Transistor Data sheet current gain (Hfe)

Discussion in 'General Electronics Chat' started by cages, Nov 12, 2011.

  1. cages

    Thread Starter New Member

    Jul 3, 2010
    19
    0
    I am using a transistor (MPSA42) as level converter 1.5V to 3.3V.
    I would like to use a minimum base current from the 1.5V signal.

    When working out the base resistor value I need to use the current gain hfe.
    The data sheet has minimum Hfe (attached) in the Electrical characteristic table but with Vce of 10V.There is also a hfe graph to Ic.

    I am not sure what current gain value I should use in the calculations of the collector (load) resistor and base resistor?

    The Hfe value measured by my multimeter (cheap and not calibrated) is 100.

    When using an NPN transistor as a switch what current gain values (from the data sheet) should be used to calculate the resistor values to ensure saturation?

    Thanks
    cages
     
  2. MrChips

    Moderator

    Oct 2, 2009
    12,429
    3,360
    Since you are using this as a level converter (switch) your calculations are not critical.
    You can use VBE(SAT) = 0.7V.
    If you choose IB = 1mA
    RB = (1.5 - 0.7)/1 = 800 ohm
    Any value between 470 and 1K should work fine.
    (seat-of-pants calculation)
     
  3. Adjuster

    Well-Known Member

    Dec 26, 2010
    2,147
    300
    It would help to know what the collector load would be, hence the collector current. 1mA Ib might be a reasonably generous value if the collector current is 10mA. That is equivalent to a saturated gain of 10.

    If the collector current is very much less than 10mA, and you are keen to reduce the input current, then reduce the base current proportionally by increasing the base resistor.

    If you are pessimistic, and located in the USA, you may want to assume a gain of 10.Optimistic Europeans may be tempted to take a gain of 20, but given recent economic issues we may be led to distrust European optimism. In no case should you use the much higher gain levels seen in the active region, e.g. as indicated by your meter. These will lead to poor saturation.

    Edit: The transistor characteristics you posted show saturation voltages for a β of 10, so you should stick to this value.
     
    Last edited: Nov 12, 2011
  4. cages

    Thread Starter New Member

    Jul 3, 2010
    19
    0
    Would the saturation voltage base and current characteristics be be best place to get the current gain from when using a transistor as a switch?

    I would like the base current to be as low as possible because the 1.5V signal is from a battery powered device (200mAH battery) so would only last 200 hours driving a single 1mA signal.
     
  5. Adjuster

    Well-Known Member

    Dec 26, 2010
    2,147
    300
    Yes, assuming that you want a saturated switch, which is usual for level conversion.

    What is your collector current though? I would be surprised if it is as much as 10mA for a level conversion, unless you are driving something like a relay. What is the collector load?
     
    cages likes this.
  6. MrChips

    Moderator

    Oct 2, 2009
    12,429
    3,360
    If you want to minimize power consumption, bare in mind that the collector current is going dominate the power draw, not the base current. Hence you want the transistor in the OFF state most of the time.
    If you want to reduce the base current you can switch to a FET instead of a BJT.
     
    cages likes this.
  7. Adjuster

    Well-Known Member

    Dec 26, 2010
    2,147
    300
    This is a level conversion circuit from a 1.5V battery circuit to a 3.3V level somewhere else. The collector current presumably does not flow through the battery (but the OP has yet to specify its value).

    If a "1.5V" battery provides the input, the circuit must continue functioning until the battery voltage has fallen to a rather lower voltage. If a FET is chosen for this application, a device with a suitably low threshold voltage must be used. Many common devices have higher thresholds.
     
  8. cages

    Thread Starter New Member

    Jul 3, 2010
    19
    0
    I will consider using a FET, the transistor was available so thought I could use it.
    I just noticed that the data sheet has a spice model. Would the current gain be available from one of the values? (not really used spice)

    The collector current is not really and issue as I will limit it using a resistor (only level converting the 1.5V signal to 3.3V) and the 3.3V will not be supplied by a battery.

    Thanks
    Cages
     
  9. Adjuster

    Well-Known Member

    Dec 26, 2010
    2,147
    300
    You are, I think, failing to understand the collector current issue. If for instance you can limit the collector current to say 1mA by using say a 3.3kohm load resistor, then a base current of 100μA would meet a β = 10 requirement.

    You still need to know what your load current requirement is if a FET is to be used, as it will affect the Rds(on) required.

    For the third time of asking, how much output current do you need? Otherwise put, what does the converted output drive - a CMOS gate needing a fraction of a microamp, or a relay coil needing tens of mA?
     
  10. cages

    Thread Starter New Member

    Jul 3, 2010
    19
    0
    A CMOS gate (PIC32 input).

    My main concern was the data sheet hFE (DC current gain) graph did not reflect the Electrical characteristics give in table form so was wandering what current gain value to use when determining the resistor values to limit the base and collector current.

    cages
     
  11. Adjuster

    Well-Known Member

    Dec 26, 2010
    2,147
    300
    Unless a very fast switching speed is required, then a collector current of 1mA with a pull-up resistor of 3.3kΩ may be adequate. It may be acceptable to have an even lower current, depending on whether a few microseconds delay could be accepted.

    The base resistor should be calculated for a gain of 10, but remember to allow for the battery voltage being a bit run down.

    The higher HFE values given in the specifications reflect operation at higher collector voltages, e.g. for applications such as linear amplification. They do not apply to saturated switching.

    A Spice model of a transistor should do a reasonable job of modelling how the gain of a typical transistor changes with its operating conditions. Do not expect miracles though, and note that the minimum base current that you may seem to "get away with" in a simulation may not work in practice.
     
  12. cages

    Thread Starter New Member

    Jul 3, 2010
    19
    0
    Thanks, The switching speed is about 90 kHz so should not be a problem.
    I will try and simulate the circuit using a spice tool (might be time for me to try and learn to use spice).

    It would be nice if a hFE (current gain) for using transistors as switches was included in the data sheet as this I believe is a common use for transistors, but maybe that is why the spice model was include so you can obtain the data for yourself.
     
  13. crutschow

    Expert

    Mar 14, 2008
    12,995
    3,229
    The information is usually indirectly included in the saturation voltage values in the data sheet, where they typically use a beta of 10, and that is the recommended value for most transistor switches.

    A spice model likely only gives typical values and not worst-case, so that's not a reliable way to determine the required base current for saturation.
     
    cages likes this.
  14. MrChips

    Moderator

    Oct 2, 2009
    12,429
    3,360
    I would side with good ol' Bob Pease (RIP) and don't trust simulators. There was a recent post where the simulator gave erroneous results.
     
  15. crutschow

    Expert

    Mar 14, 2008
    12,995
    3,229
    I had great respect for ol' Bob but his views on simulators, I believe, were excessively negative. They can occasionally give erroneous results but that is often obvious when looking at the results. And of course you should always bread-board a circuit to verify its operation. But I would not build any circuit without first simulating it to catch the obvious bugs and fine-tune the design.

    A few years ago I designed a space-grade switching regulator and Spice was invaluable during the design to verify its operation and check design margins. The first unit I built operated very close to the simulation results.
     
  16. Adjuster

    Well-Known Member

    Dec 26, 2010
    2,147
    300
    There is also the small matter that many of us may not have the same level of insight, whether intuitive or theoretically based, that may have been gifted to some famous pioneers of integrated electronics. In the absence of guidance from such gurus, ordinary mortal engineers are well advised to avail themselves of whatever predictive tools may be available.

    It is indeed unwise to rely exclusively on simulation, in the sense of believing that it can be guaranteed to give error-free results, making practical trials unnecessary. Simulation tools however naturally form part of the design and development process, backed up by prototype work to check the final result.

    Another important use of simulation is educative, in that it can allow convenient demonstrations of circuit operation to be made. Again, there are pitfalls, particularly the possibility of making circuits that cannot exist in the real world owing to impossible circuit values, say inductors of many henries with only atto-farads of self capacitance.

    I'm not sure that I would want to spend much energy arguing this with an engineer or technician of an older generation who "does not believe in simulators", but I think that to suggest that their admitted limitations preclude their use for studying simple transistor circuits is more than a little extreme.
     
  17. crutschow

    Expert

    Mar 14, 2008
    12,995
    3,229
    You said it better than I could.:D

    But I would say that simulation is not only useful for studying simple circuits but also quite complex ones, even those that are virtually impossible to analyze by hand.
     
  18. Papabravo

    Expert

    Feb 24, 2006
    10,137
    1,786
    If this circuit is a one-off then the answer is not that critical as has been mentioned. If you are going to replicate this circuit for a product then you cannot use a single number. You have to recognize that if you have "sample" of 100 transistors and you measure the parameters like hfe(small signal gain) or Hfe(large signal gain) you will get a statistical distribution. The larger your sample the more closely it will approach a normal distribution. Based on the mean and the variance the only thing you can say about an individual transistor is that it has a high probability of having an hfe that is within 3 standard deviations from the mean.

    On a datasheet you can start by considering the typical value as the mean. If they also give you a Max or a Min you use that a a measure of the variance. Be careful though because this assumption may not be valid for all datasheets and all manufacturers. As a rule of thumb it will get you started.

    The last point is that any good designer will design his transistor circuits so they do not rely on a particular value of any parameter. The design should have a specified level of performance with any value inside some range.

    For example: The nominal gain of the amplifier is 18. The gain will be at least 10 for hfe > 50 and frequencies < 25 kHz.. Clearly the overall gain of the amplifier depends on things besides hfe.
     
  19. Audioguru

    New Member

    Dec 20, 2007
    9,411
    896
    I don't believe that spec's for electronic parts are actually random and that most parts have "typical" spec's.

    If I buy 100 transistors from Digikey then maybe they recently ran out of them but recently received a shipment from a manufacturer. Maybe the shipment came from the same wafer so they will all have almost the same spec's which could be near minimum or near maximum. The spec's have a wide range so they make and sell the "fairly bad" ones too you know.
     
  20. Papabravo

    Expert

    Feb 24, 2006
    10,137
    1,786
    Just to be clear I never said that specifications were random. What I said was that from a sample of parts you will find that the parameters have a distribution of values with a mean and a variance. The variance may be large or small depending on the where the parts came from and how they were made. That is a very different thing.
     
Loading...