27 Sept. 2017
This project is a simple implementation of a basic function I have used many times in the past as part of custom-designed sensor test equipment. Inexpensive and high-quality data acquisition (DAQ) systems have been readily availbe for a long time, and where you don't need esceedingly high degrees of accuracy or precision are often useful for building custom sensor test systems. One issue, however with your typical DAQ system is that the analog outputs are limited to a +/5V or +/-10V range - and many sensor assemblies may require a higher bias or power supply voltage t(12V, 24V) to adequately. Typically, a sensor assembly does not require a highe amount of current, often less than a few tens of mA. One solution is to add an voltage amplifier that boosts both the voltage range and output current capability of your DAQ system's analog output. Some common requirements for such an amplifier are:
Stable gain in the range of 1-5.
Stable DC offset (NOT AC coupled!)
Ability to source/sink tens of milliamps.
Ability to limit output current.
Ability to measure the output current.
Kelvin force-sense connections.
Addressing each of the above points...
Stable Gain - Note that I did not say 'accurate' gain. One of the advantages of a DAQ system is that the software can be used to cure many calibration ills, but only if the gain is stable - as in it does not change over operating voltage or temeperature or phase of the moon. similarly, a stable DC offset can also be dealt with effectively by software. As for the range of 1-5, this is suitable for converting common +/-5V DAQ analog outputs into a +/-25V range that will power many types of sensor assemblies.
Output Enable/Disable - Simply driving the amplifier input to 0V does not mean that it no longer can source or sink current, so a way of disabling the output stage is desirable. The ultimate way to do this is with a relay, but there are also electronic methods for doing so, that while not guaranteeing zero output current under all conditions (e.g. high voltage applied to amplifier output) may suffice in many cases.
Tens of mA of source current - We are trying to power up an automive sensor assembly here, not crank the engine !
Output current Limiting - Sometimes the devices you test are bad. sometimes they get plugged into the tester the wrong way. Sometimes something else goes wrong that creates a short circuit at the device Under Test (DUT) connections. Limiting the output current to something reasonable (those tens of mA) can help prevent the release of the magic smoke from either your DUT or the tester itself if something goes wrong.
Ability to Measure Output Current - Measuring the sensor assembly's bias current, even coarsely can provide very useful statistical process control (SPC) data that can give you a heads up when something has changed in your production. Feeding the current drawn by DUT is relatively easy to do.
Kelvin Force-Sense Connection - Sometimes you need to apply 12.000V to your DUT. Two wires (power and ground) are typically not going to cut it at this level of precision - you get voltage drops over the wire leads to the DUT, and also any kind of connector also introduces its own voltage drops. To get this kind of precision, you actually need to measure the voltage applied at the device, and use feedback to dynamically adjust the amplifier's output. This arrangement of using active feedback is called a Kelvin or 4-wire connection.
For this project, of the above items, I tried to hit all except the Kelvin voltage connection - many time in the past I have found this to be overkill, especially when using a cheap (12-bit ) DAQ system. The block diagram below shows the key components of my Sensor Power Supply:
Key features of this circuit are:
Runs from +12V to +28V
Has switch-selectable gains of 1, 2, 5
Current Sourcing Output (transistor emitter output)
Provides output current up to ~100mA
Has switch selectable current limiting options ~200, 100, 50, 25mA
Provides a voltage feedback of output current at 50mV/mA
Output swings ~0V-24V @ 28V supply
Has a TTL/CMOS compatible enable (active low)
Has an on-board enable switch if you don't wnat to use the external enable
Has an on-board adjustable reference (0-5V) if you just want to use it as a fixed voltage source.
The following photo shows my implementation:
Control inputs and power are connected at the terminal block on the left, while the output is connected to at the terminal blcok on the right (ran out of mathcing terminal blocks :( ). The DIP-switch controls functions like enable, connecting the internal reference, the gain, and the current limit configuration. The trimpot on the upper left is for adjusting the on-board referecne, and the one int the middle of the board is for adjusting the Common Mode Rejection of the intrument amp measuring current. This is needed becuase the intrument amp is made out of one of the op-amps in the quad I used and a couple of resistors. A monolithic instrument amp would have been better, but I didn't have a suitable one in my junk box. Also, when the board is enabled, an LED (lower left) lights up to indicate that it is active.
While this project worked satisfactorily from an electrical standpoint, one thing I didn't like was that the heat-sink gets too hot to touch when you leave it in driving a dead short @ at high current limit (100+mA) @ 24V. while the transistor is probably feeling just peachy, this could be an annoyance in an application. Some simple improvements might be to put in a taller heatsink (lower theta), adjust the resisotrs in the current limiter to cut in at lower current levels (50mA?), or just not run the thing into short circuits continuously :)