Most ESR meters just detect the voltage drop caused by a square wave. This is caused by the impedance, not the ESR. This does not work. OK, it does work, because at 100KHz |Z| is small, itÕs an indication but because |Z| is higher as ESR it works a bit to well and it makes you swapping a lot more caps as needed. The problem is the ESR is rather complex. It is a part frequency indepenend resistive and a part frequency dependend caused by the dielectric losses . This meter uses a phase comparing method to detect the voltage drop that is caused by the ESR. This meter can be calibrated using resistors like most. You calibrate using a short for zero and some resistors to mark the scale up to around 40 Ohm. It measures acurate from 100nF upto as large as you want. So no problem with 1uF caps like the most meters have. It does 50-350KHz 4,7Vtt square. There is no diode protection because that influenced measurements (test signal to high) . Caps need to be discharged. Update 11-2012, a better version. 20KHz to 100KHz and some protection. The 74LS74 could become metastabile and give half the amplitude. The ESR then looked higher as it was. My meter never had that problem but it could be possible. I also added some protection. I have not tested upto how many joules it can handle, so decharge the caps before measurement but better as nothing. I do not think it survives a charged 400V, 450 uF capÉ Version MKII. This one is usable for digital readout.
Here I added two comparators as oscillator (4 in one package, do not leave the unused ones floating, tie +in to +5V through a 10K and connect the -in to ground. The outputs through 10K to +5V). The first one makes a 20KHz to 100KHz (more or less) squarewave (or much higher if you leave the 1K2 out). The second one shapes it in a nice 50% dutycycle squarewave. The inverters make two 180 degrees apart squarewaves. It is probably possible to use 1 comparatror, I used two 180 degrees apart instead of the inverters but that did not work as well, so I only used the second output just to be sure. I had 3 inverters in serie on the pcb, like on the picture because it was the oscillator so thats why I used 3 in the schematic. There are 6 in a package so who cares ;-) (my model uses all six of them, instead of grounding the rest) The opamp is now a LT1013 precision opamp. But it is not very critical. You have less noise and more resolution with this opamp. I added a divider for the minus point of the meter zo it can be set to zero Ohm by grounding it. Be carefull, I use a panelmeter with fixed decimal point. The panel meter goes upto 1.999V but reading 19.99. So 10.00 on the display is 10 Ohm but 1V on a DMM. First set the frequency at 20KHz so you minimise ESL and skineffect, short the Kelvin style test leads (4 wire methode) and zero the meter with the zero 10 turn trimmer. Then connect a 18 OHm 1% (or better if you have, or just neasure it) and adjust the second trimmer for a 18.00 reading (or 1.8V on a multimeter). If you now use it at 100KHz you will have a small zero offset caused by the resistance of skineffect and ad 18 Ohm too. But if you measure a cap you will see the resistance decreases a little with increasing frequency. Just like it should.
PA4TIM ESR Meter
uses phase comparing method (original here)