New preamplifier chip design and replacments

- notes from a meeting at University of Paisley on 22nd September 1999


Present:- 
 	Prof. R( Bob) Chapman (Paisley), Geoff Moores(Paisley), Alastair Ferguson(Paisley)
	Dr. N M (Nobby) Clarke ( Birmingham), Thierry Legou ( LPC Caen)
	


1. The aim of this meeting was to review the test measurements carried out at
Paisley, Birmingham and LPC Caen on a number of prototype preamplifier chips,
and if possible establish a strategy for a replacement chip to provide
long term security of supply for the many preamplifier channels which
are ( and will be) required for the CHARISSA project and the forthcoming TIARA
project.  LPC Caen also require 48 channels by November 1999, and a choice
needed to be made as to which preamp chip should be used.

2. The performance of the new Paisley preamp designs was reviewed; it was agreed
that there were still some lingering doubts about the asymmetry of output
response ( negative voltage swings have a much longer rise time than positive swings)
and Alastair agreed that more development work was required.
However, AF was due to retire in february, so progress beyond that date
was uncertain.
 There was also some worry that certain PCB layouts showed some instabilities 
( probably thermal) leading to gain shifts and peak wandering for the first half hour
or so, and it was suggested that these were due to the PCB layouts, and probably
arose from attempts to use very thin tracks and small spacings.
However, most Paisley designs showed very good noise levels, and spectra taken with
a 3 -alpha source and a small low-leakage detector showed resolutions of about 21keV FWHM
in GM's apparatus at Paisley - similar to those obtained with the older Oxford preamp chip.

3. TL showed spectra and noise figures from LPC, where they compared a commercial
( expensive) preamp with Oxford/Paisley designs.
 Noise (pulser) figures of ~8keV, 9.8keV, 31keV were obtained for the commercial, Oxford type, and
 Oxford+DC restorer chips - the 31keV figure was considered anomalous as NMC
 had found the chip +DC restorer improved resolution.
 It was pointed out that the commercial preamp chips come in a four channel
 package costing about 500 pounds! Estimated cost of a new preamp chip using
 GM latest design came in at about 10-15 pounds each. TL was therefore happy
 to use the older Oxford chips to meet the November 1999 deadline,
 but needed to have the DC restorer to ensure a zero output signal,
 because he would be using  50ohm terminations on the preamp outputs at the
 far end of the cables.
 
 4. GM reported on his work with modifying the older Oxford chips, and re-designing
 the chip with a new PCB. Following a suggestion by NMC, he copied a technique used
 in the Daresbury designed preamps ( made by Cooknells) to strap a DC restorer
 to an existing Oxford chip ( samples were shown) - this looked messy but worked
 fine and no difference in noise could be seen, so it was suggested
 that one way to proceed might be to re-package the existing Oxford design
 (and incorporate the DC restorer) using smaller surface mount components.
 
 5. One reason for pursuing a new preamp design was that the manufacturer of
 last batch of Oxford chips had informed DLW at York that most components for the
 Oxford design were now impossible to source. However, GM reported that he was able
 to source the components from the USA, though waiting times were long, and orders
 need to be 3000 to 6000 off.
  As a result GM had designed a new 'Oxford' PCB with double sided copper and smaller
  surface mount components, and incoporating the DC restorer circuit. GM provided 
  NMC and TL with samples of this revised Oxford design for testing.
  It was agreed that GM could not expect to make 48 channels by November for TL,
  but that a 'retro' modification to 48 Oxford chips could be achieved
  to incoporate the Daresbury DC restorer circuit, and GM will proceed on this basis.
  
6. In the long term, if the newly re-packaged 'Oxford' chip proved OK, and 
assuming the components could still be sourced, the meeting felt it would probably be
better to stick to the Oxford design, since its performance was really very good.
However, any developments by AF would be monitored for future developments.
NMC reported that he could not see a pressing need for an additional 128 channels
during the early part of next year, so that any final decision could wait until
mid-2000 when results from the re-packaged Oxford design would be available.
However, in the long run, 500 chips would be required for CHARISSA to form
the 256 channel system at GANIL, and provide spares for preamps at GANIL/ANU/Strasbourg.

7. AOB. NMC handed over  a crate of 16 HV bias supplies to GM for Paisley's use
in some compensation for all the developmentwork  and construction of 112 channels
for Birmingham. ( This transfer was agreed between BRF and NMC) NMC agreed
to supply the PC software to drive the HV crate.
TL reported on his developments at LPC using digital filtering
techniques to measure 5MeV alphas following the decay of an ion implanted in a detector
with an energy of 1GeV! Normal amplifiers saturate too long to record short lived
isotopes of super-heavy elements, but this new technique can cope with the
huge range of pulse heights.