esaj Posted May 5, 2016 Author Share Posted May 5, 2016 I could put some mosfets on a breadboard (say, 4 x IRF540's for 2 half-bridges) with floating gates, wire 2 motor phases between them and try to measure the voltages from "outside" the bridges (like those would be the rails). Then maybe try to wire up some NiMH's on the rails... Link to comment Share on other sites More sharing options...
Chriull Posted May 5, 2016 Share Posted May 5, 2016 5 minutes ago, micro said: I agree in so far that the control is not cut off as the control does still try to balance us. But there is no specific control for the brake necessary. Continuous forward currents of the diodes is in the range of 50A. And there are more than only one. ... Peak breaking currents should/could go quite up to this magnitude. With an forward voltage of at least someting like 0,5V that should kill every diode without a proper heatsink! Also braking without specific controll should lead to nasty faceplants, imho? 5 minutes ago, esaj said: ... with floating gates ... Imho not a "safe" idea - with mosfets with floating gates just anything can happen and should be imho not comparable to a controller with insufficient supply voltage... Link to comment Share on other sites More sharing options...
esaj Posted May 5, 2016 Author Share Posted May 5, 2016 17 minutes ago, Chriull said: Imho not a "safe" idea - with mosfets with floating gates just anything can happen and should be imho not comparable to a controller with insufficient supply voltage... Point taken, I wired the gates to the sources... I ran into some problems measuring though: the probes picked up 50Hz mains hum from somewhere, that's makes the signals harder to read. Could be due to the fact that I don't have grounded ("earthed") outlets in this room, this house was built in the 80's when they weren't mandatory on all rooms (only kitchen & wet-spaces). Measured with channel 1 (yellow) in "rails" (like battery + and -), channel 2 (cyan/greenish/whatever) in the phases and channel 3 in the second half-bridges middle & ground in the ground-rail: Umm... yeah, the 50Hz noise all jumps up a bit when I turn the motor, but really hard to read what's happening otherwise With 9V NiMH on the rails: I was using single-shot trigger to capture these, here the trigger-level is set at 11V (top right corner). Took a pretty hefty spin for it to trigger, but it would seem that the diodes do conduct (the battery voltage level in channel 1 / yellow jumps up a bit). Or then it's just noise... Link to comment Share on other sites More sharing options...
esaj Posted May 5, 2016 Author Share Posted May 5, 2016 Ok, the problem isn't actually the grounding, as the circuit is unpowered (so kind of like the probes were floating), the probes pick up the hum from nearby wires inside walls etc (http://electronics.stackexchange.com/questions/78920/my-scope-detects-a-50hz-signal-when-the-probe-is-not-connected-to-a-circuit-is ). I moved the motor around a bit, and found a spot a little to the side where the 50Hz-signal was "only" around 2V peak-to-peak. I redid the tests, but this time I could not get the signal to trigger. No matter how hard I tried to spin the motor, the signals didn't move. I tried single-shot triggers 0.1V above the peak voltage from the hum, with and without the battery (when the battery is connected, channel 1 stops showing the hum, as it's measuring a powered circuit), but no, now it would seem the diodes do not conduct... I don't know what to make of this EDIT: Was just a bad contact between the mosfet-circuit on the breadboard and one the phases, the wires moved when I was moving the motor around... looks like I cannot actually measure this without some sort of faraday-cage to get rid of the hum. Link to comment Share on other sites More sharing options...
lizardmech Posted May 6, 2016 Share Posted May 6, 2016 6 hours ago, micro said: This was already mentioned in my question. There is often stated that the motor types differ in their BEMF. True or not? Are our motors of the sinusoidal type? Again, my point is that the battery brakes the motor via the freewheeling diodes. Has anyone ever seen something else? The winding on PMSM and AC induction motors looks different. They overlap and fall off gradually. It's probably more suited to industrial motors that run on a fixed speed based on AC power frequency. Industrial motors also have braking circuits as a mechanical brake is inconvenient and they lack a battery to charge for regen. They usually have a 7th mosfet or IGBT in their inverter that can direct energy into a large resistor for braking. Have a look at some schematics for large IGBT half bridge modules or reference designs for industrial control. Link to comment Share on other sites More sharing options...
micro Posted May 6, 2016 Share Posted May 6, 2016 7 hours ago, Chriull said: Peak breaking currents should/could go quite up to this magnitude. With an forward voltage of at least someting like 0,5V that should kill every diode without a proper heatsink! Also braking without specific controll should lead to nasty faceplants, imho? Imho not a "safe" idea - with mosfets with floating gates just anything can happen and should be imho not comparable to a controller with insufficient supply voltage... Yes, face plants are nasty. But our control does not sleep (normally) and tries to keep us upright. There are no "floating gates". That would be the "impossible wheel" technology. Amazing artists who manage that... The control does not care for braking. It even does not accelerate us. It tries to keep us upright and does this by turning he wheel a bit till we are upright again. The body diodes are always part of FET bridges (there are no external diodes). The diode is not only part of the FET, but also necessary. This is explained in text books on motor control and inverters. The diodes are in a proper enclosure and well cooled. Or take it the other way round: Braking sharply means that you will have a high current for a moment. That is no problem as the dissipated energy is small. Riding downhill for a long time is something else. I guess on steep hills you have to get rid of some 300W. Our battery has a voltage of 60V, the RMS current is 5A. The voltage drop over the diode with this low current is around 0.5V (I was not aware of this low value before I wrote this reply) Even if this power would flow completely through the diodes, the diode dissipates only 1 % of the brake energy, namely 3W. Right? Link to comment Share on other sites More sharing options...
Chriull Posted May 6, 2016 Share Posted May 6, 2016 10 hours ago, esaj said: Ok, I got the motor and hooked it up into the oscilloscope (probe on one phase, ground on another): ... A couple of spins and screenshots later: ( ~3,68 Hz, +/- 8,2V) (~5,8 Hz, +/-12V) Can you somehow estimate the rpms you managed to turn the wheel? I assume strongly that one period of the signal is not one rotation of the wheel (like with 3 coil asyncron motors) - you would have turned it like driving about 20 km/h (with a 16 inch wheel)... So my next idea would be, that the period is a multiple of the rotation caused by the ?36? permanent magnets (the number was imho once reported for a gotway?) - but that would lead to just 0,16 rpm == ~0,7 km/h... So maybe its double of this (just half of the numbers of magnets?) so 0,32 rpm == ~1,4 km/h could sound feasable? ... or it works somehow completely different?... Or your wheel just has a different number of magnets/coils.... At least frequency/voltage seem quite linear -so this would be my reasoning behind this question to estimate the back emf once the wheel reaches 20(30) km/h... So if the second thought (18 periods per rotation) and the linear relation is true, this would mean +/-163V once reaching 20km/h - which i cannot really believe? So maybe there is really just one period per rotation (and your were really bravely spinning the wheel ) and the about +/-12V back emf is all one gets at 20 km/h? 10 hours ago, esaj said: So, the signal certainly looks sinusoidal. But, does this actually prove that it is (or isn't) a PMSM-motor? I'd expect the signal to be sinusoidal, as the coils move across the magnets (getting closer and then further away) either way... Sinusodial or Trapezodial is just determined by the way the windings of the coils are distributed along the wheel diameter - according to http://krex.k-state.edu/dspace/bitstream/handle/2097/1507/JamesMevey2009.pdf;jsessionid=856476A06A2D55274A5DA09CC9D06254?sequence=1 - there is a summary at page 300 with an overview of different "rotor-stator flux linkages for different winding functions...". I have to admit that i just very sparsly looked into it - should be imho a great lecture to get an thorough insight into this topic... On page 57 he states: " The trapezoidal motor can theoretically produce ripple-free torque when driven with ideal rectangular currents, though this is not completely achievable in practice (discussed later). Since the current in a trapezoidal motor is DC over the flat-topped bEMF, since it is switched full-on over the commutation period, and since the polarity of the current is reversed once per electrical cycle, this motor resembles a brush DC motor without the brushes. For this reason it is sometimes called a Brushless DC Motor (BLDC), although claims in the literature that it is simply a brush DC motor “turned inside out” are false, as should be apparent from prior discussion. Further, since the current is controlled by only electronic means, it is sometimes referred to as an Electrically-Commutated Motor (ECM) or electronically switched motor." He also states on page 58: " It is clear that the sine motor is capable of producing ripple-free torque when driven with properly-phased sinusoidal currents. It is sometimes called a Permanent Magnet Synchronous Motor (PMSM) since it is just like the “classic” synchronous motor, with the wound-field rotor replaced by a PM rotor." At page 59 is also a nice overview with about 12 names for these motors... But this is from 2009 and maybe the naming convetions developed in another direction or still stayed as confuse as they were in 2009... I'd bet that the KS/Gotway get there high pitched PWM sound by commuting the motor in the wrong way (sine vs. trapezoidal)... But i'd assume there is some reasoning behind this, since Ninbot also changed from "pure" sinusodial commuting (absolutely no PWM noise) to a little bit of PWM noise in the last firmware version....? Maybe one finds enough time to work through this master theses he can explain all this to us Link to comment Share on other sites More sharing options...
Chriull Posted May 6, 2016 Share Posted May 6, 2016 2 hours ago, micro said: ... But our control does not sleep (normally) and tries to keep us upright. This was in regard to your example with driving the controll board just with ?2 batteries? 2 hours ago, micro said: There are no "floating gates". That would be the "impossible wheel" technology. Amazing artists who manage that... This was in regard to @esaj's post with his experiment Anyhow - if the two inverse body diodes would be getting conductive in case of regenerative breaking ( so the voltage generated from the motor would have to be at least higher then the battery voltage) without any controller influence this would have imho hindered any EUC from happen... I (still) cannot see how these two inverse body diodes could get conductive with no mosfet switched through - there is no (voltage) potential conection between the battery and the motor in this case... Secondly i almost start to believe, that the voltage generated by the motor (normally?) stays lower than the battery voltage - but for this i am awaiting @esajanswer to the previous post and/or maybe some further experiments from his side (hopefully)... 2 hours ago, micro said: The control does not care for braking. It even does not accelerate us. It tries to keep us upright and does this by turning he wheel a bit till we are upright again. Thats imho just a difference in naming the same "process" - turning the wheel with a difference to the movement speed of the "whole system" is exactly the same as accelerating/breaking. 2 hours ago, micro said: The body diodes are always part of FET bridges (there are no external diodes). The diode is not only part of the FET, but also necessary. This is explained in text books on motor control and inverters. The diodes are in a proper enclosure and well cooled. Yes. As also written in https://en.wikipedia.org/wiki/Power_MOSFET: " Body diodes may be utilized as freewheeling diodes for inductive loads in configurations such as H-bridge or half bridge. While these diodes usually have rather high forward voltage drop, they can handle large currents and are sufficient in many applications, reducing part count, and thus, device cost and board space. " Not mentioned here is, that the body diodes have rather bad specifications beside the high forward voltage drop compared to special purpose diodes normaly used as freewheeling diodes. Also snubber circuit's are recommended in many application notes. But all this depends very much on the requirements and parameters of the design/system - special purpose diodes and snubber circuits can be anything from badly need to absolutley superfluous. 2 hours ago, micro said: Or take it the other way round: Braking sharply means that you will have a high current for a moment. That is no problem as the dissipated energy is small. Riding downhill for a long time is something else. I guess on steep hills you have to get rid of some 300W. Our battery has a voltage of 60V, the RMS current is 5A. The voltage drop over the diode with this low current is around 0.5V (I was not aware of this low value before I wrote this reply) Even if this power would flow completely through the diodes, the diode dissipates only 1 % of the brake energy, namely 3W. Right? The 0.5V were for special purpose diodes - body diodes of power mosfets are more in the range of 1.3-1.5V. About 5A RMS current is about my experience, too (http://forum.electricunicycle.org/topic/3589-9b-metrics/?do=findComment&comment=39221 - but these are values delivered from the ninebot itself and noone knows how trustworthy they are...) @esajand imho @Jason McNeilcame with different wheels and better measurment methods imho to breaking powers (regenerative braking power flowing back to the battery) in the range well above 1kW... So this 3W have to be at least 1,5*5=7,5W (~1.5V voltage drop instead of 0.5V) Should still be easy to cool with a heatsink. Just if some peak breaking loads happen (?>30A?) it could get quite thight for the body diode... So imho switching the Mosfet instead of relying on the body diode should be quite advantageous... Also for "freewheeling" it is possible to use the mosfet with his low rdson instead of relying on the body diode ("active freewheeling") But this is all just theoretic - for further thoughts we are still missing the parameters within which a wheel operates... Link to comment Share on other sites More sharing options...
lizardmech Posted May 6, 2016 Share Posted May 6, 2016 1 hour ago, Chriull said: I'd bet that the KS/Gotway get there high pitched PWM sound by commuting the motor in the wrong way (sine vs. trapezoidal)... But i'd assume there is some reasoning behind this, since Ninbot also changed from "pure" sinusodial commuting (absolutely no PWM noise) to a little bit of PWM noise in the last firmware version....? Maybe one finds enough time to work through this master theses he can explain all this to us Buzzing usually means low switching speed is being used. You can switch faster but MOSFETs get hotter and MCU needs more resources. Apparently you can use FPGAs to make motor controllers that can switch up to 100khz vs the usual 10-40khz. I think 40khz is outside what you can hear while they lower frequencies are audible. Link to comment Share on other sites More sharing options...
Chriull Posted May 6, 2016 Share Posted May 6, 2016 11 minutes ago, lizardmech said: Buzzing usually means low switching speed is being used. You can switch faster but MOSFETs get hotter and MCU needs more resources. Absolutely. But ninebot uses something about 8kHz and gotway/kingsong about 6kHz. And there is a huge difference of the PWM sound volume - of course they have higher motor powers then the E+. But it is quite for sure no "buzzing" from some a little bit loose/resonating components - it seems to be the motor making this buzzing sound by the torque ripples. This together with the articles are the reason for my idea with the "wrong" commutation PWM waveform - of course this is just some theoretical assumption from my side until someone makes real measurements... Going much higher with the PWM frequency seems not to be a possible choice with the already exisiting overheating problems - so my "last hope" is a firmware change to get the buzzing to a less annoying volume... Link to comment Share on other sites More sharing options...
electric_vehicle_lover Posted May 6, 2016 Share Posted May 6, 2016 I already analysed 3 different boards of generics that have PWM frequency of about 16khz. I shared pictures of the signal on the project page. Link to comment Share on other sites More sharing options...
lizardmech Posted May 6, 2016 Share Posted May 6, 2016 15-20khz seems normal for BLDC, no way 6khz or 8khz is going to be quiet. Link to comment Share on other sites More sharing options...
esaj Posted May 6, 2016 Author Share Posted May 6, 2016 6 hours ago, Chriull said: Can you somehow estimate the rpms you managed to turn the wheel? I assume strongly that one period of the signal is not one rotation of the wheel (like with 3 coil asyncron motors) - you would have turned it like driving about 20 km/h (with a 16 inch wheel)... So my next idea would be, that the period is a multiple of the rotation caused by the ?36? permanent magnets (the number was imho once reported for a gotway?) - but that would lead to just 0,16 rpm == ~0,7 km/h... So maybe its double of this (just half of the numbers of magnets?) so 0,32 rpm == ~1,4 km/h could sound feasable? ... or it works somehow completely different?... Or your wheel just has a different number of magnets/coils.... I think there are a lot more "electrical revolutions" per real-world (mechanical) round of the motor. Don't know how fast it was spinning, but I doubt it was anything like 20km/h Vee has taken a picture of that same motor (that exact same motor) opened, if you want to count the coils/magnets: http://aijaa.com/Qh533Z 6 hours ago, Chriull said: At least frequency/voltage seem quite linear -so this would be my reasoning behind this question to estimate the back emf once the wheel reaches 20(30) km/h... So if the second thought (18 periods per rotation) and the linear relation is true, this would mean +/-163V once reaching 20km/h - which i cannot really believe? Doesn't sound right, I've ridden above 30km/h (measured with a pretty precise bike computer straight from the wheel rim) on many occasions and for long straights, so the back-EMF cannot be higher than the battery voltage at that point? 6 hours ago, Chriull said: So maybe there is really just one period per rotation (and your were really bravely spinning the wheel ) and the about +/-12V back emf is all one gets at 20 km/h? Sinusodial or Trapezodial is just determined by the way the windings of the coils are distributed along the wheel diameter - according to http://krex.k-state.edu/dspace/bitstream/handle/2097/1507/JamesMevey2009.pdf;jsessionid=856476A06A2D55274A5DA09CC9D06254?sequence=1 - there is a summary at page 300 with an overview of different "rotor-stator flux linkages for different winding functions...". I have to admit that i just very sparsly looked into it - should be imho a great lecture to get an thorough insight into this topic... I also saw some source last night saying that "true" BLDC should produce a more trapezoidal back-EMF, whereas PMSM produces a sine-wave. Then again, the same source said that PMSM does NOT have torque-ripple, so there goes the theory of the PWM-sound being due to torque ripple. Or the source was wrong. Or it's not a PMSM 6 hours ago, Chriull said: On page 57 he states: " The trapezoidal motor can theoretically produce ripple-free torque when driven with ideal rectangular currents, though this is not completely achievable in practice (discussed later). Since the current in a trapezoidal motor is DC over the flat-topped bEMF, since it is switched full-on over the commutation period, and since the polarity of the current is reversed once per electrical cycle, this motor resembles a brush DC motor without the brushes. For this reason it is sometimes called a Brushless DC Motor (BLDC), although claims in the literature that it is simply a brush DC motor “turned inside out” are false, as should be apparent from prior discussion. Further, since the current is controlled by only electronic means, it is sometimes referred to as an Electrically-Commutated Motor (ECM) or electronically switched motor." He also states on page 58: " It is clear that the sine motor is capable of producing ripple-free torque when driven with properly-phased sinusoidal currents. It is sometimes called a Permanent Magnet Synchronous Motor (PMSM) since it is just like the “classic” synchronous motor, with the wound-field rotor replaced by a PM rotor." At page 59 is also a nice overview with about 12 names for these motors... But this is from 2009 and maybe the naming convetions developed in another direction or still stayed as confuse as they were in 2009... I'd bet that the KS/Gotway get there high pitched PWM sound by commuting the motor in the wrong way (sine vs. trapezoidal)... But i'd assume there is some reasoning behind this, since Ninbot also changed from "pure" sinusodial commuting (absolutely no PWM noise) to a little bit of PWM noise in the last firmware version....? Maybe one finds enough time to work through this master theses he can explain all this to us From what I've learned so far, it would seem that sine-wave is much more precise at slow speeds, and you really can't have much "twitching" when riding slow, otherwise it would be very hard I'll have to dig deeper to the different PWM-drive techniques at some point... 5 hours ago, Chriull said: Anyhow - if the two inverse body diodes would be getting conductive in case of regenerative breaking ( so the voltage generated from the motor would have to be at least higher then the battery voltage) without any controller influence this would have imho hindered any EUC from happen... I (still) cannot see how these two inverse body diodes could get conductive with no mosfet switched through - there is no (voltage) potential conection between the battery and the motor in this case... Purely guessing, but couldn't one phase go "below ground", thus causing the low-side mosfet to conduct (from ground to phase), and the high side would then push above the VCC to get the high-side diode conducting? I still need to do more tests with the motor, but regardless of the mains hum, I'm fairly sure that it was the back-EMF I saw last night coming through to high-side rail. Of course it wouldn't otherwise normally (with the motor being battery-driven) happen, but if the bridge(s) and the motor coil form a boost-converter... I still believe they'd open the high-side mosfet to let the coils discharge with less power dissipated in the mosfet, but of course don't know for sure. Of course even then there could be "leakage" through the body-diodes if the motor phase-side voltage is high enough. 5 hours ago, Chriull said: Secondly i almost start to believe, that the voltage generated by the motor (normally?) stays lower than the battery voltage - but for this i am awaiting @esajanswer to the previous post and/or maybe some further experiments from his side (hopefully)... I'll try to get around to do some more measurements tonight. I actually thought that maybe I should try switching the mosfets (without actually powering the motor, ie. "floating" rails) to see how it affects. A few kHz should probably do... And I don't think that the voltage goes above the battery level during "normal" riding, as I've ridden with the Firewheel with a voltage display attached & followed the values. It only goes up during braking. If I could, I'd test with the actual Firewheel mainboard, but the mosfets are SMDs, not TO-220 -cased. 5 hours ago, Chriull said: About 5A RMS current is about my experience, too (http://forum.electricunicycle.org/topic/3589-9b-metrics/?do=findComment&comment=39221 - but these are values delivered from the ninebot itself and noone knows how trustworthy they are...) @esajand imho @Jason McNeilcame with different wheels and better measurment methods imho to breaking powers (regenerative braking power flowing back to the battery) in the range well above 1kW... Well, I did get pretty high-values when I was testing the Wheelemetrics beta version with Vee's MCM2s: It's been said that Gotways have errorneus current measurement, and the actual current would be only half of that... but it would still be in 1-1.5kW range at (short-lived) peaks. 5 hours ago, lizardmech said: Buzzing usually means low switching speed is being used. You can switch faster but MOSFETs get hotter and MCU needs more resources. Apparently you can use FPGAs to make motor controllers that can switch up to 100khz vs the usual 10-40khz. I think 40khz is outside what you can hear while they lower frequencies are audible. Apparently, the mosfet switching-losses go up with switching frequency, as relatively more time is spent passing through the linear-region. Most humans have a hearing range up to around 20kHz (think Nyquist theorem about the sampling rate vs. rebuilding signal and that audio-recording is usually sampled at somewhere around 44kHz), and you lose the higher frequencies the older you get. 4 hours ago, Chriull said: Absolutely. But ninebot uses something about 8kHz and gotway/kingsong about 6kHz. And there is a huge difference of the PWM sound volume - of course they have higher motor powers then the E+. But it is quite for sure no "buzzing" from some a little bit loose/resonating components - it seems to be the motor making this buzzing sound by the torque ripples. This together with the articles are the reason for my idea with the "wrong" commutation PWM waveform - of course this is just some theoretical assumption from my side until someone makes real measurements... Do you mean that the motor would be PMSM, but driven with (more) trapezoidal signal? Or vice versa (BLDC and sine-wave)? Link to comment Share on other sites More sharing options...
lizardmech Posted May 6, 2016 Share Posted May 6, 2016 I think you can still drive BLDC with a sine wave, you just need FOC and the controllers switch to trapezoidal under various conditions, usually start up and high RPM. If the ninebot is much quieter than the kingsong despite the low switching speed it's probably likely more advanced motor control while the kingsong is running standard trapezoid signals throughout its entire range. Anyone know what MCU ninebot E uses? I had a look at pictures it appears to be a lqfp64 chip while the generics mainly use lqfp48 stmf32f1. I couldn't see an external crystal like the STM32 usually have might be a different brand with a totally different motor control solution. Link to comment Share on other sites More sharing options...
esaj Posted May 6, 2016 Author Share Posted May 6, 2016 I think I'll try something along these lines later tonight: <WHOOPS, NICE FAULTY DESIGN THERE > So basically, I switch the NPN's with PWM-outputs from an Arduino, don't remember what frequencies were available, at least something like little under 4kHz and 31kHz? Probably could emulate other frequencies with "bit-banging"... So that switches Phase 1, then every "now and then", I change between switching high-side and low-side (with some wait for DTI). Phase 2 will be tied directly to ground. I planned to leave phase 3 floating... unless anyone has better ideas? For measuring, I'll put scope-probes into "suitable spots", then turn the switching on and try to spin the wheel. My understanding is that this should cause regenerative braking then? And the voltage at the floating high-side drain should go up as the coils (try to) discharge... Should I instead add something as load there? EDIT: Wasn't as straightforward as I had hoped (the PWM-pulses are 5V and I want more volts on the gates to make the mosfets fully conductive), so I'll go ahead and breadboard one half-bridge from this (it's a H-bridge for the robot): The upside is that the voltage pump on the high-side should keep the high-side mosfet conducting even when the phase voltage rises (up to a certain point) Link to comment Share on other sites More sharing options...
esaj Posted May 6, 2016 Author Share Posted May 6, 2016 Ok, I managed to build the half-bridge, which took a lot longer than expected. The low-side worked ok (I tested it with a small brushed motor), but the high-side didn't. I first found a small mistake on the high-side, but even after fixing that, the motor didn't run. After long debugging session, I found out that some of the components and the breadboard itself had bad contacts, so annoying... Anyway, I then quickly whipped up some Arduino code (like in 10-15 minutes or so, the breadboarding took several hours ). In case anyone's interested or otherwise wants to check it for errors or whatever, here it is: http://pastebin.com/zpJ8n0qz Basically, it's driving the PWM at 50% duty cycle and changing the "direction" (high/low side mosfet) every 100 milliseconds (with DTI, first shutting down the PWM, and after wait time, changing the direction & enabling the PWM again). I didn't test it "too much", just that the motor ran (and kept a constant chirp-chirp-chirp-chirp at the PWM-frequency ) and that the voltages were at "sane" levels. Before that, just to be on the safe side, I put zeners between ground and the PWM & DIR-inputs from the Arduino (4.7V for the DIR after the diode, 5.1V for the PWM). I then proceeded to test the motor, but before hooking it up, I wanted to see how shorting the phases together brakes it. Two phases together, it was surprisingly hard to turn the tire. With all three phases together, it does turn a little, but it's surprising how strongly it opposes turning At least I now what it's like if I ever get blown mosfets and the motor resists... Any way, I then hooked the half-bridge and the Arduino to the motor. When the motor's not moving, there's no sound, but it also does that around 4kHz "chirp" when I swung it to turn and stopped pretty quickly. So it looks like the braking "algorithm" works. Then I did the measurements. First shot is with 40V trigger and only from the motor phase vs. ground: Took a few tries to hit that 40V trigger, but yes, it does boost up the voltage... a lot. I couldn't get but about 20v out of it last night (that was with no braking & the mains hum messing up the measurement...). Second shot, added second channel, it's between the "voltage rails". The + rail is actually floating, but it's "behind" the high-side mosfet, so either the voltage has to come through the body diode, or the conducting mosfet: Do note that it's now 10V per grid square. Third shot, upped the trigger to 50V, channel at 20V per square, channel 2 still at 10V Yup, max voltage hit 60V... took quite a few tries (at one point, the wire going between the other motor phase and ground got off, and I almost strained my hand yanking the tire, as there suddenly was no resistance ). Do note that all of the above are at 100µs per grid square (horizontal). I then changed the timescale, here's 20ms per grid square horizontal, 20V per grid square vertical, 40V trigger: The entire time-span here is 240ms, so the Arduino has the time to drive both high- and low-side. Not sure which is which, but I'd bet the higher voltage spikes occur when it's opening and closing the high-side -mosfet. Probably should have checked this by wiring a third channel to the mosfet-gate Another take, with 50V trigger: Max voltage at almost 90V... Some months back, I tried to ask people about the regenerative braking, and how can it work if the back-EMF "normally" doesn't go above the battery level. I was thinking maybe it's inductive spikes, but there's definitely a lot of boostin going here... if those were just "spikes", I don't think they would last around 10ms at a time. Even longer time scale, 100ms per grid square horizontal: I was already having trouble keeping the tire turning fast enough to hit the trigger and get long enough trace... It certainly resists any movement The maximum voltage here is much smaller than before, as I was more concentrated on keeping the tire moving for a longer time than as fast as possible. I then took one last shot with lower trigger and 10V per vertical square to see some more detail: That's it for now. Let me know if you want any more specific measures done, or if you think I should repeat something (like checking the gate-voltages to see whether the higher spikes occur when the upper-side is being switched or not...) Link to comment Share on other sites More sharing options...
Hunka Hunka Burning Love Posted July 15, 2016 Share Posted July 15, 2016 I ran across this TI InstaSpin BLDC motor control video which was interesting to watch. I'm sloooowly learning a few things... Link to comment Share on other sites More sharing options...
Recommended Posts
Archived
This topic is now archived and is closed to further replies.