Test of optical links on CN V.3 David M¨ unchow for the Gießen Group Belle II PXD/SVD Workshop 5. February 2013, Wetzlar, Germany 1 Test Setup 2 Hardware Modifications 3 Software Issues 4 Test Results
Test Setup In final setup input rate of 6 . 25 Gbit/s is needed Test of CN V.3 with 4 SFP cages Tested with IBERT core generated with Xilinx Core Generator David M¨ unchow (JLU Gießen) Test of optical links on CN V.3 5. Fabruary 2013 2 / 10
Hardware Modifications: Oszillator Change of oszillator Board was delivered with 200 MHz quarz Possible data rate i.e.: 4 Gbit/s (workes, error ratio not checked) Changed to 156 . 25 MHz quarz (IHEP will use this in future) Possible data rate i.e.: 3 . 125 Gbit/s or 6 . 25 Gbit/s Risk of braking conductor path during replacement David M¨ unchow (JLU Gießen) Test of optical links on CN V.3 5. Fabruary 2013 3 / 10
Hardware Modifications: SFP Speed Settings Removal of resistors SFP connector have two pins to select limitation to 4 . 25 Gbit/s or allow speed up to 8 . 5 Gbit/s We had to remove 8 resistors to be shure the limitation is off. IHEP changed resistors for future production to meet our needs David M¨ unchow (JLU Gießen) Test of optical links on CN V.3 5. Fabruary 2013 4 / 10
Software Issues Problems with IBERT of ISE 13 or ISE 14 (current version) settings were correct for 6 . 25 Gbit/s test software shows 4 . 883 Gbit/s and connection is reality 3 . 125 Gbit/s (crosscheck by counting received bits) IBERT works correct with ISE 12 (2 years old) David M¨ unchow (JLU Gießen) Test of optical links on CN V.3 5. Fabruary 2013 5 / 10
Test Results 6.25 Gbit/s Connections in test setup: X0Y3 0 ←→ X0Y3 1 X0Y4 0 ←→ X0Y4 1 Result after > 77 hours: X0Y3 0 X0Y3 1 X0Y4 0 X0Y4 1 bit error count 1 0 99 0 5 . 735 · 10 − 16 < 5 . 735 · 10 − 16 5 . 677 · 10 − 14 < 5 . 735 · 10 − 16 bit error ratio David M¨ unchow (JLU Gießen) Test of optical links on CN V.3 5. Fabruary 2013 6 / 10
Test Results 3.125 Gbit/s Connections in test setup: X0Y3 0 ←→ X0Y3 1 X0Y4 0 ←→ X0Y4 1 Result after > 24 hours: X0Y3 0 X0Y3 1 X0Y4 0 X0Y4 1 bit error count 0 0 0 0 < 1 . 851 · 10 − 15 < 1 . 851 · 10 − 15 < 1 . 851 · 10 − 15 < 1 . 851 · 10 − 15 bit error ratio David M¨ unchow (JLU Gießen) Test of optical links on CN V.3 5. Fabruary 2013 7 / 10
Check for error source for bit error SFP tranceiver switched Cables switched Result: error source at receiving side Currently using speed grade -1 Xilinx garuantees support up to 6 . 5 Gb/s only with speed grade -2 or -3 In future speed grade -2 will be used David M¨ unchow (JLU Gießen) Test of optical links on CN V.3 5. Fabruary 2013 8 / 10
Summary CN V.3 tested for 6 . 26 Gb/s Some hardware changes needed oszillator, resistors Software issues With ISE 13 or 14 the IBERT does not work for 6 . 25 Gb/s Using old ISE 12 the IBERT works fine Result for 6 . 25 Gb/s (77 hour test) 3 channels bit error ratio ≤ 5 . 735 · 10 − 16 1 channel bit error ratio 5 . 677 · 10 − 14 Result for 3 . 125 Gb/s (24 hour test) All channels bit error ratio < 1 . 851 · 10 − 15 (0 errors) David M¨ unchow (JLU Gießen) Test of optical links on CN V.3 5. Fabruary 2013 9 / 10
Test of optical links on CN V.3 Thank you for your attention! David M¨ unchow (JLU Gießen) Test of optical links on CN V.3 5. Fabruary 2013 10 / 10
Recommend
More recommend
