So the theory is that now, as the first "hit" was on them, you're going to be addicted to JMP and have to spend money to get your own license! Sigh.... No sweat. I've just finished editing the DVD that specifically covers the contents of the GPX file so it was fresh in my mind. I'd lost situation awareness that he was flying a six, so as you say, #3 is right rear. Difficult to know. It would be interesting to correlate elevation (GPS) and altitude (baro) for several flights to see how the baro performs. I think you mean my "baseless speculation™" -- note that trademark. My colleagues say, "aha, so you're engaging in your trademark baseless speculation...." Good point. I really must dig into the mixer code in the FC board to understand more how it works. I know! I'll do that in my "copious free time." (A notional concept.) Nice work. I have a feeling that knowing how to do that calculation is going to be very useful when we see an outlier motor. Andy.
As I said, this is a very useful diagnostic approach. Of course, it does raise the question now of: "What's with Motor/Prop #4?" Do you think it would be possible to build an Excel spreadsheet to do this correlation? (I'm not asking "can you do it?" -- I'm just asking whether Excel has the necessary formulae to handle it?) Andy.
Yeah it is a real powerful tool, I just recently had the licence renewed. I probably will try to figure out how to renew it when it's up I wish I had it when I did my build,. As to your question below it is possible to do regression in Excel it is a bit more cumbersome and does not provide the modeling that JMP does of course. It is possible to do all of the modeling that I did in excel though. You have to go to File tab, Options, Add-Ins and add in the analysis tool. Once you do that, you will find the regression tool in the "Data tab" and on the far right is Analysis. There is a sub Menu where you will find the regression tool. The one thing that Excel does that is cumbersome and annoying, is that all of your data that you want to regress has to be adjacent to each other. You will need to copy and paste the data so it is next to each other. I typically add a new sheet. Plug the regressor Y in the A column and all of the data being regressed in the X column to the next columns. You will see in the regressor tool, blocks to input Y range and X range. You simply just do the standard left click on the data and drag down to the range you want for X or Y. There is a hot button for keeping the labels. Then once you have you data ranges set you just find an empty spot on your excel sheet then go to Ouput range and click on an empty cell on the sheet. Once you hit OK, this will generate the ANOVA table and give you the parameter estimates. This will do the job for simple regression models. If you need to account for non-linear effects you will have to manually transform the cells performing LOG or Sqrt to the data in new columns and then regress that data. Also if you want to exclude rows you have to cut and pate to keep the columns uniform. No sweat. I've just finished editing the DVD that specifically covers the contents of the GPX file so it was fresh in my mind. I will do such a model, I think it could be tricky because of pressure differentials. Last night when I flew we were getting a frontal shift and my baro was all over the map. But it will be a fun model none the less. Most of the time the baro seems stable.
Ooof. That is a pretty ugly approach with Excel. Thanks for taking the time to write it up in such detail. I suspect if you can renew JMP, then Stanton's GPX Analysis Service could be a viable business model! Andy.
Since that motor spins clockwise it is contributing to the natural counter clockwise moment. It is possible that the motor is tilted clockwise on the boom center axis, this will cause an additional counter-clockwise moment beyond the natural counter torque force. As that motor accelerates that force is exacerbated. In my model where I regressed motor 3 against the rest, my precision was low for the first 30 data points, which is when the copter was somewhat stable. After the first 30 data points or so, is where he started to move forward. This is where the Predicted resolution of 4's effect on 3 is better. This is where there was more pitching being applied. As 4 is accelerating to pitch down it is forcing 3 to work harder to counter that. The interesting thing to me is why it would not effect 5 as much, but that could be to 3 having to compensate for CG, forcing it to bare the brunt of the load. One of the things that I had not considered in any of the models was environmentals. Although I would guess them to be negligible since the copter was in free fly. It would be interesting to know logic algorithm to the motor mixing on the Hexa. It seems that the hexa, having less freedom of movement it is going to be less tolerable to imbalances than the Octo. Shaun
I had a couple of flights today after repairs and here are the results (GPX file attached). This flight is without a camera gimbal. In the repair job I replaced motor 2 (because it was slated for replacement due to low rpm vibration) and braided wires in booms 1, 2, and 6 as well as the battery pack leads. I also did a compass calibration. As you can see there are no errors, however, it appears that motor 2 is pulling less current than the others. It was this way before replacement. Motor 5 has no vibration issues. WWJD? Just replace motor 5? Anything else? Colin
I ran one model. With all the data points all motors tested well for statistical significance. Parameter Estimates Term Estimate Std Error t RatioProb>|t| Intercept22.989644 1.850168 12.43<.0001 MC1-0.057255 0.0243 -2.35 0.0191 MC30.113666 0.023101 4.92<.0001 MC40.1201392 0.025407 4.73<.0001 MC5 -0.341294 0.018924 -18.04<.0001 MC60.0643344 0.025162 2.560.0107 To give you a quick rundown of what this means is that Prob>|t| means that the probability of the parameter estimates MC1-MC6 being error are less then 5 percent. So all of these values mean are the following. The intercept value of 22.99 A is the baseline current statistically for motor 2. The rest are multipliers of the prediction function called parameter estimates. What this model does is predict the motor current for motor 2 based off of different current values of the other motors. MC5 is the largest absolute value for current which means this has more leverage in changing the base line value. If motor 5 is predicted to use 20 amps of current then the baseline will change by 23 - (0.341*20) = 16A Of course the other motors 3, 4 and 6 will increase the baseline current and motor 1 will decrease it along with 5. 5 is the opposite motor so it may make sense on why it has the most leverage. Why is 5 possibly effecting 2 the most, I am afraid I cant answer that. Numerous factors could be causing it. Most likely the motor is tilted possibly. But whatever is going on, I dont think that your motor 2 is bad. Typically when a motor goes bad it does the opposite it pulls more current then what is needed to compensate for internal frictions. If you are not sure switch a motor and see if that makes a difference. That may give you better insight.
Thanks Shaun. I appreciate the MLR analysis to understand the correlations for the low current draw of motor 2; very appreciated! In a former life I was a product manager for demand planning and supply chain software, so I get MLR modeling. My go-to for at-home analysis was the ForecastX add-in to Excel. I just don't have acess to any of that software now. But I digress... In the rebuild I checked for tilt on all motors and replaced all props, so I'm thinking my next step is to replace motor 5 and see what happens.
No worries That might make sense. If 5 is pulling more current per RPM then it should it could be that it is not functioning as efficiently, therefore motor 2 is thinking it does not need to work as hard. That could be a possibility. An interesting test might be to switch the two motors and see if 5 becomes the lower one. Then you probably have a motor worth replacing or at least the barrings in it. If the results are the same than you can narrow down that it is not a motor issue. I might give that a try first and see what the results are.
Well. I replaced motor 5 with a brand new one and here's the result. The data below compares yesterday (before) vs. today (after). GPX file for today's flight attached as well. Tome it looks like the motor 5 current draw got worse PLUS I got a BL Limitation" (24) error at the end right before shut off. What is that? I tried looking that up on the MK Wiki, but the posts were in German and I could not get the English translation to work. Any ideas? Before: MK Version: FC HW:2.1 SW:0.88e + NC HW:2.0 SW:0.28h Flight time: 10:43:18 AM - 10:57:07 AM (829 secs, 00:13:49) Total: 825 secs, 00:13:45 Elevation(GPS) : 0 2.79 6.871 m Altitude(Barom.): -1.6 3.99 8.95 m Vertical speed : -0.9 -0.01 1.25 m/s Max speed : 9.9 km/h Max target dist.: 5.1 m Sats: 13 12 15 (min/avg/max) Voltage: min. 14.5, max. 16.6 V Current: 0.5 21 26.1 A Wattage: 7 329 414.99 W Capacity: 4961 mAh Motor1: 1.9 3.4 5.5 A Temp: 20 46 59 °C Motor2: 1.0 1.8 2.9 A Temp: 18 47 58 °C Motor3: 1.3 3.5 5.7 A Temp: 22 45 52 °C Motor4: 1.5 3.7 5.5 A Temp: 26 51 61 °C Motor5: 2.0 3.7 5.7 A Temp: 28 52 61 °C Motor6: 2.2 3.6 5.5 A Temp: 23 49 56 °C Magnet Field: 101 102 107 % (ok) Magnet Inclination: 57 59 65 deg No errors found After: MK Version: FC HW:2.1 SW:0.88e + NC HW:2.0 SW:0.28h Flight time: 11:43:40 AM - 11:49:57 AM (377 secs, 00:06:17) Total: 377 secs, 00:06:17 Elevation(GPS) : 0 0.34 7.474 m Altitude(Barom.): 0.15 5.8 8.35 m Vertical speed : -1.44 -0.08 0.9 m/s Max speed : 7.8 km/h Max target dist.: 8 m Sats: 6 8 10 (min/avg/max) Voltage: min. 15.1, max. 16.3 V Current: 0.5 21 27.6 A Wattage: 8 341 441.6 W Capacity: 2291 mAh Motor1: 0.8 3.7 5.7 A Temp: 20 43 53 °C Motor2: 1.0 1.6 3.2 A Temp: 25 42 52 °C Motor3: 0.9 4.2 5.6 A Temp: 22 49 58 °C Motor4: 2.0 3.3 5.0 A Temp: 25 54 63 °C Motor5: 0.6 4.3 6.2 A Temp: 24 60 68 °C Motor6: 2.0 2.8 5.1 A Temp: 26 45 55 °C Magnet Field: 101 101 106 % (ok) Magnet Inclination: 58 60 64 deg Errors / warnings: Error "BL Limitation" (24) occured 1 times!
That means something may be wrong with the BL on the board Can you send a picture of that BL? It seems in a few cases of people having that occur had some sort of short going on. This is the official explanation of the issue from MK in English. "BL Limitation" A BL-Controller is in limitation - either overtemperature or overcurrent Although you do not seem to be approaching any limits at 6.2 being the max for current, so It must be the temp that is out of limits. MK does a poor job explaining what those limits are other than they hint that the absolute limit is 100C and that the motor will not restart after >85C. 68 is approaching that number. That does seem a bit high for the current that the board is supplying. That leads me to the assumption that it is a BL issue moreso than a faulty motor. I would expect a high current draw coupled with an over temp if it is the motor shorting out. But in this case the Motor current draw is nominal, not even near the realm of overdraw situation. Something is a miss with the BL IMO.
Hi Shaun - Can you be more specific about the pictures you want to see. Do you want pictures of each board or just #5? Both sides of the board or just top? Colin
The top of BL5 would be great. Just a couple of other questions. What was the outside air temp? When I run models comparing each motor current against temp. I do see 5 being the one that a greater increase in temp for the current. But not that much more than the others. I would guess in the San Fran area it was not too hot of a day to overwork the motors, especially since they only were draing on average about 5 Amps. So I am wondering if there is something else going on with the board.
So, there was blade of grass on BL 5 and I removed it and flew again. This time no error. Amps for #5 seems in line with the other 'odd' motors, but still the temp is higher. Going all the way back to the beginning of this thread - even though I fixed the compass error problem - the odd motors are pulling more current than the even ones. Is that normal for a hex? I have nothing to compare it to, so if anyone with a hex can weigh in and briefly show their data please do so. Thanks much, Colin Data: MK Version: FC HW:2.1 SW:0.88e + NC HW:2.0 SW:0.28h Flight time: 1:30:56 PM - 1:43:43 PM (767 secs, 00:12:47) Total: 763 secs, 00:12:43 Elevation(GPS) : 0 4.32 7.846 m Altitude(Barom.): 0.1 5.66 7.35 m Vertical speed : -0.97 -0.03 1.59 m/s Max speed : 13.9 km/h Max target dist.: 7.5 m Sats: 9 10 13 (min/avg/max) Voltage: min. 14.4, max. 16.4 V Current: 6.8 21 30.3 A Wattage: 101 333 436.32 W Capacity: 4614 mAh Motor1: 1.2 3.6 6.2 A Temp: 22 42 58 °C Motor2: 0.6 1.5 3.3 A Temp: 20 41 53 °C Motor3: 2.1 4.5 6.3 A Temp: 23 46 53 °C Motor4: 1.8 3.5 5.2 A Temp: 27 52 60 °C Motor5: 2.7 4.3 6.3 A Temp: 28 56 67 °C Motor6: 1.2 2.6 4.2 A Temp: 25 42 51 °C Magnet Field: 100 101 107 % (ok) Magnet Inclination: 57 60 66 deg No errors found
Outside temp was a warm and comfortable 68-70 degrees. Here's a pic of the board. See anything unusual?
Have you suspended your copter and checked its balance? I made an aluminum plate that bolts to the top of the battery tray that allows me to suspend the copter and have a dead center mounting point. It wasn't until I did this did I realize how much of a difference 50 grams here or 50 grams there makes to the balance of the copter. If you are not up for this I would use MK tools with the X bee and check the balance and CG.
No not really your wires look good. The board is a little dusty, might not be a bad idea to clean the board off with compressed air or some sort of electronics wash. Pure alcohol can be used. I need to this myself, my PDB looks the same That blade of grass makes sense, because it is partially conductive so that would cause an some arcing. As to the imbalance of why the odd motors are different then the even ones what Dave is saying makes sense to certain degree. The simple average of the odd motors from the data I analysed earlier is 42 Amps and 27 for the even. That does seem to be a little disparity there. I have a Hex, its not a CS but the physics are the same and I notice motors on particular side both even and odd ramp up to compensate for CG. The only thing that makes any sense is that there is some adverse yaw being induced. I would look at neighboring propellers and see if you notice blade being higher than the other. If so that means one motor is not level. Andy came up with a neat solution for motor leveling. He glued one of the original Black Nuts that came with the QC motors on one of those disc bubble levels to do it. I have yet to try this, it seems legit. The physics to me seem like a yawing issue more so than a balance issue.
Thanks Dave. Do you have a picture of that you can share? Also, what did you do to balance out your copter? Yaw makes sense and fixing an CG imbalance couldnt hurt. I checked all props and they look to be all at same level. They are all brand new APC 14X 4.7 SFs and balanced. Is Andy's level solution in the video or is there a post on that? I searched the forum but did not see it. My next test will be the full monte - gimbal and CX760.
Hi Colin: Here it is: http://forum.freeflysystems.com/ind...motors-and-checking-for-runout.334/#post-3276 Andy.