Transistor Buffer Design

Discussion in 'General Electronics Chat' started by Bill_Marsden, Jan 14, 2009.

  1. Bill_Marsden

    Bill_Marsden Thread Starter Moderator Staff Member

    Joined:
    Mar 24, 2008
    Messages:
    19,281
    Location:
    Dallas, TX (GMT-5 w/ DST)
    OK, one another thread I came up with this...

    [​IMG]

    to which eblc1388 wrote...
    I like learning about this kind of stuff. It is said you learn more from failures than successes. So why would this circuit fail to feed 50Ω with unity gain?

    The transistor power dissipation is a problem, but that would be a failure mode (as in letting the smoke out).

    Gain is the possible answer. I figure a β would be around 100 or so, but is the software showing less, or am I missing something obvious? I wonder if they were Darlingtons and the bias adjusted accordingly whether it would make a difference? I think the output for 50Ω is pretty decent overall.

    BTW, thanks EB.
  2. Bill_Marsden

    Bill_Marsden Thread Starter Moderator Staff Member

    Joined:
    Mar 24, 2008
    Messages:
    19,281
    Location:
    Dallas, TX (GMT-5 w/ DST)
    Thinking about it, 3Ω would be ¼ less signal, even with infinate gain, since the 1Ω resistors would take some of the signal. With a β gain of 100, which isn't actually worst case, and an output of 50Ω the loading transfer would be around 5KΩ at the base. Lower resistances at the emitter would start to significantly load the input.
  3. eblc1388

    eblc1388 Senior Member

    Joined:
    Nov 28, 2008
    Messages:
    1,542
    Location:
    UK
    Hi Bill,

    The simple answer is that the transistors cannot provide sufficient current to drive the 1Ω load impedance.

    Additional stages are needed.
  4. AchMED

    AchMED Active Member

    Joined:
    Aug 5, 2008
    Messages:
    41
    Try something like the NJT4031N and it’s complementary NJT4030PT1G. The NJT4031N specs a worst case 0.3V sat at 0.3A base drive for 3A collector current,the NJT4030PT1G is similar. Power disapation would likely be a problem though.

    Zetex also makes the FMMT618 And other high gain low Vce sat bjt’s they ere pricey though. The Onsemi ones I mentioned are about half the price of comparable zetex.

    http://www.onsemi.com/pub_link/Collateral/NJT4031N.PDF

    http://www.onsemi.com/pub_link/Collateral/NJT4031N.PDF

    Models

    http://www.onsemi.com/PowerSolutions/supportDoc.do?type=models&rpn=NJT4031NT1G

    http://www.onsemi.com/PowerSolutions/supportDoc.do?type=models&rpn=NJT4030P

  5. Bill_Marsden

    Bill_Marsden Thread Starter Moderator Staff Member

    Joined:
    Mar 24, 2008
    Messages:
    19,281
    Location:
    Dallas, TX (GMT-5 w/ DST)
    I don't think it is any inherent current limiting. Transistors will burn themselves out if allowed. One way to establish it though, add another 2N2222 and 2N2907 to make a Darlington. I'll redesign the circuit for that. My preferred Darlingtons is a TIP102 / TIP107 complementary pair. Don't forget the 1Ω is going to enter into the calculations too.

    I'll get back here when I have a redraw.
  6. eblc1388

    eblc1388 Senior Member

    Joined:
    Nov 28, 2008
    Messages:
    1,542
    Location:
    UK
    Hi Bill,

    Do you use any simulation package?
  7. Bill_Marsden

    Bill_Marsden Thread Starter Moderator Staff Member

    Joined:
    Mar 24, 2008
    Messages:
    19,281
    Location:
    Dallas, TX (GMT-5 w/ DST)
    Nope, need to start evidently. All you see comes out of my head and calculator.

    SgtWookie is pretty good with his. Yours is the second I've seen used.
  8. eblc1388

    eblc1388 Senior Member

    Joined:
    Nov 28, 2008
    Messages:
    1,542
    Location:
    UK
    Nope, I'm just a novice. Been using it for a few months only.

    Mostly used it to prove a simple circuit design actually works as it should before giving out the design to others.

    SqtWookie & Ron_H (and possibly many more) are the experts on this. :)
  9. Bill_Marsden

    Bill_Marsden Thread Starter Moderator Staff Member

    Joined:
    Mar 24, 2008
    Messages:
    19,281
    Location:
    Dallas, TX (GMT-5 w/ DST)
    OK, here is the revised schematic...

    [​IMG]

    So where did you get your SPICE program (I assume its name is Booster)?

    Attached Files:

  10. eblc1388

    eblc1388 Senior Member

    Joined:
    Nov 28, 2008
    Messages:
    1,542
    Location:
    UK
    I corrected your transistor connections and adjust the bias to give 20mA through the 1Ω resistor in the darlington. Then I did another simulation.

    Surprise, surprise!!!. The traces are identical to the results above so I don't bother to post them.

    [​IMG]

    The SPICE program is the free LTSpiceIV. The circuit that I'm simulating is called "Booster", which is just any name one can use to identify it among other circuit files.

    Attached Files:

    Last edited: Jan 15, 2009
  11. Bill_Marsden

    Bill_Marsden Thread Starter Moderator Staff Member

    Joined:
    Mar 24, 2008
    Messages:
    19,281
    Location:
    Dallas, TX (GMT-5 w/ DST)
    I suspect we are seeing several factors at work. With a 1Ω output impedance feeding 50Ω we would get a 2% drop in the signal if all else was perfect. With 25Ω it would be around 4%, 3Ω would be 75%, and at 1Ω it would be 50%. So some drop is expected. 1KHz through 1000µF is very close to zero (0.16Ω reactance). There would be some drop there of course, but very tiny.

    OK, so with a 3Ω we would get 75% of the signal. This would work out to 2.25V Peak, the simulator shows 2.0VPeak. At 10Ω you are getting 91%, or 2.7V Peak, and the simulator shows 2.6V. I don't know where the extra difference is (freq response maybe?), but it isn't as great as it first appears.

    When I get a chance I'll check out the simulator.
  12. Bill_Marsden

    Bill_Marsden Thread Starter Moderator Staff Member

    Joined:
    Mar 24, 2008
    Messages:
    19,281
    Location:
    Dallas, TX (GMT-5 w/ DST)
    Something to try, use a 1Ω resistor and a 1000µF capacitor between a signal source and feed the same impedances, so see how different the levels actually are. I'm wondering if the capacitor might account for the last little bit of voltage drop.
  13. studiot

    studiot E-book Developer

    Joined:
    Nov 9, 2007
    Messages:
    4,425
    Location:
    Somerset UK
    With only 12 volts supply rail I think you are loosing too much in the Darlington connections.

    Why not try the alternative of npn/pnp and pnp/npn complementary connections for upper and lower ouput compund transistors respectively?

    This will put ony two BE junctions in series with the load, rather than 4.
  14. Bill_Marsden

    Bill_Marsden Thread Starter Moderator Staff Member

    Joined:
    Mar 24, 2008
    Messages:
    19,281
    Location:
    Dallas, TX (GMT-5 w/ DST)
    It was an experiment to see if it was a gain issue. It wasn't, looks like the push/pull work on the first layout OK.

    [​IMG]

    Part of the problem is we (I'm including me) are so used to op amps with their feedback creating super low output impedances we forget what it used to be like.
  15. KL7AJ

    KL7AJ E-book Developer

    Joined:
    Nov 4, 2008
    Messages:
    1,607
    Location:
    North Pole, Alaska

    Hiya Bill:

    This is a classic TRUE complementary symmetry amplifier. In most audio applications, the transistors were run in Class B or class AB. You CAN run them in class A for minimal distortion, but the efficiency drops precipitously.

    The whole point of this circuit was to be able to drive a speaker or other low impedance load directly. It was typically driven with a class A voltage amplifier....sometimes even an op-amp. The circuit only became popular after some fairly beefy power transistors became available....of course a 2222 is not in that category!

    But it's a circuit everyone should be familiar with, because it leads to so many other circuits....and indeed is incorporated INSIDE many op-amps.

    eric
  16. Bill_Marsden

    Bill_Marsden Thread Starter Moderator Staff Member

    Joined:
    Mar 24, 2008
    Messages:
    19,281
    Location:
    Dallas, TX (GMT-5 w/ DST)
    This I know. When EB wrote me about the dropping level I thought something was wrong with my theory. I now suspect we're seeing the capacitor and output resistance have its say.

    Since I've started this thread I've downloaded his SPICE package, it'll take time for me to wrap my head around it. While I'm about it I may come up with a decent driver using an op amp and this circuit, it is a stable for newbies.
  17. eblc1388

    eblc1388 Senior Member

    Joined:
    Nov 28, 2008
    Messages:
    1,542
    Location:
    UK
    Hi Bill,

    I added two more "power" transistors to your existing circuit and can now drive a load as low as 1Ω. I will hold off post it here in order not to spoil the fun you are having with the new simulator.
  18. Bill_Marsden

    Bill_Marsden Thread Starter Moderator Staff Member

    Joined:
    Mar 24, 2008
    Messages:
    19,281
    Location:
    Dallas, TX (GMT-5 w/ DST)
    I'd like to see it when you get the chance. Transistors will burn themselves up nicely if you let them, I knew this was a problem with the original design. With 1Ω you'll only get 50% of the signal, best case, so I would like to see the differences.
  19. eblc1388

    eblc1388 Senior Member

    Joined:
    Nov 28, 2008
    Messages:
    1,542
    Location:
    UK
    Yes, sir. Your wish is my command.:)

    [​IMG]

    Attached Files:

  20. Bill_Marsden

    Bill_Marsden Thread Starter Moderator Staff Member

    Joined:
    Mar 24, 2008
    Messages:
    19,281
    Location:
    Dallas, TX (GMT-5 w/ DST)
    I like it. I was thinking of something similar to D1 and D2, but the output transistors are new to me. Did you come up with this, or reference another source? If the current requirments aren't too heavy maybe we can reference the original post I proposed my design it. In either case I think I'll add it to my personal cookbook.

    I notice some slight clipping on the positive side, but very small. Wonder what difference it would make if R10/11 were 1Ω instead? 1Ω is off the shelf.

    What is the current through R8/R9?

    You could put a speaker after this sucker and it would be a good driver. Use a ± power supply, eliminate the capacitors, you can tie it after an op amp.

    Build up a big enough cookbook and you can save huge amounts of time designing.
    Last edited: Jan 16, 2009
Similar Threads
Forum Title Date
General Electronics Chat Do I Need a Transistor Buffer? Dec 12, 2010
General Electronics Chat About PNP transistor 2N3906 Yesterday at 10:39 PM
General Electronics Chat Diodes and transistors replacements / equivalents database Nov 21, 2014
General Electronics Chat maximum current supported by NPN transistor Nov 18, 2014
General Electronics Chat Obtaining Collector Current Vs lumens for photo transistor setup Nov 14, 2014

Share This Page