Probably best to look for something covering the bulk of the 20/120Nm range
So just about everything from M6 alloy engine threads... up to suspension and wheel components.
I am lucky to have been 'hands on' for 30+ years.. and have a 'feel' for what is required..
But still advocate a calibrated device at the extremes.. or for 'fussy' applications.
Charlie
Great advice being given here. My most used wrench is an old Britool which does 20 to 100 lbs ft
I guess it does 90% of any "torquing" I do. There is a problem with relying on torquing fixings to achieve the desired result though. Lets think about tightening down a cylinder head, which would be a typical "torquing" job. What is the objective? Well, we are trying to ensure that the cylinder head face crushes the head gasket evenly and to the required degree of "crush" against the face of the block. Yes? So why are we measuring the force required to turn the nuts/bolts? Well, if everything is optimal, so all the threads on the nuts/bolts/etc in perfect condition with no imperfections or contamination, probably lightly oiled, but check manufacturer's recommendation (and how many of us bother to do that) and the holes in the head have no problems and the faces of the nut/bolt and casting are perfect - think of anything else, there probably is? - Then, under these theoretically perfect conditions, you can probably get just about the same tensile pull registering on each fixing. If anything isn't perfect then "the games up the pole" You will not achieve the desired result precisely because you are measuring turning effort on the fixing NOT CLAMPING EFFORT HEAD TO BLOCK. Some clever people have obviously thought about this because now a days you'll find, more often than not, that head bolts, and many other fixings, are "angle tightened". When I first ran into this I didn't like it, mostly because I had got used to torquing every thing in sight! However if you think about it it's a very logical solution to the problem. Typically an angle tightening procedure will involve a reasonable level of preparation of your fixing followed by a light torquing (maybe 10, 15 or perhaps 20 ft lbs - equivalent Newton Meters are available) This is to just settle the component (head perhaps) in placed and take backlash etc out of the screw threads to give a "common" starting point for the main event. The manufacturer will have specified a very precise procedure for the angle tightening both in terms of the order in which the fixings are to be tightened and the stages in which it's to be done. A typical head may be done in two or three stages, perhaps all tightened 90 degrees, a short pause to allow settling. Then 90 degrees again and a pause. Then, perhaps 30 degrees. The great advantage to doing it this way is that it removes friction from the equation. The treads have a known pitch and the initial settling torque job makes sure the fixings are all snugged up so by turning the fixing through a certain number of degrees we know that it has advanced up its thread by an "X" amount thus bringing the head face that much nearer to the block evenly for each fixing. That, of course, is why the tightening is done in stages because a certain amount of bolt stretch will take place so you don't want to dramatically tighten one bolt more than another or it will stretch more and the whole result will be compromised. Even this is part of the equation these days with "stretch bolts" being used which go, to some extent, towards evening out the small differences as individual fixings are tightened. That's why you can only use these bolts once, If reused they will not give a predictable amount of stretch and quite possibly snap because they are already stretched before you start.
To do this angle tightening you need an angle gauge:
Which, I'm delighted to tell you costs peanuts, you could even make one but they're so cheap it's not worth it. The gauge is used with a socket and T bar or power bar:
A simple piece of wire is set up against the engine block/head or wherever to stop the scale rotating as the pointer moves with the turning of the power bar.
I like the click type of torque wrench shown above and you can buy them much cheaper, but perhaps sacrificing a little on quality and accuracy, when compared to my Britool. There is a "cheap" option which I wouldn't dismiss if you're on a very tight budget and that is the "bendy" beam type:
This is actually the one I bought at college back in the '60's. It works by having a socket at the opposite end of the bar to the handle. As you apply force the socket resists turning so the thick bar (made of very high quality steel) bends. the pointer remains straight but the scale moves with the handle as the beam bends and so the pointer traverses the scale. The Americans are very keen on this design - I believe Snap On still list one? - But it requires you to be able to see the scale in use, sometimes impossible in a crowded engine bay. The accuracy is entirely reliant on the quality of metal used in that beam so avoid very cheap examples.
All torque wrenches will "drift" (go out of calibration) in time and the more they are used the quicker that will come about. The manufacturers recommend a yearly calibration for a workshop, shared, tool (possibly used every day) I use to do mine every 3 years but now retired I think every 5 will do. It ain't cheap! I was looking for a solution to this so I could do it at home (system of levers and weights) but it all got a bit silly when I started thinking about the one I use on hubs which can go to over 300 NM. Then I saw some videos on you tube about devices like this:
It's basically an electronic strain gauge. According to the manufacturer never (well nearly never) needs calibration and runs on cheap CR2023 button cells. My two most used wrenches had just come back from calibration. I clamped the square end of the Clarke tool in the vice (with some face protection) and the torque wrench into the square socket and tested at 1/4, 1/2 and 3/4 scale. The two readings coincided so exactly (and I did each reading 3 times for consistency) I really couldn't see any divergence. So, although I could use the electronic Clarke tool as my every day tool (well, once a month these days) I am going to keep it for checking calibration. I decided this for two reasons. First, being use so infrequently and in conditions where it is unlikely to suffer damage, it should remain super accurate. Second, It's too bulky in cramped situations. See it here Compared to the Britool:
See what I mean?
So there you are. I like to torque anything which is particularly safety related, so hub nuts, caliper to carrier and carrier to hub, anything to do with suspension etc. I also like to torque engine and gearbox/drive line components. controversially perhaps, I don't usually torque wheel nuts/bolts because I like to lightly coat them with anti seize. Doing this makes them subject to overtightening if a torque wrench is used and as I've been doing this for maybe 50 years now, I've got a good feel for tightness. Never had a wheel come loose yet - Oh dear, shouldn't have said that!
PS remember to always, ALWAYS, back the adjustment back to zero when not in use on the preset "click" types. if you don't the calibration will rapidly be lost and you might as well be using a power bar from then on for all the accuracy it will give you!