EMRFD Message Archive 15060

Message Date From Subject
15060 2018-09-08 16:43:51 Tayloe, Dan (Noki... Low leakage NPN?
I am trying to work with very low power DC signals.  I would like to switch a 1 uA pull up resistor with a NPN transistor, but normal small signal types (like a 2n4401) are too "leaky".  About the best I can reliably switch is 5 uA of pull up current when driven by a 0.08 uA DC high voltage feedback signal.

The rest of the 400v switching supply for a gieger counter and everything else (Muller tube, LEDs and all) uses 6.5 uA average current drain at 6v.  I would rather not blow half my power consumption on a 1 M ohm pull up resistor to 5v.  If I could get reliable on/off to 5 or 10 M ohm (0.5 to 1 uA), I would be happier.

Any suggestions for a low leakage NPN?

- Dan

Sent from my Galaxy Tab® A
15062 2018-09-08 20:20:49 Russell Shaw Re: Low leakage NPN?
15063 2018-09-08 22:25:24 Russell Shaw Re: Low leakage NPN?
15068 2018-09-10 20:44:13 arfghans Re: Low leakage NPN?
Tried a small MOSFET?

Gary NA6O
15069 2018-09-11 06:24:18 Tayloe, Dan (Noki... Re: Low leakage NPN?
I tried a 2n7000 and found the gate leakage increased the power consumption on the feedback from the high voltage side a lot.  I normally think of the gate of a MOSFET as being an open circuit, but I went from one high voltage "top off" pulse every 2 seconds to two a second.  The 2N4401 input base current is apparently consuming around 0.085 uA from the high voltage zener feedback string,  so the 2N7000 gate must be taking about 0.35 uA.

It seems when we get down to the 1 uA level, ***everything*** leaks.  I think I have found a transistor from Mouser with 50 nA of Ice leakage at 0v Vbe.   I will order a few this weekend and see what happens.

- Dan, N7VE

Sent from my Galaxy Tab® A


-------- Original message --------
15070 2018-09-11 09:12:19 Russell Shaw Re: Low leakage NPN?
15071 2018-09-11 09:20:40 E-P Mänd VS: [emrfd] Re: Low leakage NPN?
Hello all, have you tried an old electrometer FET, 2N4117 type = MMBF4117
from Fairchild; Farnell etc. sells them. Idss (at zero Ugs) is only 42
microamperes typically, but most important is gate current, less than 1 pA
(pico amperes!) at 25 degrees Celsius. 2N4119 /MMBF4119 can sink max. abt.
0.5 mA (Idss) if you get it. Gate characteristics are same. Mostly used as
ultra low leakage diode, max 50 mA is allowed. – Kindly: EePee, OH2NFI.



Lähettäjä: emrfd@yahoogroups.com [mailto:emrfd@yahoogroups.com]
Lähetetty: 11. syyskuuta 2018 16:24
Vastaanottaja: emrfd@yahoogroups.com
Aihe: Re: [emrfd] Re: Low leakage NPN?





I tried a 2n7000 and found the gate leakage increased the power consumption
on the feedback from the high voltage side a lot. I normally think of the
gate of a MOSFET as being an open circuit, but I went from one high voltage
"top off" pulse every 2 seconds to two a second. The 2N4401 input base
current is apparently consuming around 0.085 uA from the high voltage zener
feedback string, so the 2N7000 gate must be taking about 0.35 uA.



It seems when we get down to the 1 uA level, ***everything*** leaks. I
think I have found a transistor from Mouser with 50 nA of Ice leakage at 0v
Vbe. I will order a few this weekend and see what happens.



- Dan, N7VE



Sent from my Galaxy Tab® A





-------- Original message --------

15072 2018-09-11 10:41:20 Mike Dinolfo Re: Low leakage NPN?
Dan:

Is it possible that you could share a schematic for your (working, or
otherwise) geiger-counter power supply? I've been searching for a
high-efficiency DC(battery)-to-400VDC homebrew power converter that I
can build to power up a SBM-20 geiger tube.

And if anyone wants to respond to me off-list, please feel free to do so.

Thanks!

Mike Dinolfo N4MWP

15073 2018-09-11 10:59:40 Tayloe, Dan (Noki... Re: Low leakage NPN?
Yes, I have found out the hard way about the zeners being low at low current. I have four 200v and one 56v Zener (and 22 M ohm, but I don't think it is doing much) in series my HV feedback string to get regulation a bit above 400v. I.e., that is 856v of Zener diode to regulate 430v. At half the effective voltage, it is unclear how much is Zener action and how much is leakage.

I think I would likely get closer to the marked voltage level if I had something smaller like 200 mW types. It is not a big deal since they are not harming the power consumption. However, I do see some 200 mW surface mount types from Mouser that I might try out.

The 15 mH charging inductor is charging up to 90 mA (~250 usec) for one longer pulse and up to 18 mA (50 msec) for a shorter pulse with each "top off" and each charge lasts ~ 2 sec. That is a 3v increase on a 0.047 uF 430v power supply cap (E=1/2*C*V*V). These two pulses are a total energy (E=1/2*L*i*i) of 0.0618 mW-sec (i.e., milli-Joules). If that energy gets consumed over 2 seconds at 430v, it is an average current drain ((0.062e-3)/430v)/2 sec = 0.073 uA. I am optimistically assuming this is 100% converted to current through the feedback Zener string into the base of the feedback transistor with the 1M ohm pull up. I.e., I am assuming no 6v to 430v conversion losses, no leakage in the 0.047uF 1000v ceramic capacitor or in the two series 1N4007 1000v high voltage "catch" diodes.

It is really hard working on a circuit that I cannot measure directly, nor use my 'scope to watch what is happening. I am kind of stuck analyzing this via theory and an excel spreadsheet.

I was kind of disappointed with the 2N7000 results, but as I said, without the ability to actually measure signals directly, it is difficult to understand what exactly is going on. I was deriving the 2N7000 gate drive from 30M ohm to ground from my HV Zener string feedback (0.07 uA* 30M ohm = 2.1v gate bias). I have likely overestimated the 0.07 uA feedback current available and thus used a terminating resistor that might be 4x too small to generate a sufficient gate voltage for my 2N7000 MOSFET without needing 4x more current from the HV Zener feedback string. That implies that I would need to get up to 120M ohm to "break even" with the 2N4401 from a HV power consumption point of view (since I dropped from 2 sec "top off" to 2 per sec "top off" pulses).

A 120M resistor seemed a bit excessive. The largest I have
15074 2018-09-11 12:56:19 Tayloe, Dan (Noki... Re: Low leakage NPN?

That is quite some micro power FET.  I see mouser has both 4117 and 4118 parts.  I have never heard of an N channel JFET with an Idss way below 1 mA before.

 

  • Dan, N7VE

 

15075 2018-09-11 17:45:39 jim Re: Low leakage NPN?
Does that 10 nA refer to leakage gate to source?   He was pulsing the things ...Gotta charge them Ciss/Coss etc

Jim


15076 2018-09-12 03:11:43 Russell Shaw Re: Low leakage NPN?
15077 2018-09-12 03:13:59 Russell Shaw Re: Low leakage NPN?
15078 2018-09-12 09:22:44 Tayloe, Dan (Noki... Re: Low leakage NPN?

Thanks for the tip.  Half the battle is finding out what you don’t know.  I did not realize it was possible to have a comparator so very low in power, complete with push-pull outputs (no pull up resistor!).

 

I see Mouser carries the TI TLV3691 comparator which runs on only 0.075 uA.  That will save a lot of power.

 

As far as a resistive feedback, the zeners work fine and are cheap at roughly $0.26 each.  Right now a 1G resistive divider would be 6x the average load that I am seeing right now

15079 2018-09-13 23:24:20 Russell Shaw Re: Low leakage NPN?
15080 2018-09-14 06:48:38 Mike Dinolfo Re: Low leakage NPN?
Dan:

One other reason why you might want to reconsider Russel Shaw's
suggestion is because using zener diodes in this manner (at current
levels far below their usual application) might result in considerable
variation in zener voltage drop vs. temperature. Whereas resistors
should (mostly) be immune to temperature variations that are likely to
be encountered. I am very weak on semiconductor theory, but I suspect
that the voltage drop exhibited by a zener at current levels many orders
of magnitude less than typical in-service current (based on rated watts
/ rated voltage) does vary considerably not only as a function of the
applied current, but also as a function of temperature.

Unfortunately, you are dealing with current levels far lower than those
at which the zeners are designed to be operated at. Datasheets for
zener diodes typically indicate that the operating voltage of the zener
is within the zener manufacturer's design tolerance only for currents
that are at 75% of the rated current (rated zener wattage divided by
rated zener voltage). So a 400mW 100V zener is voltage tested at 3000
uA. And although the zener voltage drop does drop only very slightly
for maybe an order of magnitude (0.1), or two orders of magnitude
(0.01), or maybe even three orders of magnitude (0.001) of rated
current, I suspect that it gets pretty "wobbly" at the sub-microamp
levels you are working with.

One way to check this out would be to measure the output voltage of the
operating circuit at the same time that you are changing the ambient
temperature (maybe with a hair blower?) of the zeners. If the output
voltage varies considerably with ambient temperature then it might be
significant in your application. If you lack a method to measure the
output voltage without excessively loading the circuit, you could try
building a voltage divide probe to work with a VTVM- a 1G resistor, for
example, used with a VTVM (typically 10 or 11 Meg input resistance)
would drop the 400V PS voltage by a factor of about 100, to about 4
volts. The 1G resistors (in 1/4 watt rating) are available from Mouser
for $0.66 (PN 279-RGP0207CHK1G0). I've done this by mounting the
resistor inside a ball point pen sleeve, although as others have stated,
you need to use something like isopropyl alcohol to wipe off any
fingerprint oils during assembly. The 1G resistor is also usable with
the common inexpensive DMM's, which are typically 1M input impedance
(giving 400 mV from a 400V source).

Your work with this circuitry is very interesting- keep us informed as
to how it's going!

Mike Dinolfo N4MWP

15081 2018-09-14 12:04:47 Tayloe, Dan (Noki... Re: Low leakage NPN?

I thought about it a bit yesterday and I agree that it is good advice that the temperature effects could be bad.  I already see that it takes ~ 10 seconds for the zeners to “warm up” and regulate the high voltage up to the point where the SBM-20 Geiger tube starts giving background detections pulses.  So there is already a fair amount zener voltage “start up” instability given the tiny fraction of current trickling through the zeners.

 

This early this morning I ordered some parts from Mouser today including a pair of $3.50 5G ohm resistors and few extremely low power comparators (TLV3691) to experiment with.

 

Getting rid of my 6v 1M pull up resistor and replacing it with a “pico-power” comparator and G ohm voltage divider ought to save me 45% of my current power consumption.  A lot of this is also due to the fact that the 8 pin PIC processor I am using for control uses only 0.4 uA in “sleep” mode waiting seconds for the voltage to sag too much. When I am looking at a pico-powered comparator and its bias circuit drawing ~ 0.3 uA, having an entire uP draw 0.4 uA in sleep mode in comparison also strikes me as amazing in and of itself.

 

It is quite the interesting time to live in.

 

By the way, my wife Vicki is already quite disturbed to learn that a very nice “reproduction” greenish-yellow Fentonware glass vase that she has appears to be *very* radioactive.  She has a few pieces of glassware that are moderately elevated, but I would not want that vase

15082 2018-09-14 13:18:38 John Levreault Re: Low leakage NPN?
When I have needed to sense a voltage or signal with minimum power, I like to
use a low-current comparator. With low-power being a design trend these days,
there are quite a few comparators available now that consume well under 1ua with
bias currents in the pA range, with push-pull outputs. The reason I mention this
last point is that push-pull means you don't need to drive a pull-up resistor,
which can sometimes be a significant design advantage. So now your 1-10G
resistor divider makes design sense. The down-side is, of course, the need to
power the darn thing, but I don't know enough about what the rest of your system
looks like to know how much of a problem that might be. I have generally found
that stealing a few hundred nA is always possible.

The tradeoff is, of course, speed, i.e. limited AC bandwidth and response time
possibly in the tens of microseconds.

FWIW.

de John NB1I


15083 2018-09-15 07:52:20 Tayloe, Dan (Noki... Re: Low leakage NPN?
Response time is not an issue.  When the processor gets feedback the voltage is too low, it charges up the boost inductor (15 mH) for 250 usec to get a 90 mA current on it.  After turning it off, it waits for another 250 usec to see if the feedback shows the voltage is too high now.  If not, it repeats the loop until it is.  One pulse increases the high voltage by about 3v.  The 250 usec wait gives plenty of time even for a very pokey comparator.

- Dan, N7VE


15084 2018-09-15 16:10:56 Will Kimber Re: Low leakage NPN?

Hi Dan,

as a teen I had luminous wrist watch.  It sent counters to full.  Still got it but last time I tried it many years ago it was going down.


Cheers,

Will


15085 2018-09-15 17:06:44 Tayloe, Dan (Noki... Re: Low leakage NPN?

My note from yesterday did not seem to make it out for some reason. It appears my wife has provided a nice radioactive source for my counter.

 

Watch dials are for sale on “that auction site” as a test source for Geiger counters.

 

---------

 

I thought about it a bit yesterday and I agree that it is good advice that the temperature effects could be bad.  I already see that it takes ~ 10 seconds for the zeners to “warm up” and regulate the high voltage up to the point where the SBM-20 Geiger tube starts giving background detections pulses.  So there is already a fair amount zener voltage “start up” instability given the tiny fraction of current trickling through the zeners.

 

This early this morning I ordered some parts from Mouser today including a pair of $3.50 5G ohm resistors and few extremely low power comparators (TLV3691) to experiment with.

 

Getting rid of my 6v 1M pull up resistor and replacing it with a “pico-power” comparator and G ohm voltage divider ought to save me 45% of my current power consumption.  A lot of this is also due to the fact that the 8 pin PIC processor I am using for control uses only 0.4 uA in “sleep” mode waiting seconds for the voltage to sag too much. When I am looking at a pico-powered comparator and its bias circuit drawing ~ 0.3 uA, having an entire uP draw 0.4 uA in sleep mode in comparison also strikes me as amazing in and of itself.

 

It is quite the interesting time to live in.

 

By the way, my wife Vicki is already quite disturbed to learn that a very nice “reproduction” greenish-yellow Fentonware glass vase that she has appears to be *very* radioactive.  She has a few pieces of glassware that are moderately elevated, but I would not want that vase on my bed stand at night!  She mentioned that as a teen, it was curious to her that fresh cut flowers in that vase never seemed to last very long.

 

---------

 

It is not really that big of a step from using a simple boost circuit like this to generate 400v at about 0.15 mA max for a Geiger tube to some that would power a 400v QRP tube final.  It seems a lot cheaper than buying a new 450v, 50 mA transformer.   It makes me think of that old 6T9 transmitter I once build out of an ARRL publicati