Speed Calibrating a Quantum Magnum Decoder
by Paul Turvill
Paul Turvill walks you through the paces using his Quantum Q2 Magnum to do the speed calibration for scale mph reporting for his LGB Mogul.
Introduction
Because the large-scale Quantum Magnum decoder from QSI Solutions has double the on-board storage capacity of its smaller scale cousins from QSI, its designers saw fit to add many new features and options to the available firmware. Among these new features are a resettable odometer and the ability to calibrate the speed readback function.
The odometer can be set to obtain its data from either a “chuff cam” in steam units, or by monitoring the amount of back-electromotive force (BEMF) generated by the turning motor over time. The Magnum-equipped locomotive’s speed is constantly monitored, and can be read back verbally by pressing F10 while the unit is in motion. By correctly setting appropriate CVs, the readback can be calibrated to yield results very close to actual measured scale speeds.
Configuration Variables
The CVs that regulate BEMF, Throttle Mode, Speed Control and other speed and distance related parameters are all indexed CVs in the CV56 Group:
CV56.0 permits the selection of the Cam Synchronization option for Chuffs and Odometer.
CV56.9 adjusts the ratio between measured BEMF and MPH reported.
CV56.24.0 and CV56.24.1 contain the loco’s driver diameter (dual precision).
CV56.25.0 and CV56.25.1 contain the loco’s scale (also dual precision).
For the odometer and speed calibration to have meaning, the Quantum Magnum needs to “know” both the scale of the locomotive (e.g., 1:29 for US prototype Aristo; 1:22.5 for LGB), and the actual diameter in inches of the loco’s drivers. This information is then used to calculate the distance sent to the odometer, and, along with the empirically determined value in CV56.9 to calculate the real-time scale speed for the verbal readout.
Calibration Setup
While it is possible to set all CV values with nothing more than a DCC system and a throttle capable of issuing programming commands, the process is greatly facilitated by the use of a Quantum Programmer and CV Manager software. To manually program the driver diameter into CV56.24.0 and CV56.24.1, for example, would require the user to first convert the measurement into thousandths of an inch (1.89” = 1890 thousandths), then break that number into values for the “low byte” (1890 modulo 256 = 98) and the “high byte” (1890 / 256 = 7), and program these values into CV56.24.0 and CV56.24.1, respectively. A similar calculation is required to program the scale into CV56.25.0 and CV56.25.1. More details on this process can be found in section 5.8.9 of the Quantum DCC Reference Manual, Version 4.1.0, available in PDF format from QSIndustries or QSI Solutions.
With the Quantum Programmer and CV Manager, all that is required is to click on the “Config” tab for the loco being programmed and either type in the diameter and scale or use the mouse or arrow keys to drag the on-screen indicator to the desired values and click the “Program” button. I’ll leave the former process to the mathematical masochists in the crowd; for me, the investment in a Quantum Programmer is well justified by the simplicity it provides.
That said, my calibration setup is a circular test track fed by a Digitrax DB150 booster driven by the output from my Quantum Programmer. This lash up is detailed in the document discussing locomotive speed matching elsewhere on this site, or the somewhat more formal approach at http://www.turvill.com/t2/train_stuff/mps.pdf. There are several advantages to this setup or one similar, since it allows the Quantum Programmer to substitute as a Command Station driven directly by the CV Manager software. Used in this fashion, the Quantum Programmer can operate one or more locomotives requiring up to the full output of the Booster, and is not limited by its own 300 ma (or optional 800 mA) power supply.
While a length of straight track could be used, I’ve found that a closed loop offers the convenience of continuous running, without having to constantly watch for the loco running off the end of the test section. For the exercise detailed below, I used my Magnum-equipped LGB Mogul running on a circle of LGB track.
After setting up the hardware, my first step was to calculate the length of the test loop in scale miles, using the nominal LGB scale of 1:22.5. One trip around the 1.2m circle works out to just about 0.053 mile, which is the figure I used for all of the speed and distance calculations that follow. I then selected a “reference point” on the loop for timing laps, put my trusty Casio atomic watch in Stopwatch Mode, and loaded up CV Manager and the previously created configuration file for my LGB Mogul.
Next, I went to the “Config” tab of CV Manager, and programmed in the driver diameter (2.561”) and the scale factor (22.50). (If the wheel diameter seems a bit large, it’s because I actually used twice the diameter of the wheels on the tender, since that’s where my chuff frequency is measured, using two axle-mounted magnets and a reed switch; two revolutions equals four chuffs, or the equivalent of one driver revolution.) Then I programmed CV56.0 for Cam Synchronized Chuffs and Cam Odometer. With these settings, the speed and odometer would be determined by the actual length of the test loop and the number of wheel revolutions required to complete each trip around it. At this point, running the loco around the loop for ten laps resulted in the odometer registering 0.53 miles, which was right on target.
Speed Readout Calibration
Since the Magnum’s speed calibration is dependent on more factors (actual motor speed, gear ratios, friction, etc.) than the odometer reading, some tweaking of CV56.9 is was required to obtain a reasonable correlation between stopwatch measured speeds and the Magnum’s verbal readout. I started with CV56.9 at its default value of 128, and began running timed laps at various speeds, using the 28 speed step mode of CV Manager’s on-screen “Test Cab.” I developed a simple factor to quickly convert the time-per-lap in seconds to scale miles per hour.
Reasoning that the actual speed would be
X miles/hour = (3600 sec/hour x 0.053 miles)/(T sec x N laps)
X is then 3600 x 0.053 or 190.8 divided by the time in seconds and lap count, or X=190.8/TN
So, a single lap time of 10 seconds gives a scale speed of 19.08 smph; a 5 second single lap would be 18.16 smph, etc. Substitute the length of your test track in scale miles, to make the equation fit your situation.
With this simple formula in mind, I set up an Excel spread sheet into which I fed the results of clicking F10 (verbal speed readout), my measured times and lap counts, and let the computer do the math. The results of my trials with CV56.9 set to various values are shown below. With CV56.9 set to its default value of 128, correlation between reported and measured speeds was fair, with errors ranging from about 21% low to 48% high. When I reset CV56.9 to a value of 160, based on the 21% error at higher speeds, the result improved for medium and higher speeds, but was much worse at very low speeds. The non-linearity of the results is apparently due to factors outside of the Magnum’s ability to resolve, such as the non-linear development of BEMF, low speed frictional effects, and other factors.
Magnum Decoder Speed Characteristics
BEMF-Derived (18.7 Volts DCC)
CV56.9=128
| SS | Time (s) | Laps | Calc. smph | F10 smph | Deviation |
|---|---|---|---|---|---|
| 1 | 47.05 | 1 | 4.1 | 6 | 1.48 |
| 2 | 32.91 | 1 | 5.8 | 8 | 1.38 |
| 3 | 24.98 | 1 | 7.6 | 9 | 1.18 |
| 4 | 20.19 | 1 | 9.4 | 10 | 1.06 |
| 5 | 16.71 | 1 | 11.4 | 11 | 0.96 |
| 6 | 14.37 | 1 | 13.3 | 12 | 0.90 |
| 7 | 12.50 | 1 | 15.3 | 13 | 0.85 |
| 8 | 11.00 | 1 | 17.3 | 15 | 0.86 |
| 9 | 19.60 | 2 | 19.5 | 17 | 0.87 |
| 10 | 18.08 | 2 | 21.1 | 18 | 0.85 |
| 11 | 16.53 | 2 | 23.1 | 20 | 0.87 |
| 12 | 15.36 | 2 | 24.8 | 21 | 0.85 |
| 13 | 14.25 | 2 | 26.8 | 22 | 0.82 |
| 14 | 13.63 | 2 | 28.0 | 23 | 0.82 |
| 15 | 12.73 | 2 | 30.0 | 25 | 0.83 |
| 16 | 12.16 | 2 | 31.4 | 25 | 0.80 |
| 17 | 11.67 | 2 | 32.7 | 26 | 0.80 |
| 18 | 22.28 | 4 | 34.2 | 28 | 0.82 |
| 19 | 21.72 | 4 | 35.1 | 28 | 0.80 |
| 20 | 21.03 | 4 | 36.3 | 29 | 0.80 |
| 21 | 20.36 | 4 | 37.5 | 30 | 0.80 |
| 22 | 19.77 | 4 | 38.6 | 31 | 0.80 |
| 23 | 19.41 | 4 | 39.3 | 31 | 0.79 |
| 24 | 18.87 | 4 | 40.4 | 32 | 0.79 |
| 25 | 18.53 | 4 | 41.2 | 33 | 0.80 |
| 26 | 18.05 | 4 | 42.3 | 34 | 0.80 |
| 27 | 17.73 | 4 | 43.0 | 34 | 0.79 |
| 28 | 17.46 | 4 | 43.7 | 35 | 0.80 |
Magnum Decoder Speed Characteristics
BEMF-Derived (18.7 Volts DCC)
CV56.9=160
| SS | Time (s) | Laps | Calc. smph | F10 smph | Deviation |
|---|---|---|---|---|---|
| 1 | 49.25 | 1 | 3.9 | 8 | 2.07 |
| 2 | 28.12 | 1 | 6.8 | 10 | 1.47 |
| 3 | 20.28 | 1 | 9.4 | 13 | 1.38 |
| 4 | 17.22 | 1 | 11.1 | 14 | 1.26 |
| 5 | 14.78 | 1 | 12.9 | 15 | 1.16 |
| 6 | 12.80 | 1 | 14.9 | 18 | 1.21 |
| 7 | 11.19 | 1 | 17.0 | 19 | 1.11 |
| 8 | 9.86 | 1 | 19.3 | 20 | 1.03 |
| 9 | 17.65 | 2 | 21.6 | 23 | 1.06 |
| 10 | 16.68 | 2 | 22.9 | 24 | 1.05 |
| 11 | 15.56 | 2 | 24.5 | 25 | 1.02 |
| 12 | 14.70 | 2 | 26.0 | 26 | 1.00 |
| 13 | 13.61 | 2 | 28.0 | 29 | 1.03 |
| 14 | 13.19 | 2 | 28.9 | 30 | 1.04 |
| 15 | 12.42 | 2 | 30.7 | 31 | 1.01 |
| 16 | 12.00 | 2 | 31.8 | 31 | 0.97 |
| 17 | 11.65 | 2 | 32.8 | 33 | 1.01 |
| 18 | 22.08 | 4 | 34.6 | 35 | 1.01 |
| 19 | 21.97 | 4 | 34.7 | 35 | 1.01 |
| 20 | 20.96 | 4 | 36.4 | 36 | 0.99 |
| 21 | 20.89 | 4 | 36.5 | 36 | 0.99 |
| 22 | 20.01 | 4 | 38.1 | 37 | 0.97 |
| 23 | 19.60 | 4 | 38.9 | 39 | 1.00 |
| 24 | 18.78 | 4 | 40.6 | 40 | 0.98 |
| 25 | 18.57 | 4 | 41.1 | 41 | 1.00 |
| 26 | 18.17 | 4 | 42.0 | 42 | 1.00 |
| 27 | 17.80 | 4 | 42.9 | 44 | 1.03 |
| 28 | 17.53 | 4 | 43.5 | 44 | 1.01 |
After runs with several other values in CV56.9, I finally settled on a value of 146, which yielded results accurate to within 10% or better in 24 of the 28 speed steps (see below):
| SS | Time (s) | Laps | Calc. smph | F10 smph | Deviation |
|---|---|---|---|---|---|
| 1 | 92.29 | 1 | 2.1 | 6 | 2.90 |
| 2 | 33.57 | 1 | 5.7 | 8 | 1.41 |
| 3 | 22.60 | 1 | 8.4 | 10 | 1.18 |
| 4 | 19.05 | 1 | 10.0 | 11 | 1.10 |
| 5 | 14.97 | 1 | 12.7 | 13 | 1.02 |
| 6 | 12.92 | 1 | 14.8 | 14 | 0.95 |
| 7 | 11.53 | 1 | 16.5 | 16 | 0.97 |
| 8 | 9.74 | 1 | 19.6 | 18 | 0.92 |
| 9 | 18.56 | 2 | 20.6 | 19 | 0.92 |
| 10 | 17.35 | 2 | 22.0 | 21 | 0.95 |
| 11 | 16.00 | 2 | 23.8 | 22 | 0.92 |
| 12 | 14.44 | 2 | 26.4 | 25 | 0.95 |
| 13 | 14.07 | 2 | 27.1 | 25 | 0.92 |
| 14 | 13.49 | 2 | 28.3 | 26 | 0.92 |
| 15 | 12.85 | 2 | 29.7 | 27 | 0.91 |
| 16 | 11.87 | 2 | 32.1 | 30 | 0.93 |
| 17 | 11.79 | 2 | 32.4 | 30 | 0.93 |
| 18 | 22.52 | 4 | 33.9 | 31 | 0.91 |
| 19 | 21.97 | 4 | 34.7 | 32 | 0.92 |
| 20 | 20.82 | 4 | 36.7 | 33 | 0.90 |
| 21 | 20.37 | 4 | 37.5 | 34 | 0.91 |
| 22 | 19.78 | 4 | 38.6 | 35 | 0.91 |
| 23 | 19.46 | 4 | 39.2 | 35 | 0.89 |
| 24 | 19.06 | 4 | 40.0 | 37 | 0.92 |
| 25 | 18.49 | 4 | 41.3 | 37 | 0.90 |
| 26 | 18.11 | 4 | 42.1 | 38 | 0.90 |
| 27 | 17.75 | 4 | 43.0 | 39 | 0.91 |
| 28 | 17.60 | 4 | 43.4 | 39 | 0.90 |
Conclusion
The Quantum line of large-scale decoders offers a number of enhancements that should make them attractive to serious modelers. Their robust design and their capability of handling and adding quality sound to even the largest of O- and G-scale locomotives is further enhanced by a range of features not previously available in any DCC decoder. Custom calibration of speed and distance characteristics by the end user is just one of numerous new features available.
