PDA

View Full Version : PWM signal to motor


Jshwaa
2015.02.07, 11:30 PM
Not sure if anyone has actually seen the signal that a mini-z sends to the motor, and I haven't checked out the mini-z's PWM signal myself until just now, but I put my Tektronix scope across the motor output while throttling at varied amounts, and viewed how the signal changes. Here's what I found...

http://i83.photobucket.com/albums/j314/Jshwaa/Mini-Z/IMG_5796.jpg (http://s83.photobucket.com/user/Jshwaa/media/Mini-Z/IMG_5796.jpg.html)

The mini-z PWM frequency is just a hair above 1KHz, or 1000 cycles per second. Depending on how hard you throttle, the duty cycle changes proportionally. If you throttle a little, the duty cycle is small, meaning that the ON time is short compared to the Period of the signal. Here is the PWM signal at the lowest throttle I could trigger...

http://i83.photobucket.com/albums/j314/Jshwaa/Mini-Z/low_duty_PWM.jpg (http://s83.photobucket.com/user/Jshwaa/media/Mini-Z/low_duty_PWM.jpg.html)

On the right side of the scope display is some measurements on that signal.

The Frequency is 1.024 KHz
The Period is 977 microseconds
The Mean Voltage(average) is 1.96 V (voltage motor sees)
The Peak-to-Peak is 5.08 V(voltage of cell composition), and
The Positive Width is 269.6 microseconds

The duty cycle is then (269.6us/977us)*100% = 27.6%
The Mean Voltage of 1.96V is the product of this duty cycle and the Peak to Peak voltage, or 5.08V*0.276 = 1.401V, which is lower than 1.96V I know, but the reason for that is because of the downward transition from 5.08V to 0V, it is sloped and not sharp which averages out to a higher mean voltage.

The motor by nature will average the signal to a DC voltage. We know this because the PWM signal is not apparent in the motor's output, although you can hear the 1KHz if you listen carefully.

Now, take a look as I throttle halfway, and the positive width time gets higher while the period stays the same.

http://i83.photobucket.com/albums/j314/Jshwaa/Mini-Z/mid_duty_PWM.jpg (http://s83.photobucket.com/user/Jshwaa/media/Mini-Z/mid_duty_PWM.jpg.html)

The positive width went to 710.6us while the period stayed at 977us, so...

(710.6/977)*5.08V = 3.69V, which is pretty close to our mean measurement of 3.98V.

And now for near 'full throttle'...

http://i83.photobucket.com/albums/j314/Jshwaa/Mini-Z/reverse.jpg (http://s83.photobucket.com/user/Jshwaa/media/Mini-Z/reverse.jpg.html)

And full throttle was a flat line 5.08V...

What I found interesting was that a full throttle in reverse did NOT result in a flat line voltage across the motor leads, instead it was just short of a flat line, about the same as near full throttle in forward...

http://i83.photobucket.com/albums/j314/Jshwaa/Mini-Z/hi_duty_PWM.jpg (http://s83.photobucket.com/user/Jshwaa/media/Mini-Z/hi_duty_PWM.jpg.html)

This explains why when you throttle forward, you can get just a tad higher RPM's than reverse. I don't know if that was intentional, but none-the-less it was the result.

LED
2015.02.08, 02:17 AM
Thx, I was looking for this :-)
Have you checked what effect the ICS frequency settings is?
You have a choice of putting the drive frequency in 1.2 Khz 2.5 khz and 5khz.
Does this actually change the drive frequency? I always wondered.
I do not know if the sports has the same ICS settings.

KWT
2015.02.08, 10:24 AM
I think you should've measured the stock output before changing the FETs. Now if you change to a different FET, you would see a slightly different waveform.

Jshwaa
2015.02.08, 10:45 AM
I think you should've measured the stock output before changing the FETs. Now if you change to a different FET, you would see a slightly different waveform.

This is true, and I would be interested to actually see this if anyone was inclined to do that.

However, the drive signal from the micro of the mini-z receiver is typically (0V~4.8V Pk-Pk), so I doubt that the waveform from stock FET's would prove to be much different if at all different. It is when you start using FET's that require a higher Vgs (10V~20V) to turn on that you start deforming the waveform, because the FET's have a slower and limited switching response. They can't turn on all the way before they are being told to turn off, especially at a 1KHz frequency when the gate only has less than 500us to open or close

At any rate, the indicator for running FET's with too high of Vgs is the heating up of the FET, and these FET's seem to run cool. It is my cells and motor that start to cook after running for a bit, which is expected.

Within the small group of FET's that are adequate replacements for mini-z, you should expect a small deviation in the switch response from one FET to the next, but as long as the FET switches 'fast enough', you are fine, and this involves heeding the Vgs parameter of any FET datasheet.

Thx, I was looking for this :-)
Have you checked what effect the ICS frequency settings is?
You have a choice of putting the drive frequency in 1.2 Khz 2.5 khz and 5khz.
Does this actually change the drive frequency? I always wondered.
I do not know if the sports has the same ICS settings.

Really? How exactly do you configure the "ICS Setting"? I am under the impression that the MR-03S elecs are simpler and less configurable than most previous mini-z revisions but I do not know this as it comes from consumer experience. If the method seems feasible I will give it a shot. Going higher in frequency does have diminishing returns though. 1KHz is plenty fast enough if "speed" is the goal you are looking for in your FET switching. Going faster creates the opportunity for 'cross-conduction' in the H-bridge, as the FET's may not open/close completely for a given 'power phase' of the PWM cycle. Trust me, higher frequency in switching does not equate to more power to the motor. If the designers of the MR-03S boards chose 1KHz as the drive frequency, it's because they knew 1.2/2.5/5KHz wouldn't be much better for the effort. I don't know of an engineer that would purposely plant a handicap into the design if it was just as easy NOT to do so, but I am still surprised by the missing 2 FET's on those boards....how stupid of them when the pads are right there. They probably pay $0.0035 a FET, and they have to run two processes where one gets the FETs and one doesn't, not to mention the inventory floor space, etc....anyway...I still think they do it knowing kids are going to attempt a FET upgrade and then come crying back to Kyosho for another board.

arch2b
2015.02.08, 11:02 AM
i have no background in this area but awesome to see it explained at this level. i'm sure others would agree that seeing this level of investigation done with the various options for FETs out there would be very interesting to say the least. no shortage of opinions out there but little of the 'science' shown behind it other than quoting specs.

Jshwaa
2015.02.08, 11:57 AM
i have no background in this area but awesome to see it explained at this level. i'm sure others would agree that seeing this level of investigation done with the various options for FETs out there would be very interesting to say the least. no shortage of opinions out there but little of the 'science' shown behind it other than quoting specs.

I understand where you're coming from. People understand when they connect a battery to a motor, what goes on, but when it comes to a FET being an intermediate connection between the batteries and the motor, people's understanding sometimes seem to cling on to magic and their imaginations. As if the FET is doing something magical to the energy from the battery and converting it to some other form of power for the motor to consume. This is where the terms 'turbo FET' comes from, and the idea that we could someday refer to a chart of a huge list of FETs and be exclusive about our FET choice for a particular race condition, and if you just pick the right FET you're going to stand out in the crowd of mini-z racers. This is just plain false. You can pick a 'good' FET or a 'bad' FET, but the idea of this grey area where each FET has its own paper/rock/scissors effect against another FET or that a FET with higher resistance would be more ideal for a particular race condition is where the hokum starts.

For any FET worth considering as an upgrade to mini-z's (ie N-channel/P-channel complementary, 8-SOIC, mosfet array), the main thing to consider is...'can my mini-z switch this FET on/off'. The mini-z cannot provide any higher Vgs than the voltage you decide to run at (ie 4.8V, 6.6V, etc) This one advantage to raising the operating voltage, along with the obvious others...
The next thing to consider is the capacitance you need to charge/discharge each time the FET is commanded to turn on/off. The Vgs could be low, but the capacitance so big that it takes a while to achieve full charge at the gate to make the switch. I don't pay too much attention to this parameter IF the FET is advertised as a "fast switching" mosfet, because the gate capacitance is the deciding factor in whether or not a FET is considered 'fast switching', so I defer to the manufacturers lawyers for this spec, BUT if anyone was so inclined to perhaps enhance the FET drive signal, it could be achieve through identifying the resistor between the main micro and the FET, or between an intermediate transistor from micro to the FET, and lowering that resistance. This would tax the main micro's output pin or the intermediate transistor, but I would bet the designers were conservative with this resistance so as to keep the main micro or intermediate transistor safe. But back to FET's...

The biggy now is R(ds) or , drain-to-source resistance. When the FET is turned on, the current flows from drain to source. The path is through this R(ds), so obviously you want this to be as low as possible, no if's and's or but's about it. If you turned your light switch on in your house, you wouldn't want a resistor (essentially a small heater) in the path of current through the switch to the light, because it is robbing the light of power, thus not shining as bright as it could. Same analogy with FET's, however the heat accumulates in a FET, and then the advertised R(ds) goes out the window because that spec is at room temperature, so now you have a snowball effect going on where the FET resistance goes up, heat goes up, resistance goes up, etc. until poof, you let the smoke out. If you are lucky to not burn up the FET, then you were still operating at 'higher than optimal' resistance and robbing your motor of power, and creating a nice little heater inside your mini-z for the little imaginary driver that must get cold from time to time :rolleyes:

So, to summarize. Low Vgs, low Rds, Low Ciss (gate capacitance) or 'fast-switching' description. There is no grey area. If you have a FET that fits the package style, and absolutely optimizes to those parameters, you have the best FET possible. Throw the FET charts away.

LED
2015.02.08, 12:10 PM
Really? How exactly do you configure the "ICS Setting"?

well, again, I do not have any experience with the sports version so it is very much possible that the sports does not have ICS. An MR02 and MR03 do have it. It simply a number of settings you can do in the car itself.
Take a look at http://www.mini-z-guide.com/ASF.htm
Altough it is not up to date for the MR03 you get the point.
Also look at http://mini-zracer.com/forums/showthread.php?t=34181
There also some threads here for making your own cable or bleutooth dongle. Should be very simple for you :-)

Anyway, I always doubted that the frequency settings in the ICS actually did something. But I do not own a scope to check myself.

Solo1
2015.02.08, 12:17 PM
OK so i'm a complete dummy to all this hi tech talk (more or less) could you tell me in laymans terms why 4 fets are better than 2 and what effect this would have on power of the car or driveability?

LED
2015.02.08, 12:27 PM
OK so i'm a complete dummy to all this hi tech talk (more or less) could you tell me in laymans terms why 4 fets are better than 2 and what effect this would have on power of the car or driveability?

Basicly, one FET does not provide enough current to drive the motor. That's why they put in 4.
Look at as it were tunnels. You can only pass a certain amount of cars through a tunnel in lets say one minute. If you have 4 tunnels next to each you can pass 4 times as many cars. Now image if you make your tunnels 2 lane instead of 1 lane you pass even more cars.
So a standard FET is like a single lane tunnel, and the upgrades are like a dual lane.
Now does you motor require 4 dual lane tunnels? Stock motors and PN from 50T to 80T, no. Anything below 50T and it is better to upgrade.

Now imagine that the road leading upto the tunnels is only suited for 4 dual lane tunnels. If you put in 6 dual lane tunnels 2 of them will never be used because the supply is lower. But the lights in the tunnel are still working; So you are actually using power for something you do not need.

Ok there is alot more science behind it, but I hope this makes it clear for you :-)

Solo1
2015.02.08, 12:38 PM
Basicly, one FET does not provide enough current to drive the motor. That's why they put in 4.
Look at as it were tunnels. You can only pass a certain amount of cars through a tunnel in lets say one minute. If you have 4 tunnels next to each you can pass 4 times as many cars. Now image if you make your tunnels 2 lane instead of 1 lane you pass even more cars.
So a standard FET is like a single lane tunnel, and the upgrades are like a dual lane.
Now does you motor require 4 dual lane tunnels? Stock motors and PN from 50T to 80T, no. Anything below 50T and it is better to upgrade.

Now imagine that the road leading upto the tunnels is only suited for 4 dual lane tunnels. If you put in 6 dual lane tunnels 2 of them will never be used because the supply is lower. But the lights in the tunnel are still working; So you are actually using power for something you do not need.



Ok there is alot more science behind it, but I hope this makes it clear for you :-)


Yes thanks! So running a 50 or 70 t motor i'm better off staying with 2 instead of 3 or 4 as i would be wasting power that can go to the motor, thats what i needed to know i wont be adding any more then as i doubt i could handle more power than a 50 even with the 70 and the new radio i have to be careful not to spin the tires coming out of corners and have been working on the abs to get it slowed down going in without sliding!

Jshwaa
2015.02.08, 12:43 PM
OK so i'm a complete dummy to all this hi tech talk (more or less) could you tell me in laymans terms why 4 fets are better than 2 and what effect this would have on power of the car or driveability?

Good question Solo, and LED provided a very good analogy for this, but let me confuse you some more. :)

I believe your question is asking, why add FET's at all, and what's the benefit?

It would take a schematic of the circuit to make things clear, but when you populate the missing FET locations on a MR-03S board, OR do an operation known as 'stacking' FET's, what you are doing is putting additional paths of current in parallel.

Imagine if you had a huge power source and a huge load, like say your 120V outlet in your house, and a vacuum cleaner. You wouldn't want a skimpy path for current, like the gauge wire you'd find in a ribbon cable. That current would cause that skimpy wire to get hot, possible melt, and quickly fail. To prevent that, you could add a LOT of those skimpy wires in parallel and improve the survivability of the circuit. Each additional wire in parallel is lowering the resistance of the sum, therefore dropping the power dissipated across that section of wire, until the resistance is insignificant like the heavy gauge wire you would actually use for a vacuum cleaner.

Same is true for adding FET's. It provides an adequate path for current under the heavier demands of higher voltages and lower turn motors.

It's easy to do a FET, cell, and motor upgrade at the same time and attribute the performance to your FET's, however the FET's are silent servants that are either providing that path for current as equally as your wiring, or they are a little more resistance than your wiring and are seen as a little resistor in the path of current between battery and FET, and dissipating heat proportionally. They don't boost, enhance, or add energy, they can only take away.

So if you were hesitant to putting a FET upgrade in your car because you fear it would give you too much power, that is hokum. It is the battery, motor, and gear composition that is causing you to lose dynamic friction to the ground. Just tame the motor and gearing, and your FET's will all of a sudden seem less potent.

I may do up a schematic of the elecs which should makes things a little more clear....or not.

Solo1
2015.02.08, 04:09 PM
I think i get what your saying and i think it goes back to if i only use a 50 70 or 80 turn motor 2 fets should be enough and if i were to add 2 more in may in fact take away a tiny amount of power (voltage) from the motor, but would be the thing to do if i were to use a 30 or 40 turn motor so as not to over heat the fets as the lower turn motor is drawing more power thru them than 2 could handle without burning out.

Jshwaa
2015.02.08, 04:18 PM
Here is a pic of the top-side of my MR-03S board for reference...

http://i83.photobucket.com/albums/j314/Jshwaa/Mini-Z/DSCN2274.jpg (http://s83.photobucket.com/user/Jshwaa/media/Mini-Z/DSCN2274.jpg.html)

In this pic you can see the FET's as well as the component in the middle labled 'TG68815'. That is the mini-z microcontroller. This component provides all of the switching to the FET's, as we will see in the schematic below...

http://i83.photobucket.com/albums/j314/Jshwaa/Mini-Z/motordrive.jpeg (http://s83.photobucket.com/user/Jshwaa/media/Mini-Z/motordrive.jpeg.html)

The part named IC1 is the TG8815. From there you can see that the N-channels are driven straight from the micro, through 1K ohm resistors. The P-channels are driven from NPN transistors with 10K ohm pull-up resistors at the collectors. And then you can see all the Drains go to the motor connections, and the Sources go their respective B- or B+ terminals. This is your mini-z motor drive circuit for better or for worse. For those of you that like to differentiate mini-z's from xmods....xmods motor drive circuit was exactly the same with the exception of maybe the component manufacturers, part numbers, and of course the xmod PWM frequency was not as high as 1KHz. If I remember correctly, it was around 300Hz.

Referring back to the PCB pic... the two tiny black parts in between FET1 and IC1, and underneath the orange rectangle, are resistors R1 and R2.

Now to the bottom-side...

http://i83.photobucket.com/albums/j314/Jshwaa/Mini-Z/DSCN2278.jpg (http://s83.photobucket.com/user/Jshwaa/media/Mini-Z/DSCN2278.jpg.html)

The little black parts that sit right at the source and gate of the P-channels are pull-up resistors R3 and R4. The transistors Q1 and Q2 are the 3-legged black components that are near top middle of the pic. And R5 and R6 are tiny black components near those transistors.

Now, back to what I was saying about optimizing the PWM signal to the gates of the FET's...

If you wanted to create stronger switching and better ensure that the gate charge is being met for a heftier gate capacitance, you can lower the values of R1, R2, R3, and R4. This will provide more current to the gates and the PWM signal will be smacking the gates harder, and opening the channels wider, thus achieving that R(ds) on resistance you so dearly need.

Solo1
2015.02.08, 04:34 PM
do you know and could you show me which components are for the receiver?

Jshwaa
2015.02.08, 04:37 PM
I think i get what your saying and i think it goes back to if i only use a 50 70 or 80 turn motor 2 fets should be enough and if i were to add 2 more in may in fact take away a tiny amount of power (voltage) from the motor, but would be the thing to do if i were to use a 30 or 40 turn motor so as not to over heat the fets as the lower turn motor is drawing more power thru them than 2 could handle without burning out.

Not exactly. If the FET is good for you to use with hot motors, it can't hurt to use with any cooler running motor. Just don't expect a FET upgrade alone to do anything for you.

Ok, taking another look at the datasheets for an Si4562, here is a little excerpt to take into consideration as it pertains to switching speed. The faster the switching the better, for obvious reasons, but especially when you are changing directions and the opportunity arises for the FET's to be in a state where you're conducting current straight through the bridge to ground, and not through the motor, thus shorting your cells through your FET's for nanosecond intervals of time, and the heat dissipated from those occurrences can cause FET's to fail in high current setups.

http://i83.photobucket.com/albums/j314/Jshwaa/Mini-Z/4562speed.jpeg (http://s83.photobucket.com/user/Jshwaa/media/Mini-Z/4562speed.jpeg.html)

Notice the spec labeled "Turn-On Delay Time" and "Turn-Off Delay Time".

For the N-channel mosfet in a Si4562 it takes 40nS to turn on and 90nS to turn off.
For the P-channel mosfet in a Si4562 it takes 27nS to turn on and 95nS to turn off.

http://i83.photobucket.com/albums/j314/Jshwaa/Mini-Z/DMC1017Nspeed.jpeg (http://s83.photobucket.com/user/Jshwaa/media/Mini-Z/DMC1017Nspeed.jpeg.html)
http://i83.photobucket.com/albums/j314/Jshwaa/Mini-Z/DMC1017Pspeed.jpeg (http://s83.photobucket.com/user/Jshwaa/media/Mini-Z/DMC1017Pspeed.jpeg.html)

For the N-channel mosfet in a DMC1017 it takes 6.9nS to turn on and 70.3nS to turn off.
For the P-channel mosfet in a DMC1017 it takes 10.6nS to turn on and 144nS to turn off.

Overall the DMC1017 is faster, but the 144nS P-channel off time is kind of steep. I'm not certain if this is causing any difference in performance as compared to stock, but I highly doubt it was for the worse, as I am using a 32T and so far so good. I tested the circuit pretty good by stalling the motor and throttling to induce max current and the FET's stay cool. I believe what the elecs do, although I haven't verified this with a scope, is that they lock the P-channel in the on state, and then PWM the N-channel to vary the output voltage, so that the P-channel off-time does not become an issue at all. The switching time is therefore dictated by the N-channel on/off time alone. This makes the comparisons between the Si4562 and DMC1017 all about their N-channel switching response, and again...

For the N-channel mosfet in a Si4562 it takes 40nS to turn on and 90nS to turn off.
For the N-channel mosfet in a DMC1017 it takes 6.9nS to turn on and 70.3nS to turn off.

Which one do you think is your better option if you are looking for switching response? Hmmm.....

But I'm also happy with the DMC1017 R(ds) and I(d) specs. The kicker is the broader body of the FET, which I believe the secret for the higher wattage spec.

Jshwaa
2015.02.08, 06:06 PM
do you know and could you show me which components are for the receiver?

http://i83.photobucket.com/albums/j314/Jshwaa/Mini-Z/DSCN2274.jpg (http://s83.photobucket.com/user/Jshwaa/media/Mini-Z/DSCN2274.jpg.html)

In the lower left of the above pic, you will find the receiver module. They compacted the entire radio into the square IC labeled "A 7105 AMR23910"

And then on the bottom-side...

http://i83.photobucket.com/albums/j314/Jshwaa/Mini-Z/DSCN2278.jpg (http://s83.photobucket.com/user/Jshwaa/media/Mini-Z/DSCN2278.jpg.html)

The part on the lower right with the white goo on it is a trimmer inductor for tuning the receiver for a specific frequency band. The receiver IC can then switch channels within that frequency band on command so that you can operate multiple mini-z's together without having to mess around with crystals. Oh and on the upper-left is the receiver IC's local oscillator. Forgot to take the opportunity to measure that frequency, but it is the receiver IC that uses a trick called 'phase lock loop' to multiply that frequency in order to be able to demodulate the 2.4GHz carrier and extract the command information.

Solo1
2015.02.08, 06:22 PM
Thanks! nothing like learning something new!:D