Monitoring System Phase II

From GlueXWiki
Jump to: navigation, search

Dear Elton, Here is the very detailed answer of Kappos to your two questions.

Cheers, Christine


Original Message --------

Subject: Re: Epi pleon erwthseis

Date: Mon, 28 Mar 2011 14:19:19 +0300

From: E. Kappos <ekappos@otenet.gr>

To: Christina Kourkoumeli <hkourkou@phys.uoa.gr>

1. The signals/wires needed for the control box are

  T1, T2, T3, T4  (4 triggers, one for each daisy chain of 10 LEDs)
  Vcc                  (for digital buffers)
  Vbias               (0 to about 25 Volts)
  GND               common (ground)
 ------------------------
  The triggers are assumed 0/5V and will produce a local trigger
  on the positive edge 0 -> 1. They will be regenerated locally
  to sharpen their edges, which will be degraded due to long cables,
  typically using a Schmitt buffer i.e. a buffer with hysteresis, to get
  rid of "bumps" (glitches) in the rising edge. The buffer will be high
  impedance. They could share the same return wire (GND), and
  supplied with ribbon cable.
  Note: the rising edge 0 ->1 is converted locally to a 100ns pulse
  (0 -> 1 ->0) and it is this pulse that triggers the LEDs on the small
  boards. Therefore the constraint is that the rising edge lasts for at
  least 100ns (i.e. each trigger T1,...T4, is longer than 100ns, say
  200ns).
  Vcc is 5V and can share the same return wire (GND) as the triggers.
  Vcc will be filtered locally, for noise.
  Vbias will be supplied externally and filtered locally.
  If there is a concern about noise injection, Vbias may have its own
  return wire (GND_B) in addition to the digital&power GND.
  This is preferrable, in my view, just in case noise crops up somehow.
  In that case, Vbias with its GND_B, could be supplied by ribbon cable.
 All signals will be protected for electrostatic build up and overvoltage.
 Vcc will be protected against reverse bias and fused (resettable fuse,
 recovers automatically, i.e. no need for replacement)
  CONCLUSION:
     A. All signals can be supplied by ribbon cable
     B. Vbias will preferrably have its own return wire (GND_B)
     C. Wires needed (8):
           T1, T2, T3, T4, +5V, GND, Vbias, GND_B

2. This is already accounted for.

   The LED capacitors are only 100pF, so 40 of them in a box are only 4nF.
   Using 4 x 100nF decoupling capacitors for each of the 4 daisy chains
   is well overdesigned to account for the re-supply of the discharged
   caps.
   In the final experiment, there will be 96 boxes, i.e. 96 x 4 x 100 =
   38.4uF. This is small by electronic standards, i.e. a 200uF  cap
   (tantalum or electrolytic) will easily decouple this in the 'external
    box' that produces Vbias.
   However, to smoothen the (small) in-rush current
   while the caps are charged up, I'm considering putting a small series
   resistance in series with each of the 4 x 100nF caps on the control
   board. In this case, I need to know the max re-charge (i.e. trigger)
   rate that will be used (1kHz, 10kHz, 100kHz ???)
 ************************************************
   BEAWARE - CAPACITOR LEAKAGE CURRENT WILL
  AFFECT Vbias ARRIVING AT EACH BOX VIA EXT WIRES
  BY A SMALL AMOUNT (TYPICALLY SMALL FRACTION
  of 1mV)
  The DC leakage current to ground of all 4 decoupling caps plus the
  40 LED caps (100pF each) will create an ohmic drop on the external
  (Vbias, GND_B) wires. If the 96 boxes are daisy-chained, this drop
  will be different for each box as they share different sections of their
  return wires. If they are in a star topology with equal-length wires,
  then the Vbias drop to each box will be identical.
 In any case, assuming
      25V operation,
      1km of ribbon cable (250 Ohm/km), AWG 28
      a pessimistic 10000  MOhms insulation resistance for each cap
  we have 2.5nA/cap leakage, i.e. 100nA per box of 40 caps
  which on a 1km cable gives a drop of 25 uV.
 If the insulation resistance drops to 1000 MOhms, say due to
 humidity, the drop over 1km cable will be 0.25 mV.
 If I add protection to the Vbias wire against overvoltage or
 electrostatic build-up, this is expected to add some microamps
 of leakage, depending on the device chosen. This may result in
 a few mV of voltage drop for 1km wire.
I assume this effect can be considered negligible,
*** unless I'm notified otherwise. ***
************************************************

3. I see basically two ways

  A.  1. The LED is soldered onto the small board at the required angle
              (e.g. 15 deg)  using an appropriate assembly jig
             (basically a 15 deg wedge)
        2. The light guide (1mm diam. fiber) is glued vertically onto the
             active area of the LED for max light coupling.
              The guide is pre-cut to the right length.
        3. The small board with soldered LED and glued fiber is slided
             into position on the plexiglass pyramid, and the light guide
              is inserted into the pre-drilled hole and held there by glue
  B.  1. The small board (without the LED) is glued into a fixed position
              on the plexiglass pyramid.
        2. A flexibe 2-wire cable (typically 2cm to 3cm) hangs over the
             edge of the small board with the LED soldered at its loose
             end.
        3. The LED is placed (glued) onto the extruding end of the
             lightguide which is already fixed in its position in the
             pre-drilled hole in the plexiglass.
  With option B, the difficulty seems to be attaching the hanging LED onto
  the pre-fixed light guide.
  If the guide is made with some "receptor" at its end, this may be eased.
  With option A, everything is pre-fabricated before put in place.
  Complete pre-fabrication can also apply in B, if the light guide is
   attached (glued) to the free-hanging LED, and then the final step is
   inserting the guide into the drilled hole and glueing the
  small board on the plexiglass.
 Note: if there is concern about the light absorption by the bottom side
  of the small board, then
    1. the bottom side can be made reflective by glueing some aluminium
         foil to it
    2. the small board could be placed on tiny spacers (if possible) so
        that its doesn't contact the plexiglass

Best regards,

Dr Euthymios Kappos ekappos@otenet.gr



Original Message -----

From: "Christina Kourkoumeli" <hkourkou@phys.uoa.gr> To: "E. Kappos" <ekappos@otenet.gr> Sent: Sunday, March 27, 2011 05:58 PM Subject: Epi pleon erwthseis


This afternoon we met and discussed various issues related to the electrical and cooling systems. A couple of comments/questions came up regarding this morning's discussion:

1. Cabling to the controller boards requires a) LED bias voltage b) low voltage +5V c) four trigger signals. A question this morning was whether we could use +5V from the other electronic boards. We are told that yes, in principle, but they are concerned with noise issue on small signals, so prefer that this be input from an external supply. Therefore, we need an additional line. Question: could we two small ribbon cables (4-pair each), one for the voltages and the other for the trigger signals? Coax could in principle be used, but connectors become problematic. We assume that the controller will regenerate the trigger signals anyway, so we assume that ribbon cable should be ok. Could you check this?

2. Comment: If all LEDs fire from one controller board fire at once, it might require considerable instantaneous charge to recharge the capacitors. Therefore, we assume four relatively large capacitors would be required on the controller board to handle the load.

3. We also had several discussions about mounting LED boards to the light guides and how to make the connections of the fiber to the light guide and how to support the board itself. This might argue for a sturdy connection for the board and a flexible connection on the LED.

Filika, X.Kourkoumeli