Is a mosfet floating source ok?

Higher voltage zeners, above roughly 7 volts, leak nanoamps below their knee.

It's the zener/avalanche thing.

--

John Larkin      Highland Technology, Inc 

The best designs are necessarily accidental.

I do not have DS3231 handy. Checked DS1307 and it seem to have low leakage, probably of order tens of picoamps.

But transistors also tend to have low leakage and voltage you see is balance between DS3231 leakage and transistor leakage (and possible measurement errors).

Well, cheap DT830 when set to 200mV actually measures current trough 1M resistor, effectivly giving 200nA current range. Similar thing can be done with many other meters, the ones with 10M resistance give 20nA current range. One needs to verify that meter actually has aproporate resistance: some newer ones use digital calibration and may contain say 12M resistance so you would need to rescale. Some other connect directly to measurent part, so input current corresponds to leakeage measurent part, with such meter one would need external resistor.

Of course, there are also fancy professional low current meters but I have none. Or, as Steve wrote, one can make sensitive meter using low-leakage opamp.

--
                              Waldek Hebisch

Well, once voltage is high enough there will be avalanche breakdown in the output fet. So it is quite likely that there is no need for extra ESD protection on this pin: fet will protect itself and probably simultaneously will protect other elements. But I do not have insider knowledge, I do not know if there is some extra failure mode or some extra protection. If it were my chip I would just use the circuit without adding extra Zener (OK, I would probably use resistive divider and BJT, but for resons that have nothing to do with protecting INT pin). But I leave it to OP do decide if he is worried about small currents.

--
                              Waldek Hebisch

I only push parts past abs max if there is a big performance payoff, and some real-life-tested margin. That doesn't seem to be the case here.

I do, but the lowest current it will display is 0.01mA.

I originally had R2 at 470K just to minimize the current flow out of the coin cell when the power is on. But it sounds like the leakage current wouldn't be enough to hut anything even if voltage goes above 5.5V. So I'm inclined not to use either R2 or the zener. It certainly works fine like that in my test setup. Nothing gets hot. :-)

Neither do I. All I know is the datasheet says the absolute maximum voltage that can be applied to the INT pin is 5.5V, regardless of vhat Vcc is. But what others are telling me is that even at higher voltages, if the current is effectively limited to some small fraction of a microamp, I shouldn't take that 5.5V too literally.

Well, the other version I looked at was using an NPN in place of the N-channel mosfet, with the base driven from the coin cell through a high-value resistor, and the emitter connected to INT. The voltage on INT in that setup would be pegged at about 2.4V when INT is inactive. But current would flow from the coin cell while the power is on, so if will work ok, I'd rather use the N-Fet, which wouldn't draw any coin cell current when either on or off.

Using the 12V rail through a divider to supply the base would also work, but that 12V might also be a battery, so I'd rather not draw current through the divider either.

Newer datasheets for MCU-s say samething like: "you can exceed voltage limits on I/O pins as long as current is limited to 5mA". I do not think that there was change in MCU-s. Rather, in many circuits people had some leakage to pins and now manufacturers recognized reality and gave official rating. Also, during ESD events there are ampers flowing trough pins at voltage exceeding "absolute max" and chip is supposed to survive this. Clock chip is probably more delicate than MCU, so I would not dare to pump miliamps at increased voltage, but microamps should be safe.

A lot depends on battery. Coin cells have long life and low capacity, so extra load matters (but if INT have low duty cycle extra load can make no practical difference). I have calculator with 36 year old batteres which still have enough charge to run calculator, so clearly long life batteries are possible. OTOH standard "long life" alcalines seem to self-discharge in about 10 years. Assuming nominally 2000mAh (for AA format) that gives almost 23uA self-discharge current. Other types (especially rechargables) seem to have larger self-discharge current.

2.7M plus 1.5M divider takes 2.85uA when INT is high and delivers about 4uA when INT is low, this should be enough base current for NPN. It depend on other factors if this current trough divider is acceptable or not. Of course, there is large parameter space to explore, there are tradoffs between static current and temperature stability, noise immunity, and switching time.

--
                              Waldek Hebisch

Does it have a 200mV setting? If so, then it can show a current of .2/10Meg = 0.2uA full scale.

0.02 uA, better known as 20nA.

Yes. Thank you.

Even so, that is 5 times better than your 50uA analog meter.

I'm confused. It's an autoranging meter. The lowest displayed current value is 0.01 milliamp. The lowest displayed voltage is 0.1 millivolt. Anything less than those just displays as zero.

Since anything on the 50uA scale on the analog meter just shows as zero on the DMM, I use the analog meter for very low currents.

Ok, I said that wrong.

Anything below 10uA on the analog meter shows as zero on the DMM. The analog meter is 50uA all the way to the right, and has individual uA divisions. So the DMM can show me 0, 10, 20, 30, 40 or 50 uA, but the analog meter shows me all the inbetween levels. And at the very low end, the smallest needle twitch I can detect is probably in the neighborhood of

200nA.

So 50uA is the full scale of the analog meter, but 0.01mA is the lowest non-zero *value* displayed on the DMM. I don't know what the scale is.

f

He is saying that the current drawn by your meter on the voltage range is e quivalent to the voltage displayed divided by the input resistance. So if the resistance is 10 Mohm on the 200 mV scale the current would be 0.2V / 1

0e6 ohms or 20 nA full scale. The only down side is the voltage drop out o f up to 0.2V. You can get less sensitive ranges by adding a resistor in pa rallel as a current shunt. Use a 1.1 Mohm resistor to get a 200 nA range o r 100 Kohm to get 2 uA.

Forget about using your meter the way it was intended, use it the way it wo rks best.

--

Rick C. 

+-- Get 1,000 miles of free Supercharging 
+-- Tesla referral code - https://ts.la/richard11209

Sorry, but I don't understand at all. I thought I made it clear in the beginning that I'm not formally trained in any of this. Can you say specifically how I should connnect the meter, and what function I should turn it to, to measure current down to 20nA? Or even to 1uA?

Also, the DMM manual says impedance is 10M-ohms but "more than 100M-ohms on the 400mV range".

I assume that means your lowest range is 400 mV. That sounds like they don 't have a terminating resistor on the 400 mV range. They are counting 1,

2, many. "More than 100 Mohms" is "many".

To easily measure up to say, 40 uA you calculate a resistor value that will give 400 mV with 40 uA of current. E = I * R or R = E / I so 400 mV / 40 uA or... 10 Kohms. So put a 10 Kohm resistor in the current path and m easure the voltage across it, divide the reading by 10 and you have the cur rent in uA.

Maybe explaining it this way makes it more clear. If the meter had a defin ed input impedance the lowest possible current range would be the voltage s cale divided by that resistance. What meter do you have? I need to buy a new meter and I get tired of looking at 50 vendors on eBay or Aliexpress se lling the same crap meter. Did I say 50? More like 500! It's hard to fin d useful variation or get any idea of quality.

--

Rick C. 

+-+ Get 1,000 miles of free Supercharging 
+-+ Tesla referral code - https://ts.la/richard11209

Ok. I have a 5W 1-ohm resistor that I use to measure power rail current, or look at any voltage drops there on my toy scope. The 1 ohms makes the math easy. I just never thought that setup would work at really low currents and a high value resistor, but I'll give it a try.

I have the Radio Shack 3 3/4 digit auto-ranging meter #22-803, which I purchased in August, 1998 for $29.99, and that was when $29.99 was real money. I don't think they make it anymore. :-)

Dave Jones sells three levels of DMMs on his store. I assume that means he thinks highly of them. You might look at the Brymen235, also available on Amazon for US$139.

Or for the kitchen sink, for $300:

It's the same deal. The problem with a high value resistor is the larger v oltage drop, but 0.4V should not impact this circuit and since the actual c urrents are likely to be much lower you will probably need a larger resisto r. I just picked the 40 uA example to compare to your analog meter. BTW, do you know the resistance of the 50 uA meter? It's probably not 10 kohm, but I bet it is something over 1 kohms.

Yea, I had a couple of Fluke meters for decades, but one is lost and the ot her is no longer of this world. I bought a couple of decent meters from eB ay over the last 10 years, but one stopped working and the other was damage d by a leaking Costco battery. That one was pretty featured, but I don't u se the fancy features often. It was ok other than the choice of colors in lettering. Light blue lettering against grey just doesn't show up. I was going to return it because it didn't have an AC range until I finally saw t he lettering. I still have a small one that I carry in my computer bag. I 'm not spending serious bucks on any more. With decent meters being availa ble for under $50 why waste money on expensive meters?

It's hard to find reviews of stuff on eBay because it changes so often.

--

Rick C. 

++- Get 1,000 miles of free Supercharging 
++- Tesla referral code - https://ts.la/richard11209