1. Discussion on the preamp and calibration circuit and layout
G. Visser, IU
7/3/2014

2. Two-stage preamp
Essentially same circuit previously discussed, prototyped
Noninverting first stage (for flattest input impedance, lowest noise), inverting 2nd stage
Transimpedance gain −3188 Ω
Add the switched cal signal into 2nd stage input (which polarity?)
+3.60 V 31 mA
14.3 W per boardstack

3. Two-stage preamp
Power switching requirements…
Have to be able to kill the preamp power dissipation
No option to run without preamp – no point to that
No need (IMHO) to selectively kill some channels power (discuss…)
Don’t use the powerdown pins of the LMH6629, just kill the 3.6V power supply
It serves no other loads (besides the preamps & cal circuitry)
Slightly simplifies routing in preamp area
One less way for spurious coupling / crosstalk to happen
+3.6 V regulator circuit…
1 A load
Drop ~1 V on linear regulator, ~1 W per carrier board
Pass transistors or integrated regulator IC’s should be distributed over several locations directly adjacent to thermal tab areas of the board
Besides the thermal advantage, that is the only way to fit it in the front section of the board

4. Calibration circuit requirements
Bandwidth at least 500 MHz; to 1 GHz probably would be preferable
 Justification: See the 508 MHz BPM signal cleanly
Cal circuit shall not be a limiting factor in bandwidth of PMT signal path
PMT signal shall not pass through a low-bandwidth mux like ADG3248
The general proposal is to sum a switched cal signal with the (always enabled!) PMT signal...
Cal circuit shall not introduce crosstalk of PMT signals from one channel to another (more than 0.1%)
Only at most one channel needs to get the calibration signal at a time
Of course, no channels getting calibration signal is also a required mode
Peak amplitude of calibration signal seen by IRSX should be supported up to 500 mV
Switching times: 400 ns (??) global enable, 1 ms (??) channel select
Calibration signal crosstalk (to unselected channels): 5% ok?? or 0.1% necessary??

5. Fanout on carrier board – tree option w/ 1:5 demux
PE42452A-Z
4 x 4 mm2
all outputs disabled mode exists
5 −3 dB
5 −3 dB
1:5
1:5
1:5
1:5
3 −2 dB
3 −2 dB
parking spot
parking spot
1:5
1:5
3 −1 dB
3 −1 dB
1:5
1:5
from SCROD
5 −2 dB
5 −2 dB
(the “parking spots” are desirable to tie down “long” lines passing near our sensitive circuitry)

6. Fanout on carrier board – tree option B w/ 1:5 demux
PE42452A-Z
4 x 4 mm2
all outputs disabled mode exists
4 −3 dB
4 −3 dB
1:5
parking spot
1:5
n.c.
n.c.
1:5
1:5
4 −2 dB
4 −2 dB
1:5
1:5
1:5
4 −2 dB
4 −2 dB
n.c.
n.c.
1:5
1:5
possibly could omit this buffer
(but too many reflections?)
there may be no room for this thing
from SCROD
4 −3 dB
4 −3 dB

7. Fanout on carrier board – four busses option
All channels receive nominally identical waveform, little skew
There is of course a reflection made when the wave hits the junction of an enabled channel on the bus. If only one is enabled, the reflection is (nominally) absorbed at the source termination.
This scheme depends on very light stub loading for disabled channels!
At the block diagram level, this plan is the simplest. For high isolation, though, “SW” block contains several discrete components...
However it’s not clear if we need that high isolation.
1.0 x 0.6 mm2
BAP51LX SW SW SW SW SW SW SW SW SW SW SW SW SW SW SW SW SW SW SW SW SW SW SW SW SW SW SW SW SW SW SW SW
from SCROD
additional switch/ isolation from external signals could be added here globally if required

8. 500 μA bias (one or other diode)  300 mW per boardstack
What is ? SW
1.0 x 0.6 mm2
BAP51LX
500 μA bias (one or other diode)
 300 mW per boardstack

11. “and now something completey different...”
4 mils thick
Indicates ~ PSI peak, concentrated in about 0.25x0.4 inch2.
Expect ~380 pounds per 6-32 screw torqued to 11 in-pounds (as these were).
The sensor film looks like a useful diagnostic. Results make sense. Indicates we are not achieving good contact though! (And surface flatness/finish is part of the issue.)