Brushless RPM Sensor Accuracy

[SpeedVision]

Until someone made a comment about "fine tuning" the brushless RPM sensor, I took for granted that it was giving a VERY ACCURATE representation of motor RPM thus an accurate head speed calculation.

Edit - Oops, there isn't a way to "fine tune" the brushless sensor...phew.

For the benefit of all who have one, can we get some perspective on the expected accuracy relative to RPM deviation? Given the high head speeds of modern Heli's I think it's essential for us to know what to expect.

In my case, using an optical tachometer to verify the brushless RPM sensor is not possible nor practical. I purchaed the Eagle Tree V3 in lieu of any other tachometer alternatives.

Some feedback from Eagle Tree and some real world examples would be very helpful!!! Thanks.

[spinup]

Was that comment referring to the "Calibrate Motor RPM" option in the eLogger application? IMO it should be labeled "Set Motor Poles and Gear Ratio" instead. It doesn't calibrate anything, unless you consider the number of poles to be a calibration. :)

I asked Eagle Tree about RPM sensor accuracy a few months ago, and they said the brushless sensor is the best one. Unfortunately my brushless sensor is not giving a stable reading (http://www.rcgroups.com/forums/showthread.php?t=832763), so I haven't been able check the accuracy. ET seems pretty responsive on these forums though. I'm optimistic that they will pinpoint the problem soon... ;)

-Jeff

[SpeedVision]

Spinup, I corrected the content of my post... Sorry for heading everyone down a dead end...

After corrections, still a valid question... I wouldn't be surprised if accuracy was within 25 RPM of actual...

[BEC]

Assuming the ESC is in synch with the motor (no commutation errors as sometimes happens especially with cheapies), the reported RPM should be exact - after all it's actually counting the commutation pulses being sent to the motor by the ESC. I'd expect it to be way more accurate - and precise - than either optical or Hall-effect-based sensors.

I agree with Jeff that calling the procedure to tell the Eagle Tree application the number of motor poles and the gear ratio a "calibration" is slightly misleading. Perhaps in the next rev. of the application Bill will change the terminology there.

I will also say that if I use a Medusa sensor and an Eagle Tree sensor on the the same motor/ESC combination at the same time and compare the readings between the PA Pro and the MicroPower, they track exactly within the averaging that the Medusa software does. This gives me a sense that they're both dead on.

[spinup]

Assuming the ESC is in synch with the motor (no commutation errors as sometimes happens especially with cheapies), the reported RPM should be exact - after all it's actually counting the commutation pulses being sent to the motor by the ESC.Hmm...I thought it worked by counting back-EMF pulses off the motor. When I spin my rotor by hand, no throttle, the sensor picks it up correctly. Do the pulses from the ESC confuse it? Sorry if this is a naive question. I don't know much about how my esc works yet :o

I am having sensor trouble with a cheap motor/esc right now. What do you think those commutation errors would look like?

One more question, I can't resist since your handle is BEC and you have 11k+ posts: What would you recommend for a good 20A ESC? The obvious Phoenix, or are there other good options out there?

Thanks!
-Jeff

[kcasser]

My Brushless Tach is bang on in my TRex 450SE as compared with my SkyTach. I think it is incredibly accurate.

[spinup]

Yeah, I'm sure it will be very accurate when it gives a stable reading. My problem is that the reading jumps all over the place, ranging from 24000rpm to 34000rpm. And it has dropouts when running off the flight battery only (no USB).

I may have gotten a bad logger or sensor unit. Hoping to hear from ET this week.

Just for perspective, how much noise do you get on your brushless rpm readings? +/-1000rpm, +/-100rpm, etc? Thanks.

-Jeff

[BEC]

Hmm...I thought it worked by counting back-EMF pulses off the motor. When I spin my rotor by hand, no throttle, the sensor picks it up correctly. Do the pulses from the ESC confuse it? Sorry if this is a naive question. I don't know much about how my esc works yet :o

I am having sensor trouble with a cheap motor/esc right now. What do you think those commutation errors would look like?

One more question, I can't resist since your handle is BEC and you have 11k+ posts: What would you recommend for a good 20A ESC? The obvious Phoenix, or are there other good options out there?

Thanks!
-Jeff

My "handle" is just my initials....

Anyway - you may well be right that the sensor is allowing the logger to count the back EMF pulses. Perhaps Bill will chime in and straighten us both out.

I know what commutation errors sound like - a screaming noise coming from the motor. I can't say as I've seen anything on EagleTree graphs I can point to and say "that's where the motor and the ESC got out of synch". A current spike at the same time is a pretty good indication, though. Also seemingly random RPM values/wide swings in a short period of time. I have seen commutation errors on Medusa graphs, and that's what they look like.

20A ESC, eh? Well, I'm a big fan of the Castle Creations Thunderbirds - so the TB-18 in this case. They're simple to use and they work and they're backed by CC's exemplary customer service.

Others I can recommend from good experience: Motortron MDrive-18, Jeti Advance Plus 18 and ECO 18. The Great Planes ElectriFly Silver Series 25 isn't bad, with two caveats: one - the minimum RPM it will run is kind of high, so it's not a choice for a really floaty airplane when you're trying to land, and it sets the LVC to 67% of the initial voltage (not adjustment), which is too low for high-discharge-rate LiPolys.

I've bench run but not yet flown one of the really cheap Suppos - and it's OK on the bench. I also need to spend more time to get familiar with the Scorpions.

You can see what sort of results I'm getting by looking at my SA SHAFT review here: http://www.rcgroups.com/forums/showthread.php?t=740836 There are three EagleTree graphs with RPM traces in the article.

[BPD]

Spinup,

I think you are in for a surprise. The brushless RPM subject has bee sent to the bit bucket several times before. Here is the one I can still locate.
http://www.rcgroups.com/forums/showthread.php?t=813317
They never did resolve my problem.

BPD

[billpa]

Hi BPD/Spinup,

We delete posts to this forum when we believe the issues are resolved, and have gotten feedback as such.

We have made a fix to brushless RPM in version 6.38. If you still have issues with brushless with that build or later, can you post what is happening?

Regards,

Bill, for Eagle Tree

[spinup]

BPD - fwiw, my brushless sensor thread is linked from the second post in this thread. My thread was deleted, BUT first Eagle Tree responded with 6.38 (fixed rpm dropouts when running on flight battery) and a replacement sensor.

IMHO it looks bad when a vendor deletes support posts in a public forum, even if the issues are resolved...but this is EagleTree's forum to run as they choose. I'm happy as long as the product is supported effectively. :)

Anyway, the replacement sensor did help. Instead of that crazy 24000-34000 rpm fluctuation, it now keeps to +/-500 rpm range at WOT. Actually it has ~500 rpm resolution, e.g., samples are exactly 35211, 35714, 36231, nothing in between those values. So my headspeed readings are +/-50 rpm. I guess I'll need a laser tach to check accuracy. :rolleyes:

This is at 10 samples/second. I'm going to play with lower rates to see if this helps the resolution. Also planning to run bench tests with a better ESC (Phoenix 25) and three more motors (including a nice Axi Gold and a cheap Suppo). Stay tuned... :)

-Jeff

EDIT: BTW, I still need to use both leads to get a reading on my current setup. And the leads on the replacement sensor were much shorter, about 3" IIRC. Perhaps that makes a difference?

[billpa]

Hi Spinup,

Thanks for the post and glad to hear your RPM sensor is much better.

The 500 rpm "buckets" you are seeing is actually a firmware limitation, which saves logging space. The buckets are usually much smaller, but tend to get larger with brushless motors with larger pole counts.

We have an option to greatly increase RPM resolution uder "Tools, Custom Hardware Options" but this has not been tested in ages and probably needs to be debugged. But if you need that level of resolution, let me know and we'll get this working if it is not.

The shorter leads do not affect sensor operation.

Regards,

Bill, for Eagle Tree

[spinup]

Thanks Bill! Yes, I really do need the higher resolution. Especially since the next two motors I am hooking up have 14 poles and much lower kV. :)

I will happily sacrifice logging space for resolution. I don't think I've used even half the memory on the v3 eLogger yet.

Do you need feedback on how that option works now? Or should I wait for another firmware rev to test this?

-Jeff

[billpa]

Hi Jeff,

Sounds good.

I would give it a whirl, but don't be too mad at us if it does not work. :-) Let me know.

Regards,

Bill, for Eagle Tree