The operation of the four variables in the DOE sets show the impact on several combinations of the performance values. The DOE design is discussed here: https://www.llxlocomotives.com/wp-admin/post.php?post=2433
The activity was initiated with grade & train length studies as the main focus. As is discussed in later sections, a number of additional performance differences were also identified. The process to run the engines/trains as to use 12 volts as the main examination voltage. The supply is a DC lab supply that delivers a pure DC signal.
There is no PWM what so ever. The power is turned on at the set voltage and the throttle is not changed during the run. All speed & current measurements are recorded at the end of the run. Any variations that might occur during the run are ignored. All data points are repeated three times. The recorded values are averaged for the point.
Maximum number of cars on a 2.5% grade
The process to determine the maximum number of cars is a iterative one. The Engine with the weight adder is first run by itself, engine only (EO). Then the second run is with the car equivalent that equals the number of engine drive axles time 2. This is 12 cars for six axle engines and 8 cars for four axle engines. The subsequent runs are done by adding cars until the engine will no longer pull the train up the grade. If the train stops on any of the three runs with a given train length, then the last successful three run set is determined to be the maximum number of cars for that configuration. Usually this last successful set is running at a speed near 1.0 smph. This is not necessarily useful, but does describe the maximum train length at the grade.
The following chart is the average result of six runs for each of the motors in the test series:
The average maximum number of cars for all tests was 22.3 cars. The exceeds the nominal train length the was described above. In fact all fifteen motors averages exceeded the nominal level.
Recasting this chart, showing the percent difference from the average for each motor clarify’s the result. Her it shows a tenth five percent variation in the motor averages. The Mashima 1833 #1 motor is the best puller. exceeding the overall average by 15+ percent. The poorest puller is the Rapido motor. fall 10 percent below the average.
On this assessment, the best motors are the Mashima 1833 #1, The Athearn high performance #2 & the Cannon C-22. The poorest are the Rapido, the Atlas Genset & the Athearn Genesis. All of these later are recent releases that are likely designed for low noise & current draw.
Along this same line, the percent impact of the Weight & wheel material variables is shown in the following figure:
Adding weight is clearly a benefit, as has been shown by others, adding 154 grams(5.4 onces) increases the train length by 25%. As second increment of nearly the same amount adds 25% more. This clearly explains the push to add as much weight as possible. The impact of the worst to best motor is equal to addition of one of these increments.
While not nearly as dramatic, the wheel size and material results do show an interesting result. The nickel silver wheels actually further help in train pulling capacity. Additionally, the 40 inch wheel result is consistent with the reduced diameter percentage. So in this set the stainless steel material is doing nothing improve the pulling capacity of the train.
Focusing on the engine variable, the following figure was developed:
Some of what is seen in this chart is due to the base engine weight variation. The NP U33C is 55.3 grams heavier than the average weight. based on the weight increments shown in the last figure, it would be expected to be 9% better then the average. Here it is close to 14% better. Something in that engine is achieving the extra five percent. The Seaboard U30B is the lightest, and would be expected to be 9% poor than the average in pulling capacity. Here it is roughly 13%. This may be due to the four axle configuration. The engine that is really off the mark is the SP SD9. Based on weight alone, it should be 3%. Instead it is down 9%. This indicates that there is more than weight difference driving these results.
Looking further into the details of the data, the next chart takes the average for each motor and adds the maximum and minimum measured train lengths for each motor. Here we start to see a dichotomy for several motors. The best average motor, the mashima 1833 #1 also had the poorest result in one test. The Canon C-22 was able to pull 39 cars in one of its configurations, but on average fell to third in the ranking. The three poorest pullers were also the bottom three of the maximum cases.
The test bed engines are show in the following chart.
Generally as you should expect, the two four axle engines, BN U28B & Seaboard U30B, nearly the poorest pullers. However, the SD SP9 as also low. As discussed earlier the why is not clear as of yet.
The BN U28B demonstrated the best and the worst results from an engine perspective. The best average engine was the NP U33C which also happens to be the heaviest base engine at 321 grams( 11.3 ounces). The lightest is the Seaboard U30B at 208 grams ( 7.3 ounces) Again, this spread may be worth 3-4 cars depending on the weight derivative. That does not account for the difference between the two engines. For now, the data is the data. The problem is that all elements weigh differently. Plus the impact of the weight is different on each parameter. Best to not read more into the results than are actually there.
The other two DOE variables impacts are shown the the following chart:
These variables are a fixed weight adder, 0, 154 & 308 grams; and wheel material & size, 42 in, stock Athearn blue box, 40 in NWSL stainless steel & 42 in NWSL nickel silver. Both charts show a large variation from min to max on all variable settings.
In the case of the weight adders, the average gain is also similar to that in the min data points. On the max side the middle weight actually loses capacity. The Max regains some benefit, but the total gain is less than half of the Min. This implies that there is a diminishing return for adding weight. While not shown in the average data as such, it would likely appear with another increment of weight. It would be and interesting thing to test.
In the case of the wheel material and size, The result varies from the low side to the high side. On the low side the Athearn BB wheels are the poorest, with the 42 in NWSL NS being the best. The same is true on the high side. However, these differences are washed out on the average. Other tests have shown the stainless steel to be much better, but not in this data.
12 volt engine only velocity on level & 2.5% grade
The engine only condition at any voltage represents the fastest the potential train could run, assuming that the motor did not lose speed with additional train load.
For this case the speed are shown in the following figures. Note all speed values are in SMPH.
The level grade is the dark blue bar and the 2.5% grade is the red bar. For these averages, there is a speed decay of the engine only from level to 2.5%. In other tests outside of this series, the engine only speed has been shown to increase slightly.
At both grades the fastest motors engine average are the Canon C-22, the Athearn Genesis and the Atlas China #2. The slowest speeds are with the Sagami 1836 #1, the Atlas Genset and the Walthers ES44 motors. These engine only speeds are consistent with the motor only tach RPM levels identified in the raw data: https://www.llxlocomotives.com/wp-admin/post.php?post=2435
When looking from engine to engine in the following chart, there is very little speed difference on a level grade. This implies that the average gear ration for the engine trucks is nearly the same. The speed then becomes the average of the motor defined speeds in the previous figure. The decay with grade increase is slightly more pronounced for some engines, but still very small, less than 10 percent.
looking at the impact of weight & wheel configuration, again the differences are small. In the case of the weight, it appears that the speed levels are starting to drop with the 308 weight delta. This is more pronounced at the level grade. Additional weight would likely show a more pronounced speed loss for the engine only.
In the case of the wheel configuration, this speed is fastest for the 42 in Athearn BB wheels. The 40 in wheels run slower. These results are similar for other grade levels. As in the other variables, the speed decay from level to 2.5% grade is small.
12 volt nominal train velocity on a level grade
As discussed earlier, the performance as measured for the nominal train number of cars on a level & a 2.5% grade. The next chart shows the nominal train speeds on a level grade. These are down from the engine only levels by varying amounts.
As was the case earlier, these averages are dominated by the 12 volt motor base rpm and the manner which the motor speed varies with load. Here the desire is for the engine to produce the speed closest to the real engine type without compromising other performance aspects. The three fastest motors are the Cannon C-22, the Athearn Genesis & the Atlas China #2. The three slowest motors are the Sagami 1836 #1, the Atlas Genset & the Walthers ES44. This is the same ranking as the engine only result above.
The following chart shows the average engine train speed on a level grade. While these results are dominated by the engine truck gear ratios, they are hovering around the average speed of the overall set, 79.3 smph.
. The four axle engines look to run slightly faster than the six axle engines.
As shown in the following figure, the impact of weight & wheel material on the speed is modest. The heavier weights both are about the same indicating that the best weight is likely between the two levels. Again adding additional weight will likely lead to a engine train speed reduction.
In the case of the wheel material, the 42 in Athearn BB still shows a slight advantage. The 40 in NWSL SS configuration is slightly slower than the two 42 in versions.
12 volt nominal train velocity on a 2.5% grade
Equally as important is the ability of the engine to maintain speed when load is applied. In this case the load is in the nominal train on a 2.5% grade. Here the average has dropped to 52.6 smph. This is a very noticeable loss in speed.
The fastest motors here are The Atlas China #2, the Canon C-22 and the Chinese motor. The Athearn Genesis has fallen back to seventh fastest. The slowest motors are The Sagami 1836 #1, the Atlas Genset and the Rapido. The Walthers ES44 is now more in the middle of the pack. The motor rankings have changed for a speed consideration.
The next chart looks at this velocity from a percent loss perspective. That is how much of the speed was it able to retain on the 2.5% grade. This is a better measure of the load impact on the motor speed.
Here the motors that lost the least speed were the Chinese motor, the Atlas China #2 & the Mashima 1833 #1, The Athearn high perf #2 shows to have just slightly more speed loss. The motors that lost the most speed were the Sagami 1836 #1, the Rapido and the Atlas Genset. Interesting how these orders changed when looking at the percent change rather than the value.
As was shown in an earlier discussion, when the maximum and minimum values are included with the averages, the motor impact takes on some new insights. From the maximum measured speed perspective, the best motors are the Canon C-22, The Mashima 1833 #1 and the Athearn Genesis. On the other hand the slowest motors, from the minimum measured perspective, are the Sagami 1836 #1, the Atlas Genset & the Rapido. The Sagami & Atlas Genset minimums were near 1 smph. Which is likely the maximum number of cars that could be pulled.
The Atlas china #2 & the Chinese motors both show the least variation from min to max levels. While not the fastest, these appear to be the most robust motors from a speed consideration.
The engine impact on the 2.5% grade speed is shown in the following figure. As was the case on a level grade, the four axle engines are the fastest at max, average and min. All of these engines show a large variation from max to min. Note, both of the min motor values from motor figure occurred on the SP SD9 engine.
As shown in the next chart, most of these min to max variations come from the weight increments & wheel configurations.
The train speed weight impact is particularly large on the min measured tests. The average and the max measured tests are about equal in speed increase, still showing considerable increase with added weight. Unlike the engine only, the nominal train speed looks like it would continue to increase as more weight is added.
In the case of the wheel material, the trend switches between the min tests and the max tests. For the min, the 42in Athearn BB wheels are the fastest. With the 42 in NWSL NS the slowest. The max points favoring the opposite trend. In the averages, the 42 in Athearn BB wheels are a little faster than the other two. This is a peculiar result and will be looked at further.
Maximum input power impact
The testing recorded current draw at a given supply voltage. This is actually the input electrical power, VI. Thus for a constant voltage the ratio of input current draw is the input power ratio. While this overstates the actual power to the tracks by the efficiency, it is still an interesting parameter to compare. For these tests the maximum input power occurs when the drive wheels a locked. In that case there is no power transferred to the tracks. It is all being canceled by the drive. This is the same as the motor stall condition. The easiest way to see this is to hold the engine with a downward force as ufficient to stop the drive wheels. The current measured here is the stall current used in sizing DCC decoders. It is also the current associated with the maximum input power. This power is independent of the engine weight.
The following figure shows this average stall current level for each motor:
The lowest stall power motors are the Mashima 1824 #1, the Rapido & the Sagami 1836#1. The Athearn Genesis is a close fourth. The motors with the highest stall current/power are the NWSL 163-4 Sagami, the Canon C-22 & the P1K #2 motors. These latter are all just under 1.6 Amps which is a bit high for a DCC decoder.
The engine variation of the stall current is shown in the next figure.
The variation in this aspect is surprising because it is hard to see the engine features influencing this parameter.
The variation due to weight and wheel material are shown in the following figure:
the weight trend is also a surprise, because the parameter should be independent of weight.
The wheel material is also showing only a slight average variation from wheel configuration to wheel configuration. The wheel resistance differences was expected to play a role in the parameter. Apparently not.
Maximum usable power
Another interesting power level, is the maximum usable input power. This is the power that is associated with the case where the engine wheels are are slipping, but the train weight does not allow movement. This maybe slightly lower than the actual maximum usable power, but the testing indicated that it is very close to the maximum measured for each combination.
This point can be simulated by attaching a car to the engine and holding the car fixed while the supply is providing the desired voltage. The current measured for this condition is associated with the maximum useable power. This power is potentially a function of several conditions. It is impacted by the engine weight.
The ratio of the maximum usable current to the stall current indicates how much of the maximum potential power is available to the system. It also indicates how far from the stall current the system could operate.
the following figure highlights this ratio for the motors minimum, average and maximum conditions.
As has been shown in other parameters the variation from min measured, to average, to max measured is quite large on some motors. Others have a very small variation. Because this directly contributes to the pulling capacity this value should be as large as allowable. Some margin between the stall parameter and the level should be maintained for decoder safety.
All of these motors on average are below 45 percent. An extremely generous margin. Two of the motors have a max measured result that stands well above the others. These are The Mashima 1824 #1 and the Athearn Genesis. These two are also in the top three on average, with the Athearn high performance #2 placing second.
Two of motors have very little variation from min to max. The NWSL 163-4 Sagami & the P1K #2 both have very little change. They are the motors that us the lowest percentage of the maximum power on average. The Mashima 1833 #1 actually has the lowest measured usable power and is third lowest on average.
The engine to engine variations in the parameter are again interesting. The lowest average being the two four axle engines. With the BN U28B the lowest. It also has the min measured result. These are also the lightest engines, but the trends do not replicate the weight variation precisely.
The max measured are the NP u33c & the Penn PA-1. These are also the largest average, but in the reverse order. The NP U33C has the largest variation from min measured to max measured.
The weight trend generally shows benefit with added weight. However the impact is around five percent for both the min measured and the average. The max measured shows a decline on the first increment and then a very dramatic increase on the second.
The wheel configuration trends are just as interesting. On the low measured side, both NWSL sets are nearly equal. The average still favors the 42 in NWLS NS configuration slightly with the 40 in configuration falling behind the other two. On the max measured side, the 42 in BB has made a large step to out distance the other two configurations. Another result that needs further investigation.
Power impact
Following along focusing on the power capacity for these tests, several current values have been examined. These are:
- Minimum sustained velocity, varing low voltage
- Engine only on a level grade, for these charts at 12 volts
- Nominal train length on a level grade, at 12 volts
- Engine only on a 2.5% grade, at 12 volts
- Nominal Train length on a 2.5% grade, at 12 volts
These results normalized by the maximum power are shown in the following figures. The previously discussed maximum usable power at 12 volts are included in the comparisons.
The figure below shows these power impact for the motor variable. The third and fifth bar in each set are the average nominal train results at level and 2.5% grades respectively. While following the trends of the max usable power, the power usage does very for each motor. The motors that are using the most their power capacity with a nominal train on a level grade are the Mashima 1824 #1, the Kato #2 & the Sagami 1836 #1.
In every case the power increased from the level to the 2.5% grade, the first two motors are still the same, but several are contending for the third spot. These are the Athearn high performance #2, the Atlas Genset, the Rapido & the Athearn Genesis
An item of concern would be the proximately of the nominal train at 2.5% to the maximum usable and their relationship to the maximum power. In both cases this would be where the motor generates the most heat. Long durations of operation at this heat generation point will impact the motor reliability. The Kato #2 motor runs at grade very close to the max usable power. This may be acceptable because it is below 30 percent of the maximum power for the motor.
The most under utilized motors in the group are the P1K #2, The NWSL 163-4 Sagami & the Mashima 1833 #1
These same power ratios for the engines ar shown in the following figure.
These engine trends seem to follow the max usable trend. The Penn PA-1 uses the most power and the BN U28B uses the least.
The results for the weight and wheel configuration are not necessarily tied to the max usable. The nominal train power ratio with weight seems to be constant with weight on a level grade and only slightly increasing at the 2.5% grade. The max usable trend is increasing with weight.
The wheel material results seem to be constant across the variable at grade.
Current impact
For those who are interested in current levels, the motor figure is repeated using the average current draw in Amps for each motor. The relation ship between the motors changes because this is no longer lookin at motor capacity. Actual AMP utilization is important when considering the Amp budget for your system.
The minimum sustained velocity bar has increased towards the others, because previously the power was low because of the low voltage where is occurs.
Looking at the current at the 2.5% grade, the motors with the lowest current draw are the Atlas Genset, the P1K #2 & the Rapido. Several other are close.
The motors with the highest current draw are the Atlas China #2, The Chinese motor #2 and the Athearn high perf #2. The Kato #2 is a close fourth.
Again the ranking changes based on the perspective.
Minimum sustained velocity
The focus in recent years has been on low power performance. In particular low speed. This has many different meanings. here the minimum engine only sustained velocity value on a level grade is identified in the following figuress.
The average motor value are shown in the nest figure.
These results all hover around 4 SMPH. The motors that run the slowest are the NWSL 163-4 Sagami at 3.5 SMPH. Followed by the the P1K #2 and the Sagami 1836 #1. followed by three other motors that run slower the 4 SMPH.
The motors that initiate running at the highest speed are the Bachmann Plus #2 at 5.6 SMPH, the Rapido at 4.8 SMPN & the Athearn High Perf #2 at 4.8 SMPH. There are four others around 4.7 SMPH.
The engine variable shows a similar variation with the Penn Pa-1 sustains speed above 5 SMPH. The two 4 axle engines achieve movement under 4 SMPH.
For the weight variation, the results are mixed with ak=ll falling just above 4 SMPH. the trend with weight is not clear.
In the case of the wheel configuration, the result slightly favors the small dimeter Stainless steel configuration.
The fastest result is the 42 in NWSL nickel silver set. That again is a surprise and will be looked at further.
Overall performance characteristics
Another way to examine these results is the use a global technique to pull togather several aspects in one parameter. This is always difficult and often will not cover all bases. On such approach I have developed in the following: https://www.llxlocomotives.com/wp-admin/post.php?post=365
This tends to over emphasize the low speed engine only aspects, but otherwise it tends to be relevant. Examining this date in the following figures draws some interesting perspectives.
In the following figure, the engine results for min measured, average and max measured for the different motors are shown. The large variation may be due to the low speed frailty. However, from this perspective the best(largest average) motors are the P1K#2, the NWSL 163-4 Sagami & the Mashima 1833 #1. The max measure results reverse the first two and replace the third with Canon C-22. The Mashima 1833 # 1 falls to eighth on the max measured side.
The lowest average motors are The salami 1836 #1. The Rapido & the Bachmann Plus #2. Followed closely by the Athearn high perf #2. Several of the motors do not show well on the lowest measured side. The lowest being the Mashima 1836 #1 with a slightly negative result. Next lowest is the Rapido flowed by the Atlas Genset. Five other motors scored at 10 or less which would be reason for concern.
The following figure shows the impact on the engine variables. For the max measured data all of the results exceed 80 which is very good. The best three are the NP U33C, the SP SD9 & the BN U28B. The SP SD9 result is interesting because on average and the min measured values are the lowest of the engines. The average results favor the NPU233C and the two 4 axle engines.
The lowest average results belong to the SO SD9 & the Penn PA-1. The lowest Min measured belong to the SP SD9 (with the slightly negative value) & the Und PA-1. Four of these engines fall below the 10 level on the min measured. Again a concerning result.
The impact of weight and wheel material in the following figure shows some dramatic improvement for weight addition. Particularly for the max measured and average data. The min measured dat also shows a improving trend, but not as strong. Only the heaviest min case exceed the 10 level.
For the wheel configuration the results are more mixed. On the max measured side, the 42 in Athearn BB wheels score the highest with the 42 in NWSL NS the lowest. On average, that trend reverses so the NS wheels are slightly better.
For the Min measured side, all are below the 10 level with the NS falling to the slightly negative value.
The Summary & conclusions will be presented here: https://www.llxlocomotives.com/wp-admin/post.php?post=2564
As of this writing, these are not complete.
