least!

Therefore we reverted to the HTD belt design - easily obtainable, not so pricey and well-known in SA.

The first step in any such project is to get something down on paper - well I am not much of a draughtsman and my engineering skills are more aligned to the shade-tree variety than anything like a degree hanging on my office wall! But wait - what are friends for? - I have a mate who was a big wheel locally in race car design and another who runs a company which makes all sorts of fancy armoured vehicles and he is a professional, qualified engineer to boot! Both these guys are Grumman aficionados, owning and modifying and even flying the same type of aircraft, albeit behind Lycoming 0-320s - and both have totally rebuilt their aircraft and made major mods - good company to have!

After lengthy discussion with both Trevor and Andrew, I had some clarity in my own mind as to what should do the trick in engineering terms. Further study of what is done and how it is done by both Belted Air Power of Las Vegas, Nevada and North-West Aero of Washington, USA in terms of the way they build their successful PSRUs was invaluable. The most valuable sources of info here were Contact Magazine, of which I have a number of back-issues from when I subscribed and from purchases of back-issues I made at Oshkosh in 1997 or thereabouts. One of these detailed BAPs Chev conversion carried out on the RV6A. Also the books 'Alternative Engines' (from Contact Magazine) and Finch's 'Auto Engines for Experimental Aircraft' have been useful.

Thanks to these publications and the installation manual available at one time from the NW Aero website, I could see how it all goes together. These units are quite simple really and after my researches I came to the conclusion that the key elements which need addressing in the design are:

1.Rigidity - there can be no flexing within the unit which can cause misalignment of sprockets or the belt, leading to the belt stripping teeth or generating excessive heat build up followed by catastrophic belt failure

2.Transmission of loads through and within the unit must be dealt with - there are forces of some magnitude generated by the rotation of the belt - calculations show that belt loads can be as high as 1300 NM in the case of my application (a function of power,rotational speed and sprocket sizes) and that load acts on the sprockets and bearings etc, which must be able to handle this comfortably.

3. Propeller reversal loads - my research shows that these loads are mostly taken care of by the belt having a "damping' or 'absorbing' effect, unlike a gearbox where these reversal forces are transmitted directly into gear teeth - just stand close to any Rotax motor at idle to hear this phenomenon loud and clear!

4. Flexing of the crankshaft under torsional loads felt by the driver sprocket attached to the crankshaft. By using a small (approx 2 in dia) sealed bearing mounted on the front plate of the assembly which accepts a "stub" shaft extending from a flange bolted to the front of the sprocket, such loads should be minimised and flexing of the driver sprocket minimised. This is important since flexing of the driver sprocket can cause main bearing or even crankshaft failure.

During these deliberations I managed to download from the internet some very handy software (from Dayco and from Gates) to calculate loads and power ratings of cog-belts. The software also established other parameters such as center distances for various sprocket combinations and proved to be very useful. (Download Drive Engineer from Dayco - CLICK HERE . Visit gates.com for Designflex software)

One key item that changed from my original thinking about the design was triggered by examining the installation manual from NW Aero - while other units appear to use bearings at front and rear of the prop shaft, mounted to the front and rear support plates, the NW Aero unit uses a cantilevered arrangement by having an extended upper shaft housing in which are fitted front and rear bearings. This has some advantages it seems to me - the main one being that one can more easily adjust the belt tension, without fear of misalignment of the sprocket, as the upper or driven sprocket and the shaft and shaft housing are essentially one unit which can be moved up or down in the vertical plane.In addition oil rather than grease can be used to lubricate the bearings on the upper shaft, since this can now be enclosed with oil seals. In all likelihood after the initial fill, only very occasional checking and top up of the oil should be necessary, unlike greased bearings where I expect I would want to be re-greasing them all too frequently!

The first thing we needed to do was to get the plates cut which would form the "box" or "frame" in which everything would run and to which almost every component is mounted. These plates are of 12 mm (approx 0.5in) aluminium and the grade available here in SA was 8082-T6 - the strength of which is adequate, but if I could have obtained 2024 T3 I would have preferred that. The front plate has to be wide enough to give space for the largest sprocket (driven) and belt and deep enough to allow for positioning of the driven sprocket according to the length of belt used - in my case we used the shortest belt available of 966mm pitch length, as this would preserve most closely the original thrust line and cowl design.

The plates were water-jet cut by a local supplier, and then began the task of mounting first the back plate on to the existing bell-housing mounting bolt-holes. Obviously for a different engine the procedure would be essentially the same, but the bell-housing bolt-holes being differently placed would require a differently designed or shaped back-plate.

The front plate needed the driver-sprocket bearing housing to be positioned accurately then fabricated and installed - this can only be done once the sprocket is in place, so we proceeded to order that item. The sprocket, with 30 x 14mm pitch teeth and made from grey iron (ductile cast iron) came as a solid cylinder of metal weighting 9 KG (approx 20 lbs)! After judicious machining this happily was reduced to 3.5 Kgs and perhaps could be made even lighter - but let's not take chances!

Of course the Chev crankshaft uses imperial threaded bolts - the originals were no use as the engine had a flex-plate so they were too short and no similarly threaded high tensile bolts of correct length could be had for love or money in South Africa! We therefore had to make the bolts by buying oversized HT metric bolts of the correct length and machining and threading these to suit!

Once we had the bolts the driver sprocket could be mounted to the crankshaft, with an alloy spacer between it and the crank flange to position the sprocket correctly between the front and rear plates of the "box".The support bearing housing (a square block of alloy) was positioned and attached to the front plate with 4 x 6mm bolts.

Next on the menu came the driven sprocket and belt - enquiries revealed that belts of the width required - 90mm - were available off the shelf only from the Gates distributor in Johannesburg - and they had only the latest Polychain design, which are extravagantly expensive - I was quoted no less than R8000 for the belt I wanted (that's about US$1150)! Since all earlier belt-drives on aircraft used the original HTD belts since the 1970s when such belts became available, I figured that if such a belt could handle the power and torque in the 1970s it should be able to do the same in the 2000's - that then would be my choice, as the straight HTD belt is vastly cheaper. Beware however - today there are some very cheap and very nasty belts coming out of India and China which are of inferior quality and WILL FAIL. Good enough for a bottling machine firmly bolted to the factory floor perhaps - not for your aerial conveyance!

Incidentally the tooth design of HTD and Polychain belts is not the same and once your sprockets are cut for one or the other, you are stuck with that belt - they are not interchangeable!

Using the downloaded software mentioned earlier showed that a standard HTD belt of 90mm, with my sprocket combination and rotational speeds, can handle 375 hp! Good enough!

So far so good - but just about here the wheels came off - on ordering the belt and sprocket from a local supplier, who would import the belt and manufacture the sprocket, I was quoted 2 weeks. Five months later the items were delivered! The delay took us up to October 2007 and with these items to hand we could finally get back to the project!

However that damned work stuff intruded - so it was November before activity resumed on the PSRU.

Once again the sprocket, this time made of alloy and the most expensive item in the entire contraption at around R3200 (US$450) needed machining for added lightness. It was my intention to connect the propeller shaft to this sprocket using splines - after a lot of time wasted running around to find some company who could provide a splined connection, I gave up when I realised that nobody was interested for one unit - they can supply if you have 20 or 50 or whatever but one - no way man!

Plan B then had to be adopted - this meant that we would be using a simple bolted flange at the backend of the propshaft - not wanting any weldments in this shaft whatsoever, I found a rear axle shaft from some ancient truck at a breakers yard. The diameter of this solid shaft was 41 mm, and unlike many modern automotive drive shafts it had no taper along it's length - and the flange onto which the brake and wheel would fit on a truck was flat and a perfect 85mm diameter - again most auto units are dished in the centre of the flange - don't ask me why! This shaft was also 6 feet long, so I knew it could be made to fit!

Now we had in our grubby paws - a shaft - a sprocket - a drive belt - and after purchasing the bearings and oil seals, plus a log of aluminium of 155mm length and 185mm diameter from which would be machined the shaft housing we were in business. Something like a month later, and many drawings printed out by my computer, out of the machine shop came the kit of parts for the top end of my PSRU!

Having assembled the unit with these (almost final) parts, we now have a "burning and turning" PSRU - no prop yet though - which looks like it will do the job and be plenty sturdy enough to cause me no worries when aloft. Weight wise we came out at just over 60 lbs - but I reckon that later I can shave at least 10 lbs by adding lightness, in the form of lightening holes or cutouts to the plates which make up the "box" - right now it's all solid.

Just a word on the drawings - at the start of this project to build a PSRU, I spoke to a fellow who does engineering drawings as a sideline to his day job as a draughtsman. He indicated some rough pricing that would make your eyes water, and I wondered how to do such drawings - obviously, buy the necessary tools and get to it! Actually no - I discovered that the programme installed on my desktop computer, which I never use, MS Word (and I have an ancient version MSWord 97!) can do drawings using lines and shapes - and they can be scaled! A little experimentation later it was clear that this software would do the job perfectly well - and it did too!

Working with an experienced machinist also means that he will be able to adjust fits and clearances in line with "best practice" as he goes along, so the necessity for super-accurate drawings falls away to an extent anyway. I suspect that if one was doing such machining on a modern NC machine then whatever is on the drawing will be whatever is produced - right or wrong - old-style craftsmanship still has a value, I am pleased to say!

Next instalment will be when the prop is bolted up and we start to blow things around!

-----------------------------------------------------------------------------------------------------------------------------------------------------------

Author's note: I received the propeller shaft and shaft housing yesterday (2007-12-16) from the machine shop after 4 weeks of waiting. Amazing how these guys work - the parts sat on his shelf for 3 weeks and 6 days - then he knocked them out in a day at the very last minute before toddling off on holiday for three weeks!

So we will have to wait until January 2008 before we can get on with the last piece which will be the propeller hub and flange. Meanwhile I can check how this assembly runs and whether belt alignment is correct., so that once the prop hub is ready we can just bolt the prop on and go fly - NOT!!

Here are some pictures! Click the image for larger (1024x768) view - I have provided the large size images as I know you guys want to see DETAILS!

General view of PSRU Currently this unit weighs in at 28.15 Kg (62 lbs) - which is a bit more than hoped for but then we can still take a bit of weight (maybe 5 Kgs) off by cutting lightening holes in the alloy and steel plates. That would give a weight of 23 Kg or 50lbs. The green coloured side channels are simply zinc-chromate painted steel channels.	General view of PSRU There are three bearings in this assembly - a small (lubed/sealed for life) bearing at the bottom in the square housing and a front and rear bearing inside the conical upper shaft housing. The top bearings are Deep_Groove Ball Bearings of 62mm (front) and 83mm (rear)	General view of PSRU - oil fill plug visible and front oil seal, behind which is a 62mm Deep Groove Ball Bearing.At the back of the housing is a 83mm Deep Groove Ball Bearing and a seal. Bearings run in EP90 gear oil, which fills the shaft housing to between 1/3 and 1/2 the depth of the bearing. There is a similar plug at bottom to drain the oil.
Showing side channels - 120mm steel sections - would have preferred alloy but unobtainable in South Africa - even these steel sections were very difficult to obtain! The steel is heavy at 2.8Kg per side but some lightening can be done later to probably halve that weight. The rectangular block on the front of the front plate is part of the adjustment mechanism, not yet completed - a simple jackscrew arrangement to raise or lower the shaft housing assembly.	Driver sprocket. This is cast iron and after machining the centre out of it, weighs 3.5Kg. The shiny bit between the flywheel and the sprocket is an alloy spacer to place the sprocket in the right position fore and aft between the front and rear plates. The funny shaped green thingy on the flywheel is the Chev balance weight, without which this engine wants to shake itself off the mount! Note that I am using a full-weight flywheel - the Chev motor came with a flex-plate for an auto-box so to match the starter teeth I am using a Nissan flywheel, machined to fit. it weighs in at close to 20 lb and according to Fred Geschwender a full-weight flywheel is essential to dampen combustion pulses.	Side of engine from front.Tubular mount is rough version for test purposes made of mild steel tube 22mm od.I bought one length of tube (6 metres) for this mount. Note "bobbin" enclosure for rubber isolator at front - this is a rear leaf-spring shackle rubber mount from a Nissan 3 litre LDV - works just fine!
Exhaust pipes (headers) 3 into 1 and (almost) tuned length and mild steel, just like on a car! Without silencers they make a terrible racket - by making an insert of a diameter to slide neatly inside the exhaust outlet pipe, packed with wire wool like a normal straight through automotive silencer, the engine noise is changed to a nice deep and smooth tone - and not too noisy.	Starter is Nissan 3 litre LDV (1.4 Kw) on fabricated welded mounting which is attached to the three threaded holes in the Chev block where I presume, the original starter lived. Pardon my rust - it is 8mm thick mild steel, not painted or plated and we have been having rain lately, so the humidity gets to the steel! Probably as a result of my poor welding skills, this mount has cracked twice already in test running - so I will eventually have something machined out of alloy.	This is what in a car would be the front of the motor but in the aircraft is the rear. Alternator and water pump (from Ford Essex 3.0 V6 motor because it was light, looked like it should pump enough volume and was cheap, reliable and turned the correct way for my setup!) hung from fabricated brackets attached to convenient existing pick up points. The stock Chev water pump cannot be used on my aircraft, as I have so little space between the firewall and the motor. If I used that water pump the motor would have to move forward two inches which is no good for my CG
Test stand on which the motor has lived for past two years - instead of making up a new instrument panel I just took the one out of the airplane - seemed the logical solution as a temporary measure	Radiator is custom built unit made from a new BMW 740 rad, cut horizontally and welded back-to-back - don't know yet whether it will do the job, as have never had any airflow through it while running the motor for short periods on test but it is of the recommended size and capacity. Without airflow the motor can run at idle or low revs for 10 - 15 minutes before getting to 100 deg C on the eater temp gauge.	Carbs are twin 13/4 inch SU type off Rover V8 motor - these were used on all UK produced Rover V8s (3.5 ltr) and they are installed on the original Chev TBI manifold after major mods to that item! SU carbs are ancient design but actually a good carburettor as they have altitude compensating characteristics due the the sliding needle design of the built-in mixture control. The motor runs well with these carbs - but over 4500 rpm there is a misfire which may be due to mixture or ignition.
Front Plate with shaft. Note the steel collar which acts as main thrust washer to transmit prop thrust loads into rear bearing thence to front plate and shaft housing	Front Plate with shaft	Front Plate with shaft and driven sprocket (48 teeth)
	Front Plate with shaft and driven sprocket (48 teeth)
Flywheel - Nissan 3.0 modified to fit Chev crankshaft - with driver pulley made of cast (grey) iron secured with 6 high tensile cap-screws. This pulley came as a solid cylinder - you can see that we have machined out the centre leaving 20mm metal at the rear fixing point.	Flywheel (Nissan 3.0 modified to fit Chev crankshaft) and back-plate with steel side channels	Front Plate with shaft and driven sprocket (48 teeth) - shaft housing machined from solid billet of alloy contains front and rear bearings.