User:Jstapko/EngLab/pstester

Power Supply Tester
This is intended to be a variable dummy load for testing lab power supplies at different current levels. Refer to its CDIO Page for the project goals and scope.

Current Hogging
refer to the Wikipedia Page on Thermal Runaway for a description of this phenomenon and its causes. This is an issue because we need to run 2 transistors of the selected type in parallel in order to pass the desired current level. One alternative for dealing with this problem might be to select a different transistor that is capable of passing the entire test current (insert part number of the one we have in the lab here, with a link to its datasheet and perhaps a picture) by itself.

Thermal Runaway
This is when various processes within the transistor cause it to pass more and more current until it burns out.

Causes
this seems to be common knowledge
 * increased ICBO (base - collector leakage current) (NRI book)
 * increased ICEO (emitter - collector leakage current, ICBO X hfe) (per the NRI book on transistors)
 * increased Hfe (if base current is kept constant, this causes greater collector current)
 * decreased Vbe (less junction voltage drop leaves more voltage to pull current through base circuit, leading to higher collector current)

Mitigation

 * use precisely matched transistors
 * use emitter stabilizing resistors (read the "thermal mismatch" section of All About Circuits.com "BJT Quirks"(accessed April 04, 2014) for a description of this technique)
 * use low impedance resistors in the bias voltage divider (add link to book that talks about stability factor)
 * use a balancing pot, per the NRI books (try to find an online version and link to it)
 * mount paralelled transistors on the same heat sink to keep them at the same temperature
 * use op amps for feedback and to help stablilze bias (eg, tie one input to a current sensing resistor, and have the output control the transistor base)
 * try [Dave Jones' idea]for an op amp stabilized constant current load (ink to dave jones' video)
 * uise old voltmeter/lcd displays from [voltmeter hack] (add link to pate)

Suggestions from Jasin: (if Jasin has a wiki page, link to it here)


 * do not design the circuit to run at exactly the maximum current rating of the transistor, leave a safety factor of perhaps 20 %
 * use heat sink grease to make sure the transistors make good contact with the heat sink
 * make sure the heat sink surface is flat and smoothely polished to make sue the transistor makes good contact with the heat sink
 * use a generously sized heat sink and a fan for forced air cooling

additional ideas based on Jasin's feedback:


 * insulate the heat sink from the chassis so that the transistor can be directly bolted to the heat sink, without having to put an electrical insulator between the transistor and the heat sink, thus reducing thermal resistance from the case to the air
 * use quadrotor or similarly high speed/power motors to move LOTS of air across the heat sink

Analog Panel Meter
Current Status:
 * Several teams have burned out meter movements while trying to determine what shunt is needed toi make them read 20 amperes full scale
 * This user has at least 2 books that describe detailed methods for determining the value of the meter coil's resistance ahnd calculating the value of an appropriate shunt
 * This user additionally believes he has several suitable movements in the personal parts collection, but still has to dig through the collection, find them, and test them
 * Experience has shown that even if the resistance value of the shunt is known exactly, factors such as contact resistance of lead wires and where ring lugs are bolted to wires has a substantial effect on the total resistance of the shunt, and resulting meter operation, so final adjustments must be made very carefully
 * leads have been soldered to a 10 turn helipot which could be used for determining the exact current/voltage required to run a meter movement at full scale

Next steps:
 * one idea for assisting with final adjustments consists of a collar with a set screw in it that bolts to a thick piece of wire in a coil. The collar can be screwed to different places along the wire to make small adjustments to the resistance/voltage
 * When trying to determine the value of a shunt, it is helpful to think of the ammeter as a voltage operated device, then use variable power supplies and rheostats to determine the millivolts required to push the meter to full scale. With that known, the resistance of the shunt can be calculated very closely because the current through the meter is VERY small compared to the current through the shunt, and from the source.  This means the shunt can be modeled as a voltage source, and loading effects of the meter can be safely ignored
 * identify and measure characteristics of a new panel meter

Overview

 * Philbrick SQ-10A op amp drives the final transistors, takes the current setting input from a potentiometer, and senses actual transistor current to automatically adjust (Dave Jones' circuit)
 * A second philbrick SQ-10A senses the difference between the currents through each transistor and automatically adjusts the drive current of the second final transistor to keep currents through each transistor equal
 * it is hoped that op amp balancing will eliminate the need to put both transistors on one heat sink, which will increase the flexibility of choices for heat sinks
 * it is additionally hoped that op amp control will reduce the tendancy of thermal runaway
 * power dissipating resistors are to be in the emitter circuit, not collector circuit as per previous designs. This should increase degenerative feedback, improving stability.

Detailed Description
The new design is based strongly on Dave Jones' circuit, but with op amp balancing of the currents between two parallel transistors. A potentiometer with suitable additional resistors in series is to be shunted across the Op amp power supply per the Dave Jones circuit. Additional resistors are to be added in series with the potentiometer to allow the maximum angular displacement of the pot shaft for the desired control, while also preventing the op amp (and thus transistor) from going any higher than necessary. The wiper of the potentiometer will go to the non inverting input of the first op amp. Current plans call for this op amp to be a Philbrick SQ-10a based on available parts in the HCC lab, but this may need to be revised if the SQ-10a proves to be defective. The output of the first op amp will be fed, possibly through a limiting resistor, to the base of the first transistor. It is unclear at this time what base current will be needed to drive the final transistors to full output. It is also not known at this time how much current the SQ-10A is capable of sourcing or sinking. If it turns out that the op amp is capable of directly driving the final transistor (2N3055's by the current design) then no driver transistor will be needed. It may however, turn out that a driver transistor IS needed, and this would need to be selected based on results of testing the 2N3055's. The collectors of both final transistors will go to the positive terminal of the power supply under test. The emitters will be connected to one end of the high power resistors, with each transistor getting its own resistor or group of resistors. It may also be necessary to switch different resistors into or out of circuit for different current settings or for testing power supplies with different voltages. A wire, possibly through a calibrating resistor or pot, will go from the emitter of the first transistor to the inverting input of the first op amp, whose non inverting input goes to the control pot wiper. An additional wire will go from the emitter of the first transistor to the non inverting input of a second SQ-10a. The output of the second SQ-10a will feed the second transistor. The second transistor stage will be essentially identical to the first, except that the wire from the emitter will go to the inverting input of the second SQ-10a. The overall effect of the second sq-10a should be to make the second transistor's current follow that in the first transistor, regardless of the cause of those changes. Calibrating resistors may be needed between various op amp inputs and the feedback voltage taps to accound for slight differences in op amp gain, lead resistance, manufacturing tolerances, etc.