Adding sensors to a Astro 3210 motor

[Burtie]

Drew,
You beat me to it! Good to see that it works without needing extra magnets.

I think this internal solution is potentialy much more practical / cheaper / verstile / easier to produce than the external setup I have been playing with.

Nice job

Burtie

[Thud]

Drew,
Gwhy,Jerremy H, & matt-n-mtl. have been di-secting the cutting out issues on our mini controllers (6FET infineoins)
& have a couple items zeroed in on. Just a few tweeks & the cut out issues are behind us. Hopefully with some of the safety features still intact.

its all in this thread:
viewtopic.php?f=2&t=16910

[drewjet]

I have been following that thread. I just got my programming cable from Lyen, just need the time to hook it up and then try some of the settings.

[Burtie]

Inspired by drewjets video, I tried fitting sensors inside the astro too.
It turned out to be a pretty fiddly job to get the timing right , but it works well on the bench now .

I had a couple of attempts at fixing halls to a circular piece of strip board that was machined to fit inside the end cap.

This one was rubbish

First attempt.jpg (84.27 KiB) Viewed 560 times

The second attempt worked better.
I needed to get the sensors further away from the stator fields.

Astro back plate.jpg (99.49 KiB) Viewed 647 times

This one worked much better

Backplate2.jpg (89.61 KiB) Viewed 647 times

Hall positioning was complicated by having to avoid a large lump of snot stuck to the end of the rotor, -presumably for balancing.

Lump of snot.jpg (85 KiB) Viewed 680 times

Now just an extra bundle of leads give the secret away.

Sensor leads.jpg (57.92 KiB) Viewed 560 times

To sum up:
wye current was 0.6 A, delta current was 4 A (was 0.6A and 3.5A with external hall setup and also with sensorless controller), so efficiency looks pretty similar so far. With This method it took about 1 hour to make the parts and the rest of the day trying to get the timing right! -sort of the opposite way round to the external method.

I can fit both internal and external sensors to the same motor, so should be easy to do a proper comparison of performance when it is fitted to a bike.

Edit update:
This internal fitment does not work well in practise, it is upset by the magnetic field from the stator
See more info later in the thread here
viewtopic.php?f=30&t=14210&start=180#p545457

Burtie.

[drewjet]

Awesome Burtie. I really like the idea of them inside. For one it elliminates the need for an extra magnet ring and also protects the Halls.

Do you think there is anyway I could get the timing done without a scope?

Also I have seen the PC based scopes, would they work and how much do they cost?

[Burtie]

I dont use a 'scope to set up the timing. I just rotate the sensor group carefully while the motor is running and look for the position where least current is drawn.
Of course you need to get the timing pretty close before using WOT or you risk drawing destructive amounts of current . I always put a 10 amp fuse in the supply when doing this now (after learning the hard way ).

I think USB scopes are a great idea, I promise myself one day when I save up I will buy one! Not sure my old craptop is up to it tho .

I dont know what they cost, anyone got any recommendations?


Edit: forgot to ask was that astro wired in delta or wye in your video? If it was delta, the current figures look ok, you would have the timing spot-on.
Burtie

[cell_man]

For the USB scopes, I used a TiePie for years and it was great but they are a bit pricey at about 1000GBP for the high frequency unit, might be more now that the pound is worth nothing... Pico Scope are also supposed to be very good and they start at very reasonable prices. The tiepie I used was every bit as good as a bench scope of similar cost IMO. They are so handy and I really wish I would have bought 1 instead of rushing out and buying a bench scope a couple of months back.

[drewjet]

Burtie, I am wired in Y, that is why I know my amps are too high. It looks like you machined away the cutouts for the power wires on the Astro, was that to make it easier to time it, by being able to rotate while it is running. I am not 100% sure I am exactly on 30 degree spacing as well. I used a 30 60 90 triangle to make my holes then drilled by hand. Fine for expermenting, but needs accuracy for real use. I am hoping you come to a final config you are happy with and then buy one from you.

Cell Man, would this scope http://www.picotech.com/picoscope2200-s ... tions.html be appropriate? I have never used one and only seen them in person a few times. I don't really want to spend $260 for something I wouldn't use much, any of them around $100?

How about something like this one? http://www.virtualvillage.com/5-channel ... ium=shcomp

[Jeremy Harris]

The cheapest 'scope I've seen that is just about fast enough for ebike controller work is this one: http://www.seeedstudio.com/depot/dso-na ... fae39e7e00 for $89. It has a USB connection, so can dump files to a PC. I built it's cheaper predecessor, this one: http://www.seeedstudio.com/depot/digita ... l?cPath=71 which is currently $33 as a kit. It's just about fast enough for general ebike controller work, but could do with a bit faster sampling rate to be able to see transient events a bit better. It doesn't have a USB port, but connects via a serial port and can only dump screen shots as BMP files.

Jeremy

[dbaker]

I have had some limited success with getting my 3220 to run on hall sensors. I made a fixture similar to Burtie's but I have it on the inside and am using the magnetic field coming out the side of the rotor. The video attached explains it.



http://www.youtube.com/watch?v=WqO2NkT_4o0

Drewjet,

Any recent developments with the Hall sensors inside your 3220? I got my 3220 and will be opening it up today. Looks like I should set the sensors up vertically spaced 30 degrees and try to null unloaded current. Did you get yours working well?

Dave

[recumpence]

One thing I have not heard anyone mention regarding running sensors is the electric RPM limit of these sensored controllers. Many controllers have a 40,000 electric RPM limit. That is only 5,000 RPM on these Astro motors. It would be far lower on higher pole count outrunners.

Matt

[dbaker]

Matt,

What is the pole count comparison for the Turnigy HXT and the Astro 3220? I remember discussions of 6 FET sensored controller problems that were eventually solved.

Dave

[recumpence]

If I remember correctly, the HXT motors are 12 pole. The Astros are 8 pole.

Luke mentioned that the Infineons are good for 80,000 electric RPM. If that is so, we are good with those controllers.

Matt

[liveforphysics]

The Kelly stripped down and de-potted on here (thanks Hal9000!) Showed the controller chip was a part made in 1998 from its datasheet(if I remember correctly)... And it was a recently purchased Kelly. I imagine they are just sticking with what works for them, and for hubmotors 40,000 electrical RPM would never be a problem.

The "infinions" that we use have a wide range of controller chips being used, some of them right up to modern 1 year old designs that run processor speeds as fast as the best RC controllers. Of course everybody has their own flavor of firmware for the chips as well, further complicating things... The bright side is, there are versions that work, and I think that includes most all of the newer ones. Lyen is the guy to buy from for the latest and greatest versions AFAIK.

[drewjet]

Drewjet,

Any recent developments with the Hall sensors inside your 3220? I got my 3220 and will be opening it up today. Looks like I should set the sensors up vertically spaced 30 degrees and try to null unloaded current. Did you get yours working well?

Dave

I have not touched it. Too many other projects and too much work. I would like to next try and make another ring. I don't think I made it all that accurate. I think that if I did it at 120 degree spacing I could be a bit more accurate.

[dbaker]

Drewjet,

Why would 120 degree spacing improve accuracy? Isn't it the issue of physically locating the Hall sensor where it will provide the same relative timing signal as it's 2 brothers? I would think that might be improved by enlarging the sensor location circle to make the sensor physical spacing delta/timing signal delta better.

My take is that without accurate analysis tools we can only adjust sensors by minimizing no load current by rotating the the sensors relative to the motor poles.

Dave

[drewjet]

My thoughts are that if you are off by 3 degrees when it is 30 spacing you are off by 10%. if you are off 3 degrees on 120 spacing you are only off by 2.5%.

Yes without a scope the only way I could see of checking validty is using no load amps. The problem is it is not so easy to do when mounting them inside the astro.

[dbaker]

Inspired by drewjets video, I tried fitting sensors inside the astro too.
It turned out to be a pretty fiddly job to get the timing right , but it works well on the bench now .

I had a couple of attempts at fixing halls to a circular piece of strip board that was machined to fit inside the end cap.

First attempt.jpg

The second attempt worked better.
I needed to get the sensors further away from the stator fields.

Astro back plate.jpg

Backplate2.jpg

Hall positioning was complicated by having to avoid a large lump of snot stuck to the end of the rotor, -presumably for balancing.

Lump of snot.jpg

Sensor leads.jpg

To sum up:
wye current was 0.6 A, delta current was 4 A (was 0.6A and 3.5A with external hall setup and also with sensorless controller), so efficiency looks pretty similar so far. With This method it took about 1 hour to make the parts and the rest of the day trying to get the timing right! -sort of the opposite way round to the external method.

I can fit both internal and external sensors to the same motor, so should be easy to do a proper comparison of performance when it is fitted to a bike.

Burtie.

Burtie,

What was your test voltage and was the current measured going into the controller?

Thanks,

Dave

[dbaker]

I got my Halls installed inside my 5 turn 3220 and tested with a 16s LiFePO4 pack and a 12 FET Infineon ebike controller. I found a good phase/Hall wire combination on my 4th attempt. It ran so smoothly I did not bother with adjusting the timing! At low throttle the motor will smoothly run at around 400 RPM; I can see the set screws on the reduction pulley! WOT current in wye was 1.5 amps with my 53 volt pack. Controller did not appear to be limiting my motor speed like the sensorless ebike controller did.

Now to get it mounted on a bike!

Dave

[Burtie]

Hi Dave,
Good to hear about your success with the sensors inside the Astro

Most of my testing was done at 24volts, the controller input current was 0.6A when the motor was in wye configuration, using a 7 turn 3220.

I did also test the setup at 50v (12s LiPo) and the no load current was still around 0.6A, so you may have some scope for improvement if you fine tune the timing.
I found I could leave the end cap slightly loose and rotate the hall sensor board by moving the connecting wires while the motor was running.

I agree with your comments about the 30 or 120 spacing having no effect on the accuracy of the positioning, and that placing the sensors at a larger radius would make accurate positioning easier.
Trouble is with the Astro, if the radius is too large, the sensors pick up the flux from the stator coils and this messes up the timing.

Burtie

[dbaker]

Burtie,

There really isn't much room inside that end cap. I found that the perf board when shimmed away from the end cap to insulate for the soldered pads would almost touch the motor windings. The benefit of the motor winding interference is keeping the Hall sensors from touching the spinning shaft (especially the balancing blob!). I fastened the power and 3 Hall signal leads to the perimeter of the perf board hoping to use that to adjust the timing but I somehow have too much friction between the perf board and the endcap bearing housing.

At .6 amp no load current aren't you seeing some of that as controller losses which might explain the same current at half the 50v test? What would you expect the difference in current for my 5 turn compared to your 7 turn motor?

Dave

[dbaker]

Burtie,

It occurred to me that the Hall sensors might function on the outside of the endcap. When you were prospecting for locations did you sweep the sensors through that area to see if you could get a viable signal there?

Dave

[Burtie]

Dave,
I guess your motor will draw slightly more no-load current than mine, because it will spin faster for the same voltage, so there will greater friction losses to overcome.

I did not investigate using the hall sensors outside the back plate except when using an additional set of magnets in a second rotor. -give it a try .

burtie

[Miles]

No-load amps go up in direct proportion to Kv (as Kt=1/Kv). Also, eddy current losses go up with the square of the speed....

[dbaker]

Miles,

Is it valid to use 0.6 amps x 7 turns / 5 turns = 0.84 amps?

Dave