Difference between revisions of "Woody Underwood"

From UConn PAN
Jump to navigation Jump to search
m (→‎Digital Control Board: formatting error)
Line 122: Line 122:
  
 
==== Digital Control Board ====
 
==== Digital Control Board ====
The semester began with the digital control board prototypes sitting on lab bench waiting to be tested. My job was primarily to design the boards, not to debug them, though I still played an integral part in the testing process. We were all relieved that none of the boards appeared to smoke immediately when we first applied power, though there were a few confusing problems that had to be sorted out. Most of these problems are described in detail on the [[Digital control board debugging notes]] page. More or less, we had to solder a pin that wasn't connected, replace the crystal oscillator with a CMOS oscillator, and rewire few traces that had were accidentally connected to pins intended for other signals. I won't repeat all of the changes we made here since it would be redundant to the page that already discusses it, but the important part is that we have one board that is fully operational, and the version 2.0 schematics/layout reflecting all these changes are complete. I anticipate that production of the version 2.0 boards will occur at the beginning of the spring 2010 semester.  
+
The semester began with the digital control board prototypes sitting on lab bench waiting to be tested. My job was primarily to design the boards, not to debug them, though I still played an integral part in the testing process. We were all relieved that none of the boards appeared to smoke immediately when we first applied power, though there were a few confusing problems that had to be sorted out. Most of these problems are described in detail on the [[Digital control board debugging notes]] page. More or less, we had to solder a pin that wasn't connected, replace the crystal oscillator with a CMOS oscillator, and rewire few traces that were accidentally connected to pins intended for other signals. I won't repeat all of the changes we made here since it would be redundant to the page that already discusses it, but the important part is that we have one board that is fully operational, and the version 2.0 schematics/layout reflecting all these changes are complete. I anticipate that production of the version 2.0 boards will occur at the beginning of the spring 2010 semester.  
  
There is still one lingering problem with the digital board design that has not yet been resolved, however. Even after all of the version 2.0 changes were made to the version 1.0 boards, only one of the three boards produced actually functioned properly. For a reason unknown to us, two of the boards ended up with their digital to analog converters burning up and smoking. We thoroughly reviewed the schematics and the layout, and found nothing miswired. Furthermore, we analyzed the placement of the ball grid array DAC to the best of our ability, and determined that a misalignment of balls is not the cause of the failure. All of our power supplies are stable, properly decoupled, and grounded as called for by DAC's manufacturer. Currently, the best hypothesis for the failure of 2 of the 3 DACs is that one or more of the absolute maximum ratings must have been exceeded at some point during the assembly process. Analog Devices lists fairly specific specs with small tolerances relating to maximum temperature and the length of time which the DAC may spend inside a reflow oven. We have run into some difficultly trying to find out from the boards' assembler whether they followed the published specs for the assembly process, but we intend to keep trying until we get an answer.
+
There is still one lingering problem with the digital board design that has not yet been resolved, however. Even after all of the version 2.0 changes were implemented on the version 1.0 boards, only one of the three boards produced actually functioned properly. For a reason unknown to us, two of the boards ended up with their digital to analog converters burning up and smoking. We thoroughly reviewed the schematics and the layout, and found nothing miswired. Furthermore, we analyzed the placement of the ball grid array DAC to the best of our ability, and determined that a misalignment of balls is not the cause of the failure. All of our power supplies are stable, properly decoupled, and grounded as called for by DAC's manufacturer. Currently, the best hypothesis for the failure of 2 of the 3 DACs is that one or more of the absolute maximum ratings must have been exceeded at some point during the assembly process. Analog Devices lists fairly specific specs with small tolerances relating to maximum temperature and the length of time which the DAC may spend inside a reflow oven. We have run into some difficultly trying to find out from the boards' assembler whether they followed the published specs for the assembly process, but we intend to keep trying until the cause of the DAC failure has been resolved.
  
 
==== Amplifier Board ====
 
==== Amplifier Board ====

Revision as of 21:58, 17 December 2009

About Me

My name is Woody Underwood. I'm a junior entering my seventh semester in the physics program at UConn. Since the summer of 2008, I've been working in the lab of Dr. Richard Jones designing electronics for the US Department of Energy's GlueX experiment. My assignment is to develop three circuit boards that work together to tag photons coming from the diamond radiator. My circuitry essentially measures (indirectly) the energy of these photons in order to determine if they are of interest to GlueX. The three boards I am designing consist of a digital board, an analog board, and a connecting backplane.

About My Circuitry

Electrons leaving the diamond radiator are deflected by a magnetic field into an array of scintillating fibres, producing photons. Wave guides carry these photons to SiPMs (silicon photomultipliers) mounted on the analog circuit board. The analog board contains transimpedance amplifiers and summing circuitry to condition the signals for digitization.

The sensitivity of the SiPMs and the gain of the amplifiers on the analog board are controlled both by the power supply VCC and bias voltages supplied from the digital board. The digital board receives commands from a computer via ethernet, and uses a 32-channel DAC to output appropriate bias voltages to the SiPMs on the analog board. The digital and analog boards are connected by means of a backplane, which is also responsible for providing power and grounds to both boards.

Summary of Fall 2008 Work

All circuitry design work was done using Altium Designer. The digital board was the first to be designed. The first step in designing the digital board was to review the list of key components that had already been selected by Igor and Dr. Jones. These components included such things as the Xilinx Spartan-3A FPGA, and the Analog Devices AD5535 DAC. I began by looking through datasheets for these components to find out their needs, including power and decoupling requirements. I reviewed the pinout diagrams, and then looked through Altium’s standard libraries to find components that matched (in many cases the particular component I was looking for was not in the library, but a similar footprint or schematic symbol was). For components without matching schematic symbols, I entered pinout information from the datasheets into Microsoft Excel, using a layout compatible with Altium’s Smart Grid Insert function. Then, I was able to literally copy and paste pin information from Excel into Altium to generate the schematic symbols I needed.

Once I had appropriate schematic symbols available for all parts, I began making appropriate connections in the schematic view in Altium. Though tedious, this task was not exceedingly difficult. I finished the schematics in several days, and then moved on to PCB design. I switched into Altium's PCB view. The footprints corresponding to the components I used in the schematics were automatically inserted by Altium. My job was then to position these components in logical places on the board and make all of the connections corresponding to the nets defined in the schematics.

Due to the large number of components being placed in the limited space available on the digital board, Altium's auto-router proved completely useless. Therefore, I routed the board manually. Despite Altium's revolutionary convergence of schematic and PCB design into a single program, this was no easy task. During the routing process, I had to take into account not only the connections that had to be made, but also things such as avoiding crosstalk and minimizing trace length for sensitive components. I was able to complete routing after several weeks of work. The digital board design has since been completed and the board has been printed. It is currently awaiting assembly.

The analog board provided a host of new challenges. The basic schematic for the transimpedance amplifier on the analog board was completed by Igor and Dr. Jones before the semester. Inputting the schematic into Altium was not very difficult. However, one problem I encountered was that the analog board contains 32 copies of this amplifier circuitry. After failing to find any way to insert multiple copies of both the schematic and its corresponding PCB layout, I decided to insert only single copies of each schematic page, and copy and paste the PCB layout to produce 32 copies of the amplifier circuitry. At first this seemed like a quick and easy way to get all of the necessary circuitry onto the PCB. However, I eventually discovered that this procedure would lead to major problems with the board assembly process (due to duplicate component designators, and for other reasons). Fortunately, this revelation came around the same time that Igor and Dr. Jones found a problem with the performance of the amplifier circuit. Making any changes to the amplifier circuit at this point will require a major reroute of all the traces on the analog board. Since the board needs to be completely redesigned anyway, this will give me another chance to find a way to match schematics with all 32 copies of the amplifier circuitry.

The backplane design is currently in progress. It should be relatively easy to complete. All that remains to be done is to add the LEMO connectors and power inputs. The board is simple enough that it can be routed completely by the auto-router, though a quick hand routing will probably be superior. I anticipate that I can complete the backplane with a few days of concentrated work over break.

Included below are links to the files I have been working on. Included in the files for each board is a "SmartPDF," viewable in Adobe Reader. For those without Altium Designer, these may be the best files to look at. They include complete schematics and PCB layout, and are also indexed by component.

Any questions about the tagger circuitry can be directed to me at mitchell.underwood@uconn.edu

Related Files

  • DigitalBoard.zip:
    • Altium Project File (SiPM Control Board.PrjPcb)
    • Altium PCB Layout File (Prototype1.PcbDoc)
    • Altium Schematic Files (*.SchDoc)
    • Altium Annotation Document (SiPM Control Board.Annotation)
      • Not used, but generated by Altium when opening the project
    • Altium PRJPCBSTRUCTURE File (SiPM Control Board.PRJPCBSTRUCTURE)
      • Not used, but generated by Altium when opening the project)
    • “SmartPDF” of the board and schematics (SiPM Control Board.pdf)
      • Can be used to explore the PCB layout and schematics without needing Altium
    • Pick and Place File for board population (Pick Place for Prototype1.txt)
      • Used by board assembler
    • NC Drill Files (Prototype1.txt, Prototype1.DRR, Prototype1.DRL)
      • Used by board printer
    • Gerber Files for all layers (in folder Gerbers)
      • Used by board printer
    • Altium CAMtastic file (CAMtastic4 FINAL.Cam)
      • Basically a composite of all the Gerbers
    • Photos and 3D rendering of populated board
      • In folder “Photos”
    • EMF Files showing different layers
      • In folder “EMF Renderings”
    • AutoCad File of PCB (Prototype1 Autocad.DWG)
    • Altium Library of Custom Footprints for Digital Board (GlueX IC Library.SchLib)
      • Current as of completion of digital board
      • This library has since been updated for the backplane
  • Analog Board 20081211.zip:
    • Altium Project File (AnalogBoard.PrjPcb)
    • Altium PCB Layout File (AnalogBoardPCB.PcbDoc)
    • Altium Schematic Files (Amplifer1.SchDoc, Summer.SchDoc)
    • “SmartPDF” of the board and schematics (AnalogBoard.pdf)
    • Altium Component Definition for SiPM (SiPM Library.PcbLib)
      • Contains part footprint and pin information for the SiPM component
  • Backplane 20081211.zip
    • Altium Project File (Backplane.PrjPcb)
    • Alitum PCB Layout File (Backplane.PcbDoc)
    • Altium Schematic Files (Analog Connector.SchDoc, Digital Connector.SchDoc)
      • Analog Connector = Eurocard to analog board
      • Digital Connector = Eurocard to digital board, +3.3V voltage regulator, and location identifier jumper
      • LEMO connections not yet included in these schematics
    • “SmartPDF” of the board and schematics (Backplane.pdf)
    • Pin layout files used to define pinouts for custom components (Pin Layout, 96 pin connector.xlsx, Pinouts.xlsx)
      • Pin Layout, 96 pin connector = pinout definition for 96 pin Eurocard connector
      • Pinouts.xlsx = pinout definitions for digital board, which were reused for the 48 pin digital Eurocard receptacle on backplane
    • Altium Library of Custom Components (GlueX IC Library.SchLib)
      • UPDATED to include new backplane components
      • An older version of this library was used for the digital board

Summary of Spring 2009 Work

At the end of the fall semester, Dr. Jones, Igor, and I determined that the amplifier/summing circuit we had designed simply didn’t have sufficient performance characteristics to be useful for GlueX. During the beginning of the spring semester, Igor came up with a new design, utilizing more transistors, to provide the high gain, fast response, picosecond resolution amplifier that we needed. The design performed flawlessly both in MatLab simulations and in a handmade single channel prototype. The first of my goals for the spring 2009 semester was to capture the schematic for this new amplifier into Altium designer, and layout a new amplifier board. The second of my goals, of course, was to complete production of the digital control board prototypes.

The populated digital board, with a tube of Chap-BlockTM for size comparison.

Since the fall semester left us with three unpopulated digital control board PCBs, getting those PCBs assembled with their components was the first priority. I began the semester by tracking down all of the components we needed (some of which were selected at the end of the fall semester), and making appropriate substitutions for components whose availability had changed since the fall. In the process of selecting these components, I noticed several places where it seemed like power consumption on the board may be somewhat high. To fix this, I developed a spreadsheet in Microsoft Excel that calculates optimal resistor values to use for to obtain a specified voltage divider stiffness. With this tool, I was able to optimize power consumption across the board, and select appropriate components. Once all components had been selected, ordered, and received, we sent the order out to Screamin’ Circuits for assembly. The boards came back several weeks later, and are currently awaiting testing by some undergraduates who will be in the lab this summer.

While Igor was finalizing his amplifier/summing circuit, I worked briefly on design of the backplane. A number of details regarding trace impedance and board dimensions were ironed out. Nonetheless, many problems still remain which I will need to tackle over the summer. The first of these problems is that we have yet to find an appropriate low cost coaxial connector to route signals off the backplane. In addition, screws with which to mount the backplane to the tagger must be selected so that appropriate holes can be created for them on the PCB.

Once we were satisfied that the amplifier/summing circuit performed as required, I shifted work from the backplane to the amplifier board. Around the same time, I began working on a poster to present my work at the Frontiers in Undergraduate Research Exhibition held during Open House Weekend here at UConn. If you’re interested in my poster, check it out here.

A prototype layout of 5 amplifier channels and a summing circuit. Actual size is ~1.6" tall. Areas in green represent problems to be resolved over summer 2009. A prototype amplifier with no green spots has been designed and will be replicated to this design soon.

To start the amplifier/summing circuit project, I searched high and low for information about how to handle multi-channel designs in Altium. Not surprisingly, the first Google result on the query “multichannel design Altium” had everything I was missing during the fall when I was trying to lay out the original amplifier circuit. Using my new knowledge of Altium’s multichannel capabilities, I captured Igor’s new design into the schematics editor of Altium Designer. With proper nested schematic sheets, the entire 30 channel amplifier/6 channel summer design was compressed into just 4 schematic sheets, vs. the ~40 or so that would have been required had I laid out the complete schematics of the old design.

I spent about a week and a half trying to figure out how to handle nesting independent nets from a repeated subsheet into another repeated subsheet which also produces independent nets from the nets of the first sheet. This sounds somewhat complicated, and I suppose perhaps it is a somewhat unique situation, since none of the ~5 sample multichannel projects included with Altium had such a construction in them. Basically, each summing circuit has five amplifier subcircuits, each of which puts out its own signal independent of the other four. From the perspective of the entire board, there are 6 summers, each of which has five independent signals coming from the amplifiers, and one summed signal. Determining how to get Altium to realize the proper connections from each individual amplifier, through that amplifier’s summer, to the main schematic was a complicated mess of naming conventions, but eventually I was able to make Altium reflect all of the appropriate connections in the PCB view. Though there are a few net naming issues still to be resolved, this problem has mostly been ironed out.

The final few weeks of the semester were spent laying out amplifier components in the PCB view of Altium. As of right now, a compact design for an amplifier measuring 0.183”x~1.3” has been completed. Making use of 0201 size components, this amplifier is approximately .2” shorter than the old design, despite incorporating an extra transistor. The amplifier design features an isolating ground trace running the length of the amplifier to prevent crosstalk between channels. A prototype layout of the summing circuit has also been completed, though some layout issues there remain to be resolved over the summer.

Related Files

  • Design Snapshot 5/5/2009
    • Backplane
      • Backplane.PrjPcb (Altium Project File)
      • Backplane.PcbDoc (Altium PCB File)
      • Analog Connector.SchDoc, Digital Connector.SchDoc, LEMO.SchDoc (Altium Schematics)
      • Backplane.pdf (SmartPDF File of the schematics and PCB layout)
    • Amplifier Board
      • Analog Board.PrjPcb (Altium Project File)
      • Amplifier Board.PcbDoc (Altium PCB File)
      • Main.SchDoc, Amplifier.SchDoc, Summer.SchDoc, Voltage References.SchDoc (Altium Schematics)
      • Analog Board.pdf (SmartPDF File of the schematics and PCB layout
        • The SmartPDF shows the entire PCB workspace, including components that have not yet been placed on the physical PCB.
        • The bottom left corner of the PCB view shows a proof-of-concept 5 amplifiers/1 summer layout.
        • Finding AMP6 using the PDF bookmarks shows the prototype amplifier layout that resolves mismatched footprint issues (indicated in green in the image farther up on this page).

Fall 2009 Work

Mid-Semester Update, 11/12/2009

We now have one fully working prototype of the SiPM Digital Control Board, and the SiPM Amplifier Board is currently being manufactured and assembled by Sierra Circuits, Inc. The backplane design is nearly complete and will be going out for quoting within a week.

Here is a zip file containing the most recent versions of the tagger files: Design Snapshot 11/12/2009

A few notes about these files:

  • There are four subfolders in the zip file. The folders for the three PCBs are clearly labelled. Each PCB folder has one project file (.PrjPcb) which can be opened in Altium Designer. Each project file refers to only one PcbDoc file, making it easy to figure out what you should look at to see the board.
  • The SiPM digital control board project has 4 shelved polygons which should be restored (Tools->Polygon Pours->Restore 4 shelved polyons) to see the state of the board as it was last manufactured. Revisions from the testing process have not yet been entered into Altium but can be reviewed on this wiki: Digital control board debugging notes
  • The SiPM amplifier (analog) board project will give an error message that it was unable to find "Test PCB.PcbDoc". Please ignore this. The file was intentionally omitted to prevent confusion, and is not related to the actual design.
  • The backplane project will give an error message that it was unable to find "Backplane_New.PcbDoc". Please ignore this. The file was intentionally omitted to prevent confusion, and is not related to the actual design.

End of Semester Update

This semester, a lot of progress has been made on the tagger electronics. The digital control board is now onto revision 2.0, the amplifier board is in production, and the backplane is awaiting completion of the purchase order for production to begin. It is exciting to say that the bulk of the work is now finally complete. Remaining for the spring are the tasks of testing the amplifier board, and seeing that everything fits into the backplane. Here are some details about what I accomplished this semester, as well as links to the most recent project files.

Digital Control Board

The semester began with the digital control board prototypes sitting on lab bench waiting to be tested. My job was primarily to design the boards, not to debug them, though I still played an integral part in the testing process. We were all relieved that none of the boards appeared to smoke immediately when we first applied power, though there were a few confusing problems that had to be sorted out. Most of these problems are described in detail on the Digital control board debugging notes page. More or less, we had to solder a pin that wasn't connected, replace the crystal oscillator with a CMOS oscillator, and rewire few traces that were accidentally connected to pins intended for other signals. I won't repeat all of the changes we made here since it would be redundant to the page that already discusses it, but the important part is that we have one board that is fully operational, and the version 2.0 schematics/layout reflecting all these changes are complete. I anticipate that production of the version 2.0 boards will occur at the beginning of the spring 2010 semester.

There is still one lingering problem with the digital board design that has not yet been resolved, however. Even after all of the version 2.0 changes were implemented on the version 1.0 boards, only one of the three boards produced actually functioned properly. For a reason unknown to us, two of the boards ended up with their digital to analog converters burning up and smoking. We thoroughly reviewed the schematics and the layout, and found nothing miswired. Furthermore, we analyzed the placement of the ball grid array DAC to the best of our ability, and determined that a misalignment of balls is not the cause of the failure. All of our power supplies are stable, properly decoupled, and grounded as called for by DAC's manufacturer. Currently, the best hypothesis for the failure of 2 of the 3 DACs is that one or more of the absolute maximum ratings must have been exceeded at some point during the assembly process. Analog Devices lists fairly specific specs with small tolerances relating to maximum temperature and the length of time which the DAC may spend inside a reflow oven. We have run into some difficultly trying to find out from the boards' assembler whether they followed the published specs for the assembly process, but we intend to keep trying until the cause of the DAC failure has been resolved.

Amplifier Board

Throughout the summer and the beginning of the fall semester, I completed the design for the SiPM Amplifier Board, version 1.0. All 30 channels have been implemented along with their summing circuits. As I mentioned in the mid semester update above, this board is being manufactured and assembled by Sierra Circuits, Inc. Some unexpected setbacks occurred in the assembly process that delayed the boards somewhat, but I believe everything is now on track for delivery of the finished product before the start of the spring semester. A number of problems relating to minimum quantities of parts, as well as part naming conventions have now been resolved, and the assembly process should be complete in several weeks.

Backplane

The backplane design, version 1.0, has also been completed. Sierra Circuits will be manufacturing the backplane, which we will assemble ourselves. The backplane is a particularly intersting PCB from a manufacturing standpoint, because it is designed to be completely light proof. Since the backplane is the only material standing between the whole of Hall D and the inside of the tagger microscope dark box, opacity of the board is particularly important.

Most PCBs with internal copper layers are already fairly opaque, because the copper blocks light from transmitting through the FR-4 and prepreg. Through hole components do not particulary compromise board opacity, because they are plugged with component pins and solder. What does have a greater affect on the opaqueness of a board, however, is the thermal reliefs by which through hole pins and vias connect to internal plane layers. These reliefs are designed to aid in the soldering process by minimizing the amount of copper that is directly connected to the plating in the hole. A typical relief consists of four 7-10 mil traces connecting the hole's plating to the internal plane. The plane itself remains about 20 mil away from the hole, to prevent conduction of heat during the soldering process. Unfortunately, this means that there is a small gap in which there is only FR-4 to stop light from passing through the board. Since FR-4 is transparent, this poses a problem for our design. We were able work with Sierra Circuits to come up with a board design that includes an internal layer of special black FR-4 to prevent light leakage through the heat reliefs. While this is certainly not a standard feature, Sierra was confident that they could implement it for us, and we look forward to testing the boards' opacity in the spring.