Hi,
Prompted by the suggestions in #7 and #9, I added a pack of AO3400 to an order via AliExpress that was already in preparation: A pack of FIFTY (in the normal SMD assembly tape) for a little over a Pound, and they arrived within a week! However, I "forgot" to order any adapter PCBs, which actually would have cost more than twice as much as these transistors; So how to test, let alone use, them in their tiny SOT23 (1.3 x 2.9 mm) packages? Also, the adapter boards look significantly larger than the TO92 through-hole transistors (BC548, BC337 and 2N7000, etc.) that I regularly use, so I decided to fabricate my own. Note that my version generally has the Gate (or bipolar Base) wire in the middle, whilst the commercial adapter boards usually have the Drain (or Collector) in the middle, as is more common with Power and RF transistors.
Veroboard (Stripboard) might be suitable, but I've previously found that the heat of a soldering iron tends to detach the copper tracks from very small boards. Therefore, I now use the double-sided, Plated Through Hole "Prototyping" Pad-Boards available in sizes such as 6 x 28 and 14 x 20 etc. holes/pads on a 0.1 inch pitch. The method that I used worked far better than expected, although I do have a fine-tipped Weller (thermostatic) soldering iron and a cheap "Helping Hand" gadget (from the Middle of Lidl), with a Magnifying Glass, LED Light and mounted/adjustable Crocodile Clip(s) to hold the PCB. However, I was relieved that the quantity / price of the transistors meant that I didn't need to fret about the one or two that "vanished" into the deep pile carpet (or wherever).
First, I cut a strip two holes wide (and whatever length was available) off the PadBoard with a junior hacksaw (which seemed more reliable than anything motorised). This will ultimately be cut to sizes of 2 x 3 holes (5 x 7 mm) or 2 x 2 holes, depending if a "3 pins in line" or a "Corners of a 0.1 inch square" footprint is preferred. Gluing the package between the pads onto the PCB might be wise, but I was impatient and the Drain pin generally lies between two pads, so I cut some approximately 5 cm lengths of 0.5 mm diameter "BTC" (Bare Tinned Copper) wire, although offcut resistor leads, etc. would probably be adequate. Then I bent each wire into an elongated "U" with sides spaced by 0.1 inch (2.5 mm). The first "U" will connect with and locate the Drain lead so it is inserted through two adjacent pads on the component side, to leave a semi-circular loop on the top side (typically along the longer axis), and perhaps pre-tinned to ensure satisfactory "welding" to the Drain lead.
At this stage the package's Gate and Source "pins" need to be only tacked (with solder) onto their pads, with the loop restraining the body in position (but obviously a spot of glue could help). If a hole gets blocked with solder, then my mainstay, manual "Solder Sucker" (or a length of Solder Wick) should easily clear it. Then freely apply solder to the U-bridge with the hope that it will connect through to the Drain pin. Next, insert a second U-link from the underside so that the ends just protrude through the Gate and Source pads on the top side, and apply a final dose of hot solder to each. Then cut (off) the "U" end of the link (which was only necessary to hold the wire(s) in place during soldering), or you may prefer to keep this uncut until the testing.
I was surprised that the solder seemed to flow easily to the Drain "pin", but this obviously needs to be tested. This can be done easily using a multimeter, preferably with a Diode (Forward Voltage) Test mode (often combined with the Continuity/Buzzer function), or a high Ohms range if nothing else is available. The basic continuity can be tested through the Substrate Diode which requires a reverse connection of the probes, i.e. probably Black to the Drain and Red to the Source. My meter indicated about 500 mV, much as expected for a forward diode at low current. With the probe connections exchanged, the FET should test as "Open Circuit", but only if the Gate voltage is close to the Source voltage. This is a reason for delaying the cutting of the second U-link until now. Also, one side of the first (Drain's) U-link can be cut off from the underside of the board now (so with experience it could start as a "J" shape), to give the desired footprint (e.g. 3 wires clockwise or anticlockwise around a 0.1 inch square).
With the Source and Gate leads separated, the basic switching of the FET can be checked (probably with the Red probe to Drain and Black probe to Source), even in a "Wet Finger" test (using the human body as a resistor) : The current from Drain to Source will be highly dependent on the Gate (to Source) voltage, so with a finger on the Gate lead and (another) on the Source lead, the FET should be completely OFF (no current). However, with the finger(s) on Gate and Drain leads, the multimeter will measure the GATE (Threshold) voltage, NOT the Drain to Source Saturation voltage. For my tests this was around 600 mV (i.e. remarkably similar to a bipolar transistor) which is surprisingly low, but of course the meter current will be around only 1 mA, whilst the FET is rated up to 5 Amps (or 30 Amps pulsed ! ).
Finally, you might wish to "dress" the wires for a specific footprint and/or coat the top side of the board with epoxy resin, because the leads will not be insulated, nor tolerant of soldering temperatures unless physically held in place (e.g. by another PCB). Personally, I plan to bend the leads parallel to the plane of the board so that the resulting "package" has a footprint similar to a TO92 through-hole component. These will be used in place of the ubiquitous 2N7000 FET, with a similar size but "guaranteed" operation from a pair of AA{A} cells near their end of life (perhaps 2 volts, but the PICaxe Brownout Threshold is actually around 2.7 volts, and the 08M2's absolute minimum supply is not much less than this). Here are a couple of photos (not a full "Instructable") which might save another thousand words here.
In conclusion, I should add that I haven't used these FETs in earnest yet, so cannot say to what extent they might be "fake", but the first five have certainly functioned "correctly" in the multimeter testing.
Cheers, Alan.
