Electric Motor Secrets

[Jetijs]

Hi UncleFester
Thanks for the explanation. I have some transil diodes across the source and drain of each MOSFET for overvoltage protection so everything higher than 400V (the MOSFET max voltage) is absorbed by them unless there is another path for the spike to go, a charging battery for example or the input capacitor.
Also it is a good suggestion about the magnetite core because I can easily cnc a good mold for the core and the result would be a solid core. It is a pain in the lower back to clean all those hundreds of plates, scrape the burnups off and glue them together so that they don't twist and so on. Also the machining would be much easier because I would not have to worry about accidentally dislodged plates. The magnetite core would be a lot easier for me to make. But then again the magnetite cores do not make as good electromagnets as silicon steel does, at least that is what my experiments show and we want here the maximum torque possible. Or am I wrong here?

[UncleFester]

Mosfet characterstics

Jeti,

RDSon is gate on resistance, which is critical when trying to minimize power usage and thus heat. The gate on a mosfet (and also IGBT) will want to be as high as possible to minimize switching losses. We want at least 12V on the gate and even 14-15Volts is ok as long as we dont exceed 20Volts. RDSon for that mosefet appears to be .20 Ohm which is good. We need to see what our peak to peak voltage is on the gate first. Mosfets will get hot if we don't get close to or exceed 12VDC on the gate signal. Most engineers would tell you that you want at least 10VDC on the gate or higher.

Also mosfets burn power in the form of heat when they are being exposed to transients as well. It appears from your scope shots the transients created from switching these coils are not massive (I am assuming the scope shots are 10V/Div). A mosfet will happily switch near its maximum amperage, or voltage, but not both at the same time. For instance, if we are switching 100VDC @ 1.5 Ampere we are ok on the IRFP360, but if we take that to 250VDC @ 12.5 ampere or more we are pushing the limits. The mosfet will get hot when it see large transients even if we are only switching a tenth of it's current carrying capacity.

Fortunately it looks like we are well within range on this device for switching those coils at 12VDC @ 3.5 Ampere as long as the transients don't exceed 250 Volts or so. It might be good to measure these transient peaks to see where we are.

It could also be that we have saturation of the core of each coil before the mosfet turns off. It would be good to test this as well. I have been using black sand cores for my Bedini, Gray and Muller projects because of the better saturation characteristics and the fact they cannot be permanently magnetized, and they are non-conductive. They can also be molded into any shape you like, and they are lightweight and cheap to make. You could easily use your CNC router to create a mold for pouring black sand (fe304) and epoxy or polyester resin mixes.

Tell me if you need help with this. I can send lab grade fe304 and send pictures of how I form these cores for my projects.

Keep up the good work!

[Jetijs]

Will do that tomorrow
There is another thing that doesn't give my mind a rest. Why is it that when I adjust the dutycycle the current waveform gets elevated? I mean this is how the current waveform should look like:



But when the duty cycle is increased the waveform starts to look more like this:



The same thing happens also if I take a scope shot from a low resistance resistor on the recovery line, like this:



I mean if there is no current flowing, we get the flat, horizontal line. When the current starts to flow, it rises to a certain point till the core is magnetized, like in the first picture and then the power is removed. So everything clear there. But what is that "elevation"? Seems to me that current shoots up almost instantly and then just starts to increase in an angle. But if the angle line is the current rise time, what is that vertical line?

[lighty]

It seems to me that when rotor is in alignment relative inductance of coils rise and what you get is basically a choke. And by their nature choke tend to lengthen current rise time and to produce more linear current curve. Also higher inductance would account for higher inductive collapse voltage spikes and better recovery.

Could you measure inductance of coil when rotor is in alignment and when rotor is not in alignment?

[Jetijs]

here is the video:
YouTube - Current waveform behavior

BTW, at certain high frequencies and using one of the V3 motors with the same coil wiring and voltages I managed to get 79% recovery.

[Jetijs]

UncleFester, I have not checked the gate signal and what is the RDSon?

Anyway, I did some experiments today and collected some data.
I used my V2 motor for testing. I used only one set of coils wired in series. There was only one MOSFET driving those series coils. For pulsing I used a separate optoswitch and I pulsed its LED with a signal generator. Here is the test data:

Pulsing at 70Hz 50% dutycycle square wave
1) the rotor is in alignment with the stator
current taken from power supply - 0,92A
current running through the circuit - 1,50A
recovery - 38%
The current waveform:


2) the rotor is NOT in alignment with the stator
current taken from power supply - 3,2A
current running through the circuit - 2,3A
recovery - 3%
The current waveform:


Pulsing at 275Hz 50% dutycycle square wave
1) the rotor is in alignment with the stator
current taken from power supply - 0,3A
current running through the circuit - 0,6A
recovery - 50%
The current waveform:


2) the rotor is NOT in alignment with the stator
current taken from power supply - 2,85A
current running through the circuit - 3,03A
recovery - 6%
The current waveform:



Pulsing at 2,5KHz 50% dutycycle square wave
1) the rotor is in alignment with the stator
current taken from power supply - 0,1A
current running through the circuit - 0,2A
recovery - 50%
The current waveform:


2) the rotor is NOT in alignment with the stator
current taken from power supply - 0,43A
current running through the circuit - 1,03A
recovery - 41,7%
The current waveform:


Pulsing at 11,3KHz 50% dutycycle square wave
1) the rotor is in alignment with the stator
current taken from power supply - 0,35A
current running through the circuit - 1,11A
recovery - 68,4%
The current waveform:


2) the rotor is NOT in alignment with the stator
current taken from power supply - 0,36A
current running through the circuit - 1,07A
recovery - 66,3%
The current waveform:



Pulsing at 18KHz 50% dutycycle square wave
1) the rotor is in alignment with the stator
current taken from power supply - 2,5A
current running through the circuit - 3,44A
recovery - 72%
The current waveform:


2) the rotor is NOT in alignment with the stator
current taken from power supply - 2,6A
current running through the circuit - 3,5A
recovery - 74,2%
The current waveform:



The power supply voltage was always the same - 12V. Interesting that at low frequencies, the circuit consumes way more current when the rotor is not in alignment with the stator and also the recovery is way down when the rotor is not aligned with the stator. But as soon as we increase the pulse frequency, the both currents (when rotor is and is not aligned with the stator) become almost equal and no matter is the rotor is or is not aligned with the stator, the current consumption stays almost the same ans the recovery also almost the same. At frequencies higher than 18Khz, the infrared LED of the optoswitch is staying ON all the time because it can not fully turn on and off at that speed, even if the duty cycle is reduced.
Also, the duty cycle adjustment is important because it will change the current draw and recovery percentage. I will upload a small video so you can see how the current waveforms behave if frequency, duty cycle and rotor alignment is changed.

[UncleFester]

So is the problem that you are getting current to flow through the primary coil even after the mosFET is turned off?

I am wondering, if when you turn turn off the mosFET, a large voltage manifests on the primary coil and causes your protection diode to be reverse biased? This might account for the current continuing to flow after you have turned off the mosFET.

-Chris Corkum

What is the actual peak to peak on the gate signal? What is the RDSon for the mosfets you are using? I notice you are using an inverted gate signal driver. Either one of these could be a clue. These mosfets should be very cool at that current unless there is a gate, duty cycle, or low voltage issue on the gate.

[Jetijs]

Thanks Eric
I will think about what you said.

Chris, the strange waveform is only in the MOSFET ON time, you can clearly see where the pulse starts and where it ends. It is this ON time when the current behaves like this.

[LowTechIsCool]

Just a thought

So is the problem that you are getting current to flow through the primary coil even after the mosFET is turned off?

I am wondering, if when you turn turn off the mosFET, a large voltage manifests on the primary coil and causes your protection diode to be reverse biased? This might account for the current continuing to flow after you have turned off the mosFET.

-Chris Corkum

[Eric]

nice work!!

Hi jetijs!

teriffic work!! allways an inspiration to me as well!
i have been stuck with working my job to save up some money for awhile
now i am back to working on my ver 3 motor it too has 2 bifillar coils
only i chose to go a more mechanical route in the switching method. but i will get back to my motor at a later time.

as for yours i think the engineering is great!
i am curious if we could compare scope shots from accross your emitter and collector of your main driving transitor. you might need to use both bands on your scope. 1 for the driving coil and 2 for the recovery coil but be carefull!! i learned the hard way that because both bands in a scope have a common ground they can short!! so you might need something like a 1 to 1 transformer to isolate 1 of the bands.

the way the picture of your current flow rose and then dipped back down looked a lot like the forth scope picture i took of my first motor across the emitter/collector found here Eric's page, Photos this is a shot i took while loading the shaft down with my fingers. notice how the vertical also dips down. normally this motor and circuit run at a stable room temp, no noticeable temp increase. but when i get that dip on the scope the tranny and resistors get pretty hot. i have learned in the case of my motor that the on time was set too long. so i shortened the on time to fix it. i am curious if something similar is going on with your motor. it seems to me that your circuit shouldn't be getting so hot in the first place. doesnt that mean that too much continuous amperage is running through your circuit? i would think that if you are only getting 17% recovery and the amount of recovery is based on the impedance of your coil as the field collapses, a rather stable amount regardless of how much your on time might be held on for to long, since the on time is based on V=IR i would think you need to dial back the on time to balance it (roughly) with your recovery amount.

hope that helps!!
Eric

[Jetijs]

Ok lighty, here is the scope shot of the MOSFET gate signal:



Looks like a perfect square wave. But the thing is that I have one MOSFET driver switching two MOSFETs in parallel. Maybe the problem is there? But then again the same happens if I connect the coils in series and use only one MOSFET. And what do you mean with "driven slowly through linear region"? These MOSFETs have heat sinks on them, but nevertheless they heat up so much that it is hard to hold a finger on the heat sinks even if only 3.5A is flowing through the circuit. And that is when they are connected in parallel so only 1.75A is flowing through each MOSFET.

EDIT: I took a scope shot of the current waveform of my V2 motor with the V2 circuit. There is also something wrong with it:


I don't get it. What is happening?

EDIT2: I put the motor away and connected an ordinary bifilar coil with welding rod core to the V2 circuit. This works just fine and I get the correct current waveform and recovery up to 70%.



Could it be that when the rotor comes in alignment to with the stator, the current consumption goes down and thus the weird waveform?

[lighty]

Another thing- if the MOSFET's are driven slowly through linear region they will heat up significantly. Could you also check out if they are heating up significantly more than in the v2?

[Jetijs]

ok, will do that tomorrow

[lighty]

It could be a number of things. Could you take scope shot of input signal to MOSFET's gate? If the signal is square than it's not the driver part and we can troubleshoot further.

Also, you significantly raised coils impedance in v3 so it could account for slower rising time.

[Jetijs]

Ok, made some tests today. But I did not get far, because when I got the scope shot of current waveform, I noticed that there is something wrong with it. Look for yourself:



For this test I only connected one pair of the coils to power so that only one phase was working. This is on my V3 motor with 55-60 turns of 0.7mm thick wire coils. Both coils were connected in parallel configuration with a MOSFET for each coil. The waveform was taken from a shunt that is connected in series of the power cables for one coil. The motor runs very fast - about 8k RPM, but the recovery is very bad and judging from the scope shot we can see why. There is something wrong either with the circuit or somewhere else. Has anyone an idea what could cause such a current waveform?
Will test this some more.

EDIT: just wired those coils in series. No change, the current waveform looks the same. No matter what I do with the recovery coils, the input current waveform does not change.