P3K Team Meeting #15

PALM-3000

P3K Team Meeting #15

Stephen GuiwitsDean Palmer

Ernest Croner

Team Meeting #15 -- 126 Cahill

October 29, 2009

PALM-3000P3K Electronics Agenda

• RTC Electronics (Computer Room)– JPL– Caltech

• Active Mirror Electronics (Cass Cage/AO Lab)– Xinetics GenIII Integration and Test– Driver Electronics Interfaces

• Active Mirror Cabling• Active Mirror Communications

• The near future

PALM-3000

RTC Electronics

3

PALM-3000

4

RTC Electronics

• PC 0 – 9 (Telemetry)– All components installed

– Rack Mounted

– Cabled (JPL Only)

– Each machine kickstarted• Suse Enterprise Linux 10 (non real-time)

– Actively using for software development (JPL Only)

• Suse Kickstart DVD– JPL version completed

– Custom RPMS built and tested

– 2 disks architecture• Disk 1 = Base + patches• Disk 2 = custom scripts and RPM’s

– Require two versions (JPL/Caltech)

– Requires custom kernel patch to see both graphic cards

PALM-3000

5

RTC Electronics

• Optical Splitter tested. Able to get full frames to all GPU’s• Quadrics High Speed Network Switch

– Installed and configured

– Suse 10 Linux driver installed

– Does not work with Real-Time Linux. Quadrics went out of business. We’re currently looking for solutions

• RAID System – Configured for throughput optimization

• RAID 0• Two partitions

– High Speed data

– All other data

• Gigabit Private Ethernet Switch– Each PC is able to communicate over the private ethernet– Separate from the Quadrics High Speed Switch

PALM-3000

6

RTC Electronics

• KVM Switch– Convenience

– Allows for graphical access to each machine while working on the racks

– Helpful while debugging an issue with any one of the 10 machines located in the racks

PALM-3000

7

RTC Electronics

Development Racks

Located at JPL Room: 306-130C

PC 0 – 8

RAID System

KVM Switch

Quadrics High Speed Switch

16 Port Gigabit Ethernet Switch

Optical Splitter

PALM-3000

8

RTC Electronics

• Production Racks (Caltech)– Located in Room 17b Cahill Center

– Machines mounted in racks

– 2 machines kickstarted for initial testing of Caltech Kickstart• JPL System Administrators are still working on the final version of the kickstart

PALM-3000

9

RTC Electronics

PALM-3000

DM Integration and Test

10

PALM-3000DM Driver Integration and Test

• July 2009 :: Xinetics :: Devens, Massachusetts• Hardware required

– Xinetics Gen III Mirror Electronics• (4U GenIII chassis + 1U fan tray) * 8

– Perle IOLAN STS Terminal Server• 8 RJ-45 9 pin serial adapters

• 8 RJ-45 cables

– Custom PCB interface boards

• Required to cable the mirror

– Bud Industry 30U Rack (2)

– Pulizzi IPC 3401 (2)

– Dell Precision T7400 Computer

– Associated cabling and interfaces required for integration and test


• Hardware


• Software– Initial non-high speed software written

– Communicates via serial terminal server• PC ethernet terminal server RJ-45 9 pin driver

– Successful command and control of all 8 driver• Command set

– Power up– Power down– Put driver in proper mode (Normal or Test)– Clear faults on the drive (Xinetics feature)– Mute the drivers– Drive actuator values– Drive bias value– Read back voltage table (needs tweaking)– Read status table (needs tweaking)


• Testing Results– Mirror had some issues. Jenny covered that during the optical bench

presentation

– PCB boards mapped incorrectly. Still able to move every actuator to bias levels

• Able to re-wire some of the cables so we can continue to use the boards while a final design is being worked on

• Cables already sent out and modified

– Software worked upon arrival. Little changes were required to deal with undocumented behavior of the GenIII driver API

– Optical results were covered in the optical bench presentation

PALM-3000

Active Mirror Cabling

15

PALM-3000Active Mirror Cabling

• 2 PCB Designs– Driver Interface Board (DIB) (x16)

• 2 per chassis• Board routes all actuators, grounds, and bias from driver 50

pin connector to 260 pin connector

– Mirror Interface Board (MIB) (x16)• Board routes actuators, bias, and ground signal from 260-pin

connector to 62 and 44 pin connectors• 2 pin connectors are used to daisy chain bias and grounds

to each PCB as needed


High Level Block Diagram

Chassis (1 of 8)10 Drivers468 Actuators(48*9) +36

10 Grounds10 Bias M irro r I n t e rf a c e 1 6

<V a lu e >

D riv e r I n t e rf a c e 1

<V a lu e >

Hypertronics 260contacts

Hypertronics260 contacts

2-pin Molex(x2)

44-pinconnectorfor Bias

62-pinconnectorforActuators(x4)

50-Pin IDC(x5)

D riv e r I n t e rf a c e 2

<V a lu e >

Hypertronics 260contacts

50-Pin IDC(x5)

44-pinconnectorforgroundsx2

M irro r I n t e rf a c e 1 7

<V a lu e >

2-pin Molex(x2)

Hypertronics260 contacts

Mirror Interface Board (MIB)(x16)

Driver Interface Board (DIB)(x16)

44-pinconnectorforgroundsx2

62-pinconnectorforActuators(x4)

x8

x8

44-pinconnectorfor Bias


Hypertronics N Series Connectors


Computer Rendering of the Mirror Interface Box


• Remaining costs to complete the cables and interface boards– Procurements

• Hypertronics (34 plugs&sockets) $29,000DM

• cables (17) $70,100 ($3800 per cable +tx)

• Driver interface boards (17) $1300

• Rack mounting panels (8) $800

• Driver-rack interface ribbon cables $1200

• IDC Connectors for PCB $500

• DSUB connectors for PCB $1050

• DSUB mating connectors $1150

• Back shells for DSUB connectors $700

• Mirror interface boards (17) $1300

• Bench interface box (1) $1500

---------------

$108,600



• Near-term milestones– MIB and DIB designed finished by November 26,

2009– 17 custom cables ordered by November 26, 2009

• Cables have a long lead time• Cables can be started without the final connectors available

PALM-3000

Active Mirror Communications

23

PALM-3000Active Mirror Communications

PALM-3000DM High-Speed Data Link

Progress to date• Created a CMC Carrier Board hardware requirements specification• CMC Carrier Board Circuit design and schematic completed.• CMC Carrier Board PCB layout begun.

– Component placement is almost complete.

– Most of the individual circuit blocks are wired.

• During the component placement, I found that we’re not able to fit all the connectors on a 6U VME card without greatly increasing the complexity of the PCB layout (blind vias, increase layer count, etc.).

– Looked at a few mitigation options: • Separate board designs for LODM and HODM• Piggy-back board with connectors for HODM• Wires from board to enclosure boundary

– Separate board designs is the best option for signal integrity and board complexity. Much will be shared between the two designs.

• Selected a COTS cable/connector combination for the interface to the Xinetics chassis.

– This requires a small adapter board at the Xinetics side– Adapter board design is nearly complete.

PALM-3000DM High-Speed Data Link FPGA

CMC Carrier Board FPGA Progress• Began initial FPGA development.• Simulated flow of data through the FPGA from

the serial Front Panel Data Port input to the Xinetics SOR422-U outputs.– Still need to work out a minor bug.– Need to enhance the state machine for the data

input. – Need to add inter-actuator checking algorithm.– Need to add support for LODM.– Need to add test mode.

PALM-3000

DM High-Speed Data Link Completion Forecast

Forecast• November

– Complete HODM Carrier Board PCB layout– Design review November 10, 2009 – Receive assembled boards.– Complete preliminary FPGA design.– Begin board testing.

• December– Create Carrier Board FPGA design specification.– Complete HODM Carrier Board testing.– Complete LODM Carrier Board PCB layout.

PALM-3000

DM High-Speed Data Link Completion Forecast

Forecast (cont.)• January

– Complete Carrier Board FPGA design.– Receive assembled LODM Carrier Cards.– Test LODM Carrier Cards

28

PALM-3000The near future

• Build up RTC racks at Caltech– Complete work on Caltech Suse Kickstart

– Incorporate new kernel findings into every verison of the kickstart

– Work on Real-Time Linux with Quadrics

• Build up DM racks at Caltech• Test the DM when it arrives from Xinetics• Measure power consumption of drivers to help

determine proper cooling method• Move motion control and acquisition hardware from JPL

to Caltech for development and testing• Move AO racks down from Palomar. They have been

modified with cassegrain cage mounting hardware

29

P3K Team Meeting #15

Documents