DIY giga-ohm resistors???

Not much, that's the appeal. You've got to worry about their photocurrent, hence the need to keep 'em dark.

Good LEDs have amazingly low forward current at low forward bias. IIRC there's a nice graph out there in one of Pease's articles.

Mike could use a couple to clamp Vgs to the rails, or 2x2 in series, back-to-back.

I think UV LEDs are leakier. I'm not sure which LED materials / colors are best--haven't measured recently.

Another junk-box low-leakage item he could use to protect the gate: bipolar transistor junctions. Much better than, e.g., 1n4148.

ent, hence the need to keep 'em dark.

there's a nice graph out there in one of Pease's articles.

-to-back.

are best--haven't measured recently.

ar transistor junctions. =A0Much better than, e.g., 1n4148.

Hi James, Knowing almost nothing about the reverse leakage current of leds, I'd guess that the higher the energy of the photon the lower the leakage... a kT vs. E-gap thing.

Leaky uV diodes don't fit this model.... which might be curious.

George H.

rrent, hence the need to keep 'em dark.

RC there's a nice graph out there in one of Pease's articles.

ck-to-back.

s?

rs are best--haven't measured recently.

olar transistor junctions. =A0Much better than, e.g., 1n4148.

Oops, and the low forward voltage current (re-reading...)

current, hence the need to keep 'em dark.

IIRC there's a nice graph out there in one of Pease's articles.

back-to-back.

oms?

lors are best--haven't measured recently.

ipolar transistor junctions. =A0Much better than, e.g., 1n4148.

I figured the same, but Phil said something a while back about excess latti= ce defects in certain LEDs (blue GaN?) that gave me pause.

I used green GaN as low-leakage clamps, mostly for the high Vf. They were = very good, but I didn't check leakage down to electrometer current levels.

--=20 Cheers, James Arthur

tocurrent, hence the need to keep 'em dark.

=A0IIRC there's a nice graph out there in one of Pease's articles.

, back-to-back.

rooms?

colors are best--haven't measured recently.

bipolar transistor junctions. =A0Much better than, e.g., 1n4148.

of

tice defects in certain LEDs (blue GaN?) that gave me pause.

Yeah I think that's right... I heard of this physics lab where you look at the 'turn on' voltage of LEDs, vs the wavelength... and things work out 'as expected' till you get to the blues...

Where it doesn't... I've not done the measurements.

ere very good, but I didn't check leakage down to electrometer current leve= ls.

Hmm, that's interesting. Are the blues 'worse' than the greens? Or do they just not fit 'expected value' as well?

George H.

photocurrent, hence the need to keep 'em dark.

there's a nice graph out there in one of Pease's articles.

back-to-back.

are best--haven't measured recently.

bipolar transistor junctions. Much better than, e.g., 1n4148.

defects in certain LEDs (blue GaN?) that gave me pause.

very good, but I didn't check leakage down to electrometer current levels.

Normal red and green LEDs are very good. I haven't measured any just lately, but a dozen or so years ago when I was doing it for real, I measured several jellybean red AlGaAs display LEDs at less than 50 fA from -5V to +0.5V. I think they were from Chicago Miniature Lamp--I got a couple of reels, and haven't run out yet.

Cheers

Phil Hobbs

hotocurrent, hence the need to keep 'em dark.

=A0IIRC there's a nice graph out there in one of Pease's articles.

es, back-to-back.

shrooms?

/ colors are best--haven't measured recently.

e: bipolar transistor junctions. =A0Much better than, e.g., 1n4148.

nt of

the

attice defects in certain LEDs (blue GaN?) that gave me pause.

were very good, but I didn't check leakage down to electrometer current le= vels.

I measured a number of eBay whites (=3Dblue GaN + phosphor) that were very = leaky in the forward direction (manifests as a high threshold). They might= 've been static damaged, or who knows what, but it was consistent. I assum= e they would've leaked backwards too. Dunno if that's common, but there ar= e behaviors in prime parts (low-current flickering) that hint it could be.

So, caveat emptor.

--=20 Cheers, James Arthur

photocurrent, hence the need to keep 'em dark.

IIRC there's a nice graph out there in one of Pease's articles.

back-to-back.

colors are best--haven't measured recently.

bipolar transistor junctions. Much better than, e.g., 1n4148.

lattice defects in certain LEDs (blue GaN?) that gave me pause.

very good, but I didn't check leakage down to electrometer current levels.

Here's some old data from Win...

------------------------ > Here are a set of I-V measurements for your edification.

• indicates affected by light pickup. I put some curves on the web,
  formatting link

Note how ordinary glass silicon diodes look like 10M resistors below +/-50mV. GROSS! but often acceptable for virtual-ground node protection, e.g. if the feedback resistor is 1M or less, etc.

------------------------

hence the need to keep 'em dark.

there's a nice graph out there in one of Pease's articles.

back-to-back.

best--haven't measured recently.

transistor junctions. Much better than, e.g., 1n4148.

Be advised the last two lines were stated in jest..

hence the need to keep 'em dark.

there's a nice graph out there in one of Pease's articles.

back-to-back.

best--haven't measured recently.

transistor junctions. Much better than, e.g., 1n4148.

Not at all. BFT25A B-C junctions are competitive with the best low-leakage silicon diodes anywhere.

Cheers

Phil Hobbs

Err..i was referring to my comments stated above.

ooms?

Yeah, I knew, but there are "black light LEDs," so it seemed worth commenting on.

-- Cheers, James Arthur

Dark Emitting Axial Diodes?

hence the need to keep 'em dark.

there's a nice graph out there in one of Pease's articles.

back-to-back.

best--haven't measured recently.

transistor junctions. Much better than, e.g., 1n4148.

Hey, Phil, do you know anything about the reverse beta of a BFT25?

Imagine a linear ramp made from a current source and a cap; say 10 mA and 20 pF, for a few ns to 2.5 volts. Now I want to switch in bigger caps for slower ranges, like 200 pF and 2 nF. I figured I could use BFT25s to do that. But when I discharge the ramp, current flows backwards into the BFT25s.

isrc | v | | diode +----------+---------+---ak----reset | | | | | | 20pF 200p 2n | | | | | | | | | gnd c c >--b >--b e e | | gnd gnd

If there's any decent amount of reverse beta, I don't need a huge amount of base current to get a decent reset time.

(I'll probably add a weak pullup to the transistor collectors to bias them up when they're off.)

I have some very rough data that suggests a reverse beta around 5 maybe, but it's not solid.

Good idea that India ink. Stir some into some plastercine to put in a pair of wires for a variable resistor. Stir into glue, wires at each end, let dry and measure. Ken

hence the need to keep 'em dark.

there's a nice graph out there in one of Pease's articles.

back-to-back.

best--haven't measured recently.

transistor junctions. Much better than, e.g., 1n4148.

Clamp diode of low pF from C to E on each one? You could remove the

Jamie

On a sunny day (Sun, 04 Mar 2012 09:54:50 -0500) it happened Jamie wrote in :

hence the need to keep 'em dark.

there's a nice graph out there in one of Pease's

back-to-back.

best--haven't measured recently.

transistor junctions. Much better than, e.g.,

Is this not normaly done with PIN diodes?

photocurrent, hence the need to keep 'em dark.

there's a nice graph out there in one of Pease's articles.

back-to-back.

are best--haven't measured recently.

bipolar transistor junctions. Much better than, e.g., 1n4148.

According to the spice model, it's about 16.

Cheers

Phil Hobbs

hence the need to keep 'em dark.

there's a nice graph out there in one of Pease's articles.

back-to-back.

best--haven't measured recently.

transistor junctions. Much better than, e.g., 1n4148.

Yes, that would absorb most of the discharge current. Tiny schottkies wouldn't add much capacitance. We have some nice diodes in stock, SMS7621, that are only about 0.25 pF.

You could remove the

The diode capacitances are nonlinear on voltage, so that would bend my nice linear ramp. That's the reason to pull up the "off" transistor collectors, to reduce the nonlinearity of their c-e capacitance.