Inexpensive nano power comparator needed

Why not sim it with the TI part and build it with something else? Comparators aren't complex.

Why bypass a voltage source?

What is R8 for? R17? C2?

Did you not read my post? Apparently not: "...and settled on the TLV3701 / TLV3702, which works very well in my simulator (Micro-Cap V12)."

'Not terribly so, but when working with microvolt differentials and microampere currents, a very small change in input capacitance, input bias currents, input impedance, common mode rejection, etc, can make a big difference in the behavior of the circuit. I did try some that plainly will not work as required. If this were TTL, or even ordinary CMOS, then your response would have some merit. The fact this entire circuit (including numerous daughter circuits) must use considerably less than 1mA, with commensurately large resistors and commensurately tiny capacitors, relatively small differences between two analog components can make a big difference. Since the various pinouts are quite different, designing a board for a part that won't work produces a serious impact to both cost and time, especially for a tiny start-up company like mine.

Do you mean R3? Without R3 the pulse width is far too narrow and the period far, far too short. Without R3, the period is only 51 microseconds, and the pulse width is 3 microseconds. The output is also not very stable, at all. With R3, the output is very stable, the period is 57ms (more than ten times longer), and the pulse width is 1.7ms (568 times longer). A 3 microsecond pulse is not enough time for the daughter circuit to reliably produce any pulses, let alone a train of 7 pulses.

R8 Reduces the negative feedback just a little, making the output just a little more square and just a few hundred microseconds wider. This helps make the daughter circuits more stable without increasing the current significantly. R17 prevents Q1 from triggering erratically and produces an even sharper edge to the output pulse leading edge. Again, this provides for greater stability with no significant increase in current. C2 allows for a very low ESR using less expensive components.

Now, do you have anything constructive to add, such as a SPICE trace using the MCP6542, or a suggestion for a low cost 3.3V nano power comparator?

Nope. You're on your own.

You will probably have to edit the file to get it to work, those Spice models don't always work too well, and they take up a lot space in the circuit diagram.

Thanks! I will give it a whirl.

You don't need to simulate that circuit.

The TLV7031 is a superior low voltage comparator for your application. Not jelly bean cheap, but less than half the price of 3701, less quiescent current, higher speed, internal hysteresis ( a big plus) and flexible output options

I think he's trying to generate free energy through muon induction from the spinning corpse of Bob Pease...

-- john, KE5FX

Something like that. There is an infinite supply of no-name obnoxious amateur googlegroopers who want free engineering and, probably, free energy.

I met Bob a few times. He had a fatal attraction for rusty Volkswagens.

I disagree. Test prototyping (as opposed to production prototyping) is very expensive in both time and money - especially for someone who hasn't had an income for more than 2 years.

Yes, it absolutely is, I would think. I have looked at both the 7031 and the 3701. Neither one is modeled in Micro-Cap

No, and that is a problem. They are not less than half the price, either. 'Closer to 75%, in the sources I checked. Both are considerably more than the MCP6541.

Lower current is certainly a plus, although most of the current draw is from other devices in the circuit. Speed is not an issue, at all. The highest speed device is the 555, and it is only a few KHz. The comparators are all operating at less than 30Hz. Internal hysteresis is very nice, but the circuit design handles that externally (costing less than $0.02). Unfortunately, the MCP6541R is not pin-compatible with the 7031 - the power pins are swapped. The 7032 is more economical per comparator, but I am using an odd number of comparators on the board. I suppose I could use a mix of 7032 and 6541 chips. That is another design cycle out the window. I guess I just have to cross my fingers on this design cycle.

No, just trying to come up with some things to sell. This is for a superior submersible flashlight.

I am not an amateur, and I don't think (hope) I am generally obnoxious. Since my funds are extraordinarily limited, I do need for things to be as economical as possible. As a former physicist, I am very well aware there is no such thing as energy for free. There is such a thing as free energy, as any physicist or physical chemist can tell you, but in that context, the word "free" has nothing to do with trying to get something for nothing. It has to do with whether a process can occur spontaneously or not.

I thought the OP was trying to flash a LED at low duty cycle? You and I might do that with 4-5 descrete transistors for less than the cost of a micropower comparator but the OP presumably has his reasons.

piglet

That is correct. This circuit is designed to turn the flashlight on when a reflective target is placed in front of a TCRT5000L retroreflector and off when a non-reflective target is placed in front of the TCRT5000L. The intent is for the battery to last at least 6 months with an acceptably low amount of battery drain (less than 20% of full charge). This means the entire circuit must draw significantly less than 1 mA when off. In order to manage this, I employ a series of 7 pulses over a period of about 2ms to trigger the unit to switch on when the reflective target is in place. These pulses are created by a TLC555. The reset pin is driven by a pulse created by a very low duty cycle circuit which draws only a fraction of a uV and produces a 2ms pulse every few hundred ms. The circuit draws an average of about 85 uA, including the high current (~100 mA) pulses.

If you can drive the TCRT500L near its maximum of 60 mA instantaneous current while still drawing significantly less than 200uA average using a few discrete transistors, I am all ears. I am using a 3.3V power supply in order to minimize the current, but 5V at about half the current would also be fine.

There are a lot of unstated assumptions in what you wrote. It seems that you want TCRT500L to pulse with about 1/1000 duty cycle at frequency of order 1 Hz. This is doable using RC oscilator and pulse forming circuit in something like 5-10 transitors. If R are high enough this control circuit will run on few microamps. But your circuit generates more complicated thing. More complicated waveforms are easily generated using MCU. AFAICS almost any MCU with low power modes should be able to generate required pulses using less then 10uA of current, some claim below 2uA for such use cases (of course you need to separately account for TCRT500L current).

Ignoring the complication of a burst of 7 pulses here is a first pass illustrative sketch. With shown values should give approx 80us pulses every 80ms with LED current approx 100mA. Average supply current should be a little over 100uA. If you are going to simulate it you may need to start the supply from 0V rather than steady-state to help simulation astable startup.

NPN are gum drop type like 2N3904 and PNP 2N3906 etc etc. Change caps to

piglet

No, there are not. I did not ask for a circuit design. I provided that. All I asked for was either a low cost comparator ( ~ $0.30 or less) that would work with Micro-Cap 12, or for someone to sim the circuit in their own software using an MCP6542. >It seems

Well, 1/1000 would be nice, but the design offers about 0.44%. Anything under 1/200 should be fine.

That's a little too low. The switch should not take a full second to turn on or off. On the order of 50 - 100ms is good.

A circuit employing 5 discrete transistors is not too bad. A circuit with 10 discrete transistors is getting to be a bit of a mess, especially since each component adds to the cost of assembly. I would absolutely rather have 2 comparators, a 555, and a handful of resistors than two or three dozen discrete components. There are a number of reasons why manufacturers employ ICs, rather than discrete components, you know. The design as-is only uses 28 components, mostly resistors. Unless you can show me differently, I submit any design employing only discrete components is going to have a much higher parts count, and probably be more expensive, provided the MCP6541 or an even less expensive comparator will work.

How is that, again?

So now you are going from one extreme to the other. Zero integration to massive integration. There are plenty of applications where an MCU is all but essential, but this is not one of them, IMO. If I *WERE* to go that route, I would employ a more complex waveform and decode it on the other side. The TCRT500L has optical filters that make such complexity unnecessary, or so I hope.

At a certain point, reducing the idle current has no significant impact on the average current, since the bulk of the current is delivered during the LED pulses. Below about 100 uA, the idle current is quite insignificant, and there is no real difference between 2uA and 100uA.

Nice. Here it is in LTspice:

Thanks Steve, that gets me to a google drive link that wants a login?

That two transistor complementary oscillator has a 1960s feel to it but works well in low power applications like this because most of the time both devices are non-conducting.

piglet

Thanks. I thought I had enabled sharing, but there is a separate procedure to follow that is new.

I checked the file. It should work now. Ignore the signin link in the upper right.

Let me know if you find any more problems.