 | Muon Detector Project Page |
||
New Project PageThis project is now aimed at measuring any pulses with a high time resolution, down to 65 picoseconds (hopefully). Sensing devices, such as photomultiplier tubes, can be connected to this unit for measurement. The project was originally intended for use in a cosmic-ray muon detector, but now is more general to allow the connection of a wider variety of sensing devices. This page is for the latest design. The origins of this design (incomplete and thus non-working) can be found here.Update 1/12/2009 More changes:
Update 11/21/2008 OK, I've decided to make many more changes:
Update 11/18/2008 I will remove all the clamping diodes from the input channels (shown in the latest schematic at the bottom of the page). They would have worked but I will probably just use single schottky diodes possibly with a varistor. The varistor will begin to conduct before reaching the varistor voltage, so I have to be careful with any voltage measurements. This unit is designed primarily for up to a 2V p-p signal. The clamp diodes do not appreciably conduct until around +/-5.3V (supply voltage plus voltage drop across diode). But this point varies depending on the current through the diode (and temperature, as usual). The point is, the voltage level measurement becomes less accurate the closer the input voltage get to the supply rails. So, the diodes and varistor are for protection, but because of their presence, the input voltage should be kept at 2Vp-p or less to ensure accuracy. Note that this device is designed to measure short-period pulses anyway, not voltage, but I still want voltage to be accurate. I am concerned that the switching converters will be too noisy, even with all of the filtering. Yet, the device is designed to be very power efficient (one of the reasons I decided to use discrete logic instead of a power-hungry CPLD). I may have to do something different. I may be able to deal with the fundamental frequency of the switching converters, but a higher-order harmonic may be a problem. Note that all switching converters produce ripple (the fundamental). Ripple is documented in data sheets. However, there are higher frequency components to this ripple that is usually not documented. The harmonics are related to the steepness of the slope of the square wave. Remember that square waves are essentially a superposition of many sine waves of different frequencies. The highest frequencies depend on the steepness of the slope (steeper slope = higher frequencies). Anyway, I have seen this higher frequency noise on my scope, and it can be larger than the ripple. This device is sensitive to noise from DC up to around 150MHz (possibly higher). Also, although I live in Albany, the KLOO FM tower near Corvallis is a constant problem for development (but not so much a problem for the actual implementation of the device, which will be shielded and have a good board layout (hopefully)). It is interesting to watch a frequency modulated signal on a scope while correlating it with what is actually heard on the radio. Try it! The schematic is still being developed so it is not yet usable. Everything below is outdated, with the most recent schematic at the bottom.  New Schematic Still needs more work, but almost finished. Click here for large image. A few quick notes: Theoretical time resolution of 65 picoseconds. This is the time resolution of the TDC. I am using the TDC-GP2, an ASIC from acam. The input electronics will limit this performance somewhat. I may eventually move over to something like LVPECL electronics and the TDC-GPX. But, I can't afford that right now. No more digital controls. It sounded like a good idea at first, but it turns out that the capacitance of digital switches and digital potentiometers is really high, way to high for our high-frequency and/or short pulse-width signals. Most everything is surface-mounted now, and the PCB will be four layers. Some parts in the new design are optional. All the various options utilize the same PCB. There is a trigger output connector, with optional square-wave output. The square-wave output can be used, for example to measure capacitance by measuring the time it takes to discharge from one voltage to a lower voltage. You may have done this in a Physics lab. This adjustable square-wave output (up to 68MHz) takes the place of the function generator. The on-board temperature sensor also utilizes this method of using time to measure a quantity using the TDC (time to digital converter). Optional SD card slot like in a digital camera for huge data logging storage capacity. No special hardware required to flash the microcontroller. All programming is over the serial port (the uC has a bootloader). 128x64 back-lit LCD display (optional) Otherwise, must use serial port. keypad (optional). Otherwise, must use serial port. Real-time clock with ability to wake up unit to take periodic measurements and the sleep again. (optional) The comparator thresholds are digitally controlled. Optional A/D converters for inputs and other optional onboard sensors. Much better, more flexible trigger section. This section I've never been fully satisfied with because of the many compromises, but I'm finally OK with it now. Many things have to explained here, which I will do later. I will say there is a programmable coincidence window, and the ability to switch directly to the TDC for predictable events. It is also capable of handling the muon lifetime experiment. Coincidence detection now uses a feature of the TDC (start masking). Power SupplyCurrently:The 5V and 3.3V supplies are via Low-Dropout linear regulators, and the -5V is via a switched supply. Using switchers would be more power efficient, but the TDC's data sheet strongly recommends against using them, regardless of filtering. The -5V supply will still have some effect on the 3.3V (which the TDC uses), but it is much smaller than if the 3.3V line itself was supplied by a switcher. The RS-232 chip also produces ripple on its supply pin (3.3V), but it is much smaller and there is a shutdown pin. The power supply also needed proper power sequencing. This has been achieved, but just in case, the diode is there to prevent the 3.3V from exceeding 5V during power-up and power-down. Simulations show that they get close during power-down. Changes Due:
Update (not reflected in schematic):
Update2 (not reflected in schematic): * I will use two TI switching regulators. One has very low ripple which I will use for the main input op-amps and comparators. It is less efficient (higher switching frequency). The second one I will use for everything else. It is noisier but much more efficient. Noise is less of a concern with the devices connected to this one. This also helps isolate load switching transients from the quiet one. The second switcher also has a comparator I will use to monitor battery level. Both switchers feature inrush current limiting. The -5V inverter switcher will hang off of the second 5V switcher and will have only a small load. All switchers are of the switched-capacitor type (charge pumps). I may have to venture into inductor-land. Update3 (not reflected in schematic): Made several more changes:
I will have time for this project during winter break at which time I should finish the schematic. Latest schematic v0.83 
I still have to make one last pass through the design yet. Some resistor values will change, I/O pad labels, power dissapation, etc. I will also seek comments on some of the design elements from an electrical engineer (particularly the new, but strange, but very effective input protection mechanism). Otherwise, the basic design is now complete. I will post v0.9 when the final pass is complete. I will print out a four-layer board from this version, which will include some extra pads/traces for filtering components near some devices in case the current filtering proves insufficient. In particular, there will be pads/traces for two capacitors and some series element (low-frequency ferrite bead, low value resistor, or inductor) for each switching power supply output and input. The noise characteristics need to be tested on the actual board to be used, so this part of the design will not be finished until then. The input filtering components will be partially combined in some way. This schematic will eventually be on several separate pages. UPDATE March 1, 2008I'm taking too many classes this term and cannot work on this project. A few more updates:I began work on the code for the GUI (graphical user interface) several weeks ago. It has branched into a separate project that will become a commercial product and MUST be complete late Summer. I am now using Atmel's AVR microcontrollers. Switching to this RISC architecture was surprisingly easy coming from the 8051's which had some RISC qualities already (R-registers). This project will probably use an ATmega-644.
|
|||
|
|||||||||||||||||||||||||||||||||||||||||||||||||
|
|||||||||||||||||||||||||||||||||||||||||||||||||
| |||||||||||||||||||||||||||||||||||||||||||||||||
| |||||||||||||||||||||||||||||||||||||||||||||||||
