A Breadboard Kit For The Bare-Ass Boost

[smallbearelec]

I have updated the How-To

https://www.smallbearelec.com/HowTos/BreadboardBareAss/BreadboardBareAss.htm

to reflect what the kit provides:

http://www.smallbearelec.com/servlet/Detail?no=1529

I know that I am going to get requests now for JFETs sorted and characterized, especially for phasers. For right now, I only have testing capacity for the few that I need for kits. To be able to efficiently handle requests for specific parameters, I really want an automated setup. I know that Vp and IDss are easy to test, but figuring transconductance needs two measurements of gate bias/drain current. Also, I would like the rig to automatically print a ticket when done. As one of the "geezer" generation, I last wrote Assembler back in the days when the 8080 was hot; I could re-learn, but don't want to. If anyone is interested in doing a project in return for a bunch of store credit, please drop a message to the smallbearelec address.

My next project is the bread-to-box for the Bare-Ass: perfboard layout, pcb and complete build.

Regards
SD

[PRR]

> figuring transconductance needs two measurements of gate bias/drain current

Bah. Long before you were coding 8080 (before ENIAC) it was done in one (AC) shot.

You bring the device up to a suitable (can-o-worms) bias point, put small AC voltage in, read AC out. Hickok had many patents and sold thousands of tube Gm testers which were little more than a power transformer and an AC meter (and a LOT of sockets and switching and side-functions).

Bias point: for most small-signal audio, 1mA is a nice round number and will be representative of actual op-points. 6V-7V is a reasonable Vds. Hickok invariably used 1V AC off the 50Hz/60Hz power transformer. The 1V causes oddities in high Gm tubes at low Vgk, JFETs have lower Vto, so 100mV may be a better tickle. 1KHz leads to smaller bypass caps but I suppose 60Hz is fine too; also you might use a DMM and summa these fall-off above 400Hz.

Plan below gets any suitable N-JFET very close to 1mA and outputs 0.1mV per uMho (100,000V per S).



Two 9V batts should last over a year if you don't leave a FET in it.

While testing you should monitor Source for at least +0.1V (opamp and LED); if not, then this JFET is not big enough to work right at 1mA. Voltage over 3V suggests the FET is too fat or too high-Vto for small 9V work.

You ideally should ear-monitor the Output for pure tone; if distorted, something isn't happy and the reading can't be trusted.

Vs is also a useful sort-number. It is the Vgs to get 1mA of flow. Parts well-matched on this number should be well-matched in multi-FET arrays such as phase shifters.

Big cap should be Bi-Polar because it goes one way with FET and the other way when FET is removed. Go ahead and do BP on the lttle cap, so you can do P-FET by swapping the battery leads.

An NPN with 973 Ohm (1K 1% || 36K) resistor in series with Emitter can be a calibration check. It should read 1,000 uMho.

[thelonious]

Combine PRR's circuit with an Arduino-based voltmeter and some basic code to make it print a ticket in Windows, and I'll bet that would fix you right up. Not that I know enough to implement that, but somebody around here surely does...

(and you could run it right off the voltage from your USB port, with a charge pump)

[smallbearelec]

Hi Paul--

I still have enough of my marbles to have figured out that programming a micro-controller to read, calculate and print would be the way to go for this job. I did not/do not know theory well enough to realize that it could be done in one step. Thank you for the circuit suggestion for making it efficient. My offer of store credit for design time remains open.

SD

[PRR]

> making it efficient.

For the narrow problem of measuring Gm, it's neat.

For the broader problem... there's issues.

Arduino does not directly measure small AC volts. The caveman style would be an AC/DC converter, semi-precision, plus maybe a booster to get up in the several-volt range. The digi-head approach might be to read the AC as samples, find the high/low, and compute the AC voltage. Having done that, you might also read the input voltage to be sure it is in a 50mV-150mV zone, and divide based on its actual value at the moment.

Which is perhaps more complicated than just driving the Gate with a small DC step and reading the DC step at the output.

Even with automated tags, there's still a lot of hand-work. Putting FETs in the jig. Getting the right tag on the FET. Going through the tags or database to find parts/sets that meet a given request.

What you really want is a box. You shovel bulk FETs in a hole. Requests come in via internet. The box finds suitable parts/sets, factors the price by how "rare" that combination of specs is, and (if confirmed) spits the tagged bag onto your packing table. I don't know how to do all that. Anything less is major labor, even if you get the test-tag routine sweet.