Okay so, I have a design that 'works' on the breadboard. But I feel like it could use some improvement. Here is my current design:


simulate this circuit – Schematic created using CircuitLab MAX9064

Now I think the solar power collection part is okay. But the pump driving part seems like it could be improved...

The solar panel will generate around 9V which is stored in a capacitor bank and converted to a 1.8V supply with a maximum current of 100mA. I haven't tested this thoroughly in real world conditions, but all seems to be working as expected.

The motor requires more than 100mA for a cold start. So the idea of the second part of the circuit is to charge a capacitor up before switching the motor on to deliver the required current. On switching the motor there is a voltage drop of over 100mV, so positive feedback is required to generate a fairly large hysteresis band to cope with that.

The MAX9064 has only 0.2V reference which I think could be an issue.

I've tried with a few different R values. I've been going for a switching voltage of 1.7+ with around 400mV of hysteresis. With R3 and R4 at 10Meg I seem to get switching problems with the motor on. Rapidly switching, it does run, but not well.

Moving down to an R3 of 100K causes big problems with accuracy of the switching voltage and I've had trouble getting it to switch at the desired level.

Currently using R3 of 1Meg, R2 of 33K and R1 of 220K has worked best. The switching accuracy is still a bit off (on the lower end than it should be). I've left R4 at 10Meg.

My goal is obviously to deliver a much current to the motor as possible, while having enough accuracy to switch somewhere between 1.7V and 1.8V and deal with the voltage drop caused by switching. So what can I do to improve on it? I think C2 at 10F is probably overkill and maybe I need to get some noise reduction going? I'm also worried that the 0.2V ref might be an issue, though I could use a different comparator if that is the case.

Thanks for any input!

Update: I made a small change to the circuit which seems to have greatly improved it as follows:


simulate this circuit The addition of a 100nF ceramic in the location shown greatly improves the output performance. I also went back to higher R values, but that wasn't doing much.

The problem is I have NO idea why adding the cap there made such a massive difference???

  • 1
    \$\begingroup\$ Decouple your comparator VIN. A 10F cap isn't going to have a good ESR -- I'd at least put a 100nF cap right at the comparator, from ground to VIN. I might even decouple the comparator with a small series R and something like a 1uF cap (since 1uF ceramics are easy to get these days). The "small" resistance should be enough to drop no more than about 50mV when the comparator is pulling its maximum current. \$\endgroup\$
    – TimWescott
    Commented May 12, 2019 at 16:45
  • \$\begingroup\$ @TimWescott Good point, I will definitely add a cap across the comparator. Do you think I might need something around M2 as well? Is 10Meg too much for M2? \$\endgroup\$
    – hekete
    Commented May 13, 2019 at 5:36
  • \$\begingroup\$ If you mean for R4, yes, \$10\mathrm{M}\Omega\$ seems excessive to me. You can get that much change in resistance by wiping a sweaty thumb across an open circuit on a PCB. My gut feel is that \$100\mathrm{k}\Omega\$ is safe, but I might check the off-state leakage of M1 before I believed it (and, of course, consider what current consumption you want from the whole circuit). \$\endgroup\$
    – TimWescott
    Commented May 13, 2019 at 15:03
  • \$\begingroup\$ Yeah, it's just important that the maximum amount of the pretty limited available current is delivered to the motor. Which is why I went with a crazy high value.. \$\endgroup\$
    – hekete
    Commented May 14, 2019 at 6:03
  • \$\begingroup\$ I think I'm having issues with R4 and M2. 1.6V is measured at the junction of R4-M2, but only ~1.2V is measured on the other side... But changing the R4 value down doesn't help. \$\endgroup\$
    – hekete
    Commented May 15, 2019 at 12:46

1 Answer 1


super late answer, but I think I know why that cap helps. I'm also fighting with a MAX9064 on a different circuit and came across this.

The capacitor has more impact than the resistor because it has a much faster AC response. When the transient from the motor comes in, the capacitor will look almost like a dead short while the resistor is still a very high impedance. You can use the same trick on regulator feedback networks.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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