Connect with us

Power supply games

Discussion in 'Electronic Design' started by George Herold, Feb 14, 2013.

Scroll to continue with content
  1. Hi guys, Doing a simple circuit.


    V(+)------+------+
    20-30V _ |
    ^ LED|\|
    +----|+\
    | | >-+--out
    R +-|-/ |
    | | |/| |
    | +---)--+
    GND |
    Vminus

    The LED is a GaP one that avalanches at ~25V*
    R is 100k
    I need V minus to be 3 or so volts below ground.
    (V+ to the opamp will get the usual 0.1 bypass cap and some R between
    it and the led.)

    I thought I'd power it from a Phihong 48 volt wall wart.
    With a LM317HV for the 20-30 (+Vminus) side.
    I first drew it up with another lm317 to make the ground above
    Vminus. Duh... it can't sink any current.
    An LM337HV would work.. but it looks to "be no more".

    Any ideas for making my ground? At the moment I'm thinking I could
    just use another opamp. (a dual opamp is easy.)

    George H.

    *So the LED is an AND114R, I ordered some other GaP LEDs from the same
    company to see if I could get them to break down. There's these
    AND113R's that have very similar specs.
    (same wavelength 700nm but a bit more light output.)
    I cranked 'em up to 50V... no breakdown, another supply,
    100V.. no breakdown, another supply... finally at ~125 V the
    AND113R's avalanched. The avalanche was also photosensitive. maybe
    more so... But 125V is way too scary! And WTF is going on in the
    AND114r's?
     
  2. Hey, I was reading the LM337 spec sheet.. looking for the maximum
    input voltage. (none listed...) And it struck me, It's only the
    voltage difference (Vin to Vout) that the chip knows about... I could
    use this to regulate -100 Volts down to -90.
    (Sorry I'm brain dead some days.)

    George H.
     
  3. Thanks Phil... I'd forgotten about the protection diode! (It's always
    the simple circuits that I seem to screw up (but only a little :^)
    I can add some resistance in the power lead to slow down the turn
    on.
    The Phihong 48 volt supply is listed as +/-5% load line regualtion.
    Does that mean I'll see output voltages from 45.5V to 50.5? (unit to
    unit variation.)

    That looks to make the design a bit 'tight' on the Vminus side.

    George H.
     
  4. This one LED turns into a single photon avalanche photodiode.
    (Here's a plot of the statistics.. I've got some better data but this
    will do,)

    http://bayimg.com/fajaJaAEk

    A colleague’s built a counter - timer that was mostly made to go on
    the end of a PMT experiment. But this is a much cheaper way to count
    single photons.
    (I wish I could get some solid state expert interested in it.. I’d
    love to understand what going on in this one LED.)

    George H.
     
  5. Guest

    Looks like all you need is a resistor in parallel with the pseudo-APD. If you want the OA IN(+) at nominal 3V above GND at V+ of 20V then that's a divider ratio of 3/20, then to get a 100K resistance that would be a a 750K inparallel with APD and 120K from APD||120K junction to GND. Then IN(+) is 0..14 x V+ ranges from 2.8V -> 4.2V as V+ ranges 20V->30V. If the V+ is variable and you don't like the IN(+) junction variation then inject a constant current into the 100K, or ac-couple the OA output.
     
  6. Hmm, the output is an RC pulse response. RC recharging determined by
    the 100k ohm resistor and LED capacitance.
    The count rate is proportinal to the light intensity. (I haven't
    looked over a lot of the bias range. At higher bias voltage there are
    more 'dark' counts but also more 'signal' I haven't plotted any of
    that out.)
    So I guess the strongest evidence is the histogram of the time between
    pulses.
    (The graph I posted.) The number of occurences of any time difference
    exponentially decreases from zero time, with a slope that relates to
    the average count rate. So it's some random process. If you want to
    suggest something other than photons then I'm all ears (or eyes). At
    some point this is just a random pulse generator to do the statistics
    with.

    There is some 'after pulsing' More counts at a few microseconds than
    you'd expect from just the count rate. Apparently this is also common
    in other SPAD's.
    Here's some better data that shows the 'after pulsing' in the first
    bin.
    http://bayimg.com/CaJdMaaEk

    I'd like to look at the spectral response too, but at that point it's
    alomst a research paper.

    I've offered to send these to a bunch of physics types and anyone on
    the SED or SEB.. Send me your name and address or order your own from
    Newark/Allied AND114R. I'm pretty sure you have the 100k ohm
    resistor :^)

    The circuit's up and running I'll spend the rest of the day getting
    some dark count vs bais voltage numbers.

    Oh I measrued some (11) 12V Phihong supplies, they had a nice tight
    spread of voltages. min was 12.08 V max was 12.14 V (50 ohm load)

    George H.
     
  7. Guest

    On Thursday, February 14, 2013 6:56:54 PM UTC-5, George Herold wrote:

    Alternative CCS if you need that:

    Please view in a fixed-width font such as Courier.

    ..
    ..
    ..
    ..
    ..
    .. 20V-30V
    .. 317 OUT--------------+-----------+-----> V+
    .. | |
    .. | |
    .. [220] APD
    .. | |
    .. 317 ADJ--------------+ +-----> IN(+)
    .. | |
    .. [2.7K] [100K]
    .. | |
    .. | _ |
    .. | /| |
    .. [2K] |
    .. /| |
    .. | |
    .. +-----------'
    .. |
    .. [560]
    .. |
    .. |
    .. ---
    .. ///
    ..
    ..
    ..
    ..
     
  8. Thanks Fred. I need the OA in(+) at 'ground'. The output signal goes
    into a counter that's got a 0 to 1 Volt comapartor on the input. I
    guess an opamp that went down to the negative rail would work. (I
    don't know a lot of those) but something ~8-10 MHz, with Cin ~3-10pF
    might be worth a look see. Oh it should have ~36V supply voltage.

    George H.
     
  9. Guest

    On Friday, February 15, 2013 4:13:56 PM UTC-5, George Herold wrote:

    This is not a precision analog application, this will be more than enough, no need for exotic OAs or dual PS:
    Please view in a fixed-width font such as Courier.

    ..
    ..
    ..
    ..
    ..
    .. 20V-30V
    .. 317 OUT----+-----------+-----> V+
    .. | |
    .. | |
    .. [220] APD
    .. | | | \ 2N3906
    .. 317 ADJ----+ +-------|+ \
    .. | | | >--+--[Re]--E C----+-->TRIG
    .. [2.7K] [100K] .-|- / | |
    .. | | | | / | B |
    .. | _ | | | | [Rpd]
    .. | /| | '--------' | |
    .. [2K] | | |
    .. /| | | ---
    .. | | | ///
    .. +-----------' |
    .. | |
    .. | |
    .. [72] |
    .. | |
    .. +-------------------------------------'
    .. |
    .. [510]
    .. |
    .. ---
    .. ///
    ..
    ..
    ..
     
  10. josephkk

    josephkk Guest

    Cannonically that means that the output will change no more that 5% over
    the entire range on input, or no more that 5% of the line change,
    whichever is largest.

    ?-)
     
  11. Guest

    LOL_ not at all...
     
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

-