The most reliable torque wrenches, in my own experience, are the click type; however, they are the most sensitive as well. Whenever one of my guys drop one on the ground, we send it to check calibration.
Dial-type torque wrenches are durable..i.e. they can survive a drop to the floor; however, these are designed so that the handle 'gives" or bends a predetermined amount of flexing at a particular point from the fulcrum. These are reliable as well, but the error is introduced when the technician reads it -specially in tight places or when it has two scales and the viewing angle is squewed. These, however, are great for determining run-on torque (the minimum torque that a nut will exert while turning on the threads). This is one way to determine if the locking feature of a self-locking nut still is present on a used nut.
The "click" type wrench has an advantage that no matter what, the pre-determined torque point is reached regardless of any lenght added to the lever...i.e. a pipe bar (don't try this on plastic handle wrenches) adding to the lenght will make it easier on a technician to apply 300 ft/lbs, but the wrench will click exactly then.
The only place that the added lenght changes things is at the pivot point. Placing an adapter 3 inches long at right angles with the wrench's handle will maintain the same torque value. This is usefull in tight places. I use this method when torquing intake manifold bolts (the center ones), for the carburator pad gets in the way of setting the socket squarely onto the bolt head.
However, if one aligns the adapter so that it's lenght adds and aligns with the handle, then one is changing the fulcrum on the wrench and the added lenght will have a tendency to multiply the torque value by a fractional amount...i.e. you set 100 lb/lbs and a 3 inch adapter effectively adding to the lenght of the handle from the fulcrum will probably torque the nut or bolt to 125 ft/lbs. ( For example only. I have not calculated the actual value) The same can be done by aligning the adapter with the handle and substracting the effective lenght between the handle and fulcrum.
We use torque multipliers made by Sweeny, and these are designed to torque a nut anywhere between ratios of 7:1 to 12:1. i.e. we apply 45 ft/lbs and the nut is torqued to any value between 315 and 540 ft/lbs. These beat any pipe bar method out there...:L
Calculating the effects of an adapter on a torque wrench is usefull when one only has one range of torque values in one's torque wrench and one must use a value outside such range. i.e. if the torque wrench's value range begins at 40 ft/lbs, then one could calculate the torque setting for torquing a nut to 25 ft/lbs and vise-versa. We get around this by having a collection of wrenches that will cover the whole torque value range in our engines.
Typically, once one is done using the torque wrench then one must set it to the lowest torque value on the scale on click-type wrenches. The most accurate range for torquing is about 15- 20% from the bottom and top ends of the range. i. e. a torque wrench with a range of 20 to 150 ft/lbs provides the most accurate torque value at 23 - 24 torque setting on the bottom and 120 - 127.5 ft/lbs. When this wrench is set at 128 ft/lbs the accuracy begins to decline. However, most manufacturers recommend a particular range of torque values for a specific bolt usage. If one tries to stay in the middle of the range, then this will eliminate any error or inaccuracy.