Design Project Help Wanted

Discussion in 'The Projects Forum' started by undertheradar, Oct 19, 2010.

  1. undertheradar

    Thread Starter New Member

    Oct 19, 2010
    14
    0
    Hello... long time reader, first time poster...lol.

    I have a "competitive" project, and I have done a good deal of research, but if there is something out there that you pros know that would benefit me, I'm all for it.

    GOAL: Design a BJT Differential Amplifier with the highest CMRR we can get.
    -Direct Coupled Input (no caps on inputs or in the amp)
    -Single Ended Output (cap on output)
    -Assume +15 and -15 power supplies are available
    -Use a 10k loading resistor as loading device
    -only use parts in kit, no IC's: 2N3904's, 2N3906's, diodes and resistors.


    Now, "by the book" I am familiar with how I should pick the best current source. We have studied current mirrors, Widlar & Wilson. We have not looked at any others, but Iwonder is there some reason I should look into the Cascode? Is there a way to make a Cascode with BJT's? We didn't study these, but they aren't off limits.

    We have studied balanced output, single-ended output, emitter follower, common emitter, balanced output direct coupled to input of second stage, etc... amplifier topologies.

    So then I hit the interwebz to see what is all out there, and I see things... things that cant be unseen. :D I also get a little creative:

    One idea I had was that even though the negative rail that is supplied is only -15 volts, I could boost it with a BJT oscillator and voltage booster (caps and diodes) to get it to say... -30 volts (small ripple that a cap might take care of, but this also drops the current a bunch). Then we could run multiple current sources in series to boost the stability. Having a wider rail-to-rail voltage seems like it would be an advantage to me at least.

    Yesterday the professor stated in class as an "exception" to the rule of no IC's was that if we needed to boost the output to see on the oscilloscope, we could use a LM741 in feedback amp mode. He seemed to suggest that this was a 'hint' as well... that to get the highest possible CMRR, that we would have a very small output signal... not sure about that.

    So, I wanted to post this here and see if anyone had some "magic sauce" they might know of that would really make for an advantage. Otherwise, this seems like a rather boring project; just number crunching with the given topologies. Hey, at least Im learning a bunch of stuff that we haven't seen in class, but Im not sure if using any of them would be an advantage. Darlington Transistor Pair, Sziklai Pair, Wilson Fudh Current Source, diodes for voltage drop rather than resistors.

    Thanks in advance for any ideas...
     
  2. R!f@@

    AAC Fanatic!

    Apr 2, 2009
    8,733
    759
    Boy U sure did layed it out.

    Welcome by the way..

    Why don't u draw up a schema, do some number crunching and we can go from there.
     
  3. undertheradar

    Thread Starter New Member

    Oct 19, 2010
    14
    0
    Thanks for the welcome.

    I have been trying many crazy schematics to see what is what... from multiple university websites and such, and found that when actually plugged into LTSpice, they dont work as well as they claim.

    I am also working on a multi-stage amplifier, but even though it works in simulations, I cant get it to work IRL.

    So I made this as a backup... gets me 116dB for the CMRR, but Im sure there is room for improvement so here goes...

    [​IMG]

    My concern is matching the betas of the 2N3904's. I tried the "R?" resistor and changed the beta above it (put in a darlington of two 2N3904's for RW3) but it didn't work (although it did boost the CMRR a decent bit from 112dB for some reason that I cant understand). What can I do to control the betas though? Add emitter resistors everywhere in the Wilson's Mirror? That doesn't work.

    There is an "improved Wilson's Current Source" there, unless I connect the "?" node and bypass QW3, but it seems both work exactly the same here.

    Otherwise, Im all ears for suggestions on how to get an even higher CMRR or suggestions on how to make this thing work well when actually constructed.
     
    Last edited: Nov 2, 2010
  4. Kermit2

    AAC Fanatic!

    Feb 5, 2010
    3,765
    925
    matched pair of NPN's

    http://www.analog.com/static/imported-files/data_sheets/SSM2210.pdf

    They are not expensive(really) and might serve you better as a component, since matching is done at manufacture, it would be a drop in part and not need to be 'tweaked' so much if replaced, as discrete transistors would have to be.

    anything over 110 db of rejection is excellent. 116 db CMRR is beyond excellent, expecially for discrete components.

    Look up(google) opamp offset adjustment circuits, as that might add a few more db's of rejection at the input pins
     
  5. Kermit2

    AAC Fanatic!

    Feb 5, 2010
    3,765
    925
  6. undertheradar

    Thread Starter New Member

    Oct 19, 2010
    14
    0
    @Kermit2,
    From my first post: "-only use parts in kit, no IC's: 2N3904's, 2N3906's, diodes and resistors."

    Matched pair npn's would be awesome if this was an IC design or I wasn't restricted by the project parameters... I am restricted to 1N914 diodes, LED diodes (2V drop each), I do have a zener (forget which one), 2N3904's, 2N3906's, etc...

    One thing that the professor said was that we could use an op amp to amplify the output if needed (he went "wink wink-hint hint" with this) because our output signal might be so small that the scopes may not pick it up otherwise. Some of my peers said that they were measuring nano-volts... perhaps this was just with reference to the common mode though.
     
  7. Kermit2

    AAC Fanatic!

    Feb 5, 2010
    3,765
    925
    The circuit is for a transistor tester. for MATCHING them,.

    :)
     
    undertheradar likes this.
  8. thatoneguy

    AAC Fanatic!

    Feb 19, 2009
    6,357
    718
    Have you looked into the trusty amplifier (audio and other) input of a Long Tailed Pair?

    High CMRR using the differential input. I didn't notice a limit on number of transistors.

    Look at the internal schematic of an opamp, though dual emitter and dual collector transistors are a bit rare outside of ICs. :D
     
  9. undertheradar

    Thread Starter New Member

    Oct 19, 2010
    14
    0
    I had considered a long tail pair prior to this... I just coundn't find resistor values that would get the thing to work right though... It would require elimination of my RCO resistor (or could I leave it in somehow?) and that makes things complicated...

    I would like to, but its something that we didn't really get into in class, so I really dont know how to go about changing the other components to make it work...
     
    Last edited: Nov 3, 2010
  10. thatoneguy

    AAC Fanatic!

    Feb 19, 2009
    6,357
    718
    This should give you a few ideas. There are many others, but I wouldn't outright copy one.

    Presented to get your mind working in "The Zone", so to speak.
     
  11. ifixit

    Distinguished Member

    Nov 20, 2008
    638
    108
    Hi undertheradar,

    BJT gain changes slightly as the Vce changes, therefore unmatched BJT's will cause a CM signal.

    BJT junction capacitance differences will cause CM signals. The higher the frequency, the higher the CM signal.

    CM range and frequency are not specified.

    Hope this isn't cheating, but if you lower the signal input level from 1 volt to 100mV and reduce the frequency to 100Hz from 1KHz then you should see an improvement in CMRR.

    Since the output is only 1.58uV with a 1 Volt CM input and a CMRR of 116dB, it will be much more difficult to measure in the real world than in the LTspice world. The LM741 has limits too.

    Regards,
    Ifixit
     
    undertheradar likes this.
  12. marshallf3

    Well-Known Member

    Jul 26, 2010
    2,358
    201
    http://en.wikipedia.org/wiki/ENIAC

    Reliability

    ENIAC used common octal-base radio tubes of the day; the decimal accumulators were made of 6SN7 flip-flops, while 6L7s, 6SJ7s, 6SA7s and 6AC7s were used in logic functions. Numerous 6L6s and 6V6s served as line drivers to drive pulses through cables between rack assemblies.

    Some electronics experts predicted that tube failures would occur so frequently that the machine would never be useful. This prediction turned out to be partially correct: several tubes burned out almost every day, leaving it nonfunctional about half the time. Special high-reliability tubes were not available until 1948. Most of these failures, however, occurred during the warm-up and cool-down periods, when the tube heaters and cathodes were under the most thermal stress. By the simple (but also expensive) expedient of never turning the machine off, the engineers reduced ENIAC's tube failures to the more acceptable rate of one tube every two days. According to a 1989 interview with Eckert the continuously failing tubes story was therefore mostly a myth: "We had a tube fail about every two days and we could locate the problem within 15 minutes."[10] In 1954, the longest continuous period of operation without a failure was 116 hours (close to five days).
     
  13. undertheradar

    Thread Starter New Member

    Oct 19, 2010
    14
    0
    @ifixit,
    yes, that design I made is really only good for up to about 1v before it starts clipping the output. The professor never really designated any frequency or voltages. He mentioned it in class once that 1v would be a good input amplitude. Frequency is not too important, so thanks for the suggestion on lowering it. I had started at 100Hz for this very reason, but in class 1kHz kept getting suggested by the professor as well as peers, so it seemed like a good benchmark. Still... if lowering it to 100 makes that kind of improvement... I'm in.

    The one thing I cant explain is that R? resistor of 100ohms. It leaves the DM the same, lowers the CM, and I cant replicate its effect with the RW, using diodes instead, or placing a resistor elsewhere. Very odd.
     
  14. jkrohn22

    New Member

    Mar 26, 2011
    1
    0
    Hello did you get your circuit to work?
     
Loading...