I could not find instruction on calibrating a Torsion Pharmacy Balance anywhere on the internet. In the US it was legally required that a Pharmacy have a torsion balance on hand for compounding prescriptions. So these scales are very common and since they do not have delicate knife edges and bearings to wear or break, they are often in usable condition but calibration is uncertain especially if they have been shipped improperly or mishandled. You can find information on the internet about use and how to ship them properly but I have never found instructions on calibration techniques.
First basic shipping considerations. All balances should have provisions for locking the balance beam for shipment. Find the mechanism for doing this and be sure to lock the beam. On the balances made by The Torsion Balance Company there are stanchions that are moved into place after loosening a knurled nut on each one on the underside of the balance and then tightening it once in the locked position. Remove all loose parts and pack separately. Turn the knob that clamps the scale and tape it in place. Also tape the cover in place and wrap the whole thing appropriately for the method of transport. If you balance has a fluid filled dashpot, you will need to remove the cover and secure the top plug and or remove the oil.
Now for calibration secrets. I developed the method on my own so anyone who does this professionally is certainly welcome to comment and suggest other or better methods.
The torsion balance has two rails that operate in parallelogram fashion. Three torsion units act as the main fulcrum and as the fulcrums on which the weighing pans are placed. These are made in the factory with special tools and are difficult or impossible to repair if damaged. So check to be sure that the parts are all in place and that the tension bands are under tension. If you pluck or tap the bands gently you should hear a tone, not a thud. There is probably a desired musical note for each brand and size of scale but I don't know what they would be. If it is a thud there is little or no tension and there is no point in trying to make the balance work.
In all you do to adjust the balance, be aware that the bands can easily be broken and that whenever loosening or tightening any screws or nuts, be sure to oppose the loosening or tightening with a suitable tool so you are not stressing other components, especially the torsion bands. (How do I know this, you may ask?)
The torsion balance is quite simple so anyone who collects and works with scales should be able to understand how it works. Basically there is a center fulcrum and outer fulcrums that the weighing pans are mounted to. Since there are upper and lower beams connected with top and bottom tension bands (actually one band on each position serves as both top and bottom) the balance acts like a parallelogram and the top and bottom fulcrum points will always be vertically inline if the balance is level and in balance. There is quite a bit of tolerance for error in level and leveling is in fact the method used to fine tune the zero balanced position.
Depending on the type of balance there will be one or more adjustable weights for balancing the torsion assembly. My balance has a dashpot and for that reason there is a lot of additional mass on the right side of the beam assembly. So, there is a large counterbalance on the lower left side. On the right top beam is a fine adjustment. If your balance does not have a dashpot, it may not have the coarse adjustment. It is pretty obvious that these weights are used to get the balance balanced.
To calibrate the balance, you obviously need to remove the cover and put all the parts back on except the cover so you can make adjustments with the balance in working order. Start by leveling the balance with the built in bubble level or a separate one if there is none. My balance has a front to back level but not a side to side one. This is because it is assumed that the balance is calibrated professionally and the side to side is how you zero the scale. Since this might not be the case it is good to level the balance side to side as well as front to back for the calibration procedure.
If you have a dashpot, make sure it is not sticking on the dip rod or internally. Best to take it off and clean it if there is any doubt. Then adjust the various rods and levers to allow it to move from one extreme to the other without any parts touching or binding.
It is obvious how you get the basic leveling of the beam done. Just slide the adjustment weights after loosening the relevant clamp screws. What is not as obvious is the fact that the right fulcrum assembly can be move in and out with respect to the center fulcrum. This is critical in getting the balance to work properly with different loads. If one end fulcrum is closer or farther from the center than the opposite one, the balance will not behave linearly. Adding weight to one side will not be the same as adding weight to the other side for obvious reasons. On my balance the adjusting screws are made with two different pitches on the one screw. The fixing point for the tension band is movable within the beam. Turning the screw cause a differential motion between the fixed and the movable nut because the pitches are different. On my balance turning the screw clockwise (viewed from the end) moves the fulcrum closer to the center. This is the intuitive choice but other balances might be different or they may have entirely different methods for adjusting the position of the tension bands.
To get the arms of the beam equal you may have to adjust the screws that move the right side fulcrum both top and bottom. You can check for the need to do this using two weights that are as close to identical weight as possible. You will get the most amplification of error if you use weights near the limit of the balance but if you are careful and if absolute balance is not critical you can get by with smaller weights. I used 10 gram weights because my set had two of them. When I was finished adjusting the beam I tested with two 50 gram weights and the error was less than 1 mg.
I happen to have an old Mettler analytical balance so i used that to calibrate the 10 gram weights and to make them identical within about .5 mg. Both were a little over according to the Mettler so i sanded them a tiny bit at a time until they were right. The Mettler uses precise rings for reference and as long as you zero the scale and the balance has been cross checked for alignment errors the absolute weight is probably pretty accurate but that does not matter as all you need for calibrating the Torsion balance is two weights that are very close to each other. Another source of very close weights is freshly struck coins from the same run from a reliable mint such as the Canadian government. Uncirculated Canadian Maple leaf coins are probably the best you can get for precision. I checked several from the same batch and they are very close but they do vary by a few mg. Wire runs pretty accurately as long as it is from the same spool. You can measure identical lengths of wire and cut it squarely. Use a lot of fine wire so the cut does not matter that much nor does absolute precision in measuring length.
Getting the balance to balance with equal weights is easy, just move the adjustment weights until it balances. However that does not insure that it will balance with 10 times the weight or 1/10 the weight on either side. For that the distance between the center fulcrum needs to be precisely the same on both sides of the beams. Critically balance the balance with no weight on the pans. You can do this with both the adjustment slider weights and also using the leveling legs. Be sure that the front to back level gauge is centered and as a final nice touch, balance the scale with the side to side also level so you will then have the most adjust ability when the scale is used.
If the right fulcrum adjustments have not been messed with you can probably get the balance calibrated by adjusting the top and the bottom screws by equal amounts. When noting fractional turns of adjustment take into account that there may be some free play due to thread inaccuracy and it may be different on the top and bottom screws due to wear or ??? Take up the slack each time you change direction of adjustment.
After balancing with no weight on the pans, put equal weights on each side. Note if the balance moves to one side. Whichever side goes low indicated the side where the side fulcrum is farther from the center fulcrum. Adjust the screw both top and bottom little by little until the balances balances with the added weights. Be sure to damp the scale each time you add or remove weights. Undamp it slowly and allow it to settle before deciding if the adjustment was correct and if it was adequate. It is almost certain that the first adjustment will not be enough. When you move the fulcrum you also change the balance point with no load so when you check again you will find that the adjustment weights will also need to be changed likely in the direction opposite to the direction you moved the fulcrum. This is an iterative process, you will probably have to go back and forth several times.
The centering of the fulcrums is critical and will certainly limit the accuracy of the balance if not correct. If the top and bottom are not exactly even and if the scale passes the adjustment procedure described above the error is probably minor and can be ignored. If you are concerned, you can probably figure out how to use a caliper to check that the distance top and bottom are the same but this will be difficult and it is my inexpert opinion that if the balance balances at both minimal and maximal loads it is not a concern if the top and bottom fulcrum distances are not exactly the same. If you can not achieve this then the difference might be too great and this should be investigated. Assuming that the torsion bands were installed precisely and that the adjustment screws were at the same positions when this was done at the factory then the degree to which the screws protrude from the end of the beam should probably appear the same, certainly not varying by a huge amount that is visually apparent. If the band clamps have been disturbed you can not depend on this.
It is a real pain to unlock move and relock the adjustment sliders each time. Every time you lock or unlock them you will move them a little as well. So it is better to add various weights to the top beam or the pans until you have it right and then use the sliders just once. If you get tired of trying to lock the slider weight in place and finding it moves a little each time, you can glue little bits of correction weight on the beam instead. Or use a small piece of wire tied on at an appropriate place. On a scale with a dashpot, the dashpot arm is a convenient place to tie on a wire.
The other adjustment you can make is the knurled weights that move vertically on threaded rods on each side of the lower beam. Their is probably a name for these and for the property that they control but I do not know it. It is difficult to describe what they do but if you move them from one extreme to the other you will see for yourself. They basically adjust for sensitivity vs stability. When the weights are moved higher up, it takes less off balance to move the beam. If the weights are on completely vertical rods then they have no effect on balance when the scale is perfectly balanced. When the scale goes out of balance the weights move the center of mass of the beam farther out thereby adding to the offsetting that occurs. If the weights are too high up the balance will never be stable, the tiniest amount of off balance will send it to the extreme position. The lower the weights are, the less sensitive the balance will be. There is probably a specified amount of off balance that one division is supposed to represent. If you know what this is or if you are after a certain number, you can carefully adjust the knurled weights to achieve this value. Either adjust each side the same amount of check the sensitivity in each direction separately and adjust the weights on the side you want to adjust but be aware that this will also slightly change the sensitivity on the other side too.
Most of the adjustments on the balance will also change other settings. So whenever you make a change, check basic operation again and know what else you might have screwed up. If you use the torsion balance as recommended, the beam is always balanced when taking measurements so all the nasty non linear errors caused by minor geometrical misalignment are likely to be minimal. There is probably a procedure for each balance with a specified order in which adjustments are to be made but that does not seem to be public knowledge. The scale companies keep that a secret so they can charge for calibration services. My Mettler balance has something like 150 points of adjustment so needless to say, i fiddle with just a few of the most obvious ones and cross check everything i do. A torsion prescription balance is a much simpler device and it should be possible to get very good sensitivity accuracy and repeat ability with no special tools and the suggestions made above. I have compared measurements on my torsion balance with the Mettler after i calibrated the torsion and agree well within the stated accuracy of the torsion balance so i must have gotten the calibration pretty good.