calculate the amount of current required to drive mosfet gate

[nour]

I know there is a lot of variables
basically I want to make a generic (sort of) bipolar stepper motor driver\controller
Max 50V\20A, min 2V for the motor
it should have microstepping (down to 1/16)

I am now trying to choose the mosfets and I will choose all of them N-channel and go with High\low side driver --->IR2101 ---> 210 mA

I think it is not that hard (I suppose) to choose the mosfets, basically low on resistance, low gate charge, high power dissipation as possible
the problem that I really don't have enough experience to say for sure that this particular driver will deliver enough current to charge the gate quickly enough at higher switching rates

can someone help me to decide about the driver or choosing the right mosfet for the thing

I am planning to make 4 drivers\controller (I am making CNC  )


EDIT:
Lets say I will use the IRFZ44 for the fets

[uncle_bob]

Hi

Ok, we have part of the info.

A mosfet gate looks like a capacitor. It's a really weird cap, but a cap none the less. It takes essentially zero current to turn it on or off at DC. The only thing you need current for is to charge / discharge the gate capacitance (and the other strange stuff from feedback). The faster you switch, the more current you need. The more voltage on the FET, the more feedback. More feedback means more C to charge or discharge.

So:

How fast is your stepper?

Bob

[nour]

So:

How fast is your stepper?

Bob

 
I really don't know how fast, as I have said, I want it to be generic as possible
which I think this is one side of the problem
the other side is I have no Idea about the specs of a fast stepper motor how fast it can spin (switching rate in the H-Bridge)
can we say we may want the switching up to 500 Khz max (Pure guessing) I hope that we won't need higher rates than 500K

[uncle_bob]

So:

How fast is your stepper?

Bob

 
I really don't know how fast, as I have said, I want it to be generic as possible
which I think this is one side of the problem
the other side is I have no Idea about the specs of a fast stepper motor how fast it can spin (switching rate in the H-Bridge)
can we say we may want the switching up to 500 Khz max (Pure guessing) I hope that we won't need higher rates than 500K

Hi

Well, you can buy off the shelf (dirt cheap) drivers for normal steppers. There isn't much justification making a custom driver for a normal part. If you do have one that steps 1000X faster than anything on the planet, then indeed you will need a custom driver for it. You also will need a bit of data on the inductance of the windings and a FET that will handle a *lot* (as in a few hundred to a few thousand) of volts.

The same thing applies to the driver current. There are a *lot* of drivers out there. With the "right" combo of output FET / speed / driver you are ok (say) at 1/10 the output current as the limit on the driver. Pick a combo that is not well suited to the frequency (to high) and a driver that runs as much current as the output may not do the job (the output device will still melt). Go to the other side (frequency really low) and you can get away with 1/100 or even less drive current.

How to figure it all out?

Simulatin.

This is why LTSpice exists. You model up your motor windings and grab the model for your output FET. Pick a driver (with a model) and a frequency. Click the mouse and you have the data. Sweep frequency and you have a plot of how well this combo does. Change any part in the mix (or the supply voltage or the current) and it all changes.

Bob

[nour]

I am sure that the ready made off the shelf modules would be tested and working just fine

well, believe it or not, building them is much cheaper than buying them specially I am going to build almost 8 (other 4 for bldc)
the idea is to be able to build cnc prototype as cheap as possible (one that I can do something with  ) using some old dot matrix printer motors to try to understand the difficulties that I might face
after successfully be able to run it without major problems, I will buy new stronger motors to build a bigger one

I have most of the parts already, only the drivers not available
so I want to have good driver from the beginning that can handle smaller and bigger loads without the need to build new one for the other bigger motors
I don't know how bigger I will buy later in the future (that's my problem now)

maybe I am over thinking about it, but that's the way I am seeing things right now
can I use LTSpice to simulate h-bridge built with mosfet and being drived with gate driver ics
sorry never used it before, maybe it is a good idea to use it(hope it is easy to learn and quick)

[uncle_bob]

Hi

Ok, if you are making the stuff from scratch for less than the $1 integrated drivers cost that's fine.

Bob

[PChi]

As a first approximation the MOSFET manufacturers normally quote a Total Gate Charge figure, Qg. So for a particular MOSFET and a given switching time it's possible to calculate the current required.

[nour]

Hi

Ok, if you are making the stuff from scratch for less than the $1 integrated drivers cost that's fine.

Bob

what 
weren't you talking about something like that


It seems like I haven't read carefully, I thought you are talking about off the shelf ready made motor driver like in the image above 
it seems like you are talking about the the small integrated circuit rather than the module in the image

please confirm it 

[nour]

As a first approximation the MOSFET manufacturers normally quote a Total Gate Charge figure, Qg. So for a particular MOSFET and a given switching time it's possible to calculate the current required.

is there a mathematical model for this relationship (hopefully easy to understand and use)?

[uncle_bob]

As a first approximation the MOSFET manufacturers normally quote a Total Gate Charge figure, Qg. So for a particular MOSFET and a given switching time it's possible to calculate the current required.

Hi

If you dig into the Qg numbers they get pretty detailed on some of the data sheets. Different numbers for different operating regions of the device and lots of graphs. If you have a device like that, the data sheet may get you close. On a typical part that has a single number / single condition .... it's rarely the condition you are operating in. Does it still apply .. maybe. Is it close ... maybe. Is it better than no number at all - yes indeed.

Bob

[uncle_bob]

Hi

Ok, if you are making the stuff from scratch for less than the $1 integrated drivers cost that's fine.

Bob

what 
weren't you talking about something like that


It seems like I haven't read carefully, I thought you are talking about off the shelf ready made motor driver like in the image above 
it seems like you are talking about the the small integrated circuit rather than the module in the image

please confirm it 

Hi

The gizmo in the picture has a lot of stuff in it.

The question asked so far has been how to connect a driver to a MOSFET. In the context of a simple stepper driver, there are indeed $1 IC's that do the whole job.

It would have been much easier if the original question included a bit more detail.

Bob

[nour]

ok, you are right, stuff mixed  up in my mind
anyway what details would you like to know?

I will try my best to minimize the range of selection

let's say this is the motor that I may use in the future
http://www.tme.eu/en/details/103h7126-1740/electric-motors/sanyo-denki/

it is rated at 24V\4A (let's imagine it is 4A per phase not for 2 phases to have a good safe margin)

what else should I consider!

[uncle_bob]

ok, you are right, stuff mixed  up in my mind
anyway what details would you like to know?

I will try my best to minimize the range of selection

let's say this is the motor that I may use in the future
http://www.tme.eu/en/details/103h7126-1740/electric-motors/sanyo-denki/

it is rated at 24V\4A (let's imagine it is 4A per phase not for 2 phases to have a good safe margin)

what else should I consider!

Hi

Inductance is your enemy when trying to get a motor to step fast. A "24V" part may need many times that voltage to get 4A running through it. In some cases that gets you to a couple of hundred volts in a hurry. There are various ways to do this or that and get the voltage down a bit.

Are you trying to do this off a 24V supply?

If this is headed to a 3D printer ... anything over 48V is a bit crazy safety wise.

Bob

[PChi]

As a first approximation the MOSFET manufacturers normally quote a Total Gate Charge figure, Qg. So for a particular MOSFET and a given switching time it's possible to calculate the current required.

is there a mathematical model for this relationship (hopefully easy to understand and use)?

As a start the the current required = Qg / Switching time.
Beyond that it's probably time to get the components out and soldering iron hot.

[nour]

Hi

Inductance is your enemy when trying to get a motor to step fast. A "24V" part may need many times that voltage to get 4A running through it. In some cases that gets you to a couple of hundred volts in a hurry. There are various ways to do this or that and get the voltage down a bit.

Are you trying to do this off a 24V supply?

If this is headed to a 3D printer ... anything over 48V is a bit crazy safety wise.

Bob

it is milling cnc machine
for my prototype machine it will be for milling wood (small areas) and doing pcb milling
later after finishing this prototype I will upgrade it most likely with motors as the ones I have provided in my last post

for my prototype I will use 12v for power supply but of course if it require more I have supplies up to 38V\3A or 48V\2A

There are various ways to do this or that and get the voltage down a bit.

if you have something to read about those ways I would appreciated(or at least the name of the process to search about it)

[nour]

MOSFET part number? Intended switching voltage? Hard or soft switch?
Expected efficiency? Intended di/dt and dv/dt? EMI/EMS requirements?

Gate drives are not that innocent and simple as they look like. I strongly recommend you to read TI SLUP169.

Part number: IRFZ44N
Intended Switching voltage:3 up to 50 volts
other parameters I really don't know(no experience, still beginner)
by the way can't find 169 app note

[uncle_bob]

Hi

Inductance is your enemy when trying to get a motor to step fast. A "24V" part may need many times that voltage to get 4A running through it. In some cases that gets you to a couple of hundred volts in a hurry. There are various ways to do this or that and get the voltage down a bit.

Are you trying to do this off a 24V supply?

If this is headed to a 3D printer ... anything over 48V is a bit crazy safety wise.

Bob

it is milling cnc machine
for my prototype machine it will be for milling wood (small areas) and doing pcb milling
later after finishing this prototype I will upgrade it most likely with motors as the ones I have provided in my last post

for my prototype I will use 12v for power supply but of course if it require more I have supplies up to 38V\3A or 48V\2A

There are various ways to do this or that and get the voltage down a bit.

if you have something to read about those ways I would appreciated(or at least the name of the process to search about it)

Hi

On a reasonable sized CNC, higher voltages are fairly common. 24V motors and multi hundred volt supplies with dedicated cage enclosures don't take up a lot of room compared to the machine it's self.

One way to get higher voltages is to use the inductance of the motor as part of boost switching power supply. Essentially you store the energy and re-use it. If you are thinking of designing an ASIC as part of this, that sort of thing is quite practical. Without at least an FPGA, you will run out of "stuff" to get the job done pretty quick.

The quick and simple way is to just use the high voltage supplies and move on. (or use a packaged controller ....). One clue of "how high" is to take a look at the torque curves on your motors. If at the bottom of the chart they mention "+/- 400V driver used", that is a clue about what your 12V stepper may need to hit the numbers they are showing. How *much* of the 400V got used ... who knows.

Bob

[nour]

Hi

On a reasonable sized CNC, higher voltages are fairly common. 24V motors and multi hundred volt supplies with dedicated cage enclosures don't take up a lot of room compared to the machine it's self.

One way to get higher voltages is to use the inductance of the motor as part of boost switching power supply. Essentially you store the energy and re-use it. If you are thinking of designing an ASIC as part of this, that sort of thing is quite practical. Without at least an FPGA, you will run out of "stuff" to get the job done pretty quick.

The quick and simple way is to just use the high voltage supplies and move on. (or use a packaged controller ....). One clue of "how high" is to take a look at the torque curves on your motors. If at the bottom of the chart they mention "+/- 400V driver used", that is a clue about what your 12V stepper may need to hit the numbers they are showing. How *much* of the 400V got used ... who knows.

Bob

WoW  ASIC and FPGA
that seems pretty advanced for my level I am still learning how to use mcu 

my intention is to use arm mcu particularly I will use atmel ATSAM3S --> 64MHZ with cortex M3 ARM very cheap around 2.6$
what you have mentioned about using the motor inductance as part of the boost switching process, I haven't actually heard about it and don't know anything about it(as I have mentioned still learning)

I will search about this concept and try to read more about it
I have a simple question (it might seems easy for your level)
if a bipolar stepper motor rated at 24v how on earth could it be powered with a couple of hundreds of volts 
for example by using BWM or current limit at higher voltage and why?

[AlfBaz]

by the way can't find 169 app note

found it
http://www.radio-sensors.se/download/gate-driver2.pdf

[nour]

by the way can't find 169 app note

found it
http://www.radio-sensors.se/download/gate-driver2.pdf

oh thanks for the search
couldn't find anything in the document saying it 169
if I have came across this file I wouldn't know it is the 169 
anyway thanks

[uncle_bob]

Hi

Running high power steppers with an MCU is sort of a recipe for disaster. One glitch and the machine eats it's self ....You find that even with months of testing on the code, glitches still happen. Running little motors is generally not so big a deal. A skip here or there and not much (if anything) noticeable happens. With a fast stepping CNC drive motor doing odd paths through material, the timing can easily get beyond what hand crafted / profiled assembly code is up to on an MCU.

Bob

[hamdi.tn]

am not sure  why you will need 500K switching !!! or any frequency over couple of hundreds hz to make a stepper spin

[uncle_bob]

am not sure  why you will need 500K switching !!! or any frequency over couple of hundreds hz to make a stepper spin

Hi

Indeed you will have to take a close look at the torque / frequency curves on your motors to work out how fast they can step. It also depends a bit on how much back pressure your tool (bit) exerts through the drive mechanism. I've seen some drive setups that go nice and fast right up to the point they touch material ....

Bob

[nour]

Hi

Running high power steppers with an MCU is sort of a recipe for disaster. One glitch and the machine eats it's self ....You find that even with months of testing on the code, glitches still happen. Running little motors is generally not so big a deal. A skip here or there and not much (if anything) noticeable happens. With a fast stepping CNC drive motor doing odd paths through material, the timing can easily get beyond what hand crafted / profiled assembly code is up to on an MCU.

Bob

I am sorry I couldn't understand exactly what did you meant by "MCU is sort of recipe for disaster" 
did you mean using MCU is a bad idea or not recommended?
if so what should I use?
I think there is a lot of open source controllers\drivers MCU based that is working fine(I suppose, never tried any)

can you please make further explanations for this point 

[nour]

am not sure  why you will need 500K switching !!! or any frequency over couple of hundreds hz to make a stepper spin

I just set a maximum frequency for the driver to handle not for the motors
for the motors I really don't have enough experience to say for sure how much switching that would make a specific motor runs smoothly and fast

may be you are right, couple of hundreds hz would be enough, still need to do tests to understand how efficient would that be

[nour]

I have decided to build a test circuit  for testing mosfet characteristics free running and under load using signal generator and oscilloscope and a couple of DMs   
I think that would help me to understand more about how different mosfet behave under different conditions and compare them to the basic theoretical calculations

I have more than 10 different discrete mosfets laying around so I will go through them one by one and will do a table for comparison later
I am not in a hurry and I think that is better than buying stuff that wouldn't work for me later

later I will decide if I will use gate driver for my H-bridge or not and decide how much current I will need for such a thing

I will open new topic for that very soon after building my basic circuit and get my hands dirty in the experiment
hope nothing smoke 

if someone would think that's waste of time, please advice me and tell me why

thanks

[nour]

later I will decide if I will use gate driver for my H-bridge or not and decide how much current I will need for such a thing

How can you drive an H bridge without gate driver? Logic shifting with high dv/dt middle point is a PITA if done yourself. 2 integrated half bridge driver takes care that issue for you.

I meant by gate driver an integrated circuit driver not discrete components, may be I used the wrong terms there 
actually if possible I would like to do the gate driver from discrete components because I have a lot of components on my shelf and want to do something with them 
may be gate driver with integrated circuit would be much simpler and easier(indend it is) but as I have said, I want to make use of the components that I have (I have a lot of discrete transistors and other stuff)

[nour]

Logic shifting with high dv/dt middle point is a PITA if done yourself.

would you please explain what did you meant by "dv/dt middle point is a pita" ?

[nour]

Logic shifting with high dv/dt middle point is a PITA if done yourself.

would you please explain what did you meant by "dv/dt middle point is a pita" ?

If it is a half bridge, then the source pin of the top mosfet is switching at a very high dv/dt, as intended. You need a method to level shift your ground referenced gate signal to the switching virtual ground, which is not easy to do.

I will try to solve this issue after finishing the test circuit that I have mentioned
thanks

[nour]

I will try to solve this issue after finishing the test circuit that I have mentioned
thanks

An integrated half bridge gate driver has this integrated. You do not have to worry about it if you use one.
This only becomes your concern if you roll out your own gate driver, or you are switching at an insanely high dv/dt.

yes I know that, but as I have mentioned before I will try first to do this from discrete parts because  I have a lot of them
also I will need a lot of this drivers IC
I am willing to build around 4 drivers for bipolar steppers, every motor has 2 H-bridge which means 4 motors X (2 H-bridge X 2 Drivers per bridge) = 16 drivers X 0.7$ = 11.2 $

If I can use my current resources to build a driver that will do the shift level without spending those 11$ it would be good
also I am willing to build after finish this another 4 BLDC drivers, so that will help later

I Don't have much information about how to do it or whether or not it is easy to build, but later I will decide about it
if you can help me with a schematic for doing that, it would be very nice

[hamdi.tn]

I think for this particular application , you may use some of the shelf circuit , a simple one and used a lot by hobbyiste is the L298 double H-bridge so you will need 4 of them, with high voltage and high current capability. should be easy to interface with any micro-controller and can get the job done.

[nour]

I think for this particular application , you may use some of the shelf circuit , a simple one and used a lot by hobbyiste is the L298 double H-bridge so you will need 4 of them, with high voltage and high current capability. should be easy to interface with any micro-controller and can get the job done.

I manage to blow up like 3 of those L298 until now  , they are so fragile, that's one of the reasons that made me decide to use external mosfets

[uncle_bob]

I just set a maximum frequency for the driver to handle not for the motors
for the motors I really don't have enough experience to say for sure how much switching that would make a specific motor runs smoothly and fast

may be you are right, couple of hundreds hz would be enough, still need to do tests to understand how efficient would that be

Hi

Ok, let's back off a bit. If you are buying a driver, simply going for overkill is fine. The cost isn't much and they have already taken care of the all the nasty issues.

For an application specific driver, that's not a good way to go:

Your end device needs to move at X rate to be useful. You have a mechanical advantage of Y between the motor and what you are moving. The device weights something and in a CNC you need force to hold it in place. Those are all numbers that are a direct result of your basic design.

You then head off to see if you need a giant motor or a little tiny motor. They all have torque and hold data as part of the spec sheets. Different motor designs have different curves. It's *always* a "runs out of torque here" sort of decision. The max step rate only applies to a motor with absolutely nothing attached. You then take a look at the price of this or that motor. If they are $900 each and you need four of them that might not be inside the budget. Back you go and take a look at the basic design.

Looping through the basic design, you find that high mechanical advantage (gear way down) gives you high resolution but forces you to a very high step rate. It also impacts torque, but you need it at a higher speed so likely you loose there as well. Back through the design, and "can you get the resolution at the price and speed" is the question.

On all of the systems I've built, the mechanical stuff has cost *way* more than any of the electronics.

Bob

[nour]

If you just want a rugged and faster L298N, why not try DRV8844? For $1.75, you get 4 half bridges (2 H bridges), rated at 2.5A peak, 1.75A RMS, 60V at 25 deg C. It is PWM ready with only 90ns dead time.

unfortunately this one is not available at TME, currently I just buy from them because they have the lowest shipping rates (and very fast) for me

anyway, I want to do it from discrete components for the sake of power dissipation and better heat transfer over time
when I was using the L298, it was getting too hot very quickly which makes it not idle for long term running (of course I used heatsink for it)

[nour]

On all of the systems I've built, the mechanical stuff has cost *way* more than any of the electronics.

Bob

can you recommend motors for me to work with in the range of sub 100$

I don't have any experience with motor's torque and how much would be enough
how to evaluate a motor torque whether or not it would be good for the job without spending money over a lot of motors to just try them
I have already a lot of small stepper motors vary between bipolar and unipolar and all of them came from various consumer printers, but I feel they can't do the job (I can't find enough info about them)
most of the time when I try to drive them with a basic drivers using arduino and discrete components they get hot after some time and soon enough they are getting ugly hot

another example the one that I have introduced, the sanyo denki one, the spices os over there
http://store.lipsiasoft.com/uploads/415640_103-H7126-1740.pdf

How I can evaluate that motor and have a good estimation about whether or not it is good enough for my application and it wouldn't be over-killing
it says in the datasheet the following
Holding torque:165 Ncm
Rotor Inertia: 350x10^-7  Kgm²
back E.M.F: 31 V/Krpm

I hope there is a guide line for something like this that would be easy enough to start with

I think we are drifting over here and the discussion is no longer related to the topic 

[hamdi.tn]

I think we are drifting over here and the discussion is no longer related to the topic 

i think coz everyone here think that you overthinking this ... did i say think 
and cause you missing some steps to the choice the drivers... including the steppers choice , you need to know your loads before choosing the switches and the drivers.
am not sure about the aspect of the project ,but i really recommend to choice the steppers and some basic H-bridge circuit, and go forward with the project. if you doing this for fun, it will be easy to salvage some steppers and circuit drivers from old printers, i personally did that before. and as been said before, the mechanical part of such project could be more challenging and expensive than the electronics.

[uncle_bob]

On all of the systems I've built, the mechanical stuff has cost *way* more than any of the electronics.

Bob

can you recommend motors for me to work with in the range of sub 100$

I don't have any experience with motor's torque and how much would be enough
how to evaluate a motor torque whether or not it would be good for the job without spending money over a lot of motors to just try them
I have already a lot of small stepper motors vary between bipolar and unipolar and all of them came from various consumer printers, but I feel they can't do the job (I can't find enough info about them)
most of the time when I try to drive them with a basic drivers using arduino and discrete components they get hot after some time and soon enough they are getting ugly hot

another example the one that I have introduced, the sanyo denki one, the spices os over there
http://store.lipsiasoft.com/uploads/415640_103-H7126-1740.pdf

How I can evaluate that motor and have a good estimation about whether or not it is good enough for my application and it wouldn't be over-killing
it says in the datasheet the following
Holding torque:165 Ncm
Rotor Inertia: 350x10^-7  Kgm²
back E.M.F: 31 V/Krpm

I hope there is a guide line for something like this that would be easy enough to start with

I think we are drifting over here and the discussion is no longer related to the topic 

Hi

Motors are part of the "mechanical" side of the system. You can easily spend $1,000 apiece on motors for some of this stuff. That same $1,000 motor runs fine with a pre-built driver that costs less than $50.

The sizing of the motors and what you "need" is mechanical engineering. You do a design, run through the math and that is what gives you your requirements.

You start with:

How fast does the head need to move in mm/s
How much does the head weigh
How much does the structure to hold the head weight
How much back force does the tool generate (you may have to measure this).

Next up:

How rigid does the system need to be
How accurate does it need to be
How repeatable does it need to be (different than accuracy)
How small a step do you need to get a good finish (different than the other two above)
All of these basically are "how good does it need to be".

Next:

How large an area does the machine need to traverse in X, Y, and Z
That's a biggie all by it's self. A machine that needs to go a meter in each of those can be "interesting".

With those inputs you can work out your aluminum budget and figure out how much that's all going to weigh. There are various drive approaches, screws and belts are the two main ones. Your accuracy budget will steer you to one or the other.

If one lead screw (or belt) does not get you to adequate positioning, you may need two (or three for interesting drive approaches).

Finally:

You can calculate the number of RPM you need to move your head a given distance. You can work out how fast you need to accelerate or decelerate. (right angle turn that is still fast enough). That will give you a torque at RPM number. That plus the friction and back force on your motor give you the likely torque range. I'd add at least 20% and probably 50% to whatever you come up with as a "first time I've done this" safety factor.

So now you run off and buy a bunch of aluminum straight piece parts, and all the fittings to rigidly attach them to each other. You come up with the drawings for your various mounting plates and send them off to the laser shop to have them cut. You order up a few (hundred) bearings. Out goes the order for belts or lead screws. If it's lead screws, they start at "expensive" and go up from there. Also get the traveler nuts for the lead screws from the same guy.

Figure that it all comes in and *something* isn't going to fit. Back for more mounting plates or some such thing.

Note that that the motors and electronics come in very late in the process. The other stuff needs to be worked out much earlier.

Bob

[Ghydda]

Whenever I wade into new and uncharted territories I find that answers come fastest  and with least effort by actually building a prototype of the subcircuit and using it to get a feeling or even confirmation on what works and what does not.

LTspice is great, I use it for analyzing switch behavior among other things, but it is very very difficult to get accurate results as the real world is never as idealized as a simulation.

Doing math is great to get in the ballpark of what values the components should have, but for finetuning math is a long and hard and tedious road on all but the simplest of circuits.

Mock-ups/prototypes is a fast way to get know new circuits, learn what parameters are more important than others (and which parameters dominate others), and how to evaluate the performance of said circuit.

Anyway, that would be my approach to nour's challenge.

Cheers
/Ghydda