1. 1 9 December 2002A.P.Kashchuk (LNF/INFN), Frascati) New approach to CPC design

2. 2 9 December 2002A.P.Kashchuk (LNF/INFN), Frascati) It is my 1-st presentation from LNF/INFN (Frascati) The scientific work is still under development and the new ideas have to be tested… Triggered by P.Campana Why results on cross-talks obtained with small chamber in May 02 are somehow better than with the large M0 prototype tested in October 02?

3. 3 9 December 2002A.P.Kashchuk (LNF/INFN), Frascati) ?? 2 problems observed in beamtests have to be explained and suppressed: - rather high cross-talks from wires to pads; - double and multiple TDC spectra Problems mentioned here were observed already in M2R1 and other prototypes built at CERN, as well as in Ferrara’s prototypes; 50% crosstalks observed in some conditions in M3R3 in October 2002 beam-tests at operational HV I’d like to add:

4. 4 9 December 2002A.P.Kashchuk (LNF/INFN), Frascati) Cross-talks ?

5. 5 9 December 2002A.P.Kashchuk (LNF/INFN), Frascati) Cross-talks along the wires (longitudal) perpendicular to wires (transverse) Longitudal crosstalks are less studied and much less suppressed

6. 6 9 December 2002A.P.Kashchuk (LNF/INFN), Frascati) M1M2M3M4M5 R3 Pad size (cm) & Number of pads Cdet (pF) Cwp (pF) CPC: 2,5 (48x4) 30.3 4.75 CPC: 2.5,12.5 (24x2) 50 15 CPC: 2.7,13.5 (48x2) 56 21.5 CPC 5.8x14.5 (24x2) 90 40 CPC: 6.2x15.5 (24x2) 140 46 According to specification (LHCb 2000-061, W.Riegler) Cross-capacitance Cwp will be increased with pad size, as shown: Cwp=0.475hw (pF) where h, w are pad height and width (cm) LNF M0

7. 7 9 December 2002A.P.Kashchuk (LNF/INFN), Frascati) Cwp measured in large M3R3 is absolutely similar Let’s measure: (very good agreement with table shown above)

8. 8 9 December 2002A.P.Kashchuk (LNF/INFN), Frascati) The original idea was following: Cross-talks from wire strips to pads (longitudal) will be reduced with grounding wire strips through HV-capacitors

9. 9 9 December 2002A.P.Kashchuk (LNF/INFN), Frascati) RC-model (Rwire =90 Ohm/m) current source HV-capacitor grounded on one side, as shown Wire strip with 4 wires Cathode pad Only capacitive coupling is taken into account in this model Assumed that width of wire strip is equal to cathode pad Signal from particle

10. 10 9 December 2002A.P.Kashchuk (LNF/INFN), Frascati) Wire strip as a transmission line, i.e. LC-model (first proposed by LNF group) Each wire can be considered as a transmission line Wires in strip are connected in parallel: L reduced, C increased (product LC is the same) Cathode pad current source

11. 11 9 December 2002A.P.Kashchuk (LNF/INFN), Frascati) f=60 MHz f=95 MHz The ringing frequency depends on inductance in series to HV-capacitor (what the reason?) HV-capacitor 680pF directly grounded Recently it has been found: the wire strip is ringing (response of wire strip in small LNF prototype made with injector)

12. 12 9 December 2002A.P.Kashchuk (LNF/INFN), Frascati) Cross-talk profile from wires to cathode pads in the large M3R3 prototype Volt. step on strip (strip is floating) 20ns/div Central Pad Pad-1 (15%) Pad+1 (20%) Ringing 18 ns Ratio 2-nd/1-st peak 20%

13. 13 9 December 2002A.P.Kashchuk (LNF/INFN), Frascati) Fine waveform structure at voltage rise time 1.5ns: One can see 4 ns oscillation High frequency is due to LC of the transmision line itself

14. 14 9 December 2002A.P.Kashchuk (LNF/INFN), Frascati) Response of wire strip in small LNF prototype made with Current Injector Ringing 13ns Strip is grounded through 680pF, with adding inductance period is increased

15. 15 9 December 2002A.P.Kashchuk (LNF/INFN), Frascati) Equivalent circuit 1.Inductance blocks HV-capacitor effect High cross-talk:Compare to ideal case: 2.Another parasite effectAssumed Wire strip Cathode padTerminated end But!!!

16. 16 9 December 2002A.P.Kashchuk (LNF/INFN), Frascati) Peak=0.5uA Ringing= 3ns Peak=1uA Ringing= 8ns with HV-capacitors grounded at inductance 3nH in series (perhaps, can be acheaved) 2-side (peak less factor 2) What LC-model shows? 1-side strip termination with 0 Ohm

17. 17 9 December 2002A.P.Kashchuk (LNF/INFN), Frascati) Green peak=1uA Red peak=1uA Ringing=30ns Stray inductance (printed traces) in series to HV-capacitors and full capacitance of the wire strip mainly specify the ringing frequency 1-side strip termination with 0 Ohm2-side termination Green – 3nH Red – 300nH (can be if width of traces 0.25mm, see M3R3) Green peak=0.5uA Red peak=1uA Ringing=20ns

18. 18 9 December 2002A.P.Kashchuk (LNF/INFN), Frascati) Correct strip termination with 377 Ohm can not be used 2- middle 3- near end 1- signal at far end to capacitor No ringing, waveforms are independed to position of the signal source, but the highest crosstalks will be in this case 2-side termination through 680pF At 1-side termination amplitude will depend on signal position along the strip (see next slides) Ohm

19. 19 9 December 2002A.P.Kashchuk (LNF/INFN), Frascati) Inductance in series to HV-capacitors (1-side termination with 680pF and 50 Ohm) Same schematics No ringing Peak independ on Lstray due to 50 Ohm Green=3nH Red=100nH Scale +/-2uA 10ns/div, but not enough cross-talk attenuation 12 3

20. 20 9 December 2002A.P.Kashchuk (LNF/INFN), Frascati) Inductance in series to HV-capacitors (1-side termination with 680pF and 0 Ohm) Same schematics Green=2nH/Ringing 200MHz Red=100nH/Ringing 70MHz Scale +/-2uA 10ns/div 123 Better attenuation, but ringing at R=0

21. 21 9 December 2002A.P.Kashchuk (LNF/INFN), Frascati) Inductance in series to HV-capacitors (2-side termination with 680pF and 20 Ohm) Same schematics 1 Red=100nH/Ringing 100MHz Green=3nH/No ringing Scale +/-2uA 10ns/div 2 3 Good cross-talk attenuation factor at 20 Ohm High inductance leads to ringing even at R=20 Ohm and drastically reduces cross-talk attenuation

22. 22 9 December 2002A.P.Kashchuk (LNF/INFN), Frascati) if w=0.25mm trace width (in M3R3 prototype) h=1.5mm pcb thickness l=3-10cm length of trace (in M3R3 prototype) then L=100-1000nH if w=1.2cm then L=5nH can be achieved Inductance of printed trace (example): @ C=100pF

23. 23 9 December 2002A.P.Kashchuk (LNF/INFN), Frascati) Ringing on wire strip can double/multiple signals and TDC spectra Threshold defined experimentally: for wire readout –7fC for cathode (single) – 5fC Dynamic range of signals in CPC is large (100) Average signal 50fC FEE noise 50e/pF is not the first reason for threshold choice, mainly cross-talks define threshold, at efficiency 95%/gap So, high probability for after-pulsing can be found at bad wire strip termination and imperfect layout in CPC at any frequency of ringing (it depends on design)

24. 24 9 December 2002A.P.Kashchuk (LNF/INFN), Frascati) If Cstrip=Cwire is high? Excellent result +/-2uA +/-0.2uA No ringing and high cross-talk attenuation

25. 25 9 December 2002A.P.Kashchuk (LNF/INFN), Frascati) CPC design can be improved following the way: No ringing must be on the wire strips at perfect design 1.No wire segmentation is needed in CPC design - one of the effective way No HV-capacitors and resistorrs (cheaper and much easy design), wires are connected to one HV-resistor. Minimisation of the trace inductances has to be done in the Combined readout chambers, which dumps effect of low impedance. 2.Double Cathode Readout scheme, perhaps, can be used in some cases below M3R3 also effective way at large Cwires (it allows increase threshold at fixed HV) already tested in M1R1 with excelent results (because very low Cwp and good attenuation of the cross-talks from wires) ! Conclusion

26. 26 9 December 2002A.P.Kashchuk (LNF/INFN), Frascati) M1M2M3M4M5 R3 Pad size (cm) & Number of pads Cdet (pF) Cwp (pF) CPC: 2,5 (48x4) 30.3 4.75 CPC: 2.5,12.5 (24x2) 50 15 CPC: 2.7,13.5 (48x2) 56 21.5 CPC 5.8x14.5 (24x2) 90 40 CPC: 6.2x15.5 (24x2) 140 46 LNF M0 Double Cathode readout (CRO) below M3R3. No wire segmentation Single CRO. No wire segmentation ? Summary Voltage zero must be on the wire strips at perfect CPC design, i.e.