checking MOSFET performance

Discussion in 'General Electronics Chat' started by EmmKay, Apr 22, 2007.

  1. EmmKay

    Thread Starter New Member

    Apr 19, 2007
    9
    0
    Hi all,

    I have a circuit in which a MOSFET (irf510A) is being driven by a 2n2222 transisitor. The MOSFET is getting very very hot. The supply voltages are 32V for the Mosfet with a current load of 2.8A. This is a simple unipolar stepper motor driver which takes pulse input from the parallel port.

    I want to ultimately have a pulse frequency of 25KHz. How do I check if the MOSFET can handle this type of a frequency. ALSO more importantly how can I check if the MOSFET is performing to its specs if I do not have a scope to test its switching functionality.

    Appreciate the help

    regards

    emmkay
     
  2. Papabravo

    Expert

    Feb 24, 2006
    10,140
    1,790
    I don't know about the MOSFET at that speed, but 25 kHz for a stepper motor is a complete fantasy. Where did you get the idea that the motor could run at this rate?

    To get an idea of what is happening with power dissipation, you should look at the waveform on the gate. You will probably note that there is a plateau in the transition where the transistor is trying to add or remove charge from the gate capacitence. At this time the MOSFET is in it's linear mode and may be dissipating large amounts of power. It would help if we could see a schematic and point you in a more productive direction.
     
  3. EmmKay

    Thread Starter New Member

    Apr 19, 2007
    9
    0
    Hi,


    This is the set up I have...

    20Khz means 50rps on a 400 pulse per rev motor. Unipolar motors running at 3000rpm on a home made cnc machine. driving a 1mm thread ballscrew. The controller is a GECKO and the software is MACH3.

    The PSU I am using is 80V.

    This morning I changed the circuit based on a a diagram I found on the net
    http://www.pminmo.com
    I have been toying with a circuit to pump more Volts into the motors to see what kind of throughput I can get. The circuit is based on L297 (setup at 25Khz) + 74HC09 + (simple Gate driver home made 2n3904 (2n2222 no good) based) + IRF640 replaced IRF510A...(Cannot get gate driver iIC's or Logic Level Mosfets where I am before someone suggests using those!!!)

    Based on these components I have only managed 15Khz on my home made controller. The IRFs get pretty hot and the motors start skipping. I DO NOT have a scope (cannot afford even a 2nd hand one!), I solely rely on the software (MACH3 and TurboCNC) to tell me what pulse rate I am using. I compare that to actual travel times I encounter on my machine and calculate the rpm of the motors.!

    I also wrote a basic prg to generate appropriate pulses removing the L297 for test purposes. Again after 15KHz the irf's started getting hot and at 18Khz V. V. Hot.

    So in short I am running my old set up at 20Khz.. the motors I have have been working well for the last 10 months. (No idea how long will they last!!!)
    Am getting greedy for more torque so am looking at higher rpm with gear/belt reduction to get the torque I want.. Hence the new controller. Hence the questions...

    Hope it helps!!!!!!!
     
  4. thingmaker3

    Retired Moderator

    May 16, 2005
    5,072
    6
  5. Papabravo

    Expert

    Feb 24, 2006
    10,140
    1,790
    Well sufficient torque to avoid stalling at that speed is the name of the game. Avoiding resonance is also handy, as is half-steping, microsteping, and velocity ramping.

    Trying to do this kind of design without basic tools will result in a great deal of trial and error, frustration, and blind alleys. I doubt you can get a stepper to 50 rps. Designing circuits based on other peoples work is hardly impressive. Upping the voltage to improve risetime is OK but you don't want to be pulling power from a HV supply when you are standing still. If you had a two level supply I might be more impressed.

    BTW where in those datasheets do you get to 25,000 steps per second? I still think that speed is a fantasy.
     
  6. thingmaker3

    Retired Moderator

    May 16, 2005
    5,072
    6
    Charts show speed vs torque.

    There's nothing wrong with building on someone else's idea. Impressing pessimistic old engineers is not the goal here. The goal is getting a hobby machine to work to the satisfaction of the hobbyist building it.
     
  7. John Luciani

    Active Member

    Apr 3, 2007
    477
    0
    Average power dissipation is your FET problem not frequency. What are you using for a
    heatsink? The thermal resistance (Rja) of the MOSFET is 62.5 DegC/W without
    a heatsink. The on-resistance at 125 DegC is around 0.8Ohms. If you had
    100% duty cycle (DC) you would dissipate apx. 3W meaning an 180 Deg *rise*.
    You would use the transient thermal response curves to adjust for duty cycle.

    Also how are you driving the gate? The on-resistance increases as the Vgs
    decreases. The 0.8Ohms at 125DegC is with a 10V drive. Does your gate
    drive signal look like a square wave or is it a trapezoid. You will dissipate
    more power if you do not turn the FET on and off quickly.

    The maximum operating of the Fairchild IRF510A is 175DegC which is quite
    hot (to the touch). Running it hot will decrease your reliability. If space is not
    critical use a large heatsink. I usually derate by at least 20% for a MOSFET.

    (* jcl *)

    ---
    www.luciani.org
     
  8. Papabravo

    Expert

    Feb 24, 2006
    10,140
    1,790
    Au contraire. They do not go to 25,000 steps pers second, they top out at 5000 steps pers second in one case and 10,000 in another. You guys are playing fast and loose with the facts and I'm not buying your bumpf!
     
  9. beenthere

    Retired Moderator

    Apr 20, 2004
    15,815
    282
    Well guys, this site - http://www.excitron.com/servlet/Detail?no=202 - has steppers with specs up to 25K sps. I have reservations about torque characteristics at this speed, and I would expect a ramp being necessary to get to that speed, but Excitron says they'll do it (be interseting to see the drive circuit).
     
  10. thingmaker3

    Retired Moderator

    May 16, 2005
    5,072
    6
    By golly you're right! I misplaced a decimal point somewhere. Naughty me!

    So PapaBravo, what are your helpful suggestions for improved performance with MOSFET controlled steppers on EmmaKay's hobby CNC machine?



    EmmaKay, if you can post a schematic of your MOSFET driver (how the 2N2222 is used) we might be able to spot the heat problem.
     
  11. Papabravo

    Expert

    Feb 24, 2006
    10,140
    1,790
    Steppers generally have at least two resonances. The low frequency one is mechanical and involves the moment of inertia of the load and friction. This resonance is normally avoided by ramping the velocity up and down in such a way as to avoid the known resoance.

    The second resonance is electrical in nature and is derived from the L over R time constant of the motor coils. A motor coil has an inductance which is determined from its geometry and the core material around which it is wound. The R can be as small as the DC resistance of the wire. Small R is no good for high speed operation though. Large R makes the L over R time constant smaller. But large R and large I make for large power dissipation.

    What to do? My solution is to use two power supplies. One supply is low voltage high current for holding purposes. The other supply is high voltage medium current for producing a large di/dt when a coil is first energized. Monitor the current as it builds up in the coil toward the holding current value and switch off the high voltage supply.

    Next I would control the flyback spike by charging a capacitor that sits between the high voltage supply and about twice that value.

    Last I would use a class AB push-pull switch to charge and dischrge the MOSFET gate. Seeing the Vgs plateau on a scope will bring into sharp focus how important it is to charge and discharge the gate in absolutely the least amount of time you can imagine.

    MOSFETS were still pretty expensive when I used this approach to build engraving tables. So I used TIP31C bipolar transistors. They were beasts. They took a licking and kept on ticking.
     
  12. EmmKay

    Thread Starter New Member

    Apr 19, 2007
    9
    0
    Thank you all for the various inputs. Encouraging and not so Encouraging and all! :)

    I am still absorbing the wealth of information presented here. And given the opportunity would thank each one individually.

    I know this much that the GECKO drive I am using does categorically run my steppers at 50rps and has been running it for nearly 10 months now. So the 50rps is certainly not a figment of my imagination.

    Phil at pminmo.com has been very kind and helpful by making his designs available to any and all for non-commercial use. And is very supportive of any variation/change in the design of his controllers.

    In the mean time something fundemental in the design of my controller has cropped up!. A Unipolar motor is driven by four inputs, All synchronised and phase shifted (is that the right word!). Each mosfet drives one input. Now if I wish to rotate the motor at the fantasy speed of 5000 rpm I would have to send signals at 25Khz rate. Will I be correct in thinking that This would not be the frequency for each mosfet as I would have to divide the 25KHz by 4. This would mean that each mosfet driving circuit should be designed for 6250Hz or thereabouts and not 25KHz.

    In case of the controller that I am trying to build/copy etc etc the L297 IC takes the step and direction input and generates the appropriate signals for each phase. I assume that it will be doing the divisions internaly.

    With regards to the problem of over heating. I have been given a document to look at heatsink specing in detail that will no doubt help me.

    I did manage to figure out why the IRF was over heating. Not having a scope makes life very difficult (a 2nd hand 10Khz old single channel costs a months salary for me!). So I made a small non-inductive load. i.e. arrays of led's with known resistor values and based on the data sheets of the IRF it transpired that the IRF wasn't being turned on Hard at higher frequencies. Someone suggested a Totem Pole gate driver circuit(have to find out what that is and work on that!)

    This is all part of the learning experience for me as my background is strictly mechanical and if it turns out that stepper Motors do not run at that speed atleast I would have learned how to make a high speed on/off switch and what a Totem Pole is:)

    cheers all..
     
  13. beenthere

    Retired Moderator

    Apr 20, 2004
    15,815
    282
    Here it would be helpful to have the schematic to look at. For full saturation - on - a FET has to have the gate 10 volts above the source. Anything up to 20 volts is ok.

    If your FET's are running as grounded source, this is trivial. If you are switching the high side (from V+ to the stepper and then to ground) a driver circuit has to be made to keep the gate at least 10 volts above the source. A totem-pole circuit will not necessarily do this, but there are dedicated high side driver ic's available to do this.
     
  14. EmmKay

    Thread Starter New Member

    Apr 19, 2007
    9
    0
    What is the best way of attaching a schematic...
     
  15. EmmKay

    Thread Starter New Member

    Apr 19, 2007
    9
    0
    Duh....
    ignore the previous post please.
     
  16. thingmaker3

    Retired Moderator

    May 16, 2005
    5,072
    6
    A "totem pole" is a variant of the Class AB Push-pull amplifier which PapaBravo suggested.

    Laszlo Balogh wrote a good article about MOSFET driver basics and driving MOSFETS at higher speeds: http://focus.ti.com/lit/ml/slup169/slup169.pdf
     
  17. EmmKay

    Thread Starter New Member

    Apr 19, 2007
    9
    0
    Following is what I have been using to drive the Irf's.
     
  18. John Luciani

    Active Member

    Apr 3, 2007
    477
    0
    Are you sure about the position of R2? It probably should between the Q1 collector
    and the +12V supply.

    (* jcl *)
     
  19. EmmKay

    Thread Starter New Member

    Apr 19, 2007
    9
    0
    can someone please review the attached schematic and tell me if I am on the right path to draw up a Totem Pole gate driver for Mosfets at an input pulse frequncy of 25KHz!!!

    thank you
     
  20. Papabravo

    Expert

    Feb 24, 2006
    10,140
    1,790
    I don't think so. It will take literally forever to charge the gate of the IRF540 through a 4.7K resistor. The discharge will be faster because of the diode.

    In the previos diagram that you posted it does seem as though the bipolar NPN transistor is shorting the +12V supply to GND. Can you explain this anomoly?

    A 74LS09 is a very obscure, nearly obsolete part with an open collector output driver. This means the output driver has an active turn on and a passive pullup. Open collector devices were use in bus systems before the invention of tristate drivers. They serve no useful function in this application. Can you explain why you are not using a CMOS part with more drive and a symmetrical output stage?
     
Loading...