Most dyno's fall into two categories. Inertia and applied load (?). There are several ways to apply load using a water brake, electric brake, generator, and so on.
The Dynojet is an inertial dyno that calculates HP based on the acceleration of the weight and the time it takes to do said work. From that, you can extrapolate torque using the HP=Torque * RPM / 5252 equation. For RPM you need to have either a pickup from the engine or calculate it based on tire size, gear ratio, transmission ratio, and speed of the roller. Obviously, the engine input would be the best.
There are some good and bad of the dynojet. First the good. Traction is rarely a problem and operation is easy. And operator error is minimized because of it's simplicity. The bad is that you can't change load to get maximum power at all levels of the graph. You are assuming that you are going to be at maximum power when you put your foot in it, but a diesel's power lags behind it's acceleration. So the numbers may become skewed and may not give a representative example of the engine's true potential.
An applied load dyno works like this. You spin a set of rollers that has a brake attached to it. You load the brake and the brake deflects against a counter force (spring, lever, or fixed weight). As the brake resists the motion of the rollers, it moves against this weight. The distance it moves is measured by a strain gauge. This is converted into torque against the rollers (tractive effort). The machine then calculates HP based on the speed of the rollers and the tractive effort being applied at that speed using the equation HP=tractive effort * speed (mph) / 375. Based on HP you can extrapolate torque using the equation above. Of course, getting your input for RPM in one of the above manners.
The good of this dyno is that you can fully load the engine to it's maximum power for a given speed (RPM). You will have full boost and fueling and be able to hold it there to allow the reading to stabilize for accuracy. The bad is that traction can be a big problem. The brake may function, but the friction required may not be sufficient and the wheels will slip. Obviously, this will screw your numbers up and ruin the test. Then you'll have to do it again. Our observation with our dyno (GB Transmissions RAM 11) is consistant that our trucks will start to slip around 350-375 HP (actually around 1800 Ft-lbs of TE). The other bad of this dyno, is that the operator must know what he is doing. We used a pre-loaded curve to simulate road load (same as they did in Muncie) and found that the power lags behind the curve (for the same reason as the Dynojet does). Also, every problem we ran into in NE IA, the guys in Muncie ran into. Not to mention, I don't think that the calibration was accurate for that brake, IMHO.
A good test with this dyno would work this way. Bring the engine to the governor at full fueling, and then apply the brake to take the enigne down to your idle speed (or something close) and then bring the engine RPM back up by releasing the brake taking data on the way back up. That way you'll have full fueling and boost for a given RPM when you take your data point. But again, you run into the evil traction issue and that make this test very hard to do. The whole test, as described above, should take no more than 30-45 seconds. But having not done a successfull test as described, I can only speculate.
So how's that? Did that answer your question adequately? If not, I'm sure someone else can correct me or add their 2 cents. Hopefully, I didn't muddy the waters on this too much.
BTW, thanks to docdyno and Sun diagnostic systems for explaining this stuff to me while we were getting our dyno "modernized".