How can I find NPN's and PNP's that match Hfe?

lardomlardom Posts: 1,570
edited 2019-12-16 - 23:19:59 in General Discussion
I built a triple half H-bridge to experiment with a sensorless BLDC. I can make the motor spin but I can't make it run closed loop yet. One of the problems I'm having is that the BEMF from the three phases don't match. In closed loop the BEMF from the floating phase is fed to a comparator and is compared to a common virtual ground.
At the moment I'm trying to match NPN collector current and match PNP collector current by swapping resistors individually. For example, I'm using pn4401 small-signal transistors to drive the bases of three TIP31's and three TIP42g's. One pn4401 has a gain of 139. Another pn4401 has a gain of 226.
I don't have a lot of experience biasing transistors so that's why I'm asking if variation in transistor Hfe is common. Are mosfets easier to balance?

Comments

  • Peter JakackiPeter Jakacki Posts: 8,854
    edited 2019-12-16 - 23:55:21
    You best buy them as a complementary transistor array such as this

    EDIT: Now that I have time to actually read your post I would off the top of my head suggest that you might handle this with opamps regulating the current so then it doesn't matter that the driving transistors are mismatched. Variation in Hfe is very normal, and no, mosfets will not be easier to balance and they are very different beasts.
  • @Peter Jakacki, thank you. I'll order some complimentary transistors and mosfets.
  • Peter JakackiPeter Jakacki Posts: 8,854
    edited 2019-12-16 - 23:56:57
    Note my recent edit - I'm not really recommending matched pairs, just a better circuit.
  • EDIT: Now that I have time to actually read your post I would off the top of my head suggest that you might handle this with opamps regulating the current so then it doesn't matter that the driving transistors are mismatched. Variation in Hfe is very normal, and no, mosfets will not be easier to balance and they are very different beasts.

    That's great news. (that made my day!) I just bought a supply of LM339's to experiment with. I'd heard about opamps being used as constant current devices but it didn't occur to me that I could use them to balance mismatched transistors.
  • Peter JakackiPeter Jakacki Posts: 8,854
    edited 2019-12-17 - 00:58:20
    lm339s are comparators designed to handle analog in but switch high or low out very quickly. Mind you, you could force them into some kind of linear region but opamps such as the LM358 dual opamps have a standard and simple pinout that you can swap a better opamp for later. Negative feedback is what opamps do well as the open-loop gain is huge and they could be used like a slow comparator in that fashion :) but with feedback resistors you can set the gain. By feeding this back from the output of the transistor that the opamp is driving rather than from the opamp itself, will ensure that the opamp compensates for the output path. An opamp can drive the base of a NPN for instance with a resistor in the emitter to set the current where a voltage on the input of the opamp is matched (or amplified depending upon feedback configuration) on the emitter of the transistor, so despite the transistor's variable Vbe etc, the resistor will have a constant voltage and from Ohm's law R=V/I --> I=V/R, so therefore constant current.

    So the opamp with feedback will automatically compensate for the transistor which provides the actual drive current. If you tie the inverting input of the opamp back to the emitter of the drive transistor you will have unity gain where the voltage on the + input of the opamp will be matched on the emitter. For PNP, just think of the voltage rail as being the "ground" so that voltages are relative to that.

    To save me having to explain it all, here is a quick link
  • I took a look at the article and it looks like it's just what I need. I want to set up a test circuit with three transistors. I'll order some LM358's tomorrow AM. Thanks again.
  • Phil Pilgrim (PhiPi)Phil Pilgrim (PhiPi) Posts: 22,521
    edited 2019-12-17 - 02:00:01
    Conventional wisdom is that you never rely on an hFE spec to determine the parameters of a given circuit. From Horowitz and Hill, The Art of Electronics:

    "Warning: hFE is not a "good" transistor parameter; for instance, its value can vary from 50 to 250 for different specimens of a given transistor type. It also depends on collector current, collector-to-emitter voltage, and temperature. A circuit that depends on a particular value for hFE is a bad circuit."

    -Phil
  • Just put your load in the collector and the emitter resistor is responsible for setting the current so that a 1V "signal" across a 1 ohm resistor will try to sink a 1A current through the collector load. Obviously this depends upon the load and the supply voltage too. Avoid too much voltage drop across the resistor for heavy currents which would also result in requiring a resistor to handle the power, so this might mean reducing the control voltage range (i.e. 1Vmax with 1ohm = 1A max or 1Vmax with 0.33ohms = 3A max). Typically I'd use 0.1R and reduce the control voltage via a resistor divider into the opamp which is set for unity gain.
  • Op amps are great devices. While waiting for my LM358's to be shipped I tested the constant current circuit with an LM741 that I had in my parts bin. I plugged three different TIP31's into the same spot on my breadboard and the output current matched. :smile:
Sign In or Register to comment.