1
\$\begingroup\$

I have a signal that is 0-500mV when low and 600-900mV when high. I'd like to buffer this signal to an open collector output. The output must withstand a maximum of 3.3V be capable of sinking 60uA.

The power for the detection circuit is provided by a (CR2023) battery that ranges form 2V to 3.2V.

The goal is to find a simple, cheap, low power solution. (By low power, I am shooting for on the order of 1's-10's of uA.)

One solution is to use a low power comparitor like the TS881, but this would also require an external voltage reference to compare to, adding complexity, cost, and power usage.

Another solution would be to use a reset voltage detector like the NCP303, but these are expensive and don't seem to come with thresholds lower then 0.9V.

I feel like there should be a way to do this simply with discrete parts. Maybe somehow taking advantage of the pinch voltage threshold on a JFET? Or the bandgap on carefully chosen LED?

Ideas?

\$\endgroup\$
2
  • \$\begingroup\$ If you don't have a stable supply, you will need some voltage reference (e.g. diode, bandgap, etc.) otherwise you won't be able to determine what 0.5V is. \$\endgroup\$ – Tom Carpenter Jan 1 '17 at 23:26
  • \$\begingroup\$ Agreed! Looking for a solution that hopefully uses the physical properties of a device(s) to get in the correct voltage range! Know any? \$\endgroup\$ – bigjosh Jan 2 '17 at 1:13
1
\$\begingroup\$

A bit more detail about our chip (the one that bigjosh is referring to): Find the latest datasheet here, this link won't change. bristol.ac.uk/voltage-detector

We don't advertise this, but we have different threshold levels, UB20M is 0.6V, UB20L is 0.46V, the former has a current draw of below 10 pA, the latter around 100pA, so with some very high impedance sources, the former will trigger sooner.

We have some videos (a layman's video and some demos) here.

\$\endgroup\$
2
\$\begingroup\$

Here is a simple circuit using only discrete transistors and a few resistors.. total BOM cost should be less than 1 cent in quantity.

schematic

simulate this circuit – Schematic created using CircuitLab

It's a bit trickier than it might be otherwise because your margin between on and off is rather small. Threshold voltages on inexpensive JFETs and MOSFETs are not that well controlled. This circuit will pull OUT low (say with 10K load to +3.3) when the input is higher than about 550mV. Below that voltage, Q1 hogs the current flowing through R1 and Q2 is starved, thus Q3 does not get hardly any base current.

Or just use an IC comparator of your choice with R3/R4 on one input.

If you want to reduce current further and your load will be no lower than 50K, consider increasing the resistor values by 5:1 (say 470K 5% for R1/R2 and 100K 1% / 20K 1% for the R3/R4 divider). Or use 499K 1% for R1/R2 but precision is not important there. Total current drain should be less than 35uA not counting the load, which is comparable to (or less than) self-discharge of many batteries.

\$\endgroup\$
6
  • \$\begingroup\$ What resistor series uses 4k? \$\endgroup\$ – Neil_UK Jan 1 '17 at 21:30
  • 2
    \$\begingroup\$ @Neil_UK (from memory) 4.02 is a standard E96 value, but the OP can use two 2K in series if s/he likes and wants to use E96 or some other series. As "low power" has not been defined, nor has the supply voltage or tolerance etc. I'm not going to waste a lot of time selecting precise resistor values from standard series! Consider this a starting point. \$\endgroup\$ – Spehro Pefhany Jan 1 '17 at 21:34
  • \$\begingroup\$ I tend to have the E24 series memorised so I can pick a 5:1 ratio like 15k/3k easily, 10k/2k would work as well. \$\endgroup\$ – Neil_UK Jan 1 '17 at 21:52
  • \$\begingroup\$ @Neil_UK Sure, and I remember most of E96 (even for 0.1% & 0.05% resistors E96 seems to be the values most stocked, with a few outliers). In this case I started with the 100K resistors for a 10K load. If the OP uses a CMOS comparator then the divider is hardly constrained at all, except by ratio. \$\endgroup\$ – Spehro Pefhany Jan 1 '17 at 22:03
  • \$\begingroup\$ Sorry, I should have been more clear - there is not a regulated 3.3V rail available - 3.3V is the maximum voltage that will be on the open collector output. I think a voltage reference like a bandgap would be required to reference the comparitor. \$\endgroup\$ – bigjosh Jan 1 '17 at 23:09
1
\$\begingroup\$

You don't want to use FET threshold voltage, very unrelaible.

A low power comparator like LM339 would be ideal, or you can build it the hard way from several transistors.

Your threshold has to discriminate 500mV from 600mV. It's easy enough to generate a threshold with that accuracy from a 'normal accuracy' logic rail like 3.3V +/- 5%, with a 6:1 resistive divider, like (for instance) a 15kΩ and a 3kΩ resistor, which would give you 550mV from 3.3V. If the rail had +/- 5% tolerance, then the threshold could vary from 525mV to 575mV. Make it worse by two 1% resistor tolerances and it still doesn't breach the 500mV to 600mV specification in your question.

\$\endgroup\$
1
  • \$\begingroup\$ Unfortunately any practical resistor ladder will blow out the power budget. For example, using your 15kΩ and a 3kΩ resistor values across a 3.3V supply would draw 180uA. You also have to take into account that the battery output is not 5%- it ranges from 3.3V down to 2V (or even hopefully 1.8V), so I don't think a solution based on dividing the output voltage is workable. \$\endgroup\$ – bigjosh Jan 28 '17 at 21:30
1
\$\begingroup\$

I ended up using an AVR ATTINY microprocessor with a built in analog to digital converter. The internal 1.1V band-gap reference gives more than enough accuracy and precision for discriminating the 0V-1V input signal.

The MCU spends the vast majority of its time sleeping, only occasionally waking based on the watchdog timer to sample the analog input and update the digital output. Net power usage is less than 10uA.

The cost of the chip is less than $1 and no other parts are needed - the chip can run directly off the CR2032.

While there is a bit of software involved, I'd still rate this solution as simple and it meets all the other requirements.

Still, I do feel like there is a better (cheaper, lower power, simpler) analog solution out there, so please add you answer if you can come up with one!

\$\endgroup\$
1
\$\begingroup\$

17 days after I asked this question, the part magically appeared!

UB20M High-Voltage, Low-Threshold, Ultra-Low Power Voltage Detector

The voltage detector is an ultra-low power, input-powered device, that does not require connection to a power rail. When the input voltage rises to a typical threshold of 0.65 V, the voltage detector triggers an open-drain output.

http://www.bristol.ac.uk/media-library/sites/engineering/research/eem-group/zero-standby/UB20M_Datasheet_Rev.1.1.pdf

This is exactly the part I was looking for, and targeted for exactly the use I had planned. So much for the patent I was going to file for a novel method of zero-power wake-up. :)

\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.