Device Testing Circuits

In the old days, you could count on the specifications, in a datasheet, for a part with that part number, stamped on it. Now, with parts routinely being re-stamped, you need to be able to see that parts meet your needs.

For measuring general operating specifications, such as resistance, capacitance, or gain, there are plenty of low cost instruments, out on the market. However, for measuring limits, there are few affordable options, that also yield unambiguous results.

Since I make kits, I need to know, the parts, I ship are good, and will do the job. I have made several test circuit jigs, that can remedy this current parts supply dilemma.

Rectifier Reverse Recovery Time Tester

Below, is the circuit, for verifying that ultra-fast rectifiers, are truly that, or just re-stamped 1N400x common variety standard recovery rectifiers (click on drawing to enlarge):

Recovery_Test

Rectifiers are used to block reverse current, when “rectifying”, and AC signal. What most people don’t know, is that NO rectifier works instantaneously. The table, in the above drawing shows what those times are. Your “standard rectifiers” intended to rectify household AC (50, or 60Hz. 100 or 120Hz full-waved rectified) have recovery times of ~30uS. During that time, current flows backward, for that recovery period. Over an 8.3mS (8333uS – 120Hz) period, that’s only 0.36% of the total time. However, in a switching supply, pulsing at 50KHz (20uS), a 1N4007, is as good as a simple wire; no rectification at all. That’s why “fast” and “ultra-fast” rectifiers exist. These rectifiers, nominally cost 4x the prices, of standard recovery rectifiers, tempting some to pass off the slower parts, as their more expensive cousins.

If you have a good scope, and signal generator, you could build the test setups, used to measure the actual reverse recovery times. These are found in google searches. But these also need user interpretation. This circuit, however, accumulates these pulses, and presents them as a simple voltage. The longer the recovery time the higher the voltage. Its not linear, but still yields distinct results. A very fast rectifier, like a BAV21, has very short reverse pulses, and yields voltages in the 50mV range. Often used UF4007 gives readings around a half volt, but your “jelly bean” 1N4007 has long reverse periods, giving voltages over 6.5V ! A big 3 Amp rectifier, like a 1N5408, is even slower, yielding almost 10V readings.

When measuring, allow about 10 seconds, for the voltage to stabilize. Always insert the D.U.T. (Device Under Test – rectifier to be tested) in the proper orientation. If not, you may damage the test jig. Also, as shown, test only rectifiers rated for 200PRV (or PIV), or higher. See the “Breakdown Voltage Test”, below, if you suspect the voltage rating, too.

Breakdown Voltage Test

Simple jig for verifying component breakdown voltage, such as bipolar collector-emitter breakdown voltage (Vceo) or diode/rectifier peak reverse voltage (PRV) (click on drawing to enlarge).

Breakdown_Test

This is intended as a NON-destructive test ! The 100K resistor (R1) limits the current. Also, its important to monitor the current, so that it stays under 500μA, while raising the voltage. Some people have reported damaging parts, while doing a similar test. I don’t know why (?). Maybe they should check if there’s a capacitor on the D.U.T. side of R1 (?). If so, then, YES, you just toasted the part ! Keep all capacitors, on the other side.

Requires only an adjustable high voltage power supply, a voltmeter, and a simple jig using 2 resistors and a view connectors. Below is a photo of the same jig:

HV_test03s

Here “spring clips” are used to connect to the power supply. 2mm tip jacks for connecting the meter, and alligator clips for holding the D.U.T. (device under test).

To test a device, insert in orientation, as shown. Raise voltage slowly, from zero, and watch the meter. It should normally read zero. Raise the supply voltage, until you get a non-zero reading. This is the breakdown level. Note the supply voltage where this occurs. It should be higher than the published rated value, for that part. In the photo above, a MPSA42 NPN transistor is being tested. Its rated Vceo is 300V. That specific part tested, actually broke down at 450V. 150% of the published spec. This is typical. Most transistors are never tested. Statistical methods are used for production, so most parts greatly exceed the rated value, such that they can guarantee almost 0% failures.

Measuring Very Low Value Resistors

Measuring resistors under 1 ohm, even resistors under 10 ohms, can be problematic with the ohmmeter settings on most DMMs (Digital Multimeters). I once received a message from a customer, that stated that the 0.5 ohm resistors, included in one of my kits, measured over 2 ohms, but the circuit still worked fine. He used some slick piece of test gear able to read such small values. That only works if you know how to properly use such a meter. These are 4-wire (or Kelvin) resistance meters, which both have to be hooked up properly, and are quite pricey items. Unless you need real precision (finer than 2%), there are cheaper ways. Below is basically that same method using relatively common parts, and any cheap (think Harbor Freight) Digital multimeter.

R_measure_low

For example, if Vx measured 150mV (0.15V) and Vr measured 2.9V, then (Vx/Vr)  0.15V/2.9V = 0.0517, then multiplied by Rref (10 ohms) = 0.517 ohms. Its a 0.5 ohm resistor, that measures 3.4% high, which is fine, if its a 5% part.

The beauty of modern times, is that we have inexpensive meters, and calculators. many of the methods found in text books, come from the days, when neither digital voltmeters, nor calculators existed.

Measuring Very High Value Resistors

On a continuing note, the same method for measuring very low resistors, can also be used, with some modification, for very high resistances. Most DMMs are fine for measuring resistors between 5 ohms and upto 1M ohms (cheap models) or 10M ohms (mid priced models and up).

Other than the obvious change of the battery voltage (100V vs 3V) and Rref (10K instead of 10 ohms), Vbat is measured, here, instead of Vx. The reason for that, is that the meter has its own characteristic resistance, that will load down Vx, when measuring very high resistors. This is not and issue with low value resistors. A meter with 1M resistance, in parallel with a 1 ohm resistors, changes the combined value by less than 1ppm.  That same 1M meter across a 1M resistor yields a parallel value of 500K. A 50% variation ! With the rig, below, a 10K Rref, only introduces a 0.1% error for resistors 10M or greater. Still only 1%, for resistors down to 1M.

R_measure_high

For example, if Vr measures 2.1mV (0.0021V) and Vbat (from batteries, or a bench power supply) measures 100V, then Vbat/Vr = 100V/0.0021V = 47,600.  47,600 x Rref = 47,600 x 10, 000 = 476,000,000 = 476M or 1.3% higher than 470M.

These two methods can be trusted than a straight ohmmeter measurement, since the lead resistance is not a factor, nor are the measurements loaded down by the meter. even an inexpensive meter. The limiting factors, are the precision of the resistors used for Rref, and the resolution of the meter.

Part Mishandling

There are many eBay sellers who simply don’t know, or don’t care, how to handle semiconductors. MOSFET transistors are very sensitive to static, and most ICs now made, are made mostly of MOSFETs. Even some bipolar based ICs are sensitive to static, to a lesser degree, but more so, than in the past. An old part number like a 555 timer, is not the same device, as it was, when first issued, 50 years ago. Semiconductor processing changes regularly. Silicon wafers have increased in size, while they are cramming more parts, on the same sized chip. These smaller transistors, while doing the same job, can’t tolerate the static spikes, the larger older transistors could survive. ALL SEMICONDUCTORS, then, need to be handled with anti-static measures. For ICs, I only buy from major distributors, or qualified eBay sellers, who always pack their parts, in anti-static containers (black or pink foam, anti-static rails, silvered anti-static bags).

PCB Testing

Sometimes its necessary to check the traces of printed circuit boards. Surprisingly, they can handle a considerable amount of current. Even thin 10mil (0.01″/0,25mm) trace intended for carrying only small signals.

(click on drawing below, to enlarge):

CS1_I-tst_NT2RA-1

The test here, was performed to mainly see if a 10mil trace could handle 600mA, needed to drive a tube filament. A secondary test was to measure the voltage drop, to see if was acceptable.

In this case, since the purpose was simply to drive a tube heater, the results were satisfactory. For other purposes, such as instrumentation, or an IC’s power rails, needing minimum variation on the power rails, this may not be the case. As always application determines suitability.

There was a noted difference in the measured resistance, versus the calculated value. 0.25 ohms versus 0.18 ohms, respectively. That’s over a 3rd difference. This could be cause by both the board being over-etched, resulting in traces that are closer to 8 mils, versus 10 mils, and copper being thinner. Not the 1.37mil thickness defining “1 oz” copper; the standard thickness.

Advertisements