NC700 camchain tensioner

[embee 7,288]

As a preventative maintenance measure, while doing other work I took the camchain tensioner out of my 20k mile Integra. There have been a handful of reports of failed tensioners (exact failure mode not known), so an inspection was warranted.

As far as I'm concerned this looks exactly as I would hope at this mileage, essentially no wear evident, just bedding where the chain runs. All the other parts, the links and springs etc, were as new.

In usual Honda fashion, the designer who put one of the tensioner fixing bolts (with copper sealing washers) directly beneath the thermostat cover should have been shot. No big deal to remove the cover since you've already drained the system and removed the rad to get the cam cover off, but really!

Incidentally, it has done 12k miles since I last checked the valve clearances. This time 3 of the 4 "positions", all exhausts and one pair of inlets, were a shade tight, the other pair of inlets were still spot on. Shamefully I don't have feelers in the exact values given, 0.17mm inlet and 0.28mm exh, so I use the "go-no go" principle of 0.15-0.20 for inlets and 0.25-0.30 for exh, the smaller should be a slack slip fit, the larger should be too tight, so the clearance must be pretty close to mid way, near enough anyway. A bit of engineering feel required!

 

[Guest Hati]

On 16/04/2017 at 18:11, embee said:

Shamefully I don't have feelers in the exact values given, 0.17mm inlet and 0.28mm exh

 

Before I did the first major for the 750, I was doing a lot of research to find the right feeler gauges (in metric), but had no luck. I did come across however seemingly odd sizes, that just happened to have the exact dimensions Honda prescribes in imperial. This is what I got but there are other options as well:

 

Edited April 19, 2017 by Hati

[embee 7,288]

Someone posted that I think Wurth do these feelers individually in 0.17 and 0.28mm. For all practical purposes you could use 0.007" and 0.011" in imperial. but again the feelers I have are .006/8" and .010/12" so would use the same principle. If I come across them I'll get some, but meantime the go/no-go principle works.

[Dunnster 1,650]

Thanks for the report Murray. 

[Guest sykospain]

The pairs of single-size Würth colour-coded feelers are exactly the same ones as used for the tappets on the BMW oilheads - which is handy if you have one of each of the Honda and Beemer bikes !

I use two pairs of each, 'cos when you adjust one valve clearance, with the opposing feeler inserted, the light single-size feeler gauge doesn't annoying fall to the ground.  As Wunderlich Spares in the Fatherland says :

 

"These feeler gauges are the best and most accurate for adjusting the valves on your 4-valve Boxer. When adjusting the valves, it's very important to use two feeler gauges. If you were to adjust one valve at a time, the small amount of play that exists in the rocker arm bearings will make it nearly impossible to get an accurate and consistent adjustment on both valves. The trick is to have two feeler gauges, one for each valve. You'll be able to evenly match the drag that you feel on each set of valves."  

Ja, mein Herr.....

Checked my S-DCT tappets this morning with the bike at 16K kilometers - absolutely spot-on to spec.

[kayz1 2,928]

Thank you for that snip-it kind Sir, i had not even looked at my BM feelers..I have several sets of them from my BM days.

Lyn.

[embee 7,288]

5 hours ago, sykospain said:

"These feeler gauges are the best and most accurate for adjusting the valves on your 4-valve Boxer. When adjusting the valves, it's very important to use two feeler gauges. If you were to adjust one valve at a time, the small amount of play that exists in the rocker arm bearings will make it nearly impossible to get an accurate and consistent adjustment on both valves. The trick is to have two feeler gauges, one for each valve. You'll be able to evenly match the drag that you feel on each set of valves."  

 

The main difference is that the NC uses independent rockers for each valve, whereas the BMW uses a forked rocker with two valves working off one rocker, which to make things worse is heavily offset so the feelers will tilt the rocker. Doing the NC valves one at a time is fine, no need to do them as pairs.

 

 

Edited April 19, 2017 by embee

[Tex 36,817]

Man, that NC system is substantial or 'built to last'. Sooo much better than the oil head method (which has since been changed, obviously). I wonder what delights await me inside my Triumph cam cover when the time comes?!

[embee 7,288]

A very nice feature of the NC is that the rockers have a roller follower running on the cam lobe to reduce friction. Interestingly on this pic from Google the inlets seem to use a different roller on two of them. According to the blurb the valve timing is slightly different on the 2 cylinders to compensate for the uneven firing intervals with shared siamesed inlet and exhaust ports (pulsations affect the 2 cyls differently so different valve events are used to cancel it out), but to be honest I can't really think why the rollers should be different on the 2 inlets on one cylinder. The plot thickens ........................

[Tex 36,817]

Murray, surely if the rollers are slightly bigger (or smaller) it will advance (or retard) the timing on that cylinder? Cheaper than making a camshaft with different lobes, I imagine. Triumph (Meriden) used to use racing (large radius) cam rollers on the inlet to tweak the TR6 into a 'Saint' police bike.

 

Honda have worked the same trick on two cylinders of the CB1100 retro to give the engine 'more character' by uneven firing. 

[embee 7,288]

I don't think they're different diameters, the reflections look the same, but 2 of them are wider. As far as I'm aware from the workshop manual data the timings are different on the 2 cylinders not on the 2 inlets on each cylinder. From the first pic comparing the BMW/NC you can see that the inlet rocker nearer the camchain is  different to the rocker away from it on both cylinders, and the "wider" one has an oil squirt hole in the top.

Making different cam lobes along the shaft is no problem these days, it's not done from master cams on the grinding machines any more like it used to be.

[TheEnglishman 401]

On 4/20/2017 at 09:14, Tex said:

....I wonder what delights await me inside my Triumph cam cover when the time comes?!

 

 

Watching a video of shimming a tiger800 on youtube put me off them for life - I can see why there's quite a few of those engined bikes for sale with 10-11k on them!

[djsb 493]

Can the tensioner be replaced with a manual one? I have only found Ape and krieger who make them but not for our bike. Are these manual tensioners better?

[embee 7,288]

15 hours ago, djsb said:

Can the tensioner be replaced with a manual one? I have only found Ape and krieger who make them but not for our bike. Are these manual tensioners better?

Why?

The aftermarket "manual" tensioners are usually to replace oil fed telescopic type plunger tensioners which can sometimes become sticky, depending on manufacturer and design. Early SV650 tensioners for example could sometimes be sticky and the design changed, the later ones solved the problems. You have to be a little careful to allow for thermal expansion effects making the chain become over tight with a manual tensioner.

The NC tensioner is a spring linkage mechanism with a wedge check device to stop it moving back too far. As mentioned, there have been a few reports of NC tensioner "failures" but I have yet to see the exact failure mechanism, what is it that causes the problem? No-one to my knowledge has posted any photos of a failed item. Reports are very few and far between so I'd not worry about it unless you have reason to suspect something not right.

I had gone in there to do the routine valve check but also to replace the gear selector pin which has also had a few reports of failure apparently concentrated on early build 700 Integras, which mine is, so as a preventative measure to avoid failure when I'm off on a long tour I decided to change it. I decided to have a look at the tensioner just to see what condition it was in, and as far as I'm concerned it was perfect (20k miles).

[djsb 493]

What about this

 

http://www.clems-garage.com/CCT/about_tensioners.htm

 

Does this article have any truth to it? I don't know enough about these tensioners (or any in fact apart from the manual one on my previous MZ RT125 bike ( a single cylinder DOHC engine)) to dispute it. That is why I'm asking as you clearly know a lot more than I do. Thanks.

[embee 7,288]

I've seen that article before and IMO it's a well written sensible account by a clearly competent engineer. There have been a number of engines which have had issues with cam chain tensioners, it can be a tricky problem to get right because of the dynamics of what is going on (been there). The NC tensioner works in a totally different way to those described.

As far as I know the NC tensioner is not an issue, as said there have been a few reports but until I see what the issue was with the "failed" units I reserve judgement as to whether it is simply isolated instances or something more systematic, or even mis-diagnosis which can sometimes be the case. So far NC field service would indicate it isn't a problem, and as the article says, if it ain't broke .................... (not something I always take notice of admittedly, but that's just me).

One thing (not very nice feature) with the NC design is the requirement to use a locking pin when removing/installing it (held in the backed off position). If someone forgets to remove this pin .....................

[SteveThackery 3,090]

On 16/04/2017 at 11:11, embee said:

Shamefully I don't have feelers in the exact values given, 0.17mm inlet and 0.28mm exh, so I use the "go-no go" principle of 0.15-0.20 for inlets and 0.25-0.30 for exh, the smaller should be a slack slip fit, the larger should be too tight, so the clearance must be pretty close to mid way, near enough anyway.

 

 

Actually, embee, you raise a very interesting question, because "when I were a lad" I was taught that all feeler gauge measurements are, or should be, of the "go-no go" principle, because it isn't possible to make an exact measurement with them.  The reasoning goes that the amount of drag on the feeler is subjective, and thus automatically unreliable.  On the other hand, a go-no go test is binary, yes/no, not subjective.

 

It comes from the same principle that no dimensions can be specified exactly, either.  All engineering drawings state both a dimension and a tolerance for each part, because zero tolerance is impossible to achieve. 

 

The valve clearances on our bikes are specified to +/- 0.02mm, which is to say the inlet is supposed to be in the range 0.15mm to 0.19mm, and the correct way to measure this is using a 0.15mm feeler and a 0.19mm feeler.  The 0.15mm should be free, and the 0.19mm should block (i.e. a go/no-go test).  If it passes that test, the clearance is within the range specified.

 

Obviously you could use 0.16mm and 0.18mm feelers if you wanted to, which will give you a measurement (or adjustment) to a tighter tolerance than Honda require.  But if you use a 0.17mm feeler, exactly how much drag is enough?

 

Now, I'm not saying this is right - only that it is what I was taught, and it is logically consistent in terms of engineering tolerances, specified dimensions and measured dimensions.

 

Embee: you are clearly a highly experienced and dead clever engineer.  Is what I was taught wrong?  (PS: this is not a challenge - I'm genuinely interested in hearing an argument that differs from mine, on the grounds that I might learn something).  

 

My aim is to learn something new every day.

Edited May 1, 2017 by SteveThackery

[embee 7,288]

I'd agree with all you say Steve, we obviously have very similar engineering upbringings. I'm quite happy using the go-no go feeler principle, though go-no go gauges are generally no longer used in the auto manufacturing industry because it doesn't give any process control feedback (you don't see trends or drifting etc) so can't correct the process until it goes out of range, whereas you want to see the trends to diagnose the variables (tool wear, loose fixtures etc) and correct it before it reaches the end of the tolerance spec. to keep it nearer optimum. This is why you tend to see dimensions as a nominal and a plus and a minus, not necessarily symmetrical. If you have just a max and min it doesn't tell the process engineers what the ideal size is.

 

The issue of tolerances is a very interesting engineering topic, and when releasing engine component drawings is a critical part of the job. If you adopt the simplistic "stack tolerance" method you can end up with pretty big ranges and this can make for poor fits and assemblies, think about 5 components fitted in series and plus/minus tolerances on each one. This gives very variable quality products (think old Brit bikes). A statistical approach gives a more realistic condition, it's extremely unlikely that all parts are top or bottom limit, but in theory it can mean that a component might not fit. The modern principle is much more an individual measurement system and selective fits (I remember when the company I worked at had 3 main bearing grades, and it turned out Lexus were using 20 grades). My Yaris for example has bucket tappets which are selected thicknesses, no shims are used. If you need to change the clearances you need new buckets, but there is actually no service interval for them, fitted for the life of the engine.

Edited May 1, 2017 by embee

[SteveThackery 3,090]

13 minutes ago, embee said:

This is why you tend to see dimensions as a nominal and a plus and a minus, not necessarily symmetrical. If you have just a max and min it doesn't tell the process engineers what the ideal size is.

 

Indeed, yes, and in fairness Honda specifies its valve clearances like that: 0.17mm +/- 0.02mm, so in this case it's a symmetrical tolerance.  In my limited experience I've found asymmetrical tolerances in things like shafts fitting in holes.  Going back to my college notes I found this:

 

[SteveThackery 3,090]

22 minutes ago, embee said:

If you adopt the simplistic "stack tolerance" method you can end up with pretty big ranges and this can make for poor fits and assemblies, think about 5 components fitted in series and plus/minus tolerances on each one. This gives very variable quality products (think old Brit bikes). A statistical approach gives a more realistic condition, it's extremely unlikely that all parts are top or bottom limit, but in theory it can mean that a component might not fit. 

 

Have you heard of "Monte Carlo Analysis?"  I ask because when I was a Reliability Engineer it was one of the tools we used.  Each item has a known tolerance range, and a known shape to its tolerance curve (usually a normal distribution).  You then make a computer model which selects one of each item in the assembly, and the exact size of each item is random (but within the tolerance range and weighted by its tolerance curve).  It then "assembles" all the bits together and tests whether the assembly actually works or not.  Having recorded a "success" or "fail", it then "discards" that assembly and starts again with a new set of components, each of which has a randomly assigned measurement somewhere within its tolerance range.  This is repeated thousands of times.

 

For anyone still reading (sorry!) there is a good reason for this.  If you have lots of parts, each with a range of possible dimensions, there may be many millions or billions of possible combinations of sizes for all the parts combined.  If you want to know what proportion of your assemblies won't work, you may find there are too many possibilities to compute.

 

The Monte Carlo method is so-called because it's like drawing random cards from a deck.  Or drawing parts at random from a parts bin.  Usually, after a few thousand test assemblies, you get a good indication of what proportion of your assemblies won't work.  This makes it much more realistic than working through the millions of possible combinations in sequence.  The result you get from Monte Carlo analysis starts out fairly vague, and as the sample size gets larger the result becomes increasingly reliable until you are happy enough with it.

 

Sorry, guys, I'm not sure why I just told you all that, except to say that this stuff fascinates me!  

 

Yeah, yeah, OK - I'm a deeply sad anorak.  

[djsb 493]

Monte Carlo analysis is also used in Electronic circuit simulation as well. Not used it myself though. Good to see it can be used in other engineering fields as well.

[SteveThackery 3,090]

1 hour ago, djsb said:

Monte Carlo analysis is also used in Electronic circuit simulation as well. Not used it myself though. Good to see it can be used in other engineering fields as well.

 

I should emphasise that when I used it, it was in modelling complex telecommunications networks, not mechanical components.  Before using Monte Carlo I researched it, and that's when I read about its (theoretical, at least) application in any circumstance where you have a complex system with tolerances (or, in the case of telecommunications, component failure rates).

 

I should have made clear that my description above - with reference to physical parts with tolerances - was taken from the material I read, not from personal experience.  I decided international telecommunications networks weren't a good way of explaining Monte Carlo analysis!

[embee 7,288]

I don't remember the name but that's essentially what's behind what I was referring to as the statistical approach, assessing the probability that random assemblies would meet certain criteria. Fortunately I didn't have to do the grimy bit, I had staff!  I just had to make sure I knew what it was they were doing so I could convince my bosses that we had it all under control.

I was in product design/development, but I always had a huge admiration for the production folk who actually made the stuff. That's a whole different ballgame of process control and logistics. It was always fascinating to learn about process capabilities which would then determine how you went about specifying components and assemblies.