Issues for Adding a Third
PCB Layer
The primary issue is whether there is sufficient space, although how the cooling is done would have to be revisited. Based on an examination of the present drawings it appears that there is room to add the additional 6 mm required.
StatusBoards are currently being manufactured both at CERN and in Taiwan. It is expected that they will be available by the 1st or 2nd week in February. CERN is interested in doing or managing the assembly of the boards they are producing. The company in Taiwan may be interestd in doing the mounting also.
Outstanding Questions and Action List
What level electrical tests are done on the boards? What is the yield of the boards produced? What is the cross talk with the present protection resistor layout? Is it possible that the 0805 resistors are "ok" for discharges from straws in the real setup? It would be interesting to find out layout of actual resistance element, understand cause of failure. Revisit layout to see if vias can be made larger (Bjorn says no but a fresh look...?)
First PCB has protection resistors on the bottom layer, additional protection resistors and a connector (NAIS 30 pin) on the top layer, Second PCB has two ASDBLR's and mating connector (NAIS 30 pin) on the bottom layer, connector (NAIS 50 pin) and support (R's and C's) for ASDBLR's on the top layer, Third PCB has mating connector (NAIS 50 pin) and support R & C for DTMROC on the bottom layer and DTMROC and connector for snake cable on the top.
| Length of pin above the tension plate | 2.06 mm |
| 1st PCB | 1 mm |
| Connector (30 pin NAIS) | 5 mm |
| 2nd PCB | 1 mm |
| Connector (50 pin NAIS) | 5 mm |
| 3rd PCB | 1 mm |
| Connector to snake cable | 5 mm |
| Allowance chip address jumper | 1 mm |
| Total Height to top of Snake cable (not including cable bends) | 21.06 mm |
NB: Total height to top of components on DTMROC board is 16.5 mm
Note that ASDBLR and DTMROC must be mounted either with chip-on-board (COB) or with new, smaller BGA packages. The present estimate of "bond pad frame" (on the PCB) for the ASDBLR is 6 x 6 mm which "just fits". For the DTMROC, it is 12 mm x 12 mm, which also just fits. For the ASDBLR, the ASAT BGA package which is 6 mm x 6 mm (64 pin, 0.5 mm pad pitch) does not appear to take a chip larger than 3.3 mm x 3.3 mm which is smaller than the ASDBLR die size (about 3.6 mm x 3.6 mm).Cooling - The ASDBLR and DTMROC chips no longer "face one another". Thus the cooling must be investigated. Options are: (1) cooling tab, similar to present arrangement, between the two boards. There
will be less area because of existence of connector. DTMROC must be cooled "through" the 3rd PCB. (2) Cool the DTMROC chip from the top, with thermal pads to carry heat from ASDBLR to the DTMROC board. While
less efficient than the present plan for the flex board, this is almost exactly analagous to what is done for the endcap boards (except inverted..cooling is on DTMROC boards rather than ASDBLR boards). Cooling for endcap boards work as far as we know. An issue is how to press the DTMROC/ASDBLR boards together if thermal pad is used.The increased cost of the additional connector (5 CHF) is not thought to be a big issue. Even though an additional PCB would be required, being able to use conventional PCB technology (without kapton flex) and possibly more relaxed design rules (larger vias) might result in decreased PCB costs.
Verify size required for COB pad frame by sending die size and bonding diagram to several assembly houses and asking for their recommended pad layout. Confirm maximum pad size for ASAT 64 pin package; investigate other similar packages. Perform layout of components to check feasibility of overall design Investigate cooling
The basic layout is the same as for the above option, except that the DTMROC chips would be mounted on "large boards" which include a significant number of DTMROC chips, 4 - 12. The advantages are:LV power routing can use power plane resulting in much lower voltage drop. Eliminates the need for the Snake cable (but jumpers between boards required if more than 1 Somewhat more space for components.
Same as for above: 21.06 mm
Approximately 52 different board types would be required. Thus there is a rather large amount of layout work that would be required, as well as a need for more spares. The actual production cost is not expected to be much larger since (1) most of the boards could be made in "panels" and the Snake cable, which may prove to be rather expensive, would not be required.
It may be useful to make at least a few such boards, in the near future, so that the feasibility (can one plug into may ASDBLR stamp boards at once) and potential advantages can be investigated.
Kapton Snake Cable
Some consideration was given to the present design of the Snake cable. The major observation was that the traces presently used for the LV power would result in large (> 1 volt) power drops. A large increase in the trace width, to order 3 mm, is required to achieve tolerable voltage drops (order 200 mV). It was also suggested that the power and power return should be "on top of one another" if possible, and that the sections under the connectors should be made stiffer.
Increase width of power supply and power return lines Increase stiffness under connectors Replace connector mating to cable with Robinson connector presently used Include termination resistors
Measure signal quality on existing sample Where can the cable connector be located (thickness of PCB plus connector + cable is approximately 17 mm) What is likely production cost?