The torque measuring sensor on a dyno is just a test instrument, and it needs to be calibrated like any other test instrument. It will have a perfect value +/- some tolerance for acceptable results. Since it is measuring torque, all you need to calibrate it is a lever to act against the sensor, and a known weight to hang on the lever. Length of lever in feet X weight in pounds will give you the correct torque in lb-ft, and this can be compared to what the torque sensor is reading on the dyno's display to determine if it is accurately calibrated.
On my Superflow 901, Superflow provides a bar that bolts onto the absorber and acts against the torque sensor. The bar has a notch exactly three feet from the center of the absorber where you can hang additional weight. The weight of the bar itself should cause the torque reading on the dyno console to be 17.5 lb-ft of torque. To calibrate my dyno, I built a steel frame that hangs on the notch in the bar, and that can hold weights from a Sears barbell set. I took the steel frame and weights to a certified scale, and weighed them all to an accuracy of 0.05 pounds. Now when I hang the steel frame and the weights on the bar, I know exactly how much weight is on there, and can multiply this weight by 3 feet to get lb-ft of torque on the torque sensor. Over the last five years I've regularly calibrated my dyno, and it is typically accurate to within 0.5% of the total torque on the torque sensor.
I think that intentionally mis-calibrated dynos are much more common than a lot of people realize. I've seen plenty of dyno data sheets where the numbers just don't pass the smell test. For example, one shop operated by a VERY well respected NHRA racer gave a friend of mine obviously bogus dyno results. Anyone familiar with dyno operation could tell, because the barometer reading on the dyno printout indicated that the shop was located at around 5500 feet of altitude, when in fact it was probably only at 1000 feet or so. Changing the barometer reading built into the dyno is only one way to fudge the numbers; there are other hardware and software tricks that can be easily employed to make the dyno read however you want.
In order to protect yourself when taking your engine to a dyno, you can always ask the operator to calibrate it for you while you watch. This will take about 15 minutes, and will involve having the operator attach the bar and his calibration weights to the torque sensor, to show you that the reading is accurate. Any reputable dyno operator will do this, especially since you are paying for his time to do it. Of course, if his calibration weights are labeled incorrectly you could still get snookered on this, but it is more likely that the dyno operator will give you some lame excuse why you should trust him, and he doesn't need to perform the calibration. Buyer beware at that point. You can also do a quick check of the dyno's barometer reading to see if it is accurate, along with the other weather conditions monitored by the dyno. These are usually available from the printout, and should be a good match to the day's ambient weather conditions.
There's more detail to this, but none of it is particularly complicated, and anyone should be able to go to the dyno and be confident in the accuracy of the results by checking a few simple things. My book has all this information in excruciating detail, for anyone who is interested. FWIW, when I hear a dyno operator say his dyno is conservative, I consider it a big red flag. What you want is a dyno that is
accurate. Then you've got useful data that can be used for comparison purposes.
Jay Brown
FE Power - Dedicated to Building and Racing the Ford FE Engine, 390, 427, 428, SOHC