Connect with us

Classic Chips

Discussion in 'Electronic Design' started by Paul Burridge, Aug 28, 2003.

Scroll to continue with content
  1. Great. So every time I have a problem connecting to my ISP (all too
    frequently I regret to say) one of these little bastards is to blame?
    I'd like to get my hands on the guy that designed them...
    ;->
     
  2. Jim Thompson

    Jim Thompson Guest

    No warranties express or implied ;-)

    Actually I no longer use serial ports myself... USB and Ethernet are
    so much faster.

    ...Jim Thompson
     
  3. Daniel Haude

    Daniel Haude Guest

    On Thu, 28 Aug 2003 11:24:41 +0100,
    I wouldn't call it "classic chip of all times", but a chip that at least
    to the hobbyist is of "exceptional merit or usefulness" is the 555. It's a
    comparator, a flip-flop, an open collector output, and a hefty push-pull
    stage all lumped together in a cheapo 8-pin DIP package. Apart from timer
    applications this circuit can be used and abused in all sorts of ways.

    When as a kid I first started to play with those mysterious things called
    "integrated circuits" (that was probably around 1981 or so, but the books
    I had were a bit outdated at the time, and spoke of ICs with a certain
    awe), it was the first chip I really understood. It was well within my
    price range and virtually undestructable. It made lights blink, relays
    click, and it was a good noise-maker.

    --Daniel
     
  4. Dana Raymond

    Dana Raymond Guest

    Just to show the versatility of the venerable 555 I used a cmos version in
    the early 1980s to protect a gellcell battery from deep discharge. Both of
    the comparators were used and it did the job at a time when there weren't
    dedicated chips available. The cmos version was used to limit the discharge
    current to a few tens or hundreds of uA (according to memory).

    Dana Frank Raymond
     
  5. Bill Sloman

    Bill Sloman Guest

    Hmm. I used them - back in 1973 - and they worked, but it was two
    chips, not one, and the bipolar amplifier in the the low pass filter
    filter had a lower input impedance than the CMOS amplifier in the
    4046, so the capacitors were bigger. The 4044/4046 application note
    was excellent.

    My vote still goes to the 4046, though I'd use a Philips 9046 these
    days, if I could get my hands on one.
     
  6. Bill Sloman

    Bill Sloman Guest

    The RS-232 serial link was originally a telecoms standard, for
    Teletypes and other telex terminals. It wasn't the only standard. The
    guy who originally "designed" it must be long dead, but "design" is
    probably not the right word for the process of compromise and
    negotiation that went on between the PTT and the manufacturer(s) who
    actually built the telex machines.

    I got involved with this around 1980 for the "teletex" standard, which
    was intended to be an upgrade to the telex standard, for linking
    IBM-PC like communicating word processors, and the process was a mess,
    with each manufacturer trying to protect their particular investment
    in development work, and the PTT insisting on backwards compatibility
    back to the Ark.
     
  7. Jim Thompson

    Jim Thompson Guest

    I'm not sure. It was after Motorola opened the Mesa Dobson/Broadway
    facility (1966?, 1967?) but before I laid myself off in 1970 ;-)

    ...Jim Thompson
     
  8. Jim, is the "beta fix" you mentioned earlier possible to explain
    easily?

    Best regards,
    Spehro Pefhany
     
  9. Jim Thompson

    Jim Thompson Guest

    I posted it on one of these newsgroups *many* moons ago. I'll see if
    I can find it and re-post.

    ...Jim Thompson
     
  10. N. Thornton

    N. Thornton Guest

    Hi.


    Lets try another angle. If we take the usage of the term classic usual
    with classic cars, a classic chip would be one that:

    was ahead of its time, back when, was clever and innovative
    looks kind of neat
    is horribly inefficient
    no-one would touch them now
    and is full of, well, 'personality'.

    I can think of one clear contender for that: the Russian DIL
    thermionic valve IC devdeloped in the late 1980s, which unforfunately
    sunk without trace before it even surfaced. But it sure was different!


    Regards, NT
     
  11. Phil Hobbs

    Phil Hobbs Guest

    Oh, well, if _that's_ the definition, how about the AD639 sine wave
    generator chip? It was a thing of great beauty, but it cost a lot and
    was slower and less accurate than a lookup table and a couple of 16-bit
    DACs. The AD630 monolithic lock-in amp was a good one too--slow as
    molasses, but it did the job for a lot of instruments people. The LM

    Others of this ilk are the LH0063 "damn fast buffer amplifier" and the
    LH0101 power op amp--an LF351 with a discrete power buffer. For that
    matter, do any of you remember the huge number of weird hybrid parts
    that National came out with in their 1989 Linear Data Book? They all
    disappeared very soon. The similarity to the 1969 Camaro Z-28 or the
    Mako Shark hadn't struck me before, but now I get it.

    Cheers,

    Phil Hobbs

    Oh, I forgot the LF456 op amp, which was an LF356 with an external
    current limit resistor for the output stage. Am I the only one who
    still has a Lista bin full of these?
     
  12. Phil Hobbs

    Phil Hobbs Guest


    Oh, well, if _that's_ the definition, how about the AD639 sine wave
    generator chip? It was a thing of great beauty, but it cost a lot and
    was slower and less accurate than a lookup table and a couple of 16-bit
    DACs. The AD630 monolithic lock-in amp was a good one too--slow as
    molasses, but it did the job for a lot of instruments people. The LM11
    picoamp _bipolar_ op amp was a great design except that its noise was
    horrible--150 nV, 4 pA in 1 Hz.

    Others of this ilk are the LH0063 "damn fast buffer amplifier" and the
    LH0101 power op amp--an LF351 with a discrete power buffer. For that
    matter, do any of you remember the huge number of weird hybrid parts
    that National came out with in their 1989 Linear Data Book? They all
    disappeared very soon, and no wonder. The similarity to the 1969 Camaro
    Z-28 or the Mako Shark hadn't struck me before, but now I get it.

    Cheers,

    Phil Hobbs

    Oh, I forgot the LF456 op amp, which was an LF356 with an external
    current limit resistor for the output stage. These were really things
    of great beauty. Am I the only one who still has a Lista bin full of them?
     
  13. Jim:

    [snip]
    [snip]

    I believe I bought about a ton of them in 1967 - 68 time frame.
     
  14. Jim:

    [snip]
    [snip]

    Agreed, same here... but...

    It seems that like the floppy disk drive, the RS-232 port won't die, like
    old soldiers it will just "fade" away, every system design I do these days
    still seems to have to have at least serial port somewhere.
     
  15. Jim Thompson

    Jim Thompson Guest

    [snip]
    In the S.E.D/Schematics page of my website I have posted an *updated*
    version of "EnhancedCurrentMirrors.pdf" with added pages showing the
    "Thompson" current mirror compared to the other types when transistor
    beta is very low (as in the lateral PNPs found in integrated
    circuits).

    As with many of my circuit creations this was the result of sharing a
    cubicle with Tom Frederiksen for many years at Motorola. We would
    daily challenge each other with "bet you can't". The "Thompson"
    mirror went on to be included in many of Motorola's products... a good
    example is in the MC1494... see page 6 of
    http://www.onsemi.com/pub/Collateral/MC1494-D.PDF

    For my efforts I won a cup of coffee from Tom. Likewise I garnered
    another cup of coffee for the all-NPN output stage of the MC1554 audio
    amplifier. But I then lost a bet with Tom as to how much power you
    could get out of it... I came to work one morning to find that Tom had
    mounted an MC1554 (TO-5) in a hole in a block of aluminum and was
    piping liquid CO2 through other holes in the block. I can't recall
    exactly how much power, but I think it was around 10 watts ;-)

    ...Jim Thompson
     
  16. Jim Thompson wrote...
    I've often used your "Thompson mirror" in my designs, where it seemed
    an intuitive and obvious choice for certain purposes. Are you sure
    that nobody had used it before you? Anyway, I've found it difficult
    to accurately match the NPN and PNP Vbe voltages, placing an extra
    emphasis on the need for two current-matching resistors. For me one
    advantage of this configuration is its ability to rapidly shut off
    high-capacitance output transistors in a dynamic-mirror circuit. A
    corresponding disadvantage in circuits with large output devices is
    its need for high standing bias currents to achieve quicker turn on.
    This aspect proved to be a painful issue in a high-voltage amplifier
    I designed recently.

    .. quick turn off quick turn on
    .. -----+-----+-----+------------+-----------+----
    .. | | | | |
    .. Rx | Rx Rx Rx
    .. | |/ | | |
    .. +---| | V\| |/V
    .. | |\V | |---+---| ---- etc
    .. | | |/V /| | |\
    .. in +---| --- etc | |/V |
    .. | |\ +---| out
    .. wasteful | | |\
    .. bias out in |
    .. gnd

    The PNP mirror shown in the MC1494 "complete schematic" is different
    from the "Thompson mirror" you show (left above). It's also a form
    (right above) that I often use in my own designs, and have seen many
    times in various ICs, but I don't recall its name, if it has one.
    (Incidentally, it's good at the turn on I problem mentioned above,
    but poor at turn off.) Anyway, it's not clear to me which one of
    the circuits in your list is supposed to be the one in the MC1494.

    Can you find a circuit example of a common IC (or one you designed)
    using the "Thompson mirror?" I know it shouldn't be hard to find one
    but I can't think of an example right now. IIRC, the IC engineers at
    Transitron used it in the 60s in a complex power-supply regulator IC.

    Thanks,
    - Win
     
  17. Jim Thompson

    Jim Thompson Guest

    Nope, see below.
    In the *true* Thompson mirror there are other non-published (until my
    demise :) tricks to eliminate the Vbe match requirement. Transistors
    are cheap in ICs.
    Win, you erred when viewing my circuit. The "Thompson" mirror
    connects the NPN *collector* to the PNP *emitter*, such that it
    behaves as a *Kirchoff machine*... what goes in must come out, thus
    the beta insensitivity. Did you actually look at the curves versus
    beta? The "Thompson" mirror dies at a PNP beta of ONE! Thus its
    usefulness with Lateral PNPs.
    "(left above)" is NOT my mirror, please study p4 of
    "EnhancedCurrentMirrors.pdf" on the S.E.D/Schematics page of my
    website.
    Once you recover from last night's drinking ;-), and your normal
    skills of observation return, please go back to the website, press
    refresh since I've added a cut from the MC1494 showing my mirror.

    The MC1494 was done as a Master's Thesis (Arizona State University) by
    Maurice Free, under my tutelage, pre-1967 (I'll have to track down the
    correct date, I don't remember right now... I ultimately had to fire
    him, he was prone to fist fights :)

    ...Jim Thompson
     
  18. Jim Thompson wrote...
    Why wait around. We won't be interested and you won't be there to
    provide explanations then, it'll just be too late. That's why I've
    long since stopped keeping my specialized knowledge to myself alone.
    I'm glad to see your following the same course. But don't worry, it
    won't have a negative effect on your income, probably the opposite.
    .. "Thompson" mirror
    .. -----+-----------+----
    .. | |
    .. R1 R2
    .. | |
    .. | ,-----+ R1
    .. | |/ | out = in1 -- - in2
    .. +---| | R2
    .. | |\V |
    .. | | |/V
    .. in1 +---| --- etc
    .. | |\
    .. in2 |
    .. out

    OK, got it now, and I like it. I_out = in1 R1/R2 - in2, roughly?

    I've often used a similar concept in my designs. For example, it's
    especially useful for precision high-voltage current sources made
    with low-gain transistors. In the drawing below Q3 is a high-beta
    low-voltage transistor, and Q4 has the 300V rating. The high base
    current of Q4 (plus the resistive divider current) is ignored with
    the same 'what goes in must come out' Kirchoff-machine property.
    A mismatch is introduced into Q1 and Q3's Vbe, but in high-voltage
    circuits one can simply use higher Rx values to minimize the effect.

    .. high-voltage current source
    .. -----+-----------+---- +320V
    .. | |
    .. | Rx
    .. Rx |
    .. | +-----,
    .. | | |
    .. V\| |/V |
    .. |---+---| |
    .. /| | |\ Q3 |
    .. | |/V | R3
    .. +---| | | out = in
    .. | |\ V\| |
    .. in | |---+
    .. | Q4 /| |
    .. | | R4
    .. gnd | |
    .. +-----'
    .. |
    .. out
    Thanks, that was useful, I suggest you leave it in!
    That would be interesting.

    Thanks,
    - Win
     
  19. Jim Thompson

    Jim Thompson Guest

    On 1 Sep 2003 13:33:39 -0700, Winfield Hill <>
    wrote:

    [snip]
    [snip]

    Why not.......

    high-voltage current source
    -----+-----------+---- +320V
    | |
    | |
    Rx Rx
    | |
    | |
    V\| |/V
    Q1 |---+---| Q2
    /| | |\
    | | |
    | | |
    | |/V |/V
    +---+-----| Q3 out = in
    | |\ |\
    in Q4 | |
    | |
    | |
    gnd out

    Q3 and Q4 are HV
    Q1 and Q2 are LV

    Zero volts collector-to-base is quite legal and is often overlooked by
    non-IC designers.

    If you really want to be exotic add a diode-connected transistor
    between the base of Q4 and the collector of Q1, then all mirror
    devices are operating in complete symmetry with VCB=0.

    ...Jim Thompson
     
  20. Jim Thompson wrote...
    Hey, Jim! Thanks for replying with an ASCII schematic, it
    really smooths the conversation and leaves a better record.
    Indeed, although having read out book, and being well aware
    of saturated-switch operation, they have little excuse!
    I've used such an approach, but the method I showed allows one
    to go to much higher voltages, e.g., 2x if both Q3 qnd Q4 are
    HV types. For example, consider +/-250V supplies, with the
    mirror connected to the top rail, driving a +/-240V output
    stage, and its input from an opamp near ground operating on
    +/-15 etc, via a BJT level shift. My cascode circuit allows
    using common 300v transistors all round, with a 50-volt margin.
    Otherwise one has to come up with a small low-capacitance 550
    to 600V pnp transistor, not easy to find. This scheme can be
    easily extended with more cascode stages for higher voltages.

    One issue with your circuit, for use with accurate mirrors,
    etc., given that Q3 and Q4 are probably low-gain high-voltage
    transistors. This form can be helpful to improve the accuracy,
    given that high Q3 Q4 base currents hurt you twice: first when
    taken from the programming input current and second when robbed
    from the proper output destination.

    .. high-voltage current source
    .. -----+-----+-----------+---- +320V
    .. | | |
    .. Rx | Rx
    .. | | |
    .. V\| | |/V
    .. Q1 |---------+---| Q2
    .. /| | | |\
    .. | Ry | |
    .. | | |/V |/V
    .. | +---| ----| Q3 out = in
    .. | | |\ |\
    .. | |/V Q4 | |
    .. +---| | |
    .. | |\ Q5 | out
    .. | | gnd
    .. | |
    .. in gnd

    Note, Ry doesn't have much current flowing, but this helps
    turn Q3 Q4 off faster, and reduces high-temp leakage, etc.
    If within an IC, a coupl'a diode-connected transistors in
    Q1's collector, natch.

    Thanks,
    - Win
     
Ask a Question
Want to reply to this thread or ask your own question?
You'll need to choose a username for the site, which only take a couple of moments (here). After that, you can post your question and our members will help you out.
Electronics Point Logo
Continue to site
Quote of the day

-