Is a mosfet floating source ok?

I should have looked at the data sheet:

---------(quote)-------------

Active-Low Interrupt or Square-Wave Output. This open-drain pin requires an external pullup resistor connected to a supply at 5.5V or less. This multifunction pin is determined by the state of the INTCN bit in the Control Register (0Eh). When INTCN is set to logic 0, this pin outputs a square wave and its frequency is determined by RS2 and RS1 bits. When INTCN is set to logic 1, then a match between the timekeeping registers and either of the alarm registers activates the INT/SQW pin (if the alarm is enabled). Because the INTCN bit is set to logic 1 when power is first applied, the pin defaults to an interrupt output with alarms disabled. The pullup voltage can be up to 5.5V, regardless of the voltage on VCC. If not used, this pin can be left unconnected.

--------(end quote)-----------

So, my idea can't possibly work. Possibly a zener diode instead of the resistor to keep the max voltage on pin 3 under the 5.5VDC max (irregardless of Vcc).

At this point I'm going back to stuff I do know...

John ;-#)#

I should have looked at the data sheet (RTFM):

---------(quote)-------------

Active-Low Interrupt or Square-Wave Output. This open-drain pin requires an external pullup resistor connected to a supply at 5.5V or less. This multifunction pin is determined by the state of the INTCN bit in the Control Register (0Eh). When INTCN is set to logic 0, this pin outputs a square wave and its frequency is determined by RS2 and RS1 bits. When INTCN is set to logic 1, then a match between the timekeeping registers and either of the alarm registers activates the INT/SQW pin (if the alarm is enabled). Because the INTCN bit is set to logic 1 when power is first applied, the pin defaults to an interrupt output with alarms disabled. The pullup voltage can be up to 5.5V, regardless of the voltage on VCC. If not used, this pin can be left unconnected.

--------(end quote)-----------

So, my idea can't possibly work. Possibly a zener diode instead of the resistor to keep the max voltage on pin 3 under the 5.5VDC max (irregardless of Vcc).

At this point I'm going back to stuff I do know...

John ;-#)#

It may well tolerate 10 or 12 volts, enough to drive the pfet without an additional n-fet. If your test pullup is current limited to below

But a small n-fet costs a few cents.

Maxim tended to EOL stuff without warning, and sometimes shipped flakey parts. Now that ADI owns Maxim, I hope that will change. TI seems to be best about keeping stuff available.

Them and Microchip.

Dirty old Renesas just EOLed my fave CFA, the EL5167. The 5166 is still available but doesn't come in SC-70. It's a pretty cool part: >1GHz,

I use it in my time stretchers---one of these guys running three pHEMT sampling gates with 8-pF hold caps hung right off the inputs of a

Cheers

Phil Hobbs

NXP is pretty bad.

Why don't people bake a box full of wafers and raise the price? Causing major distress to customers can't be good for buisness. Maybe they don't care for small or mid-size users.

Yeah that's what the manual says. and it doesn't seem to say why

If it said that there was an internal zener or similar on that pin then the circuit would be fine.

the DS3231 datasheet seems entirely silent on protection of the pins, but doesn't seem to say that pin must (or even should) be connected. Which is a hint that that there is some protection on that pin.

Without it you risk exceeding abs-max Vgs on the "logic mosfet". For parts that work well at 3V Vgs, Vgs_max is often 7V or similar.

The 470K resistor is cheap insurance against a wrong guess, probably not needed, but to my thuinking not provably not needed.

put 5.8V on the ~INT/SQW pin and measure the current flowing into it, if it's measurable then the 470K is not needed.

Dunno. Maybe they can't be bothered--lots of local micro-optimizations go on in large organizations.

Rochester seems to be upping its game, which is a hopeful sign.

Cheers

Phil Hobbs

If the max voltage is not due to protection diodes turning on and possible latchup, then the concern is just plain punching through the oxide gates mo st likely. It takes very, very little current for that to happen, so a res istor can't protect it by limiting the current.

I would say a 3V Vgsth (max) would be adequate given you only need to pass a few microamps. There are manufacturing tolerances involved, but this num ber also factors in temperature, so if it doesn't get too cold you should b e ok. To be certain you could bump up R1 to 470K so it only requires 25 uA to pull it to ground.

The Maxim thing is about their propensity to announce products and even pro duce data sheets right up to the date of production before they cancel the part. Otherwise they seem to make ok parts... well, as good as anyone.

I picked a National voltage regulator one time (now TI) for a test fixture. I read every detail in the data sheet. When it didn't work I contacted N ational about it (or was it already TI then?) and they pointed me to the we b page where it says it was only good for outputs of 7V and higher. What?? ? The web page supersedes the data sheet??? That messed me up a bit but I ended up getting one of those 78xx replacements that is a switching regul ator. Works well with no measurable noise in the analog section. But then the UUT was designed to work with a noisy supply, on board ferrite/cap fil ter, 5>3.3 regulator and good power planes.

Of course they don't care for small and mid size users. There's no money i n selling inexpensive parts in small volumes. My main customer has talked about dropping a line of equipment with a market in the $10's of millions b ecause it isn't big enough to warrant attention from the company heads. Bu t the people on the street know having that capability lets them bid on con tracts that are much larger, where by being able to bid this piece of the c ontracts as well they stand a better chance of winning the whole enchilada. They are telling me to expect a million or two in business on one contrac t over a five year period. That's just my little slice of the pie.

Selling to Highland isn't even on their radar. The geese flying overhead s how a bigger blip.

ss

There's no reason to include a zener when a limit in the manual is adequat e protection as far as the maker is concerned. Even if there is, the circ uit still won't work properly as the resistors form a voltage divider and r educe the range of Vgs swing on the pFET. You get a choice of turning it o n properly or turning it off properly, but it would be hard to manage both.

I thought they gave a max voltage rating? Why does an unconnected pin requ ire protection? It's the drain of a FET, not the gate. A drain should no t be terribly sensitive to stray voltages.

I'm lost here. What is the logic mosfet?

I don't know why anyone would consider saving a two cent transistor. I'm n ot following.

I did a bit more experimenting with the DS3231.

Using a multiturn pot I applied a rising voltage to the INT pin, with my meter installed as an ammeter, to see if I could detect the presence of a zener diode to ground. I gradually increased the voltage from 5V, past the

Then I went back to the original 12V circuit, without R2, and using an NDP6020P as the P-channel mosfet, and a BS170 as the N-channel. I measured the voltage on the INT pin using the nulling method, and got about 5.2V. I assume that's leakage through the BS170. I don't know why it isn't 12V, except maybe there's also some leakage out through the INT pin open drain.

Then I installed a 5.1V zener diode from INT to ground, and measured the current flowing through it. It just barely registered on my meter, so it was some small fraction of a microamp. A 5.6V zener had no current flow.

So the leakage confirms there's no need for R2 at all, and in fact it would in effect be trying to charge the primary lithium coin cell, which we probably don't want.

It seems the safe thing to do is install a 5.6V zener, which matches the datasheet limit, and which would probably never come into play. But I think it's probably going to be ok without the zener.

ld

Not sure what circuit you are describing, but I think you mean to drop the nFET and R2 from your original drawing and add a zener from INT- to ground as protection. But this WILL draw current through R1 and limit the pFET tu rning off which will also draw current through the pFET.

Or are you talking about adding the zener diode to the original design and omitting R2? In that case I thought you had determined you need neither th e zener nor R2?

Zeners/protection diodes to rails are to protect gates. Since INT is an output presumbably there is no gate connected to it, so no need for protection Zener. Once voltage is high enough current trough output transistor will start growing and at some moment will destroy it. In IC there are also other possible failure modes.

Surly there is leakage trough INT.

Hmm, I tested my Zeners. At 1.5V applied voltage 5.1V Zener leaks about

Actually, Zener leakage is likely to be much higher than leakage on INT pin, so it will lower voltage on INT. OTOH I would expect the circuit to work fine without Zener. To put things in some perspective, let me add that stray capacitance (say via DMM probe) can easily insert to the circuit 100nA from the mains. CMOS IC (with their ESD protection circuits) seem to survive OK from such treatment.

Most ICs have ESD protection on all pins, even ones without internal diodes to Vcc. The makers would get into a lot trouble if they didn't. You can generally pull up a pin, through a big resistor, and see when it starts to voltage limit, without damaging anything. It's common for parts to start pulling current at maybe 2x rated abs max.

No.

Yes.

I had come to the conclusion that R2 almost certainly wasn't needed to pull of the source up to the gate and keep the mosfet off - both logically, and from the behavior of my test circuit.

But that left the question of whether leakage into the source might raise the voltage there above the 5.5V limit. It turns out it doesn't, but it's close at 5.2V. But that might change at a rail higher than 12v, or if using a mosfet more leaky than the BS170. I think a 5.6V zener would be a very safe addition to make absolutely sure the source doesn't go above 5.5V, but I think in most cases it isn't really necessary.

The DS3231 datasheet says leakage through INT is typically zero, with a maximum of 1uA.

The only thing I have to measure low currents is an analog meter with a 50uA scale. I can barely detect maybe around

I did that test, hoping to see at what voltage current starts to flow through INT. But I stopped at 7V. Maybe if I get another RTC module, I'll test at higher voltages.

Here is the fallacy of what you are saying. The tiny leakage current of th e nFET will not be enough to do any damage. The zener only works if it dra ws current. If it draws current the transistor must be on at least some. If the nFET is off, there won't be enough current to activate the zener at anywhere near it's rated voltage. So the zener isn't going to protect agai nst anything real. It will be an elephant charm. Remember, the nFET has s ome tiny leakage, but the FET in your chip also has some tiny leakage.

I think I'd add the R2 resistor before I'd add the zener. A few nA won't h urt the coin cell and will help counter the discharge rate. R2 can be pret ty high value to minimize the drainage when active.

Peabody wrote:

See the LMC660 (MOQ 1) at Digi-Key

Long post on wiring low current op amps

Message 5 in thread

From: Paul Grohe ( snipped-for-privacy@galaxy.nsc.com >

Subject: Re: sub-picoampere op amps

Newsgroups: sci.electronics.design

Date: 1999/01/25

On Mon, 25 Jan 1999 00:11:53 +0100, in the newsgroup sci.electronics.design, "Luis de Funes" from TIN thoughtfully posted:

Almost all our "LMC" series should fit your requirements. The LMC660/2 and LMC6482/2 series would be the best choices for low cost.

On the bench, the LMC660/2 or LMC6482/4 series is typically hovering around 2-3fA. Just remember that the bias current approximately "doubles" for every 10'C. LMC660.pdf Quad LMC662.pdf Dual LMC6482.pdf Dual RRIO

The LMC6001 is hand tested and guaranteed to be less than 25fA - but you pay extra for that.

Your best bet for a "low cost" op-amp is the LMC662 or LMC660.

The only difference between the "A" and "C" grades is the offset spec. The leakages should be the same.

For lowest leakage currents, use the dual version and use the second channel ("right side", pins 5,6,7) for the critical circuit. Lowest leakages occur near the center of the input common mode range.

If you are looking for a commercial unit - the "King" of electrometers is the Keithley 642. We use the 642 for bench measuring the bias currents of our CMOS op-amps. With an elaborately guarded/shielded fixture - we can get resolutions down to 0.05fA. keith642.pdf

We also use the Keithley 617 for less critical or automated measurements (0.1fA resolution).

Hope this helps.

Cheers, Paul Grohe

--------------------------------------------------------------- Paul Grohe National Semiconductor Corp.

Assoc. Analog Apps Engineer 2900 Semiconductor Drive Standard Analog Apps Group Mail Stop C2693 mailto: snipped-for-privacy@galaxy.nsc.com Santa Clara, CA. 95052-8090 USA Usenet Disclaimer: Any opinions expressed are mine, not NSC's

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

Message 6 in thread

From: Gerry Schneider ( not snipped-for-privacy@sympatico.ca >

Subject: Re: sub-picoampere op amps

Newsgroups: sci.electronics.design

Date: 1999/01/25

Paul Grohe wrote:

mailto: snipped-for-privacy@galaxy.nsc.com Santa Clara, CA. 95052-8090 USA

Why's that, Paul? Is this standard for all NatSem parts?

Gerry @ Change "not_here" to "lsb" ##_/_\__[( The Hi-Tech Homestead

Message 7 in thread

From: Paul Grohe ( snipped-for-privacy@galaxy.nsc.com >

Subject: Re: sub-picoampere op amps

Newsgroups: sci.electronics.design

Date: 1999/01/27

On Mon, 25 Jan 1999 11:17:18 -0800, in the newsgroup sci.electronics.design, Gerry Schneider from Bell Solutions thoughtfully posted:

Nope. It is due to the layout of the pins and not the properties of devices. The singles, duals and quads are identical to each other. The same is true for the "other guys" parts, too.

Lemme' explain:

Most "low leakage" circuits apply the critical signal path to the non-inverting input.

Note the layout of the typical dual:

A Out V+ A -In B Out A +In B -In V- B +in

On channel "A" - the non-inverting input is located right next to the negative supply pin. It does not take much "Gook(tm)" between the adjacent pins to create some healthy leakage paths.

A little household dust, combined with some summertime moisture, creates a nice leakage path between the pins that can easily exceed the leakage of the op-amp itself.

However, channel B's non-inverting input (pin 5) is on the other side of the package - well away from the power pin. You will need to get a lot more "gook" on, and across, the package to create an equivalent leakage path.

It is also easier to guard and shield pin 5 than it is to guard and shield pin 3.

On a quad, all the non-inverting pins are located around power supply pins, making guarding tough. Even a single has it's non-inverting input next to V-.

Hope this clears up the misunderstanding. There is no difference between the parts.

Cheers, Paul Grohe

--------------------------------------------------------------- Paul Grohe National Semiconductor Corp.

Assoc. Analog Apps Engineer 2900 Semiconductor Drive Standard Analog Apps Group Mail Stop C2693 mailto: snipped-for-privacy@galaxy.nsc.com Santa Clara, CA. 95052-8090 USA Usenet Disclaimer: Any opinions expressed are mine, not NSC's

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

Message 8 in thread

From: Tom Bruhns ( snipped-for-privacy@lsid.hp.com >

Subject: Re: sub-picoampere op amps

Newsgroups: sci.electronics.design

Date: 1999/01/27

Paul Grohe wrote (concerning which op amp of a dual to use to get really low leakage):

Useful comments, Paul!

I discovered another sort of leakage problem in an op amp that may help someone else out. I needed an amplifier with extremely low input offset voltage as well as low input bias current to amplify the output of a zero-bias RF detector diode, HP HSMS-2850, for a ham radio signal sniffer I was working on. I chose a chopper-stabilized CMOS op amp in an 8-pin plastic DIP package. My layout carefully guarded the input pins. I built it, and it worked great in the environment inside my home: I could detect signals down to below 30uV with it. (The op amp is part of a logarithmic circuit that lets me display several decades of input voltage range on an analog meter movement.) My first "field trial" went OK, too, until, on that misty, rainy day I got out of the car and met up with some other hidden-transmitter hunters. When I opened the box to show them, the meter went to half-scale (a really big signal on the log scale!) and only slowly went down. Huh? Well, it turned out to be leakage from the + supply on pin 7 to the chopper capacitor on pin 8! Even lifting the pin from the board and tying the high-quality cap directly to the pin didn't help much. Finally, washing with aerosol-can flux remover very carefully in that area, drying in a barely-warm oven for a while, and applying a couple coats of clear acrylic lacquer which were baked on seemed to cure the problem. As they say, "ymmv." But at least be aware that maintaining proper operation of a chopper-stabilized op amp may present some leakage current problems you perhaps weren't expecting.

Cheers, Tom

Message 9 in thread

From: Jonathan Bromley ( snipped-for-privacy@brookes.ac.uk >

Subject: Re: sub-picoampere op amps

Newsgroups: sci.electronics.design

Date: 1999/01/27

Paul Grohe wrote:

Wow! That's obvious, sensible, useful info that simply hadn't crossed my mind before. One for ye olde storehouse of handy tips. Many thanks

Jonathan Bromley

Message 10 in thread

From: Gerhard Hoffmann ( snipped-for-privacy@berlin.snafu.de >

Subject: Re: sub-picoampere op amps

Newsgroups: sci.electronics.design

Date: 1999/01/26

Paul Grohe wrote: (2d15b004.htm)

..and there (

from rainy Berlin,

Gerhard

on the air: DK4XP in the air: D-8551

Messages 11-20 from thread

Message 11 in thread

From: Jeff Stout ( snipped-for-privacy@ncon.com >

Subject: Re: sub-picoampere op amps

Newsgroups: sci.electronics.design

Date: 1999/01/26

Flame ionization detectors (FID), photo ionization's detectors (PID) in Gas Chromatography (GC) instruments. The lower the charge you can measure, the more sensitive is your instrument.

Jeff Stout Luis de Funes wrote in message ...

Message 12 in thread

From: Mike Monett ( snipped-for-privacy@csolve.net >

Subject: Re: sub-picoampere op amps

Newsgroups: sci.electronics.design

Date: 1999/01/26

Jeff Stout wrote:

How about ordinary household smoke detectors? They commonly use one microcurie of radiation, which should be around 37,000 ions per second. The trip point seems to be about half the normal current, so they must reliably detect 18,500 ions/sec. How many femptoamps is that?

Best Regards, Michael R. Monett,

Message 13 in thread

From: Jeff Stout ( snipped-for-privacy@ncon.com >

Subject: Re: sub-picoampere op amps

Newsgroups: sci.electronics.design

Date: 1999/01/27

Mike Monett wrote in message ...

Assuming single ionization:

I = (18500 ions/sec) * (1.6 x 10^-19 Coulombs/electron) * (1 electon / 1 ion) = 2.96 x 10^-15 = 3 fA.

Jeff Stout

Message 14 in thread

From: John Woodgate ( snipped-for-privacy@jmwa.demon.co.uk >

Subject: Re: sub-picoampere op amps

Newsgroups: sci.electronics.design

Date: 1999/01/27

Regards, John Woodgate

Message 15 in thread

From: Bret Cannon ( snipped-for-privacy@pnl.gov >

Subject: Re: sub-picoampere op amps

Newsgroups: sci.electronics.design

Date: 1999/01/27

There is no such thing as a microcurie of radiation! The Curie is a unit of radioactive material.

The alpha particles from americium that is used in smoke detectors are emitted with about 5 MeV of energy. This energy ionizes molecules in the air (mostly nitrogen and oxygen) taking about 100 eV per ion pair. Thus the current in a smoke detector with a 1 microcurie source could be as much as 50,000 times higher than Mr. Monett's estimate. Even with ion pair recombination and less than half the alpha particles entering the air, the currents are much larger than 3 fA.

Bret Cannon John Woodgate wrote:

Message 16 in thread

From: James Meyer ( snipped-for-privacy@worldnet.att.net >

Subject: Re: sub-picoampere op amps

Newsgroups: sci.electronics.design

Date: 1999/01/28

On Wed, 27 Jan 1999 17:56:36 -0800, Bret Cannon wrote:

An dat's a fak, Jack.

With the little source that I removed from an old smoke detector held between the leads of my multimeter with a tiny space between the active surface of the source and one lead, I get about one microampere.

Jim

Message 17 in thread

From: John Woodgate ( snipped-for-privacy@jmwa.demon.co.uk >

Subject: Re: sub-picoampere op amps

Newsgroups: sci.electronics.design

Date: 1999/01/28

, Bret Cannon wrote:

But Mr. Cannon thought that his input was so important that he e-mailed me anyway. GRRR!

Regards, John Woodgate

Message 18 in thread

From: Mike Monett ( snipped-for-privacy@csolve.net >

Subject: Re: sub-picoampere op amps

Newsgroups: sci.electronics.design

Date: 1999/01/28

John Woodgate wrote:

Could be an innocent mistake. The "Re:News" and "Re:Both" buttons on Netscape are adjacent. It is easy to hit the wrong one.

Check the email header. If it shows Mozilla, there is a 50-50 chance you got it in error.

Best Regards, Michael R. Monett,

Message 19 in thread

From: Bret Cannon ( snipped-for-privacy@pnl.gov >

Subject: Re: sub-picoampere op amps

Newsgroups: sci.electronics.design

Date: 1999/01/28

I would like to apologize to Mr. Woodgate. I was not paying enough attention to which "Reply Button" I hit and inadvertently emailed Mr. Woodgate. I will try to not repeat this mistake.

Bret Cannon

Mike Monett wrote:

Messages 21-28 from thread

Message 24 in thread

From: Mike Monett ( snipped-for-privacy@csolve.net >

Subject: Re: sub-picoampere op amps

Newsgroups: sci.electronics.design

Date: 1999/01/28

Bret Cannon wrote:

I could not find specs for the chip used, but the Motorola MC145017 Smoke Detector IC seems to have similar pinouts. The input current spec at 40% R.H. is +/- 1.0 pA.

One would expect the operating current level to be much higher than this, so current levels in the microamp region do seem reasonable. I guess the reason for using air insulation is to minimize problems with household dust and grime.

Best Regards, Michael R. Monett,

Message 25 in thread

From: Jonathan Bromley ( snipped-for-privacy@brookes.ac.uk >

Subject: Re: sub-picoampere op amps

Newsgroups: sci.electronics.design

Date: 1999/01/27

Mike Monett wrote:

can we settle for FEMTO please?

well, it's not too hard... assuming single ionisation, each ion carries

It wouldn't register on my Avo 8, anyhow.....

Jonathan Bromley

Message 26 in thread

From: Mike Monett ( snipped-for-privacy@csolve.net >

Subject: Re: sub-picoampere op amps

Newsgroups: sci.electronics.design

Date: 1999/01/29

Jonathan Bromley wrote:

Not sure what you mean. Do you want it capitalized?

Good to see you got the same answer as John. Now we add the multiplier because the alpha particle is about 5 mev, and each ion pair takes about

It's pretty good to detect 150 nanoamps in high volume, fairly reliably, and real cheap! Best Regards, Michael R. Monett,

Message 27 in thread

From: Mike Monett ( snipped-for-privacy@csolve.net >

Subject: Re: sub-picoampere op amps

Newsgroups: sci.electronics.design

Date: 1999/01/29

Mike Monett wrote:

Correction:

It's even better to detect 150 picoamps in high volume, fairly reliably, and real cheap!

Best Regards, Michael R. Monett,

Message 28 in thread

From: Jonathan Bromley ( snipped-for-privacy@brookes.ac.uk >

Subject: Re: sub-picoampere op amps

Newsgroups: sci.electronics.design

Date: 1999/01/29

Mike Monett wrote:

And better yet to have my grotesque faux pas (forgetting the ionization) picked up so elegantly and civilly!

Retires gracefully to wipe egg from face (:-o) Jonathan Bromley