H-bridge motor control FAN73832

Discussion in 'The Projects Forum' started by rlwoodjr, Jun 30, 2010.

  1. rlwoodjr

    Thread Starter New Member

    Mar 19, 2010
    8
    0
    Hello all. I am a newbie to this forum and a newbie to electronics. I have read a lot about h-bridges, and would like some feedback to some basic questions.

    I have 50 VDC motors that I want to use PWM to drive them in a CNC router.

    The motors have a 6 amp stall current (tested with 50 VDC) so I have not planned any current limiting....is this bad?

    The Mosfets are rated for 55 VDC....is this enough since I am driving them on with a high side driver?

    What should the bootstrap capacitor and diode be rated for?

    I hooked it up on a breadboard. The scope without the 50 VDC looked like I expected, then I added the 50 VDC...still looked good on the scope, I hooked up the motor and smoke came up from the fan73832.

    Any comments or suggestions would be appreciated.
     
  2. rlwoodjr

    Thread Starter New Member

    Mar 19, 2010
    8
    0
    I built another circuit on another breadboard and tried 20 VDC. It worked so I moved to 50 VDC...it works too.

    I pulse 24 to 1000 (0-1024). Less than 24 or more than 1000 does not give the bootstrap to charge....? Can I do better with a larger capacitor? At 512 it stops, less it goes one way more the other.

    I have 2 voltage supplies. One at 16 VDC and one at 50 VDC. Is it possible to use the FAN73832 with out a bootstrap setup by applying 66 VDC to the chip?

    I only ran for a few minutes. I suppose I should setup a test bed and run it back and forth with a load.

    Any advice on the capacitor or diode sizes or voltage ratings?

    Should I get higher voltage rating mosfets? Mine are rated at 55 VDC.
     
  3. SgtWookie

    Expert

    Jul 17, 2007
    22,182
    1,728
    Use larger caps in parallel with C1/C2.
    I'd use Shottky diodes for D5, D6, but you'll need 'em rated for >=60v.

    UF4003's are too small for the reverse-EMF diodes. They're only rated for 1A average.
     
  4. SgtWookie

    Expert

    Jul 17, 2007
    22,182
    1,728
    I forgot - you need caps from the anodes of D5 & D6 to GND. They should be several times larger than C1 & C2 (and the caps in parallel with them)

    For example, if you parallel a 10uF cap with C1, then use a 33uF or 47uF cap from the anode of D5 to GND. Otherwise, the inductance of the wiring/traces from the supply to the diode will slow the charging of C1 & co-cap.
     
  5. rlwoodjr

    Thread Starter New Member

    Mar 19, 2010
    8
    0
    I am confused. Isn't C1/C2 the bootstrap capacitor(s). Would more capacitance allow more charge or clean up the signal?
     
  6. SgtWookie

    Expert

    Jul 17, 2007
    22,182
    1,728
    If the bootstrap caps are small, the charge will bleed off rapidly when the high-side MOSFET is turned on. This will cause the MOSFETs to burn up because their Vgs drops, giving a higher Rds and increased power dissipation. Having a larger cap in parallel with a small cap for the boost circuit will give it more "staying power".

    This also means that you need caps near the anodes of the bootstrap diodes to provide a ready source of current when the high-side MOSFET is turned off for a moment. Otherwise the inductance of the wiring between the power supply and the anode of the bootstrap diodes will cause a delay in current flow through the diode.

    Don't forget that during initialization, in order to charge the boost caps, you must have both high-side MOSFETs turned off, and both low-side MOSFETs turned ON for a period of time sufficient for them to reach normal operating voltage level. If you don't turn the low-side MOSFETs on, there will be no charge path for the boost caps, and risk fried MOSFETs. If the MOSFETs fail, they will likely take your FAN73832 with them, as a typical failure mode is a dead short from drain to gate.
     
    Last edited: Jul 3, 2010
  7. rlwoodjr

    Thread Starter New Member

    Mar 19, 2010
    8
    0
    I think I understand (see picture).

    Is one capacitor at 10.22 uF the same as two (one at 0.22 uF in parallel with one at 10 uF?

    Also, can I use an electrolytic capacitor for the 47 uF?
     
  8. rlwoodjr

    Thread Starter New Member

    Mar 19, 2010
    8
    0
    I have removed the hex inverter and run 2 pmw channels from the Arduino, one to each of the FAN73832's. One of the pwm channels is inverted.

    I start them both at low so that the low mosfets are on. Or at least I think they are.... (see picture)
     
  9. SgtWookie

    Expert

    Jul 17, 2007
    22,182
    1,728
    Yes, you're getting it. :)

    You can't buy a 10.22uF cap!
    The point is that you want a small ceramic or metal poly cap in parallel with an aluminum electrolytic. The small cap takes care of the high freq transients, and the larger al electrolytic takes care of the low frequency transients.

    Yes, but you should also have a >1uF metal poly or ceramic cap in parallel with it, as close to the diodes' anodes as possible.
     
  10. rlwoodjr

    Thread Starter New Member

    Mar 19, 2010
    8
    0
    Since D5 and D6 are close on the board and both anodes are at the same node, can I use one set of capacitors (a 47uF and a 2.2 uF) for both diodes?

    As for the bootstrap capacitor. If I add a large capacitor and there is a limited time to charge will the small one charge first? I guess I could put this into a circuit simulator and see....

    Also, since I am at 50 vdc, are my 55 vdc mosfets okay, or am I too close to the edge? Should I go with 100 vdc mosfets?

    I really do appreciate all the advice. I have been learning by the magic smoke method. It tries your patience and gets expensive.:(
     
  11. rlwoodjr

    Thread Starter New Member

    Mar 19, 2010
    8
    0
    I was re-reading the FAN7383 data sheet and it says

    "2. Under-Voltage Lockout (UVLO)
    The FAN73832 has an under-voltage lockout (UVLO)
    protection circuit for high- and low-side channels to
    prevent malfunction when VDD and VBS are lower than
    the specified threshold voltage. The UVLO circuitry
    monitors the supply voltage (VDD) and bootstrap
    capacitor voltage (VBS) antepenult."

    If I am understanding this correctly, it means that if the bootstrap cannot supply enough voltage (10.7 high side and 10 low side from earlier in the datasheet), the the IC holds it High and Low outputs low. So the mosfets are not on in the "none saturation region."

    Does this sound correct?
     
  12. SgtWookie

    Expert

    Jul 17, 2007
    22,182
    1,728
    Yes, because you will only be operating one high-side MOSFET or the other at any given time (that is, unless your program has errors, which is another can o' worms...)

    Very generally, small-value metal poly or ceramic capacitors have very low parasitic R/L values compared to aluminum electrolytic caps.

    Unless you use accurate models of real-world caps, you won't see any difference. Many SPICE simulations just use ideal caps, which is one of the problems of using SPICE. In order to make an accurate model, you also need to have all of the parasitics modeled, which is difficult and very time-consuming to do.

    Understand that an ideal square wave is the sum of the fundamental frequency, plus the sum of ALL of the odd harmonics.
    See this Wiki entry: http://en.wikipedia.org/wiki/Square_wave
    Click the image halfway down on the right to see the animation.

    If you're sure that it's clamped to 50v, you should be OK. That's a 10% margin. You could go to 60v MOSFETs if you wanted to.

    If nothing else, the three basic things you need to watch on MOSFETs is the Vdss, the Rds(on) and the Qg, or total gate charge. You want a low Rds(on) and low Qg. Very generally, increasing the Vdss will cause either/both Rds(on) and Qg to increase. This is because for a higher Vdss, the channel thickness has to be increased. In order to maintain a low Rds(on), the area of the channel has to be increased, leading to a higher Qg. If the area remains the same, the Rds(on) increases.

    One tends to learn faster when it's costing them money. ;) It can get expensive. Asking for help may help stop the bleeding of cash.

    Keep in mind that if you are making this circuit on a PCB, that you need to have your traces both short, and wide. It really is not something that you can breadboard.

    PCBtemp is a very useful freeware utility. It is no longer supported by it's authors, but is plenty good enough for hobbyists.

    Roman Black has thoughtfully made it available for download from his website:
    http://www.romanblack.com/pcbtemp.htm

    Download and use it.
     
  13. SgtWookie

    Expert

    Jul 17, 2007
    22,182
    1,728
    Yep.

    It would help a good deal if you increased your Vcc from 12v to 15v.
     
  14. SgtWookie

    Expert

    Jul 17, 2007
    22,182
    1,728
Loading...