jon928se

If we have time tommorow after dismantling Paul (UKKid) 's clutch. We'll check Pauls belt wih my Kempf Gauge.

IIRC the instructions for the JFK gauge say to use the high side of the aperture for the 32V Engines ?

Jon
Black SE

Vilhuer

OK, here goes at last.

We had a test session few weeks ago where purpose was to find out how different belt tensioning methods compare against each other and to check if all four Jay Kempf developed tools on hand give same reading.

Test personel:
- Three local 928 owners ('86.5 US S Aut., '87 euro S4 5sp & '92 GTS 5sp) with combined experience of maybe three 928 belt changes and several checks on 16V ja 32V 928's using Kempf tool but no previous experience with factory tool.
- One Porsche owner (911, 924x?, 930, 944 and 951x? but no 928's) with experience of several 944/951 cambelt changes using hand measurement to adjust both cam and balancing shaft belts (no disaster yet).

Test subject:
- '87 S4 5-speed euro model registered in Finland.
- Approximately 180k kilometers (115k miles) driven.
- Entire cambelt system just rebuild with new parts where necessary including new cam sprockets etc.

Equipment:
- Factory 9201 tension tool and it's calibration bar 9201/2. Almost new, only used on two 928's before this measuring session.
- 4 Kempf tools, one used on several 928's and 3 new ones almost straight out of the box.

- WAV-file from Jager Engineering website loaded to laptop and headphones (good quality) for listening it.
- Multimeter (low quality but adequate) for checking when loose belt warning alarm disappears.
- Assorted metric tools for crank rotation and belt adjustment.

Conditions:
- Constant 18 Celsius throughout the two hour session
- Humidity, normal Finnish spring day = no rain, snow or hail
- Test subject on jackstands waiting for new engine mounts to go in

Measuring:
- Test subject was very clean without any oil or other residue anywhere to hamper proceedings
- Tensioner rebuild but oil not yet in, all tests were done without oil in it.
- Belt warning mechanism checked with multimeter and working as intended.
- Factory tool calibrated (to 4.0) and used according to instruction by Wally Plumley before start and once in middle of measurements to make sure it's still calibrated right (it was). There's barely room to place factory tools measuring rod on three different valleys between two adjacent belt teeths. One used on measuring results was one where tools position looked same as in various instruction pictures in web. All three positions were tested to see what difference they make.
- Kempf positioned like it's instructions show but other ways were also tried.
- Both 9201 and Kempf's used like lower belt cover was on when in fact it was of.
- When measuring engine always at TDC and turned to it with long clockwise rotation. No short last degree turns and certainly no counter clockwise rotation at all.
- Always after each adjustment, engine is rotated at least two turns to spread the difference in tightness throughout the whole belt. Affect of not doing it was measured on one occasion.
- Measuring done on all meathods severan times on very step by more than one person to establish correct and repeatable results. Many measurements done by 3 persons 3 times on every tool. Single factory tool number was derived from up to 9 measurements on a basis that at least 2 users had to be able to replicate same result on 2 out of 3 tries.

Results:
Situation at begining, belt installed correct but very loose.

#1 Belt adjusted tighter until belt warning disappears.
- 9201: 4.3 but can't be sure of result as it varies several tenths between measurings and depending on who's doing it. On this first set of measurements we did check factory tool also before turning engine two rounds, result was 5.3 instead of 4.3.
- Kempf: All 4 at lower limit or 1mm under it.
- WAV: Sound closer to 'second' sample but not exactly same.

#2 Tighten adjustment bolt exactly 360 degrees.
- 9201: Very hard to get same result 2 out of 3 but probably about 8.0.
- Kempf: All 4 tools 1 or 2 mm's above upper limit.
- WAV: Sounds tighter than 'correct' sample.

#3 Loosen adjustment exactly 180 degrees.
- 9201: 6.0, slightly easier to get repeatable reading.
- Kempf: All 4 just slightly (about 1mm) below upper limit.
- WAV: Sounds slightly tighter than 'correct' sample.

#4 Loosen adjustment exactly 60 degrees using adjustment bolts head to measure degrees, 60 = 1/6. In other words losen one side of bolts head.
- 9201: 5.3, still not easy to get repeatable measurement, depends on user.
- Kempf: All 4 dead center, not mm more or less.
- WAV: Very close to 'correct' sound but not quite same.

Notes:
- Tiny movement of adjustment bolt makes large change in tighness.
- Factory tool 9201 way too sensitive. Large differencees depending on who's using it and how. Likely to work better if user practices enough.
- Kempf is very reliable, geting so similar results for all four prove that. Measured differencees (when outside center of the window) between four in my opinion isn't meaningful as factory tool results vary also. This reliability was no surprise as one of the Kempf's (the previously used one) had already been tested against 9201 on '90 S4 and '92 GTS 5sp (not mine, this one has engine together) euros by their owners. They found very similar results as we. Kempf was at middle of window when 9201 showed 5.3. Only surprise to me was how little above upper limit 8.0 (what must be very unhealthy for exhaust cam bearing surfaces) was.
- WAV-file worked to an extend. Last tenths are difficult to get right. 5.3 result is more luck than skill in my opinion.
- Using just warning system to adjust is something I wouldn't recommend. Final position was only 120 degrees from end of tighness warning, not 180 to 240 degrees (1/2 - 2/3 round) as stated in Jager Enginering site. 180 (1/2) above would be 6.0 and 240 (2/3) maybe 7.0, so both way too tight.
- Using just hands is even more quesswork than any of the above methods. Despite this, we noticed that on 5.3 it was fairly easy to turn belt 45 degrees at Kempf measuring point. After 45 if got considerably more difficult.

Writer and other measuring staff do not take any responsibility to any financial, mental or any other damage above text might cause. So do not send engine work and therapy session bills to us. However we would very much like to hear any and all comments why we got results we did. What we did wrong/differently when our 5.3 wasn't at tighter end of Kempf's window.

One thing that might have affected on our results was fact that we ended to 5.3 by releasing tension and not increasing it. In retrospect, maybe we should have returned to about 4.0 and started upwards again. Not sure if that would make any difference though as same results were found on two other 32V 928's in totally independend measurements. We do not believe explaining factor to be 9201 tools flats as they were unintensionaly rotated several times over the test and probably were in different position(s) during second calibration. Will check if they have affect on calibration when 9201 is on hand next time. If not, can rule that out.

9201 we used has since been used on 944 Turbo and 968. In 968's case, cambelt was first set by cars own 'automatic' tensioner and then belts value was measured to be exactly what it's supposed to be (4.0). This leads me to believe our 9201 is working properly. This leaves our measuring method as possible cause of difference compared to others results.

Kempf seems to give same result from several adjoining belt slots. 9201 on other hand varied significantly when measuring rod was placed to other than normal valley. Results were about 1.0 higher in next valley and completely of the scale in third.

All in all I'd say Kempf is by far the best tool for the job. Would like to know what's correct position to aim for in 16V and 32V. We'll likely do 9201 vs. one or more Kempfs test on 16V euro S soon, so this might give some answers.

Regards,

Erkka

1992 928 GTS 5-speed experiment
Oak Green Metallic 22L
Classic Gray MX

EDIT: Correct link to Wally P's writeup

Garth S

Good results: question - what TB did the car have installed? - the conventional round tooth HTD, or the newer 'dimpled tooth' HTD profile?
This would not have any impact on the torque/tension measurements, but may on Paul J's sound file technique: at constant tension, the harmonic frequency will vary with the mass/unit length of the 'plucked string'. Belts from various manufacturers and differing profile detail may well have differing masses ( and we trust length to be constant). (( I never weighed a belt to check - there may be no signifigant difference)).

UKKid35

Excellent review Erkka, many thanks! It seems churlish to complain that you didn't include the Phil Risby tool, although from personal experience I would say that although it is sensitive it is not easy to use because the arc through which the "pointer" moves is difficult to measure.

Garth, good point about the belt type and frequency, I'd forgotten about the new dimpled tooth profile.

Vilhuer

Very good points guys

Belt was newer dimpled tooth type, can't remember by which company but can check it from owner. True that difference in design could affect sound. All more reason not to use it as method of checking. Unless someone makes complete list of all possibilities. Even then it leaves user as huge variable where as Kempf seems to be almost totally user independent.

We left perspex gauge out as we didn't have time to make it and car wasn't perfectly straight on jackstands. Front was bit higher that rear. Don't think this had any effect on any of the measurements we did but it could have affected perspex gauge. Should have mentioned this on previous message and a fact that there was old Toyota present at same space at same time. If car would have been level I suspect perspex would have given similar results as sound file due to unknown variables like measuring scale. In ballpark but who knows where there exactly. Would be interesting to know anyway. Maybe next time.

Regards,

Erkka

1992 928 GTS 5-speed
Oak Green Metallic 22L
Classic Gray MX

Niklas Kampe

Erkka, the belt is manufactured by Dayco.

Best regards
Niklas Kampe
928 S4 1987

H2

Erkka,

As you did, I found the 9201 tool to be difficult to use in measuring the Tbelt (but only because of the very cramped quarters). Conversely, on the alternator belt, it was easy to get dead-on, repeatable settings every time with the 9201…but not so with the Tbelt. So I don’t think it’s so much the gauge as it is the experience of the user in finding the sweet spot and method for getting T-belt readings that are consistent. I spent 3 hours playing around with the 9201 before I felt comfortable that I had it pretty well figured out…and even then, it’s a challenge in that I’m not satisfied with until I make several repeatable measurements. But in my experience the settings for the Kempf Gauge was always at the upper end of the window as compared with the 9201 set at 5.3.

One thing I learned about my 9201 tool is that the results can vary slightly depending on which side of the “ flats “ you use to measure with against the flat side of the belt. I know a 944/928 mechanic in Boise who painted one side of the flats so he would always use the same (shiny) side each time he used the gauge because of this variance in his gauge. Using the tension check bar verifies that there is also a reading variance in my 9201 depending on which side of the flats you use.

My method is to set the tension with the Kempf gauge….then double-check it several times with the 9201. That way, I feel that I’ve covered both bases. I’m entirely satisfied that the Kempf gauge is the best tool for quick checks….but, for my 85’ 32V, it’s at the top end of the window. I understand that there are others who also use the upper end of the window of the Kempf gauge. Perhaps it’s just the way I check my belt? I assume no liability for anyone else's car and suggest you follow the directions that came with your gauge (playing lawyer).

Generally speaking, one thing you confirm that is absolutely true…The Kempf is by far the easiest for the job. If I were having anyone else work on my car…I’d trust their use of the Kempf gauge more than their use of the 9201…unless they were a highly skilled mechanic and used the 9201 regularly on 928s in their work. Even then, if they get in a hurry, they could get a wrong reading.

Perhaps someone who has connections with a highly skilled mechanic that uses the 9201 tool on 928s (and who does not sell the tools) could have them do a side by side test? Experience and getting consistent results in using the 9201 seems to a common factor. Also, repeated tests at the same time without turning the belt for a “ fresh start “ can have an effect on the measurements, I find.

Harvey
85S’
======
“What we did wrong/differently when our 5.3 wasn't at tighter end of Kempf's window.” Erkka

ew928

I forwarded this discussion to the honorable Jay Kempf.
Included is his reply:
____________________________________
Ernest,

To the 4um types talking about settings with the factory vs. my tension
gauge:

1. I haven't ever suggested using the high side of the notch for 32v 16v
cars. That is from people that have done comparisons. Not that there is
anything wrong with it. I calibrated the tool to the original factory
tool. this tool was designed for the 16v variants but they have exactly
the same belt system (minus the electrical warning as the later cars
give or take a couple small details. From my tests and experiments as
well as the prerelease tester findings the slot in my gauge will keep
the belt light off and won't overtension the belt. That is what I would
call a "safe" range of tension.

2. The early cars can and should be tensioned exactly the same as the
later cars. There is a mention in older manuals about setting the
tension on an old water pump to less than the 5.0+.3 setting. The early
cars are a little less apt to slip a belt when a bit loose because the
single cam arrangement generates a little less torque and because the
square belt profile is a little less likely to jump.

3. Some engines will crash if they spit a belt. People who crash an
engine In all but a couple cases had been neglecting the belt for a long
time or had extenuating circumstances like oil or coolant leaks or
mechanical issues. The tensioning number is somewhat arbitrary and
skilled people can do an ok job of tensioning with their fingers (it is
way better to be consistent and use a calibrated tool of some sort). The
important part is to be monitoring quick changes and looking at the
system somewhat frequently especially on older cars. No one to my
knowledge has ever spit a belt that was monitoring belt health
occasionally and the act lets you see leaks before they get out of
control.

4. My gizmo monitors tension based on a very large deflection of the
belt, like more than 45°. The factory tool deflects the belt something
like .010 inch and then uses a micrometer measurement to amplify the
reading. This is why it is so sensitive. My tool has a small percentage
of the overall movement as its calibration.. Like say 5% with a direct
reading, no amplification. It looks simple but it is actually elegantly
designed. It should be. It was designed by the engine team that designed
the 928! I just updated and reissued it. I was worried about calibration
so I made it all stainless so it wouldn't corrode and throw the reading
off. I was worried about damage so I used to literally throw the
prototypes against my garage wall and recheck and recheck them. It is
VERY hard to knock them out of calibration.

5. I really appreciate the discussion!!! The Finland comparison team
get's my utmost respect and admiration. Careful, you might get a visit
from me next winter

There have been quite a few comparisons and originally there was some
skepticism using the gizmo with the 32v cars. One of the original
testers (who wrote the 32v portion of the instructions) is still
tensioning his own 32v (and 79) car with the second original prototype
gauge. That started in 1999. Still going. There are 1000+ of these
things in circulation thanx to the teams at 928intl and 928 specialists.
There have been some questions about calibration but there has never
been one belt lost on a car that the owner made the effort and bought a
gauge (mine or factory) regardless of technique or the skill of the
perpetrator.

If anyone has any questions ever you can reach me at jkempf@tds.net
But alas you won't find me on the forums.

Cool stuff, great discussion and thanks for all the support.

Kempf
79 (the original tension test mule)

JP Rodkey

I haven't surfed this forum as much as intended, but this discussion got my attention. Just wanted to add a little to Jay's message.

My buddy and I were part of Jay's pre-release test. Gary has the 32V (before I added mine), and I have the 16V. The goal was to compare the 9201 tool with readings from the Kempf tool. We spent an afternoon performing the check on both cars, logging the procedure, and comparing results. Bottom line is that for the the tools we had, the readings were consistent for both 16V and 32V engines. We loosened, reset, and ran the engine between each session to verify results. Based on our very simple 'real world' procedure, I was convinced of the accuracy and repeatability of the tool. We got proficient with the 9201 tool, but along with it's precision you get a rather sensitive process to get a consistent reading. Although the reading you get with the 9201 is precise, it is much more difficult to properly set up the entire system to get repeatability. As Jay explained, the 9201 amplifies a very small sample. It follows that any variation (error) is also amplified (position on the belt, measuring from the same tooth, etc). This is not meant to suggest the factory tool should not be used or trusted. It's just a bit more complex to utilize.

Some of us will want to be as precise as possible and stick with the 9201. My feeling is that it doesn't require that level of accuracy. This belt was marketed at the time as the world's longest used in an automotive application. It's under heavy duty with varying loads. You've got normal wear, temp variations, mechanical complexities (several rollers), and the tensioner. Under normal conditions, that belt's setting varies while in use. It would be interesting to be able to monitor tension while driving! Maybe the next challenge for the Finnish group?

Hat's off the the Finnish experimenters. Great writeup! And I'll fall in line with everyone else expressing appreciation to Jay Kempf.

heinrich

My tests with my 9201 and the Kempf took (mine) bear these results out, *especially* the part about 9201 giving non-repeatable measurements. I use the Kempf tool and it yields consistent results.

bigrob

I used mine yesterday! Worked flawlessly! Provided that the spring is at the right tension My TB was installed last May & the car has clocked 2000 miles since. The belt checked out fine. Should it have been a little loose or is this good?

Rob

Bill Ball

This is good.

fraggle

How timely for this to pop up again. I just installed my belt and have the Kempf tool!

Of course, I forgot to put in the lower belt cover so I'm going to have to put the fly wheel lock back on and pull the damper off again. sigh.