machinists don't use torque wrenches

spokewrench

Well-Known Member
Region
USA
Except rear axle nuts, every fastener on my Aventon bike has a hex socket, but the supplied folding hex key set is terrible. More than 50 years ago, I discovered that the ISO hex keys that came with BMW motorcycles took the guesswork out of torquing, so I ordered a complete set. The H4, H5, H6, and H8 each supplied the specified torque to a fastener.

A couple of days ago I got a digital torque wrench with a 3/8" ratchet drive. I had assumed that an ISO key was designed to let you feel when you had reached a certain amount of torque. By using hex keys to apply torque to sockets on my new wrench, I discovered that I didn't have a feel for how much torque I was applying.

To double check the faster I'd torqued with allen keys, I used tape to make flags on hex bits. I'd insert the bit in the screw so that the flag was close to a reference point, use the wrench to back it off, and then to tighten to spec. In each case, the flag showed that the wrench had tightened the screw to the same angle as the hex key. I conclude that a proper hex key helps you torque correctly by feeling the "knee" in the increasing stiffness.

My neighbor is a machinist making truck transmissions at Eaton, a company that pulls in $25 billion a year. I proudly presented my new wrench. As soon as I handed it to him, he asked why I would want a torque wrench that small. That confused me. Didn't he ever torque anything to less than 60 Nm (45 foot pounds)?

He said they use only allen keys. During his training period, a supervisor would often check his work, telling him how close he'd come to perfection and explaining how the handle should feel. (I guess at the plant, torque wrenches are used only for fasteners so big that an allen key would have to be several feet long.)

With a torque wrench, an error in specs could mean a serious mistake. If machinists use allen keys, the tool maker need only specify bolts with the right size sockets. Torque requirements depend on friction, which can vary from what was anticipated. With an allen key, you're sensing the elasticity of the fastener, which is the point of torquing. I've never seen an allen key get out of calibration.

Privately, I love my torque wrenches, but if anyone might see, I'll use my ISO hex keys. I've got my reputation to consider! :)

flag.jpeg
 
Last edited:
I was observing Specialized technicians and salesmen at work. Whatever they do on the bike (like setting the saddle height for a customer), they always use fixed torque wrenches similar to this one:

1743397986909.png

A 5 Nm fixed torque wrench.

These guys work with expensive bikes and e-bikes that do not belong to them. Working with torque wrenches is the part of avoiding responsibility if anything goes wrong. Bicycles are a way more delicate than trucks, don't you think so?
 
Three years ago I bought a torque driver that works on that principle. I tested it today with a strain-gauge driver. The first time, it clicked at 4.07. The tenth time, it clicked at 3.80. For the 8 between, it clicked between 3.85 and 3.90. The whole spread was only 7%: pretty good. For the middle 8, it was 1.2%: great!

Precise, but not at all accurate. When I hadn't had it long, it clicked at 4.8, so now it's 20% low, due to 3 years of compression on the spring. What's more, it was erratic when I tested it a few weeks ago. Friction affects these tools. Maybe there was a bit of internal oxidation the last time.

I think the issue is not how delicate a machine is but how much leeway there is between "tight enough" and "too tight." With metal bikes, I think most fasteners have lots of leeway. Some bike mechanics say they use torque wrenches only to avoid being blamed in case of a mishap, as you say. My seat post clamp has plenty of leeway. With an ISO H4 key, I can feel what I'm doing much better than with a bit or folding keys. When it's tight enough to keep the post from moving, I can feel that I haven't nearly taken up the elasticity in the bolt. It's doing its job, and that's tight enough for me.

If a bolt in a factory has to handle a load near it's strength, there won't be much leeway between tight enough and too tight, and there will be more at stake than the price of a bicycle. If machinists are like commercial pilots, tool makers are like airline pilots. Besides designing parts for bicycles and motor vehicles, tool makers take charge of factory machinists. I was surprised that the tool makers at Eaton still find allen keys better than torque wrenches.
 
I previously worked at a off-road vehicle manufacturer. On the assembly line, they not only used torque wrenches, they used digital torque wrenches. They not only used digital torque wrenches, they used connected digital torque wrenches, and the torque applied to each bolt is stored in a database tied to the vehicle's VIN.
Now, not all bolts were torqued this way (eg dashboard trim pieces), but anything that mattered for safety was.
 
I believe the machinists like my neighbor build and maintain the machinery used for production. I hadn't thought of assembly workers. I'll have to ask him if torque wrenches are used there.

Henry Ford's assembly line sped up production but made work so boring that there was a high turnover. OTOH, it made tasks so simple that with attractive wages, he could grab somebody off the street to keep the line going.

If I had to to keep an assembly line going, I'd want a wrench that told an untrained worker how tight to turn each fastener. What's more, a connected wrench could let a supervisor pull up a list of how much torque an employee used each time and if he missed any fasteners.
 
Last edited:
I do not like gloves and I do not like torque wrenches. I want to feel. Given that, some employers want to hear the click to know it is right. I used a torque wrench on a stem and the next day it was sloppy, that is when I walked. It was sabotage by a violent, power hungry service manager, who wanted me to be under his thumb. He wanted to point to an error to gain power and control. I fixed it before others could detect it, then quit.
 
I now have two digital torque tools: 0.3 to 6 Nm and 3 to 60 Nm. Measurement doesn't seem to involve any moving parts, which I think may mean great accuracy now and in five years. Each can be set to record peak, which means I can tighten by feel, then see what torque I used.

That's fun, but I prefer the feel from proper allen keys. Even for a grade-school novice, I think it would be hard to get a screw much too tight with an allen key, or to leave it much too loose.
 
When we got ebikes, I bought torque wrenches. I am confident I was over-torqueing before.
 
At one point I read an article about the development of torque specs, that said most experienced mechanics will torque fasteners correctly even without a torque wrench. I have at least four torque wrenches and I use them, but not everywhere a torque spec is given... like others, I know what "feels" right. That said, for some things, like where the proper clamping force is critical for friction or sealing, I use the torque wrench. Things like cylinder head bolts, or the bolts holding a wooden airplane propeller (which is driven by friction against the flange, not the bolts). Or the handlebar clamp bolts.
 
If you want to get technical, throw away the torque wrench and measure bolt stretch instead.
There is a lot of if, ands, and buts in the proper use of a torque wrench.
 
You can use the stretch method on rod bolts. Other than that there aren't t many places you can torque by stretch.
I suppose if you work day in and day out on the same bolts and nuts one could get pretty darn good at the feel method for torquing properly. My thought on bolts that go into aluminum is better to be safe than sorry and use a torque wrench.
 
If you insert the short end of a Park Tool 4mm wrench every day into hex screws of face plates of stems, it will give you the same deflection on the end of the long side every time when it is properly torqued. It is a kind of torque wrench in that way.
1746128728664.jpeg
 
Hi all,

I registered here for the single purpose of commenting on that topic.

Torque to apply goes with screw diameter in third power, so an M6 hex wrench would need to have a more than three times longer handle than an M4 to get operated with the same hand force -- which is not the case, obviously. Thus, I find it difficult to guess proper torque when changing between, say, M5 and M8 screws. Admittedly, above of M6 there are no threads on a cycle in need of full power, axle nuts e.g. 3/8" or M10 come with hand force; full torque would do damage.
On cars, at latest, I cannot rely on feeling; an M14 10.9 strut bolt goes with 200 Nm which would need a seven feet lever for to work with the forces I'm used to.
Moreover I find myself depending on daily performance or stress level or what else. Last week I had to build a wheel which had failed two times already (not under my reign, luckily), so I started with some steam pressure and "I need to tighten these spokes really well". When I first checked tension, I had come out a major third above Sheldon Browns recommendations; Mersenne's formula told me I just had missed 200 kgf .... oops. Even when the material did fine with that, on the long term I think the spoke holes would have cracked, so I reduced strain.

Thanks PedalUma for your hint to watch the hex wrench's twisting angle, but consider that a bigger wrench needs to be longer as well to twist in the same manner.
Thanks again for having mentioned stems since these were precisely the occasion when I started to use a torque wrench on bikes.
With a stem plate improperly tightened, the handlebar, while force is applied to the grips, infinitesimally takes off from its seat and only rests on the stem's edge, suffering a stress peak. The alloy material doesn't like such, so it's only a matter of time until damage is done.
When the component fails, we use to think that of course a bad producer let us down, unaware of the possibility that a bad mechanic disregarded the poor innocent thing.
When it's not the bar, it will be the bolted connection. Too loose screws holding the plate, highly changing loads apply on them plus an infinitesimal bending -- a prime example for failure. Again users will hate the supplier instead of the machinist.
Overtorquing does damage otherwise, no explanation necessary.
Sorry, Biplaneguy, I just repeated what you were saying, just with more words.

Imo there is not much leeway on modern stems when used on MTBs, and because I don't want to feed surgeons and dentists, I use a torque wrench there.
When a fender screw comes loose, just retighten, when a rack bolt fails, it's rarely more than a nuisance, but that safety relevant connection is something that makes me cringe thinking I could miss to properly install.
Also when unusually high torques get required, as e.g. on the clamping bolts of Hollowtech cranks, this causes me to whip out the torque wrench. On the one hand, when the designer bothered to use a high grade screw, he may have had good reason, and do I really want to find out the hard way if he had or not .... rarely. On the other hand, that hard bolt will snap off suddenly when overdone -- yikes ....

I agree that torque wrenches are no exact science, the tool itself as well as the constraints of the bolted connection open a wide tolerance window.
Bolt stretch measurement is no alternative on short screws as they are common on bikes, apart from the fact that it's a badly expensive method.
When tightening by feel, there are two 'knees' as spokewrench calls it, first when the connection's gap closes and the torque rises quickly, but to meet this is no success. The second knee turns in the different direction when torque starts to rise slower with turning angle, this is where either screw or support yield and either perfection has reached or the whole thing is short of failing -- possibly both. Unfortunately you can't torque like that without slightly mashing the connection, as can be seen on many seat post clamps on cruisers and roadsters. Formerly, this was a quite safe way to torque mild steel screws (coming with a slot head, mostly), keeping racks or fenders from getting loose, but it's also a safe way to pull threads out of an alloy frame or to fail hard, short bolts.

So I see no way of doing right without using common sense, with a torque wrench help a lot by doing so.

I don't torque cable clamps because stainless cables fray out when torqued too often, I don't torque brake and shift levers, just keep them safely from moving while operated but making them able to twist on the handlebar tube when touching ground during a fall.
Similar with seatpost clamps.
I don't torque brake pads but test instead if they twist when hand force is applied.

My five cents, no offence meant.
 
Hi. No offence meant here, either. Just my five cents!

An ISO 2936 H8 hex key is only 12.5% longer than an H6, but I can apply twice as much torque with the H8. The slightly fatter handle has a big effect.

The screws for my stem, riser, and handlebar clamp are marked for maximum torque only. By sticking the keys into sockets on a torque wrench, I found that it would be very difficult to exceed those torques with ISO keys that fit. If I don't have to worry about overtightening, my concern is to have all bolts on a clamp equally tight. Adjusting one will change the tension on the others. I think getting them equal is easier with a key than a torque wrench.

With allen screws a toolmaker can specify a screw with a socket sized for the appropriate torque range. I think M5 screws normally have H4 sockets, but the M5 screws that secure my polyethylene chain guard to sheet steel brackets have H3 sockets. That application requires a delicate feel and maybe 0.4 Nm of torque. The smaller key does it better.

I don't remember when I last needed 200 Nm, but my 2-foot breaker bar will apply 150 Nm with ease. I love that bar because with a 12-point socket, it gives me the rigidity I need to torque by feel. An ISO hex key also gives a rigid feel. If your strut bolts have sockets, the designer may have intended them to be tightened with hex keys. It seems the ISO "long" series is for 10.9 bolts. I guess the handle would be about 2 feet long.

I remember when soft-steel fasteners were common because at one time they couldn't mass-produce harder stuff. It would be Class 5 ISO and Grade 2 SAE. You might be fastening a barn door to a hinge, maybe with square nuts and bolt heads. You'd tighten until you felt the threads start to schmush. Permanently deformed, they wouldn't easily unscrew.
 
…The screws for my stem, riser, and handlebar clamp are marked for maximum torque only. By sticking the keys into sockets on a torque wrench, I found that it would be very difficult to exceed those torques with ISO keys that fit. If I don't have to worry about overtightening, my concern is to have all bolts on a clamp equally tight. Adjusting one will change the tension on the others. I think getting them equal is easier with a key than a torque wrench.

With allen screws a toolmaker can specify a screw with a socket sized for the appropriate torque range…

you‘re making a general statement about best practices of working on bicycles based on a handful of parts of specific type and manufacture. while It’s generally true that larger fasteners correspond to more torque, it’s way more nuanced than that. of my two most recent bikes, one has 6mm head fasteners that range from 5 to 15nm. applying 15nm to a part specified for 5 (an expander plug in the headset) is a horrible idea, which is why actual torque wrenches exist. similarly, there are 4mm head fasteners raging from 3.1 to 6.2 and little tiny 2.5mm ones ranging from .8 to 2.0

engineers make decisions about the size and material of fasteners and the things they fasten based on many criteria, including hardness, stiffness, weight, resistance to corrosion, accessibility, and the physical amount of space available. to just say you can properly put together a bicycle by applying similar amounts of torque to fasteners based on the size of their tool heads is firstly wrong and secondly borderline dangerous to those without as much experience in the “feel” of fasteners.
 
you‘re making a general statement about best practices of working on bicycles based on a handful of parts of specific type and manufacture. while It’s generally true that larger fasteners correspond to more torque, it’s way more nuanced than that. of my two most recent bikes, one has 6mm head fasteners that range from 5 to 15nm. applying 15nm to a part specified for 5 (an expander plug in the headset) is a horrible idea, which is why actual torque wrenches exist. similarly, there are 4mm head fasteners raging from 3.1 to 6.2 and little tiny 2.5mm ones ranging from .8 to 2.0

engineers make decisions about the size and material of fasteners and the things they fasten based on many criteria, including hardness, stiffness, weight, resistance to corrosion, accessibility, and the physical amount of space available. to just say you can properly put together a bicycle by applying similar amounts of torque to fasteners based on the size of their tool heads is firstly wrong and secondly borderline dangerous to those without as much experience in the “feel” of fasteners.
Yes, for example, an M5 x 0.8 is typically torqued at 5 Nm for a clamping force of about 1400 pounds, but my designer used them to clamp a polyethylene chain guard without washers under the screw heads. That would be like driving on polyethylene in a Cadillac Escalade with studded tires.

The screws clamp the chain guard to brackets of soft steel that’s only thick enough for one turn of thread. That explains why the designer didn’t use M4 or smaller; the thinner screw would have stripped the sheet metal threads too easily.

I can apply almost 5 Nm with an H4 ISO key. In this application, that would both damage the polyethylene and strip the threaded holes. That’s why the designer used screws with H3 sockets.

The bike came with a folding set of keys. Using one handle for all sizes would make it too easy to overtorque an H3, and everything was too springy for feel. I would have been better off with a torque wrench but for one big problem: the bike manufacturer specified 2 to 4 Nm, which would have been damaging. He was probably publishing how much torque it took the snap those stainless screws loose.

With an ISO H3 key, I use about 0.4 Nm because the feel tells me I’ve compressed the polyethylene enough to keep everything tight. That’s not close to the 2 Nm I could apply with that key. It’s certainly not close to what I’d apply if I followed published specs.

My M15 x 1.6 front axle is what you would call nuanced. Judging by the size of the head, I used a 6” ratchet handle. I couldn’t feel it taking up elasticity, which was a first for me. Lacking a torque wrench for the 8-10 Nm range painted by the bolt, I could only estimate how much force I was applying. That’s not accurate.

Remembering how well Heyco keys used to work, I bought a set that seemed to be ISO. I used a scale so I could calculate how much torque I applied when I worked by feel. It came out to 8.9 Nm, right in the middle of the specified range.

Actually, I’d felt for the maximum torque I could comfortably apply with that key. I hadn’t felt the bolt taking up elasticity, and that bothered me. I bought a digital torque wrench for that range, to complement my smaller and larger ones.

Eventually, I realized I hadn’t been using a real ISO 2936. I bought a set and found the steel less springy (more rigid) than the ones I had already. With a better key, I could feel the bolt taking up elasticity. One reason I’d missed it before was that the increasing torque was very low for a bolt that size. That’s because it’s an aluminum alloy with little thread engagement.

The other reason I’d missed it was that instead of taking up elasticity over a revolution or more, it went from no load to full preload in about 60 degrees, or 10 minutes on a clock face. The bolt was about 150mm long, but it was really a 15mm bolt with a 135mm head. The depth of elasticity was only about 0.25mm.

The genuine ISO key lets me feel what I’m doing. I find that quicker and more reliable than using my torque wrench to apply somebody’s specs. The bicycle maker’s spec for that bolt is 10-14 Nm, which is too high.

The two biggest fields for tool and die makers are designing and maintaining industrial machinery and designing vehicle parts. Among mechanical engineers, they are like airline captains among civil pilots. Typically, to become one you start with a bachelors degree in mechanical engineering, then find a company offering a five-year apprenticeship, which includes classes.

I don’t know if tool and die makers design all bicycle parts, but I think they designed my axle. They had the spec marked by the bolt because a published spec could be wrong, and a mechanic might not have it, anyway. The H6 socket seems small for a bolt that size, but an H6 ISO key is just right for the task. It lets you feel if you haven’t taken up the elasticity, and pushing the key beyond 10 Nm barehanded would be painful.

I’ve had 4 e-bikes whose cranks were held by M8 x 1 bolts. On one, I had to pull the cranks several times before I decided to replace the BB. They were easy to torque by feel with a 6” ratchet handle. The elasticity came from stretching the hole in the crank arm as it slid onto the tapered shaft. When the resistance increased more steeply, I’d know I’d taken up the intended elasticity.

I happened to see 3 youtube videos where 3 mechanics used these bolts to put crank arms on 3 brands of bicycles. Each one advised the viewer to buy a torque wrench the was good to 50 Nm and use it to torque to spec, which they said was about 45. I knew that was far more than I used. Each of the 3 said he would do it by feel.

That “do as I say and not as I do” sounded like something demanded by youtube lawyers in case a viewer torqued by feel and got hurt. I don’t think a loose or broken crank bolt will cause anything worse than the annoyance of a wobbly crank arm.

One of them filmed himself torquing the bolt. He wasn’t applying anything like 45 Nm.

Indeed, my bike company specifies 40 to 45 Nm on those bolts. I still didn’t believe it.

I saw that tightening my left crank arm bolt would apply torque to the left crank arm without resistance from the freewheel. I loosened the bolt, tied a hanging digital scale to the pedal with a cord at right angles to the crank arm, and used an H8 key to tighten the bolt by feel. The scale read 103 newtons. I multiplied by the 175mm crank arm length and got 18 Nm, less than half of what was specified.

Who was right? The torque spec for an M8 x 1.25 is easier to find: 23 Nm. The spec for an M8 x 1 should be 20% less, or 18 Nm. Right on! That’s where an M8 x 1 should perform best.

The minimum tensile strength is reached at 45 Nm, so you could torque that far without snapping the bolt. However, the minimum yield strength occurs at 36 Nm, so if you torque to 40 or 45 Nm you’re likely to deform the threads, making the bolt more likely to work loose.

When I replaced my BB, the shaft was a different length, so I needed a new crank set. It came with M8 x 1 bolts which I used. The left crank arm kept getting loose. It would stay tight for hours. I’d check with an allen key to be sure it wasn’t starting to unscrew in spite of thread locker. Suddenly, the crank arm would wobble and it would have moved out 6mm. It was as if the bolt threads were stripped, but a caliper showed they had the proper outer diameter.

When I replaced it with a bolt that had come with the bike, there was no more problem. I guess a previous customer had ordered the crank set, tried it, and returned it. The damage wasn’t visible, but he’d deformed the threads of one bolt by torquing it to commonly stated specs. The bolt came with a socket for an ISO hex key that would torque it more reliably than somebody’s published specs.

I don’t regret buying torque wrenches, but I think good hex keys usually do better. Most hex keys aren’t very good, though.
 
Back