Millenium Bypass note

[R.G.]

Peter Urso sent me email noting that there is a twist to the Millenium that might help some people.

Background.
The essence of the Millenium is that certain signal diodes have leakages in the nano-ampere region, which is high by silicon diode standards. This provides a fixed source of very, very low current to bias JFETs or MOSFETs on when not pulled to ground by a high value resistor.

In all the semiconductor switched LED schemes, there is some source of bias to a switch device; the bias is shunted to ground by the output resistance of the effect. The size of the bias current determines the largest output resistance the effect output can have, because the bias current causes a voltage to appear at the effect cap output. Using resistors is a losing battle because even a 10M resistor lets too much current through.

Germanium leakage is micro amps, too big; normal silicon junctions are picoamps up to maybe a nanoamp, which is too low. Gold doped signal diodes are about 10 nanoamps, which happens to be just about right for good switching.

The quirk about "gold doped" is that back in the 60's people found they could make diodes switch off faster if they doped the junction with gold to make the charge carriers recombine faster inside the junction. There were whole families of diodes made like this, of which the most popular were the 1N914 and 1N4148. These were perfect for Millenium use.

Modern silicon processing has gotten so good that some makers of diodes can make NON-gold-doped junctions switch fast enough to make the 1N914 and 1N4148 specs, so they sell these as those part numbers. They sometimes have lower leakage.

The Trick
Use more than one. Peter suggested using two 914's in parallel for the "high leakage" diode and one 914 for the "low leakage" diode. This is a good trick if you happen to have gotten some new-manufacture diodes that don't leak enough, and is worth a try if normal methods won't get a Millenium to work.

Before you all rush off to stick varying numbers of diodes into the top and bottom of the Millenium, let me make some notes.

The low leakage diode is there only for protection of the gate in the MOSFET under bad conditions. It is NOT needed for proper operation of the Millenium. If you are having trouble with it, leave it out. The low leakage diode is not mysterious. A 1N4001 probably works. I don't recommend them because they are big-junction devices and with larger area comes larger leakage. The slam-dunk solution is the collector base of a common NPN transistor, which is specifically designed for LOW leakage. So is the gate junction of a JFET. Either works. Or none.

If you have no low leakage diode in there and the MOSFET or JFET switches v  e  r  y    s  l  o  w  l  y  then you may have a low leakage modern 1N914 or 1N4148. In that case, put another of your signal diodes in parallel with the first one. This nominally doubles the leakage and may make for faster operation.

Peter's idea of using two diodes in parallel at the top and one at the bottom is good if you have both modern signal diodes and no access to any other diodes.

The commonest problem
The commonest problem with all Millenium 2's is that the LED is on all the time. In 999 out of 1000 cases, this means that you have your MOSFET soldered in backwards and that the substrate diode is conducting and letting the LED light. You *****MUST***** have the right pinout on any semiconductor device to get it to work right. Mother Nature is quite picky about this.

[brett]

An educational and entertaining post.
Thanks RG!
PS as the proud owner of about 20 boxes with M2s in them, I also suspect that the ones with big pulldown resistors (e.g. 1M) turn off more slowly too.  Or maybe it's all in my head?