|Elliott Sound Products||Project 177|
The standard way to measure transistor hFE (DC current gain) is to introduce a known and constant base current, and measure the collector current. For example, if you inject 500µA into the base and measure 45mA of collector current, the hFE is 90. This is fairly simple to implement, and works very well. The Project 31 transistor tester works this way, and also provides other tests, such as breakdown voltage with and without a resistor between base and emitter. A meter monitoring the collector current can be calibrated in hFE, because a known base current is used and there is no calculation needed.
Most multimeters have the facility to 'test' transistors, but this is only useful to determine if the device functions (maybe). The gain displayed is (probably) reasonable for small signal transistors that work at low current, but multimeter testers are absolutely useless for checking power transistors. Even something like P31 might not provide the ability to run the test the way you want, at least not without a great deal of messing around.
While the constant base current method is by far the most common, in some cases the transistor(s) will ideally be tested with a constant collector current. Unless you are willing to experiment, testing transistors at a constant collector current is a great deal harder with the standard test circuit. Using constant collector current means that the collector current will always be fairly close to the value you set, and the base current is then measured to determine the gain.
This isn't available on the vast majority of testers (including P31), and a dedicated test setup is needed because it's an unusual way to run the test. Ideally, it would be possible to vary the collector voltage as well, because the gain does change as the voltage is varied. When the collector voltage is low, the gain is also lower, increasing as the collector voltage increases. For example, you may measure a gain of 288 with a collector voltage of 1V, rising to 639 at a collector voltage of 20V. Although that's based on a simulation, a 'real life' test will give similar results. Allowing a variable collector voltage adds considerable complexity, and is not allowed for in the design that follows.
There are quite a few ways that a constant current test can be performed, but not all are easy to implement. To be useful, a tester must be able to test NPN and PNP transistors, ideally with the minimum amount of switching. This is both in the interests of cost and reliability, particularly as often quite high current is involved when matching power transistors. There is already a project that (almost) meets our needs - see Project 106. It's a contributed project, but it was designed specifically for NPN transistors and changing it to allow both NPN and PNP is difficult to achieve.
This is not as simple as the 'traditional' test method, and with the technique shown here there's an in-built error factor because the base current flows in the emitter circuit. However, the error is small and can be ignored if the goal is transistor matching (one of the most likely reasons you'd want to test this way). While a transistor's base in a real circuit will only draw as much current as it needs to bias the circuit properly, you can't simply use a low value resistor from a power supply to the base, because the base-emitter junction is normally forward biased. If you were to use a 5V supply and a 10 ohm resistor, a BC549 transistor will attempt to draw close to 200mA, and the transistor will probably be destroyed. This is many times more base current than the ratings allow. Fortunately, the impedance can be greater than 10 ohms, so all is not lost.
To test with a constant collector current, we have to use a constant voltage for the base, and a current sink (or just a resistor) in series with the emitter. The base supply voltage (less the 0.65V base-emitter voltage) appears across the emitter load, and the base current can then be measured. The emitter current is the sum of the collector and base currents, and if a 6V base supply is used with a 100 ohm emitter resistor (for example), the emitter current will be around 53mA. The collector current is very slightly less than this, because it does not include the base current.
For most transistors with reasonable gain, this small error can be ignored. For example, if the hFE is 100, the error is only 1%. This is insignificant compared to the changes that occur with differing temperatures or even collector voltages. The general scheme for testing with constant collector current is shown below. What we are really testing is constant emitter current, but it's close enough that a correction will only be needed if the transistor's gain is particularly low.
Figure 1 - Constant Collector Current Tester Principle
It's something of a misnomer, because it's the emitter current that is constant, but without excessive complexity this will work well enough for 99.9% of the tests you might wish to run on a batch of transistors. Although just a resistor is shown above, better results will be obtained with a constant current sink, but this makes the tester more complex, and it's not warranted due to greatly increased parts count and cost. It should be apparent (but perhaps not to novices) that the circuit shown will draw a base current that's almost entirely due to the transistor's current gain. It's a simple emitter follower, and the fixed emitter voltage and resistor ensue that the voltage dropped across 'Rb' is directly proportional to the base current. 'Rb' must be a low enough value to ensure that the voltage across it is limited, preferably to no more than 100mV.
With the circuit shown above, install the transistor with base, emitter and collector in the right places. Beware - an incorrect connection may destroy the transistor. To test, press the 'Test' button, and read the voltage on the connected DMM (digital multimeter). It should be between 10-100mV, and the accuracy depends on your meter. If the voltage is less, 'Rb' needs to be made larger and vice versa. The emitter current is determined by the voltage across 'Re', which is (roughly) the negative supply voltage less 0.7V (the transistor's forward biased emitter-base voltage), and less the voltage dropped across the base resistor ('Rb'). The transistor will not conduct until the 'Test' button is pressed.
There are several things you need to decide, one of which is the collector voltage you want to test with. It should be at least 5V, and although 12V will work well, the dissipation of everything becomes excessive. You could use an external variable supply so that any voltage within the supply's limits can be used, but you'd need to make sure that the polarity is correct. The base can be supplied from a separate supply too, but that gets messy and potentially dangerous to the transistor if you make an error. The design shown uses its own ±6V supplies, and is somewhat safer (for the transistors) than external bench supplies.
Emitter current is set by Re, and Ohm's law is all you need to calculate a resistor value for any desired current. For example, to test at 20mA (Ie), Re will have (about) 5.3V across it (VRe), so Re becomes ...
Re = VRe / Ie
Re = 5.3 / 20m = 265 ohms (use 270 ohms)
The value of the base resistor (Rb) depends on the expected gain of the transistor. For a typical gain of around 100, make Rb 1k. At a voltage of 100mV across Rb, the current through it must be 100µA, so the transistor has a gain/ hFE of 200 (20m / 100µ = 200).
It's very important that the transistors being tested are all at the same temperature. This can be monitored with a thermistor and an ohm meter so that the tests are comparable. If you don't manage the temperature properly, the results will not be useful. BJTs change their VBE (base-emitter voltage) and hFE with temperature, and a higher temperature means a lower VBE and higher hFE. High current tests can either be timed (e.g. wait 5 seconds and take a VBE reading, or use a thermistor and take a reading at the same temperature for each device being tested.
A worthwhile question to ask is "why?" There's no good reason that a 'traditional' tester with constant base current won't give good results, but a high gain transistor will draw more collector current, causing the die to heat up a little, and thereby increase the gain some more. It's rare to reach thermal runaway when testing, but if the collector current varies with each device (which it will), then the test may not show you the information you're after in a meaningful way.
The advantage of testing with a constant (or nearly constant) collector current is that each device you test is subjected to the same heating effect, so, at least in theory, the results will be more predictable (or perhaps less unpredictable). To determine the gain, you have to measure the base current and make a calculation, and that throws up interesting challenges as well. If you use your multimeter to measure current, it has an internal shunt for current measurements, and for low currents it will be a fairly high resistance. It will also change (perhaps unpredictably) if the meter is auto-ranging.
That affects the collector current unless the emitter uses an accurate current sink to ensure that the emitter current is truly constant. You can use a resistor, but the voltage across it will cause the emitter current to be reduced. This is the method chosen, and it will keep the emitter voltage constant to within 100mV or so (depending on the minimum voltage you can measure). Another potential uncertainty is introduced as well, because if the emitter voltage falls, the transistor has a higher voltage between collector and emitter.
Remember that the gain will change if the collector-emitter voltage changes. This is called the Early effect, after the man who first discovered it. As the voltage between collector and base increases, so does the transistors gain. I'm not going to provide all the formulae that describe this, so if you want to know more, look it up for yourself. The net result is that it's important to keep the base voltage the same, regardless of gain. That means that the base circuit should be a low impedance, and including a meter in series may cause errors that will be very difficult to quantify.
Now we are faced with a new dilemma - how can we measure the base current without having a reasonable value resistor in series. For example, if you need to measure 100µA and the base is fed via a 1k resistor, then 100mV across the resistor equates to 100µA. As it turns out, this is easily manageable, and a change of 100mV of collector-base voltage isn't a big deal. That's fine for testing at medium current, but if you want to test at (say) 1mA and the transistor has a gain of 100, the voltage across a 1k resistor is only 10mV, so getting an accurate reading is a great deal harder. Fortunately, if we decide that 100mV variation is alright, we can simply change the resistor to suit. The meter will always be reading millivolts, across the base resistor.
With this arrangement, the emitter current (and therefore the collector current) will vary depending on the transistor's gain. However, the change is not great, and with similar transistors the error is negligible. When matching transistors, if two (or more) devices are well matched, their operating conditions in the circuit described will be almost identical. Many approaches were examined, but this is the simplest and only requires simple circuitry throughout.
The tester should be simple, but with enough flexibility to cover most common requirements. It's easy enough to make changes if you need a specific current that isn't catered for, but that usually will not be necessary. Current ranges need to start from around 1mA up to 3A or so, following a 1, 3, 10 (etc.) sequence. This should cover most test requirements. The meter used to monitor the voltage across the base resistor will indicate 0-100mV for each range shown below. The base current is measured in decade ranges. By measuring the voltage across a known resistance, the internal resistance of a current meter is no longer an 'unknown' factor (especially true for an auto-ranging meter). The required emitter resistance has been rounded to the nearest standard value, but you can change it easily (you only need Ohm's law). The base-emitter voltage has been assumed to be 0.7V in each case, but it will vary from one device to another.
|Parameter||Resistors & Currents|
|Collector Current, Amps||1m||3m||10m||30m||100m||300m||1A||3A|
|Emitter Resistor, Ohms||5k1||1k8||510||180||51||18||5.1||1.8|
|Actual Current, Amps||1.04m||2.94m||10.4m||29.4m||104m||294m||1.04||2.94|
|Base Current, Amps||100n||1µ||10µ||100µ||1m||10m||100m|
|Base Resistance, Ohms||1Meg||100k||10k||1k||100||10||1|
|DMM Voltage, Volts||0-100m (all ranges)|
It's important to understand that the absolute value of gain is unimportant. For this reason, there is only a token attempt to ensure that the collector current is as indicated. The (theoretical) actual values are included in the table. It will vary somewhat with the device being tested, but if two transistors show the same base current on any given range, their gain is the same. This tester is intended primarily for comparative tests, and high accuracy is simply not needed. Even as a 'normal' tester, it will be more than good enough to show that a transistor is within specifications. Transistor hFE measurements are not 'precision' tests by any stretch of the imagination.
Figure 2 - Complete Tester Circuit
Rather than mess around with current sinks, the circuit shown will work just fine, and it has the advantage that nothing is polarity sensitive - except the device under test (DUT) of course. The emitter-base junction voltage is compensated (more or less) by resistor selection, and although there will be some inaccuracy in most cases it should be less than ~6% (as seen in Table 1). This is more than acceptable for measuring the absolute gain of a transistor, but for matching it's as accurate as your base current measurements.
While no transistor can be damaged if the base resistance is set too low, the same does not apply to the emitter current. If a small signal transistor is installed with the 3A range selected, it will likely be damaged regardless of the base current resistor setting. The very low ranges provide some protection, but even 0.5mA base current (10µA range) into a small signal transistor will cause a collector current of at least 50mA and a dissipation of over 250mW (it will get very hot).
The overall circuit has been simplified as far as possible. This makes it easier (and cheaper) to build, and the lack of any active components in the measurement section (other than the power supply regulators) means that there is nothing that can change with time or temperature to upset the readings taken. Since the transistor being tested is connected as an emitter follower, it's highly unlikely that it will oscillate or do anything else to upset the measurements. The high power resistors will cause some grief, and make sure that they can't heat the DUT as that will cause serious errors.
The emitter resistor switch has to be rated to handle the full current from the transistor. For the 3A range this may be a limiting factor, and it might be necessary to use a relay to switch the highest current range(s). Many rotary switches can't handle more than 200mA, so unless you can get one with higher current ratings, relays might be needed for the three high ranges. Alternatively, you could use separate toggle switches for these ranges (less convenient, but simpler and cheaper). If you do that, Sw2 needs an 'open circuit' position so no other resistor is in circuit.
Figure 3 - Relay Switching Example
Relays can be activated using the circuit shown. When the switch is set to a relay switched range, the switch wiper connects the relay coil, and the relay switches the emitter resistor. The two zener diodes are to suppress the back-EMF from the relay coil when the relay is switched off. Although the relay coil is in parallel with the current setting resistor, for the 1A and 3A ranges the extra current is of little concern. If you use a relay for the 300mA range, you'll need to adjust the resistor value slightly to account for the coil current (typically around 60-80mA depending on the relays you use).
Note that the selected relay will not activate until the 'Test' button is pressed.
The power supply needs to be kept simple, but it also has to be predictable. Regulated supplies are essential. While using a constant current sink is better than a resistor (and only a single supply is needed), it actually creates more complication, because separate current sinks are needed for NPN and PNP tests. This becomes silly quite quickly. By far the simplest way to get a 3A supply is to use two 7812 regulators in parallel. You can use three in parallel if preferred, which will keep the temperature lower and make the heatsink more effective. This minimises the dissipation in each to about 6W, and by using balancing resistors it's easy to make them share the current evenly. The bridge rectifier is shown using 4 × 1N5401 diodes, but a 10A bridge can be used if preferred.
Figure 4 - Power Supply Circuit
Using 7812 regulators is convenient, because they require no other external parts. Use thermal 'grease' to ensure the best possible heat transfer to the heatsink (which will have 6V on it). It's a low voltage, so insulating the heatsink from other metalwork is easy. The supply can be simplified if you don't need the 3A range. A single 7812 will be fine for 1A, although a heatsink is still necessary. Lowering the maximum test current also means the transformer can be smaller, and a 20VA unit with a 15V AC secondary will suffice. You may also be able to reduce the filter cap from 5,600µF to 2,200µF and use smaller diodes (1N4001 or similar will be adequate).
How you build the supply (and the tester itself) depends on your intended usage for the tester. If you only intend to match/ measure small signal devices, then you won't need more than 100mA, which simplifies everything and will be much cheaper. However, if you think that you may wish to test at higher currents 'some day', you might choose to include the extra ranges and high current supply - just in case.
The transformer is specified at 15V, because at 3A DC output, the ripple voltage still has to be within the range that lets the regulators maintain the regulated output voltage. The 7812 regulators need an absolute minimum of 2.6V more input than output at all times, or ripple will break through to the output. That means that the minimum unregulated voltage (including ripple) should be at least 14.5V - preferably more. Because the current can be up to 3A, ripple becomes a real problem unless the voltage is high enough, or C1 is much larger. The transformer should be rated for 50VA, but a higher rating will have better regulation. A smaller transformer might be alright (tests are intermittent), but regulation may become an issue, requiring more capacitance for C1.
Figure 5 - Centre Voltage (Split Supply) Circuit
The centre voltage ('artificial ground') is derived from a medium current buffer, using U3 (a µA741 opamp) and a pair of transistors. You can use pretty much any opamp you like - it's not critical. A simple resistive divider cannot be used, because the load is unequal. The emitter current will always be greater than the collector current, and the difference could be as much as 100mA - a power transistor with a gain of 30 on the 3A range for example. The buffer will maintain a predictable centre voltage with base current up to around 150mA. This arrangement is simpler and cheaper than building a pair of supplies, because both must be regulated and rated for the same current. The buffer will maintain the centre voltage within a few millivolts, regardless of load. C2 is optional, and the circuit will work fine without it.
There's no reason that you can't use either a single 12V switchmode supply (or external bench supply), or even a pair of 5V switchmode supplies. With separate supplies, you can make them different voltages, and switch them to provide (for example) 12V on the collector, and 5V for the emitter circuit. The rated output current needs to be at least 3A. I'll leave the switching up to the constructor, but if the emitter supply is reduced to 5V, the emitter resistor values will need to be re-calculated because there won't be 5.3V across them. It will be around 4.3V instead, and to get (say) 1A, the resistor will need to be about 4.3 ohms. All emitter current ranges need to be re-calculated (Ohm's law is all that's needed though).
Another option you may consider is to use rectified but un-smoothed DC for the measurements. This adds some uncertainty because the mains voltage is not a fixed quantity, and it can vary by ±10% from the nominal value (i.e. 230V or 120V), sometimes more. This means that if the mains voltage changes while you test, the measurements are no longer useful. Accurate matching will not be possible due to the variable supply voltages. For basic tests this won't matter, but if you only need general testing facilities, Project 31 is a better proposition.
Always make sure that the emitter current range is set to an appropriate value for the transistor being tested. Quite obviously, selecting a current range above the device ratings is not helpful. Your multimeter connects to the 'DMM' terminals, and should be set to the 200mV range. The voltage measured across the base resistor indicates the current, with the voltages for the nominal currents as shown in the table above. The polarity of the meter is unimportant, but if the +Ve input is connected to Gnd, it will read +mV for NPN and -mV for PNP.
Select a base resistor setting that gives you a voltage up to 100mV. For example, if the base switch is set to the 100µA range (10k resistor), a voltage of 85mV indicates a base current of 8.5µA. If the collector current is set to 3mA, the transistor has a gain of 352. As noted above, the absolute value is unimportant when you are matching devices, but the figure obtained will still be pretty close to reality (at the voltage and current used for the test - it will change if either is varied).
It's very important to keep the test for each transistor to the same duration, or allow enough time for the temperature to stabilise. The latter is fine for small signal devices, but will take far too long with a power transistor and heatsink. A heatsink is essential for high current tests, because dissipation can reach around 18W on the 3A range, and even on the 1A range it will be 6W. Most testers are unforgiving if you set the wrong current range, and this is no exception.
The base current range can be set anywhere you like to start. If the range is too low, the voltage will be much greater than the 100mV maximum we're looking for, so simply switch to a higher range until the voltage measured is between (say) 10mV and 100mV. It's not even essential to work out the base current and calculate the gain if you are simply matching transistors. Just check each device in turn, and note the reading on the meter. Any devices with identical (or very close) readings are matched to the accuracy of your measurement.
When matching devices, make sure that you don't hold them in your fingers, because that will affect their temperature and the base-emitter voltage and hFE will change. It's not common that extreme matching is needed, but if that's what you need, then the device temperature is critical. In use, matched transistors need to be in intimate thermal contact, something that isn't easy with plastic cased transistors. Even the PCB layout can cause a mismatch if tracks are different lengths and/ or go to other parts that run hotter than ambient.
There are none, with the exceptions of the two projects mentioned in the text. There is almost nothing to be found elsewhere that looks at constant collector current testing. While there are many forum posts that ask about or recommend using a constant collector current, few (none that I could find) seem to have arrived at a suitable method to do so.
The only thing that comes close to the tester shown here is a tester that uses the common base configuration, and measures collector and emitter currents - alpha (α), tests rather than the much more common beta (β) test (roughly equivalent to hFE). Even looking at numerous resources based on alpha testing yielded little of any real value to anyone.
|Copyright Notice.This article, including but not limited to all text and diagrams, is the intellectual property of Rod Elliott, and is © 2018. Reproduction or re-publication by any means whatsoever, whether electronic, mechanical or electro-mechanical, is strictly prohibited under International Copyright laws. The author (Rod Elliott) grants the reader the right to use this information for personal use only, and further allow that one (1) copy may be made for reference while constructing the project. Commercial use is prohibited without express written authorisation from Rod Elliott.|