Recently 200 pairs of subject audio mosfet with 2011 datecode are acquired for a project. Excess parts will be sold and price is listed in the “Store” page item 5.0.
The matching is using a tektronix curve tracer 577. Match Vgs at 5A pulse mode 300uS (this is to prevent heating up of the mosfet too much). Gain of this part is about 5S (A/V). For a change of 0.005V Vgs, the change in Id will be about 25mA, that is 0.5% of 5A. In the adjustment of the Vgs in Tek 577, the resolution is 0.5V (range that used to measure the mosfet) x 0.01 scale = 0.005V.
In the vertical scale of tek 577, there is 80 scales which represent total 8A, thus the resolution of the screen is 1/80 = 1.25% or variation of 100mA. So 100mA against 5A (measurement point Id) is about 2%. Adding these two uncertainty, the best accuracy of the measurement is about 2.5% with this instrument. In terms of Vgs variation, it will be about 0.025V.
Note also that there is a thermal drift of the Id due to temperature increase of the mosfet during measurement. If DC is used for measurment, the mosfet will need heat sinking but heat transfer will not be consistence due to mounting variation of mosfet to heat sink (thermal contact resistance, tighting of screw, ventilation and room temperature etc…). Thus this method is not recommended as it further increase the uncertainty of measurement.
In power amp application, mosfet is used in parallel and heat should be evenly distributed among all mosfets in parallel. I would say a variation of Id within 5% is no problem and within 10% is still good to use.
After sorting, below is the raw data and analysis of 200 pcs of 2SK1530 and 2SJ201 for Vgs at Id=5A respectively. The bin is divided into 0.01V wide and a histogram is plotted to show the distribution. At the bottom of the table, max, min, range, average, standard deviation is calculated. Note that Id measurement is a quick check and the data is not that accurate and can be ignored. It can be seen that there are not much N to P match on Vgs. Re-sample check on the Vgs of matched mosfet finds that the matching is better than 2% on Id variation.