Signals crosstalks and their ratios in the LHCb
Signals, crosstalks and their ratios in the LHCb muon detectors 10 September 2003 A. Kashchuk (Roma 2, PNPI) 1
Definitions: N – number of cathode pads along wire strip in CPC (‘wireless’ GEM segment will be consider as a strip); S=1 – width of strip equal to width of pad; At S=1 shift by half strip w. r. t. pad; S>1 – superstrip (S=2, 3, etc. ): strip width double, triple etc. ; S<1 – sub-strip (strip subdivided on parts ½, ¼, etc. ); Cx – Xtalk capacitor ‘strip-to-pad’; Cblock – blocking capacitor used to reduce strip impedance; Zblock – blocking impedance if Cblock has L, Rdamp in series; Rdamp – damping resistor; ‘Floating’ strip, if no load as Cblock, Zblock; Equivalent blocking capacitor: Fundamental frequency of the signal: A (d. B) -20 d. B/dec fs f (log scale) 10 September 2003 A. Kashchuk (Roma 2, PNPI) 2
CPC – single gap Cathode 1 Wire strips Cathode 2 Cx Cx S=1/2 Wire strip S=1 S=2 Sub-strip Width of strip equal to width of pad Superstrip S=2 N=4 CPC active area with N pads along strip Illustrations for definitions 10 September 2003 A. Kashchuk (Roma 2, PNPI) 3
Signal-to-Xtalk Ratio (SXR) and Xtalk-to-Signal Ratio (XSR): XSR=1/SXR 10 September 2003 A. Kashchuk (Roma 2, PNPI) 4
Illustration for CPC-2 (N=2) This approach can be applied to any multi-channel system, it is correct for AC at f. S and any harmonica Xtalk from each I-source separately: 2 I-sources (equal) Cx Cx Cx Ic I’x Cx To find signal, first find missing fractions: 2 N (N=2 here) ‘elementary’ capacitors Cx (pure capacitive system) Ramp does not change ratios -> Real signal is I reduced by 2 missing fractions: the first through own capacitor Cx and the second, as Xtalk from another I-source Note: here ratios do not depend on capacitor value Cx 10 September 2003 A. Kashchuk (Roma 2, PNPI) 5
Consider ‘floating’ strips (strip unloaded) Table: An intrinsic feature of the CPC design CPC SXR CPC-2 N=2 -1 CPC-4 N=4 -3 CPC-8 N=8 -7 XSR -100% -33. 3% -14. 3% Already good for TH=20%S Bad for TH=10%S Note: Negative sign means opposite (negative for CPC) polarity. At high rates polarity can be ignored: one can consider corresponding base line fluctuations by Xtalks after BLR in front of discriminator. 10 September 2003 A. Kashchuk (Roma 2, PNPI) 6
Cross-talks measured in M 4 R 1 (CPC-8) at ‘floating’ wire strips Time distributions in log scale at HV=3. 15 k. V TH=13 f. C 20%Signal CD AB 0. 3%? 10 September 2003 0. 5%? A. Kashchuk (Roma 2, PNPI) 7
Strip loaded by blocking capacitor (N=2) n. Cx Cx Cx 2 N+n ‘elementary’ capacitors Cx Note: here ratios depend on capacitor value Cx, see n=Cblock/Cx 10 September 2003 A. Kashchuk (Roma 2, PNPI) 8
Results with ‘ideal’ blocking capacitor Table: Cblock=1000 p. F, Cx=20/10/5 p. F CPC SXR CPC-2 N=2 n=50 (25) 26 CPC-4 N=4 n=100 (50) 53 CPC-8 N=8 n=200 (100) 107 XSR 3. 8% 1. 9% 1% TH=10%S Good ratios for SCRO, less good for DCRO, see (n) Attention: LC-loop is created at some design, Resonance! damping is needed 10 September 2003 A. Kashchuk (Roma 2, PNPI) 9
Equivalent blocking capacitor n. Cx Cx Cx Note: neq reduced, as Ceqblock reduced 10 September 2003 A. Kashchuk (Roma 2, PNPI) 10
Results with equivalent blocking capacitors Table: Cblock=1000 p. F, L=50 n. H, Rdamp=20, Z=39, Ceqblock=320 p. F, Cx=20/10/5 p. F, CPC SXR CPC-2 N=2 n=16 (8) 9 CPC-4 N=4 n=32 (16) 19 CPC-8 N=8 n=64 (32) 39 XSR 11. 1% 5. 3% 2. 6% TH=10%S Much worst ratios for SCRO, even worst for DCRO, see (n) Partial damping Rdamp=20 is not sufficient Still resonance! 10 September 2003 A. Kashchuk (Roma 2, PNPI) 11
Consider resonance effects in details 10 September 2003 A. Kashchuk (Roma 2, PNPI) 12
SPICE: resonance effects on wire strip: rise both voltage (V) on strip and impedance (Z) are observed CPC-2 N=2 V I Z 10 September 2003 A. Kashchuk (Roma 2, PNPI) 13
Rise of cross-talk is a product ‘Voltage x Impedance’ X-talk rise Ratio=14 V Ratio=2. 5 I Ratio=3. 7 Z Ratio=6 Ratio is defined to f=10 MHz: beginning of the FEE bandwidth and fundamental frequency of the signal 10 September 2003 A. Kashchuk (Roma 2, PNPI) 14
In time domain 10 MHz and 56 MHz are shown Note: At L=100 n. H, C=80 p. F Resonance freq. f 0=56 MHz Correct damping Rdamp=70 Ohm V I X-talk rise Ratio=14 at Rdamp=20 Ohm 10 September 2003 A. Kashchuk (Roma 2, PNPI) 15
Measurements Apply sin to HV-connector Measure sin on each cathode pad: find resonance (define inductance) and compare amplitude to 10 MHz Eq. diagram 2 Cx Cx Cx LC=loop f 0 is found from measurements, L is calculated by SPICE at known Cx 10 September 2003 A. Kashchuk (Roma 2, PNPI) 16
Results of measurements of the resonance effects in crosstalks Measurements were made by author with Sin generator at Roma-2 (M 4 R 1) and at LNF (M 3 R 3) September 2003 10 September 2003 A. Kashchuk (Roma 2, PNPI) 17
M 3 R 3 prototype Crosstalk amplitude A(a. u) Bar 1 Bar 2 at 10 MHz Ratio about 30 f (MHz) Measurement: f 0=108 MHz 10 September 2003 A. Kashchuk (Roma 2, PNPI) (M 4 R 1) shows similar effects 18
Crosstalk amplitude SPICE shows same resonance frequency at L=25 n. H f, log scale 105 MHz – single gap at L=50 n. H, Rdamp=20 Ohm, Cx=10 p. F; 83 MHz – double gap at L=50 n. H, Rdamp=20 Ohm, Cx=20 p. F 10 September 2003 A. Kashchuk (Roma 2, PNPI) 19
Must be suppressed completely It would be incorrect and risky to work with resonance Note: no margin already in CPC-2 (M 1 M 2 M 3 M 4 M 5 R 3) 10 September 2003 A. Kashchuk (Roma 2, PNPI) 20
Excitation from LHC clock can be at any frequency (see pulse width effect) Time-domain periodic function (LHC clock): Frequency-domain: Slope=-20 d. B/decade -40 d. B/decade 10 September 2003 A. Kashchuk (Roma 2, PNPI) 21
Resonanceless alternative CPC design Ø Correct damping Ø Superstrip Ø Sub-strip (further strip subdivision) Ø Microstrip termination (inductanceless blocking capacitors) Ø Combinations 10 September 2003 A. Kashchuk (Roma 2, PNPI) 22
Compare various damping factors Table: Cblock=1000 p. F, L=50 n. H, Rdamp=20, Cx=20/10/5 p. F, Z=39, Ceqblock=320 p. F CPC-2 N=2 n=16 CPC-4 N=4 n=32 CPC-8 N=8 n=64 SXR 9 19 39 XSR 11. 1% 5. 3% 2. 6% TH=10%S Rdamp=50, Cx=20/10/5 p. F, Z=69, Ceqblock=231 p. F Resonance 6. 7 n=11. 5 14. 5 n=23 30 n=46 less 14. 9% 6. 7% 3. 3% TH=10%S Table: Cblock=1000 p. F, L=50 n. H, SXR XSR Not enough only increase Rdamp, see n reduction 10 September 2003 A. Kashchuk (Roma 2, PNPI) 23
Shift by half strip w. r. t. pad with new Rdamp helps Table: Cblock=1000 p. F, L=50 n. H, Rdamp=50, Cx=10/5/2. 5 p. F, Z=69, Ceqblock=231 p. F CPC-2 CPC-4 CPC-8 N=2 n=23 N=4 n=46 N=8 n=92 SXR 13. 6 28 57 XSR 7. 4% 3. 6% 1. 7% TH=10%S Good ratios for SCRO see n (Cx are reduced by factor 2) 10 September 2003 A. Kashchuk (Roma 2, PNPI) 24
Shift by half is an intermediate solution. Sub-strip is more general one: it can be ½, ¼, etc. ) Neighbour strip/strips Cx Cx Cx n Cx Cx reduced neq increased 10 September 2003 A. Kashchuk (Roma 2, PNPI) 25
Superstrip M 1 R 1 GEM Without blocking capacitors N=8, S=4 (NS ‘elementary’ capacitors Cx in GEM) M 1 R 1 CPC (NS+N+n ‘elementary’ capacitors Cx) N=8, S=4 With blocking capacitors 10 September 2003 A. Kashchuk (Roma 2, PNPI) 26
Results for M 1 R 1 See A 1 Table: Cblock=680 p. F, L=5 n. H, Rdamp=10, Cx=2 p. F, Z=32, Ceqblock=498 p. F, n=249 M 1 R 1 GEM version CPC version N=8, S=4, n=0 N=8, S=4, n=250 SXR 31 155 XSR 3. 3% 0. 6% TH=10%S Bad ratio at TH=(2 -3)%S If superstrip S=4 is allowed in M 1 R 1 it can be allowed in regions R 2 and R 3 10 September 2003 A. Kashchuk (Roma 2, PNPI) 27
Superstrip (resonaceless) can be used alone CPC SXR CPC-2 (R 3) N=2, S=6 11 CPC-4 (R 2) N=4, S=4 15 CPC-8 (R 1) N=8, S=2 31 XSR 9. 1% 6. 7% 3. 3% TH=10%S Compare: Table: Cblock=1000 p. F, L=50 n. H, Rdamp=20, Cx=20 p. F, Z=39, Ceqblock=320 p. F, n=16 CPC We are here with resonances SXR CPC-2 N=2, S=1 9 CPC-4 N=4, S=1 19 CPC-8 N=8, S=1 39 XSR 11. 1% 5. 3% 2. 6% 10 September 2003 A. Kashchuk (Roma 2, PNPI) TH=10%S 28
But if superstrip S=4 is not allowed at high rates in M 1 R 1, then problem with GEM at S=1 Cblock per strip is need in GEM (CPC will need 4 times more capacitors with Rdamp) M 1 R 1 SXR XSR 10 September 2003 GEM version N=8, S=1, n=0 7 CPC version N=8, S=1, n=250 14. 3% TH=3%S 0. 8% TH=10%S 131 A. Kashchuk (Roma 2, PNPI) 29
Microstrip termination (inductanceless blocking capacitor) Consider GEM only: Find: If we want SXR=100, then at Cx=2 p. F Cblock=180 p. F Microstrip capacitor: 10 September 2003 A. Kashchuk (Roma 2, PNPI) 30
Elegant solution for low thresholds: ‘ 2 -side microstrip termination’ Table: microstrip capacitor implementation (here 90 p. F per side) Kapton FR 4 h thickness (microns) Area (cm 2) Size (cmxcm) 25 0. 85 0. 9 x 0. 9 50 1. 7 1. 3 x 1. 3 100 3. 4 1. 8 x 1. 8 Note: both 25 and 50 microns are realistic for GEM at pad size 1 x 2. 5 cm 2; 10 September 2003 A. Kashchuk (Roma 2, PNPI) 31
SXR/XSR panorama: what could be achieved without resonances M 1 M 2 M 3 M 4 M 5 R 4 60 1. 7% >100 <1% R 3 12. 8 7. 8% 15. 5 6. 5% 13. 6 7. 4% 7 (10/16) 14. 3% (10/6%) 6. 4 (10/16) 15. 5% (10/6%) R 2 >100 <1% 45. 5 2. 2% 30 3. 3% 29 3. 4% R 1 >100 <1% 64. 5 1. 6% 57 1. 7% 51 2% TH=10%S Note: M 1 R 1 R 2: assuming GEM with ‘ 2 -side microstrip termination’; CPC with C block; Most CPC with ‘shift by half’ + Cblock=1000 p. F, Rdamp=50; M 4 M 5 R 3: Superstrip or Sub-strip + Cblock=1000 p. F, Rdamp=50, see ( super/sub ) 10 September 2003 A. Kashchuk (Roma 2, PNPI) 32
Conclusion v Partial damping is not sufficient, still resonance: At present design there are no margin in CPC-2 (all stations). It means, few % of crosstalks in ‘hits’ at some conditions can reach 100% in real environments of the experiment. v Resonanceless damping must be implemented with blocking capacitors, but it is possible only with redesign in order to get margin; 10 September 2003 A. Kashchuk (Roma 2, PNPI) 33
Conclusion (cont. ) v ‘Shift by half’ strips w. r. t. pads can be proposed for SXR/XSR margin improvement in most CPC at Rdamp=50, as the simplest technical solution; v As shown, M 4 M 5 R 3 are most problematic CPC-2: superstrip or sub-strip in conjunction with blocking capacitors only help; v M 1 R 1 must be redesign to S=1 (both GEM/CPC); v 2 -side microstrip termination is unique and elegant solution for GEM, operating at low thresholds (2 -3)%S. It can be implemented also in CPC. In this case no discreet HV-capacitors, as well as Rdamp. 10 September 2003 A. Kashchuk (Roma 2, PNPI) 34
A 1 M 1 R 1 on GEM (eq. diagram @S=1) Assuming 10 September 2003 A. Kashchuk (Roma 2, PNPI) 35
- Slides: 35