mounting hall sensors to R/C brushless motors

Matt, the Infineon controllers are programmable, too. I've been playing with the programming on mine today and it's really easy to adjust things like the current limit, low voltage cut-off etc via the PC interface. The software for changing the settings is a bit "Chinese", but works OK once you've decoded what the various settings really do. The only tricky bit is making up the programming lead, but philf sorted that for us in the thread I linked to yesterday.

Jeremy
 
Jeremy- Can you find out what to do to disable the turn-on delay? I've started playing with mine, and I find that makes it totally un-usable. 100% not ok to have delay, and a complete deal breaker.

Also, I'm running my 0-5v pot throttle rather than a hall effect throttle. I have a hall effect throttle, but it feels very cheesy compared to my magura throttle. It seem to not turn-on for the first half of the throttle twist with the pot. And ideas on that? When you find the solution to remove the delay, please post it up ASAP. It's a paper weight to me otherwise.

Thank you!
-Luke
 
Luke,

There doesn't seem to be an obvious setting in the software to control turn on time, so I guess we'll need to look at the circuit to see how it's controlled externally. There are some timing settings in the software, but it's not clear quite what they do. One is marked "block time (s)" but the lowest setting is 1, which may be the default. There is also a setting marked "auto cruising time (s)" which seems to default to 15. It'd be nice to have a translation for all of the controller programmable settings, as the only ones that are obvious are the LVC and current limit ones.

The reason that your Magura throttle isn't working at the low end is down to the way the controller is set up on the throttle input. It seems to ignore the first volt of the control signal, presumably because the Hall throttles sit at around 1V when they are "off". The Magura goes right down to 0V, so the first 20% of the throttle movement won't do anything. I believe that the Magura can be adjusted to compensate for this, by adjusting the pot position inside the twist grip. You want to aim for a throttle off voltage of around 1V, although you may be able to experiment with this by twisting your throttle until the motor just starts and measuring the voltage at the throttle input. If you then adjust the throttle pot so that zero throttle is just a naffigravit less volts you should get a better response. You may find that this takes out some of the apparent throttle response delay, too.

Jeremy
 
That's good to know it can be rotated inside. I was thinking of maybe adding a second pot that I could adjust like a pull-up resistor between the tap and the high side to tweak it so it climbs to 5v much faster, but it would still sit around 0v at the bottom. I haven't actually mounted anything, or made finalized wiring, or even put my chain back on yet. It's ghetto rigged with alligator clips and hall wires twisted together everywhere, but it makes the motor spin. Seems kinda herky-jerky though, so I don't think I've got the sensor leads vs the phase leads in the perfect combo yet. I wish I had more time to play with it, but I have a ton of real work to do at the moment. I will post up some pics when I get it better sorted and looking more presentable.

Please find a solution to solving that turn-on delay. I will instantly paypal anyone $50 to gets the fix to that figured out, because I can't use something that doesn't respond when I need it to, and I will have to dump the Infinions for something else if it can't be fixed. That would suck, cause I beefed up all the high current traces with a double pass of solder wick, added a ton of cap, and swapped all the FETs to IRF4110's all ready.
 
Wow, great thread. I had thought it would be possible to do a sensored motor out of the RC stuff. Glad to see it being done. I will definitely be doing the same with my next build. We now have all the componenents to build a most excellent RC ebike. Seems like the only thing to fix now it the infineon delay. Is this typical of most ebike controllers or just with the infineon? Any other choices for a controller that people have been happy with.

Kin
 
Luke,

I've just been running my motor/controller on the bench again to look at the throttle problem. I don't seem to have any appreciable delay on mine, it seems to respond pretty quickly when I turn the 5k pot I've got hooked up to the throttle connections. I can confirm that for my controller zero speed is actually 1.28V at the throttle input. The safe setting is probably around 1V.

I may try some different software settings tonight, but for reference these are the settings I have at the moment (remember this is just a baby 6 FET Infineon I'm using as a test set-up):

Phase current = 70A
Rated Current = 15A (kept low for test purposes)
Limit Voltage = 20.1V
Tolerance = 1V
Block Time = 1S (the fastest setting)
Auto Cruising Time = 15S
Slip Current Charge Mode = 1
EBS Level = 2
Speed Mode = 0
Limit Speed = 99%
Speed 1 = 33%
Speed 2 = 66%
Speed 3 = 100%
Converse Speed = 35%

Maybe your controller is set up for a longer "block time"? I'll see what effect this has later tonight by re-programming my controller.

Jeremy
 
Mine lags for at least a full second from the time I open the throttle until it starts to make the motors turn. You can actually go WOT, and come back quickly, and it won't move the motor. The RC controllers always responded instantly. Anything with a lag/delay is useless for technical riding.

Can you make that block be zero?

Also, Methods told me there was something with the speed setting, that you could set to over 100%, and it boosted torque everywhere, epically on top end. He thinks it may be an electronic timing advance. Might be something worth taking a look at while you have it on the scope.

Thanks for the tip on where the throttle threshold voltage is. I wonder if that is something that could easily be changed in the hardware of the board, like removing a resistor or a jumper somewhere.
 
Well, I've just re-programmed my controller with a longer "block time" and it doesn't seem to make any difference. You can't set the "block time" to zero, 1S is as low as it will go.

Mine definitely doesn't lag to any noticeable degree, the motor spins up just as fast as I can turn the speed pot. I cannot get anywhere near winding the pot to full and back without the motor following it near-instantaneously.

Maybe there's a difference between the 6 FET controllers and the bigger ones? I've not yet built one of my 18 FET boards up, so haven't had a chance to play with it.

I've read that setting the "Limit Speed" to over 100% changes the motor timing, but I've not had any confirmation of that.

What's your test set-up? I'm wondering if there is some other effect causing the throttle lag you're seeing, like a low current limit? Have you tried changing the internal controller settings using the serial link?

Jeremy
 
That's fantastic news that your 6-fet doesn't lag. These 12fets have an awful delay. I'm about to post up some photos, and you can let me know if your uController is the same chip. If it's something in the chips firmware that can't be changed, I will just buy 6-fet controllers with no delay, heatgun off the chip, and swap them onto the 18fet boards that are comming. I can't imagine why they put a damn delay in these things. 1/2 the whole performance advantage to electric power should be instant torque response and delivery.

I don't have programmers yet :( Methy has a couple 18fet boards and some programmers that he is building into them so they juts have a USB jack on the side of the case, and then sending them up to me. Then I'm going to throw IRF4110s, as much cap as the case can hold, and 8awg all the current traces. I've got something clever up my sleeve for cooling :) Hopefully they will be able to burn my motors up, because I'm dying to have an excuse to do a custom re-wind on them :) I have so much motor cooling though, the controllers are going to have their work cut-out for them :)

I'm working on finalizing some connections and things right now, and I will post up some pictures as soon as things look somewhat presentable.

-Luke
 
I want to share my idea on why I think having timing control could be important. If you take a look at the spec sheet for the ss41, you can see it has a gauss threshold to make it latch.

http://parts.digikey.com/1/parts/948156-sensor-ss-hall-effect-bipolar-ss41.html
Click the view datasheet link near the bottom



So, if you stick them physically in a neutral position, you actually have no idea how retarded the sensor is latching. In fact, odds are, if it can latch from 1/4" away from your magnet, then it likely latches as the magnet is just creeping up on the sensor. This could have the effect of advancing timing, but the extent the timing is advanced would depend on the square root of the distance that the sensor happens to be from the magnet.

What I'm thinking, is that your timing with your sensors very close to the magnets on the inside could act more advanced than my sensors on the outside, placed in identical locations. Or maybe the stator laminations re-direct the field so effectively that it shields your sensor, and my outside sensors reach the latch threshold first. I have no idea, but I'm looking forward to getting some data up on the effects of timing VS power/efficiency.
 
If the delay is built in as firmware and isn't adjustable with the programming software, then it might be possible to get hold of the full code for this controller (not just the dealer set-up software) and change the settings that way. Reading some of the documentation for this controller, it looks as if it has been deliberately designed to have a soft start. It may be that I'm just not seeing this on the bench because my motor is so lightly loaded.

The best thing would be to load the chip with a known good set of parameters using the programmer, adjust the throttle to get rid of the dead movement at the start and then see how it goes.

If the lag is a still a big problem, then maybe a call to Keywin Ge might just help, as he might know a way to re-programme the soft start.

Jeremy
 
Another thought came to me a while ago, whilst reading the delta/wye thread. Reading up on this, it looked like these RC model outrunners are wired in delta, for ease of manufacture. Delta gives a higher rpm/voltage constant (Kv) and also has a slightly lower efficiency, due to circulating currents in the unenergised windings.

A low Kv value is a pretty good thing for us, as it means less gearing down and slightly less transmission power loss. Add in the slightly better efficiency of a star (or wye) configured motor and it just seemed sensible to look at hacking a motor to this configuration. One of my poor Towerpro hack motors has been subjected to a a bit of disection. I found it only took me ten minutes or so to tease out all the wires and get access to both ends of each winding. Here's a picture of progress so far:

3438669199_6f44a3e068.jpg


The next stage is to wire the thing up in star configuration, fit some Hall sensors to this motor just like I did with the other one and see what happens.

Jeremy
 
I'd be very surprised if most RC outrunners were wired up in a delta configuration. The whole reason outrunners came into being was because they had high torque at low rpms, so they could drive a prop directly, without a gearbox.

-- Gary
 
I can correlate, most outrunners are terminated Delta. It is faster to manufacture, you have one less solder joint and 1.78 less winds to wrap for a given KV.
 
I'm a bit surprised, as well, for the same reason. However, that photo shows that this particular big outrunner motor is definitely wired delta. It's easy enough to check any outrunner, as you just need to look for the centre tap join for the three windings. If there isn't one, then it has to be wired delta, AFAIK.

Jeremy
 
Looks like I've just turned my 215 Kv motor into a 124 Kv motor then...............

I just need to mill the stator for Halls, like the one I made the other day and see how it compares. I really like the idea of lowering the Kv like this, it seems such a simple mod, too.

Jeremy
 
Wow, that is very suprising. I'll have to check my Hacker A60-18L. :)

I just talked to Bob Boucher at AstroFlight. He actually just put together a 3210-12T that has all six wires coming out, so that it can be externally setup for wye/delta switching. He did it to test a theory that this would stop a lot of controller failures that some were seeing with supposedly 50A setups. He said that the phases could be as high as 200A, which is why the FETs are popping. He said that with external switching it could start out in wye, and be swtiched into delta and that would reduce the low-end current requirements. I told him what we've been looking at here, and that it would simplify the gearing setups. Anyway, he's going to button up this motor and send it down to me this week.

The second thing we talked about was adding the hall sensors. Ten years ago, there weren't any sensorless controllers, so all his brushless motors had sensors. He said although he has some leftover boards, they are all way too small to be used with the 3210, so he needs to do a new PCB. He agreed, though, that using the inexpensive controllers, known to work well with typical ebike loads, is probably our best option at this point, so he will look into getting a new sensor PCB made.

-- Gary
 
I'm also going to take my 3210 apart, to see how hard it is to add sensors. Jeremy, do you have a part number for a suitable hall sensor? I'd like to get some ordered from Mouser or Digikey. Also, since I think I have some blown sensors in my 5304 I blew up on my Townie, would they be the same as what would be needed for a 3210 mod?. Actually it was the phase wire adapter plug that melted/shorted out when I ran 104A through the controller that Bob Mcree modified with 4110s, so I don't know if the sensors are still good or not. :)
 
OMG! This is so awesome! Jeremy, give me your paypal adress, I'm sending you 50bucks just for being awesome!

I can re-terminate my windings, and then mount larger motor sprockets (for less chain noise), and then a smaller rear sprocket for less weight. Very cool!
 
Hi Jeremy;

Thanks for the info, years ago I had a dc motor (pwm) variable speed project, for a client, I am a mechanical type person with some electronic design. I put the project together, worked fine- but I did not test it under full load, I was not turning off the power mosfet hard enough. When looking at the waveform going into the motor there was not a square wave turn off (a tail) which caused the mosfet to overheat. The first power mosfet I tried did not have as much capacitance in the gate as the 2nd one (double ~), needed to change the turnoff transistor. (pull it to ground harder as in "n" type mosfet)

I was thinking of replaceing the mosfets in a controller also (going to a 24v/30v system @ 1500~ watts, using a 30v~ 500watt controller as the driver) - in your replacement of the mosfets did you check if they are being turned off hard enough under full load?

For another project I was going to try a IXYS 14 amp noninverting power mosfet driver. IXDD414, from Digikey (to drive a PRX KS621K40A41 triple darlington).

Jim
 
Jeremy Harris said:
Mine definitely doesn't lag to any noticeable degree, the motor spins up just as fast as I can turn the speed pot. I cannot get anywhere near winding the pot to full and back without the motor following it near-instantaneously.

Maybe there's a difference between the 6 FET controllers and the bigger ones? I've not yet built one of my 18 FET boards up, so haven't had a chance to play with it.

I have the 1 second delay with my 6fet.
It does have a soldered shunt and 4110's

I have a suspicion. . . When I look at your settings it looks like your controller is not set for max speed. I think it is set up for 33%
Would you please change the three speed settings to 100%, 100%, 100% and repeat your test?

On the bench I get the 1second delay like a screaming monkey. . Just cant believe that we are running the same test on the same hardware.

Spin the wheel up to 30mph
Twist to zero then immediately twist back up to WOT
At least a 750mS delay before the power comes back on full again
You can even hear the motor spool up trying to catch up

I think your motor may be set at 33%, which would explain why it tracks so linearly as that would cut the delay by 66%

Thanks,
-methods
 
Gary, the sensors I used were pretty much the same as the ones in the Crystalyte motors, which are Honeywell SS41As. I actually used the slightly more expensive SS411A sensors, just because they were the only ones in stock here in the UK. Digikey stock the SS41A sensors, part number 480-1999-ND.

Methods, my controller is set for full speed, I just have the three optional switched speeds set for 33%, 66% and 100% and the limit speed set to the maximum, 99%. I've been running it with the high speed wire grounded, so am getting full speed. Playing about with it late last night, I can detect some delay in the response, I think. It's partially masked by the stiffness in my throttle pot though, so it may well be that I'm just not feeling the lag the same way as you are. I'm pretty sure that there has to be a way of changing this in the controller firmware somehow. It's a great pity that the programming interface doesn't allow user control of the soft start ramp, like the AXE controller programmer does.

JEB, I did a rough check on the 'scope after swapping out the FETs, but didn't see anything awry. The new FETs (IRFB3077) are pretty close to the original ones in terms of gate charge, so I wasn't really expecting to see any difference (other than much reduced resistive loss from the lower Rdson).

I should finish getting the motor modded tonight (UK time!), with luck, so will report back on how well it works wired in wye.

Jeremy
 
Back
Top