Because sensors in the 300 series of DROs from BW Electronics use a flexible stainless steel wire to measure with, this can be wrapped around a rotary table to measure angular distance. And because the sensors can be re-calibrated, this angular distance can be set to read degrees and hundredths of a degree. Note that the exact diameter of the rotary table is irrelevant; the only restriction is that the measuring range of the sensor should be greater than the circumference of the rotary table. The maximum diameter of rotary table for a 361 sensor is 6", while a 12" maximum is possible with the 362 sensor.
The measuring wire needs to run in a groove around the outside of the rotary table. The ideal groove shape is a Vee with a flat bottom. Because the wire will need to overlap for a short distance, the flat needs to be at least twice the diameter of the wire, which is 0.016" or 0.4mm. Note that this flat must be exactly vertical, parallel to the axis of rotation of the rotary table. The easiest method of making this groove is to use an HSS Vee tool and gently stone a flat at the tip.
It is essential that the groove is exactly concentric to minimise errors. One practical way of achieving this is to machine the groove in place on the rotary table. First clamp the Vee tool to the table (use toolmakers' clamps). Then gently tap the free end of the tool so that it digs into the table. Now rotate the table and the tool will cut the groove, which must end up exactly concentric. Keep feeding in the Vee tool and cutting the groove until the required depth is achieved this needs only to be 0.020" or 0.5mm.
Now a fixing for the wire ring is needed. The easiest method is a tapped hole which will clamp the ring, allowing it to be positioned so that the wire enters the groove without masking it. (The groove must not be masked otherwise after a full turn of the table the wire will catch on the ring before going into the groove, thereby causing errors.)
The body of the sensor can be fitted to the rotary table. The wire needs to come straight out of the sensor to the groove, so some type of bracket on the side of the rotary table is needed. The sensor can be clipped, or fixed more permanently, where any built-in display (as on the 391 sensor/display) can be read.
With the sensor fixed and groove machined, now you can calibrate the sensor to read in degrees.
The readout is now calibrated to read in degrees and hundredths of a degree.
If an angle of less than 360° was used in the re-calibration, it is wise to check it. Therefore unwind or move the wire out of the groove and carefully clean the groove. Now replace the wire in the groove. Rotate the rotary table more than 360°, so there is a reasonable amount of overlap (say 12", 2550mm) at the 360° point. Very carefully, identify a zero datum point on the table, using a magnifying glass if necessary, and zero the readout. Now rotate the table to unwind the wire until the previously identified zero point is reached. The reading on the readout should be 360.00. On the offchance that it isn't, the discrepancy needs explaining. Possibilities are:
With the rotary table and readout now calibrated, it can be used. In all cases, follow the procedure given below to eliminate any problems of swarf getting trapped between the wire and groove.
Because the readout sees angles on the rotary table simply as numbers, it makes no difference if the 360° is calibrated as 121, 53, 487 or any other number. This allows the rotary table and sensor to be used as a universal dividing head, being able to cut any number of teeth to a practical limitation of several hundred teeth.
The procedure is exactly as described previously, except:
When calibrating it might be easier to follow this procedure:
With the wire wound on you can start cutting the teeth.
The sensor accuracy is ±0.05mm or ±0.08mm depending on sensor type. If the wire is wrapped around a 4" or 100mm rotary table, then because the circumference is very nearly 360mm the accuracy will be ±0.05° or 0.08°. In this case the sensor resolution, approximately 0.006mm, is less than the display resolution. This means that every value can be obtained.
If a 12" rotary table is used, then the circumference is 940mm. This means that every 10mm of wire movement has been rescaled to display as (360 ÷ 940)mm = 3.8mm, or 3.8°. Because the value displayed is less than the distance moved, the sensor resolution of 0.006mm is the same, but it is now 0.006 × (360÷940) = 0.0023°. This also means that the accuracy, ±0.08mm, has now become ±0.08 × (360÷940) = ±0.03°.
What if inches rather than millimetres were used? In this case the circumference will be 360", and every 1" of wire has been rescaled to display as (360÷38), about 10" or 10°. The value displayed is now greater than the distance moved, by the factor (360÷38) or about 10. This means that the sensor resolution has also been multiplied by this factor when it is displayed, so the display will change in steps of 0.000 25"×10, or 2.5 thou. Similarly the accuracy will also have changed by 10, to ±0.030". This may seem to be far worse than previous, when the rotary table was re-calibrated in millimetres, but it isn't in reality. The important point is that the display is in degrees, not inches or millimetres, Therefore the previously calculated resolution of 0.0023° and accuracy of 0.03° are exactly the same (given that the maths has been rounded to one decimal place) as the 'inch' values of 0.0025" resolution and 0.030" accuracy. All that has changed is that by using inches an extra digit of resolution has been obtained. Is this important? No, not really. Since dividing is a process of moving between sequential positions, repeatability does not come into it. Because the error is larger than the least significant digit (±0.03 error to display 0.01) then nothing can be gained.
All this may seem of little importance, and even less interest. After all, one hundredth of a degree is is equivalent to one inch around a circle of diameter nearly 1000 feet. However, where it is of importance is when the rotary table is used as a dividing head. Because the number of divisions, teeth, per 360° is changing, then there are benefits to using either inches or millimetres as the basis for re-calibration. As the number of a teeth reduces there comes a point where the extra digit of resolution on an inches display becomes useful.
For example, suppose you need a 45-tooth gear and you are using a 12" table:
With metric, the scaling factor is 21, so the sensor resolution is (0.006÷21) = 0.0003, and accuracy (0.08÷21) = 0.04. The accuracy is less than the display resolution so it is better to use inches to gain better accuracy because of the additional digit in the display.
As a rule of thumb, if the number of teeth divided by the diameter of the rotary table in inches is greater than 10, use millimetres; if less than 10, use inches.
There is another restriction on the minimum number of teeth that can be cut because of increasing rounding errors in the readout as the number of teeth decreases. This puts an effective minimum limit of the number of teeth of twice the diameter of the rotary table in inches; e.g. 24 teeth for a 12" rotary table. There are two possible workarounds:
This is a different approach to dividing. A dividing head, because it is circular, will always come back to zero errors after one revolution. The rotary table and sensor method uses linear measurement and is therefore not guaranteed to come back to zero after one revolution. So if you are cutting an important gear it is wise to do the following:
BW Electronics, 12 Mussons Close, Corby Glen, Grantham,
Lincolnshire, NG33 4NY, Great Britain