Millenium Bypass C, diodes question

[bioroids]

Hi!

I'm wondering what's the reason of using more than one leakage diode on the Millenium Bypass C (the one with CMOS inverters).

http://www.geofex.com/Article_Folders/Millenium/The%20Next%20Millenium.pdf

Is it because the inverters need more leakage current to be able to switch? Can I get away with only one diode? Is it because there are two gates in parallel that have to be driven?

Also, the gates are in parallel to have a higer current source capability to drive the led?

Sorry to bother with these questions, I should give it a try on the breadboard anyway to find out for myself.

Luck

Miguel

[SaBer]

I have one 1n914 and three gates in parallel and it works fine 

[R.G.]

Is it because the inverters need more leakage current to be able to switch? Can I get away with only one diode? Is it because there are two gates in parallel that have to be driven?

The 1N914 and 1N4148 were originally made from gold-doped silicon to cut recombination time and hence switching time. This had the side effect of making them leak much more than ordinary silicon diodes, but less than germanium. That side effect is what makes them useful for this "high leakage" duty. However, some modern silicon processing is so good that some makers can make diodes that meet the switching speed without gold doping. They sell these as the same part numbers, since they meet the same spec, but may not have the same leakage.

Some people complained that their leakage diodes were not working well, and an alert reader suggested using multiples. That works, and the picture was intended to show the idea of "one or more, as necessary". The inverters do not need more leakage current to switch. Some high leakage diodes are higher leakage than others. So try one. If that doesn't work, use two.

Also, the gates are in parallel to have a higer current source capability to drive the led?

Yes, and because there is a plethora of gates in any single package.

[bioroids]

Thanks a lot!

Don't you hate it when they change the device in some way and still use the same number? 
Let's see how the 1N4148's I have behave.

Luck.

Miguel

[Bernardduur]

I used new and old 4148's and 1P60's I had laying around and all worked fine.

[bioroids]

Great!

What about using the base-collector diode of some cheap transistor, like a 2N3904.
On the datasheet it says "collector cut-off current" on 50nA (max). Well I'm not sure if this is the right parameter to look at  This is with a  Vcb of 30v.

The reverse current of a 1N4148 at 20v reverse voltage is 25nA... that's half of the transistor. All this looking at Fairchild datasheets, may be different for other manufacturers.

I think an old GEOFEX article mentions this option, but it seems I cant find it anywhere.

Regards

Miguel

[R.G.]

I recommend using the collector-base of an NPN silicon transistor for the LOW leakage protection diode, not the HIGH leakage biasing diode.

What you need to look at on the datasheet is the maximum Icbo - the collector current with the base grounded and the emitter open. That's what the junction actually leaks. Iceo, the collector current with the emitter grounded and the base open is Hfe times the Icbo.

I hadn't thought about it until this moment, but shoot, I suspect that using one 2N3904 from collector to emitter, cutting off the base lead, would work for the high leakage diode and another with the emitter lead cut off would work for the low leakage diode. The upper one will conduct roughly the same current into the base region, but what comes out the emitter is Iceo, Hfe times that Icbo, so it's maybe 10 to 100 times as much as Icbo. Then the lower diode is only conducting Icbo, no multiplication.

The only fly in this ointment is that Hfe is not well defined at these low currents, and so it's hard to tell what current you actually get for the biasing diode, and it may vary with temperature. Varying lower within reason is not too bad, but varying too high could lead you back into the lower end of click and pop stuff that the diode leakage stuff got you away from. I'll have to think that one over.

[bioroids]

Hi!

I recommend using the collector-base of an NPN silicon transistor for the LOW leakage protection diode, not the HIGH leakage biasing diode. 

That's why I couldn't find it... 

The only datasheet (I have) that mentions Icbo is for the 2N5088 (they call it collector cut-off current too) and it's a max 50nA. Multiplied by the minimum Hfe (300) it gives 15uA, I think this is way too much.

In fact, if you have 50nA and the DC resistor at the effect output is 100K, then you develop 5mV at that point. This can cause a little pop I think, so I wouldn't go for anything higer than 50nA. Am I right?

On the other hand, how low can this current be and still action the switch correctly?

Luck and thanks for your help!

Miguel

[bioroids]

I just thought of this: why not use the leakage current of an electrolytic capacitor?

http://www.analog.com/library/analogDialogue/Anniversary/21.html

According to this is in the 5-20 nA per uF (it should vary with the voltage rating also I think).

I dont know how reliable can this be (will check some capacitor datasheets if I can find them) and if it varies with temperature also.

If this works maybe you can get away using the leakage of the output capacitor of the effect ??  That would probably cause some pops

Luck

Miguel

[bioroids]

Of course this carries the problem of charging the capacitor first...

I just measured about 1nA leakage from a 1uf 50v electro, and about 10nA with a 47uf  16v.
With a lower quality 4,7uf x50v I measured about 100nA of leakage! But it took a while to stabilize on this values.

Luck

Miguel