I just changed the timing belt and water pump on my 98 Mazda 626.

After I got everything reassembled I started up the engine. I had been running the engine for a few minutes, revved it up to around 3k rpms a few times and started and stopped it a few times. At some point I noticed a strange sound which seemed to be coming from the timing belt area. If I had to guess, I would guess it was the sound of the timing belt flopping around due to it being loose. I never heard such a sound from that area before. I reused the tensioner, the spring and the idler pulley as they all seemed to be in good condition, although I've never done this before so what do I know.

The things that somewhat stood out to me where that it was fairly easy to get the new belt on ( most people said it was a real PITA ), I simply used the allen wrench to turn the tensioner left until the allen wrench hole was exactly to the left of the tensioner bolt. I put the belt just a little bit over the cam sprockets first, then worked it fairly easily a little onto the crank sprockets, then went around with a rubber mallet gently tapping it into place, a little at a time. It took under five minutes.

The other thing I noticed was that after I removed the zip tie, the tensioner moved back quite a bit, and the tensioner spring was maybe stretched to 2.5 times it's resting length. I don't know what's normal here, I don't have any basis for comparison.

I'm thinking that if the noise really is the belt flopping around, then maybe even though there was 100% no noise like this before, and even though the tensioner spring looked fine, maybe it's just not strong enough to deal with a brand new belt?

I'd really like to hear what people think as I'm a bit afraid to drive it till I figure out what this noise is and I don't want to just throw parts at it.

Here's a picture of how I got the tensioner spring on by holding the tensioner with a zip tie:

enter image description here

EDIT Nov. 13th 2016

So I pulled off the valve cover and upper timing cover and took a look at the belt:

enter image description here

I've run the belt maybe ten minutes total, and the outside part which contacts the tensioner and idler pulley already has this kind of smudged look to it ( don't know if that shows up so well in the picture ).

I also took an approximately one minute video of the belt with the engine running and there is no wobble in the belt and no noticeable movement of the tensioner or tensioner spring. However you can hear the noise very clearly in the video.

So here is my theory at the moment. When I tested the tensioner and idler I spun them and listened for any noise. I didn't hear any noise so I figured they were good. However, when I spun them they only turned a few times before stopping, which is in contrast to what I'm seeing in this video in which the guy spins a t-belt pulley and it spins freely for quite a while.

So I'm wondering if the bearings in my tensioner and idler pulleys are somewhat worn out to the point that the the pressure from a brand new belt is sufficiently greater than the pressure from the old belt and that's making them make noise where they didn't make noise before.

However, I ran across this other post which says the bearings should not spin freely and in fact should stop almost immediately.

So now I'm completely confused. As a side question, can I safely drive this thing around till I get this noise figured out?

EDIT Nov. 14th 2016

OK, so now I'm onto something. Only had about twenty minutes to mess with this today, but I noticed before starting the engine up that the t-belt was hanging about a millimeter off the edge of the idler pulley:

enter image description here

So before starting the engine up I pushed the belt toward the engine a bit both at the idler pulley and at the cam sprocket. So there was no noise until about ten minutes after I started the engine, when I could start to slightly hear the same click clacky knocking kind of noise, which seemed to be coming from the bottom end. At that point I noticed that the t-belt was slightly oscillating back and forth on the idler pulley ( video ) in the plane tangential to it's point of contact. After another five minutes or so ( 15m total since starting the engine ) the noise was again very distinct.

It may be that I didn't get the crank pulley quit on as far as I should, and that there is a little back and forth play in the crank sprocket which only shows up after the belt warms up and loosens a little bit.

So I'll probably test that theory out tomorrow.

EDIT Nov. 17th 2016

OK, so the noise is still there ( although not as bad ), and here's what I found and did.

When I pulled off the lower timing cover I noticed a few things:

enter image description here

The first thing I noticed which I didn't really pay attention to before is that the t-belt cover is severely warped. So much so, that it's contacting the back of the crank pulley body and the crank timing teeth and has burnished them silver.

The second thing I noticed was that the belt was sitting at the very edge of the sprocket and the edge of the belt had been rubbed off enough that you could see the copper strands in the belt itself. You can see it clearly here after I pushed the belt as far onto the sprocket as it would go:

enter image description here

After I pushed the t-belt all the way onto the crank sprocket it also automatically lined up properly right in the middle of the idler pulley:

enter image description here

I also noticed similar wear ( although not as bad ) on the back of the belt:

enter image description here

So remember what I said about the timing cover being severely warped? Well it looks like both the belt and the tensioner pulley have at some point been rubbing against the inside of the timing cover:

enter image description here

The thing is of course that I don't know when this happened, as I didn't pay any attention to the inside of the cover when I originally took it off, and the size of the circular scratch is about half a centimeter smaller than the diameter of the tensioner pulley.

This just gets more and more confusing. I'm thinking that maybe my first course of action should be to replace the timing cover, since it's clearly warped and has contacted the belt and tensioner at some time, either in the present or the past.

EDIT Nov. 22nd 2016

OK, now this gets really interesting and goes off in a direction that I think no one expected.

I decided to see if I could improve a little on Zack's truly excellent answer with some more exact measurements and data. So I took the video and loaded it into Audacity audio software, and zoomed in a bit to look more closely at the audio data:

enter image description here

As can be seen in the highlighted area, the clicking noise heard in the video is extremely uniform and occurs at a rate of exactly 16 clicks per second, or 16 Hertz. Looking through the entire audio file confirms this rate, even if an occasional click is muffled or missed here and there.

Now the question is what's making a clicking noise at a rate of 16 Hertz? Here is where I'll differ a little in my approach. In order to determine the RPM's of any gear or pulley in the system, it is sufficient to know the RPM's of any one gear, and the diameter of that gear and any gear whose RPM's you want to find. The calculation is:

(Driving Gear Diameter / Driven Gear Diameter) * Driving Gear RPM's

This works because the timing belt and accessory belts cause any gears or pulleys they are connected to to behave as if they had a direct physical connection. If gear A moves through an arc measuring one centimeter, the belt moves one centimeter and gear B also moves through an arc measuring one centimeter. How much faster or slower the driven gear turns is determined by the ratio of the gears circumferences, which is the same as the ratio of their diameters since you can simply factor out PI from 2*PI*R. You could factor the 2 out as well and just be left with the ratio of the radius', but I think it's easier to work with the diameters.

So anyways, I went ahead and measured or looked up the diameters for everything in the system and did the calculations. There are two groups, the first group is the things driven by the crankshaft sprocket, and the second group is the thing driven by the crankshaft pulley / harmonic balancer:

|Gear / Pulley    |Diameter in Cm.|RPM     |Hz    |
|Camshaft         |      10       |375     |6.25  |
|Tensioner        |      6.2      |604.84  |10.08 |
|Idler            |      5.2      |721.15  |12.02 |
|Crank Sprocket   |       5       |750.00  |12.5  |
|Ball Bearings ?? |      0.5      |7500.00 |125.00|
|                 |               |        |      |
|Harmonic Balancer|     14.11     |750     |12.5  |
|Air Conditioner  |      12       |881.88  |14.70 |
|Power Steering   |      12       |881.88  |14.70 |
|Water Pump       |      11       |962.05  |16.03 |
|Alternator       |      5.5      |1924.09 |32.07 |
|Ball Bearings ?? |      0.5      |21165.00|352.75|

So what jumped out at me was that the Water Pump pulley is rotating at almost exactly 16 Hertz, the exact frequency at which the click was occurring.

Why was my brand new GMB water pump making noise?

I say was, because the sound has since disappeared for no apparent reason I can think of. Maybe the bearings in the new water pump just needed a breaking in period?

Anyways, I'll wait for people's comments, but I think Zach definitely deserves the the bounty on this one even if it turns out the conclusion was slightly off.

  • I would have replaced both the tensioner springs and idler wheels. Both wear out. Commented Nov 12, 2016 at 22:35
  • @spicetraders I'd like to have a specific reason for replacing something before I start throwing money and parts in an old car. Commented Nov 13, 2016 at 4:28
  • 2
    @RobertS.Barnes The exposed "cooper strands" you see there is the aramide/kevlar/whatever lining of the belt. AFAIK you see them also on a brand new belt. On production the belt gets cut of some large tube of "belt" material.
    – Martin
    Commented Nov 17, 2016 at 13:35
  • 1
    @RobertS.Barnes The following link shows a brand new SKF belt. You can clearly see the fibers. auto-bitner.com/ebay/SKF-BMP/SKF-BMP-6.jpg For your timing cover problem: I would suggest you to put marking fluid (engineers blue / something else) on the inside of the cover, let run the engine for a while and then look if the belt has any traces of it.
    – Martin
    Commented Nov 17, 2016 at 23:48
  • 1
    Hahahahaha! So, the problem went away on its own? Oh well, we had fun. Thanks for improving the method. Hopefully, this work can be used by someone in the future to actually fix a problem, haha. Commented Nov 22, 2016 at 18:27

3 Answers 3


I don't know if I deserve all this credit. My idea was sound, but my measurements were pretty bad so my numbers are way off.

If you try this, use the more precise software that Robert S. Barnes ended up using and don't rely on a YouTube video to measure a belt spinning at 200 RPM!

In your first video, the timing belt is going through 1 revolution per second, aka 1 Hz.

(I slow-mo'ed your video to .25 times speed. Watch from about 7s to 15s. You'll see 3 quick, white streaks. That's the writing on the belt. Then the 3 streaks won't show again until 1s later. And then it repeats.)

[After reviewing the video, I've concluded I have no idea what it's frequency is, but it is much faster than 1 Hz. The video was blurry and I'm thinking the white flashes were too fast for the camera.]

The tapping is much, much faster. This tells us that it is not a particular spot on the belt that is causing problems.

I analyzed the audio waveform in your first video: Audio Waveform

See the two bigger spots? Those are cars passing by. Look at the first bigger blob; it has has two peaks within it. The first peak is at 14.5 seconds. I counted the number of ticks (or where there should have been a tick, but camera movement interrupted the mic). I got 205 ticks in the first 14.5 seconds.

(# of ticks) / (time in s) = (frequency in Hz)

205 / 14.5 = 14.14

The tick occurs 14.14 times per second, aka 14 Hz.

The time between ticks is (almost) perfectly uniform. This is not a random slapping.

(belt Hz) / (tick Hz) = (length ratio)

1 / 14

The problem is a pulley that's circumference is (roughly) 1/14th of the length of the belt.

Your timing belt is 42.53". (Btw, you should include the engine model when asking questions about the engine. ;)

Remember that circumference is measured in units of length, e.g. meters, inches, etc. For every inch the belt moves, the pulley moves exactly an inch too.

(belt length) / (length ratio) = (circumference of pulley)

42.53 / 14.14 = 3.00" pulley circumference

Make sure that you stick with one unit system throughout these calculations; don't divide mm by cm or in.

(circumference) = [Pi] * (diameter)

(circumference) / [Pi] = (diameter)

3.00 / 3.14 [Pi] = .96" pulley diameter

Your timing idler pulley's diameter = 2.05"

Your timing belt tensioner pulley's diameter = 2.44"

I don't think any engine pulley has a diameter of only .96". I assume this means the tick is occurring twice per pulley revolution. (Why twice? Eh, symmetry, intuition, etc.) [Nope. I just started with bad measurements, that's all.]

.96" * 2 = 1.92" pulley diameter.

There were several opportunities for human error in this calculation. I believe that 1.92" can be rounded to 2.05". The error is only .13". (That's small!)

I would start with replacing the timing idler pulley because it is (almost) a perfect multiple of the .96" predicted pulley diameter.

  • 6
    Your next level. Deduction Is as much art as it is a discipline. Commented Nov 18, 2016 at 2:39
  • 4
    seriously good answer! wish I could double upvote this.
    – Cc Dd
    Commented Nov 18, 2016 at 2:40
  • 1
    Dude, you are my kind of guy. I just woke up I'll see if I can respond to this later in the day. Regardless of whether or not it's right props for awesome methodology. By the way everyone here knows my 626 so well that I sometimes forget to write that it's the two liter model. Commented Nov 18, 2016 at 4:41
  • 3
    I am heavily impressed. I wish I could express it adequately
    – Martin
    Commented Nov 18, 2016 at 8:09
  • 1
    Regardless of any errors in measurement, the main idea was very innovative and I think you do deserve the credit people are giving you. I only reached the correct conclusion because of your idea which I just built upon. Commented Nov 23, 2016 at 7:55

well you either have bad bearings in your tensioner or a really fast telegraph operator in there.

Change out that tensioner and its springs which is what should almost always be done when changing belts. the skinny is there are certain things to toss money at and certain things not to. timing belts and associated parts are not something to skimp on. If anything breaks it may mean a new engine especially if it is an interference engine. here is a good video on why you want to change this and not let it break.

Yes if it does not spin for some time the bearings are probably bad.

Make sure after installing the belt that the cams and crank still line up correctly after letting the tensioner loose. Sometimes the tensioner can turn the cams a little and if the belt is too loose possibly a lot.

well that changes things. still if the tensioner didn't spin more than a couple revolutions you may want to replace it anyway but your update tells me that the noise was caused by one portion of the belt bulging a tiny bit further out than the rest and this was tapping the warped spot on the timing cover.

You can replace the timing cover else if you want to save a buc you could heat it up and unwarp it. Its primary job is to keep dirt and larger debris out of the timing belt.

That ticking sound from your video can be made if one of the bearings becomes smaller than the rest. This can be caused by one of the bearings getting jammed and not spinning causing a flat spot to occur or disintegrating causing a pocket. Each time this bearing, or lack thereof, passes the point the races are being squeezed towards each other, in this case it would be near the center of the bend in your belt, it will click.

Also this mangled deformed or broken bearing can cause the entire "bearing" to slow down due to debris flattened to the races or jammed in the other bearings or something dragging on the broken one. If it doesn't have a cage then the bearings may be rubbing on the non rolling one.

This probably started with the new belt simply because either the old belt was narrower or had already been warn. My guess is it was a narrower belt.

  • What specifically makes you think it's bad bearings in the tensioner? I just added some more info to my OP. Commented Nov 17, 2016 at 12:35
  • @RobertS.Barnes updated my answer.
    – Cc Dd
    Commented Nov 17, 2016 at 20:12
  • What I don't get is why this only started after I put on a new belt. Commented Nov 17, 2016 at 20:38
  • @RobertS.Barnes edited my answer with my best guess if you have the old belt you can put them together and see.
    – Cc Dd
    Commented Nov 17, 2016 at 20:54
  • 1
    Even a 1/2 degree cant on a bearing can take the belt for a shift in tracking. Which is why one simply replaces tensioner and bearing with the belt. Old dog old tricks. Commented Nov 22, 2016 at 19:08

This is a bit of a shot in the dark, but I'm going to point my finger at the tensioner spring. To me, the sound in the video is very... dull. Kind of thudy for lack of a better term. It sound like one of the pulleys is not having the teeth align quite right which is causing the belt teeth to slap into the groove.

The thing is that the belt can appear perfectly tight when you're observing it manually, but once the crank is putting rotation into it, it will create a tight section and a slack section. If your crank rotates clockwise your right side will be taught and the left will be slack. This is true in all rotary systems to some degree because there is no such this as perfection.

Now if your tensioner pulley is not putting enough pressure on the belt as it is rotating, it will be able to developed a little slack, and even 1mm of slack can create a sound like that.

One that no you could do to help, listen to the sound as you rev the motor. It may be hard to discern, but if the sound gets louder then I'd say that would add salt to my "slack" theory since it would develop more slack as rotational forces increase and press harder outward.

I'm basing this off the sound you get on a bicycle when the derailed is not working correctly. :) Also, my motorcycle chain made a similar sound when the chain was too tight... so theoretically it could be too tight too... 🤔

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .