Instead of the LM2936, you might consider an LP2981; it has much lower dropout voltage and much lower minimum Vin.

 

Oops, that's embarrassing! I think I must've copied the formulas wrong when I tried to work out the timing values. Besides this mistake, I also have formulas telling me that the 555 can only do ~0-50% duty cycle instead of ~50-100%, which I've since found is apparently backwards.

Actually, there's another approach that will allow you to control the HIGH and LOW parts of the cycle completely independently.

First, disconnect pin 7 and discard resistors R1, R2 and R3. Leave pins 2 and 6 connected to C3.

Now, connect a diode and a resistor, in series, between pin 3 and pins 2 & 6, with the anode of the diode connected to pin 3. The value of this resistor will determine the duration of the HIGH part of the cycle. Connect a second resistor and diode, in series, in parallel with the first diode/resistor pair, with the cathode of this diode connected to pin 3. The value of this second resistor will control the LOW portion of the cycle.

And that's all there is to it.

 

My fear with the diode trick for truly independent high and low time adjustments was that the drop across the diodes would be dangerously close to preventing proper operation with 3.3V operation and around 0.7V drops. But, as I was writing this I remembered that I could use diodes with much lower drops (Schottkey? I always mix up the various names!) and it wouldn't be a real issue. Probably not necessary for this project since your first proposed solution does exactly what I want, but a good trick to keep in mind for the future. Thanks!

 

Also, regarding my last LDO question, I realize now I was getting really close to LMGTFY territory: a simple Google search found me at least 3 papers from TI alone on the subject. Haven't had time to read and digest them all yet, but I should be able to find what I need. Sorry for the lazy questions on that one!

 

Here is a very good LDO doc from TI...

eT

 

That's a great doc! If I understood it correctly, I think I can get away with saturated LDO, dropout-voltage level performance for this very simple circuit. I don't expect the incoming power to be especially noisy, and I don't think this circuit should be especially sensitive, so if the Vin is too low and the regulator just drops it by the dropout voltage, that'll probably be fine. Regardless, it should be keeping the voltage in the ballpark range I need it, I think.

Having said that, when I re-read the specs on my chosen 2963 regulator I discovered that it was only rated to 50mA, and I could've sworn I was filtering for a minimum of 100mA capacity just to be safe, so I'm picking a new LDO regulator. The LP2981 mentioned above won't work for me because it's surface mount and I'm looking for an easy-assembly through hole solution. So I'm looking now at the LE33CZ-TR, the BA033CC0T, or the L4931CZ33-AP. All are through-hole, appear to be available in large quantities, can handle more than enough current, and should have sufficiently low dropout voltages in the expected 50-75mA current range.

I'm busy this weekend, but plan to revise my schematic yet again with the new resistor/trim pot arrangement OBW0549 recommended and with an updated LDO regulator and get the updates posted soon. Thanks to everyone who's helping out. Learning new stuff all the time (even if it's sometimes embarrassing when I see my mistakes!)

 

I'm afraid I'm beating a dead horse here, but I appreciate everyone's patience! I've revised the plans again:

1) Updated LDO regulator for one with 100mA capability and lower dropout voltage.
2) Re-arranged resistors and variable resistor to get consistent 555-low time and variable 555-high time.
3) Switched from n-channel to p-channel MOSFET to invert logic so relay and LED are only on for short pulses of 555-low time instead of long 555-high times.

I've never worked with FETs before and I keep getting the symbols, polarities, etc. mixed up. It's really embarrassing! I'd love a second set (or more) set of eyes on all of this, but especially on the orientation/polarity of the MOSFET. As if I weren't turned around enough, DipTrace (which I otherwise love) seems to show MOSFET symbols backwards of how I'm finding them drawn everywhere else. I flipped the one here to match drawings I've seen and to make the standard high voltage at the top of the schematic arrangement work cleanly. I think in the end I got it right, but I'd love to know for sure.

Thanks!

 

You don't need that expensive and specialized NDP6020P FET for Q1...

Use an one-cent 2N3904 NPN transistor instead.

There is an old saying that 'an engineer is someone who can design something for 50-cents that any other dam fool would require 2-dollars to do'

Keep this in mind

 

You don't need that expensive and specialized NDP6020P FET for Q1...

Use an one-cent 2N3904 NPN transistor instead.

There is an old saying that 'an engineer is someone who can design something for 50-cents that any other dam fool would require 2-dollars to do'

Keep this in mind

Q1 is a P-channel MOSFET. If you want to replace it with a BJT you need a PNP like a 2N3906, not an NPN.

With a BJT, however, I'd want to check the Vce(sat) specs of the transistor and the relay minimum pull-in voltage. With the MOSFET, the relay would see the full supply voltage, minus only a couple of millivolts; with a BJT, the loss from Vce(sat) is going to be a lot more than a few millivolts. The spec sheet for the 2N3906, for instance, gives a maximum Vce(sat) of 0.4V for Ic = 50 mA and Ib = 5 mA. I don't know what the specs of the relay are, but my guess is the relay might not have enough voltage left to reliably pull in.

Yes, a 50-cent solution is better than a 2-dollar solution-- but only if it actually works.

 

Q1 is a P-channel MOSFET. If you want to replace it with a BJT you need a PNP like a 2N3906, not an NPN.

With a BJT, however, I'd want to check the Vce(sat) specs of the transistor and the relay minimum pull-in voltage. With the MOSFET, the relay would see the full supply voltage, minus only a couple of millivolts; with a BJT, the loss from Vce(sat) is going to be a lot more than a few millivolts. The spec sheet for the 2N3906, for instance, gives a maximum Vce(sat) of 0.4V for Ic = 50 mA and Ib = 5 mA. I don't know what the specs of the relay are, but my guess is the relay might not have enough voltage left to reliably pull in.

Yes, a 50-cent solution is better than a 2-dollar solution-- but only if it actually works.

Precisely the reasons I ended up with this MOSFET. I was totally oblivious to these issues until they were raised earlier in this thread (thanks for the good catch!) Since the main challenge in this circuit is variable, potentially very low supply voltage, this is a core issue.

Even looking at logic level MOSFETs, a lot of them looked dubious with Vgs of around 3V (555 output may not reach full 3.3V supply voltage,) so I tried to choose options with especially low Vgs requirements for the Vds and current expected, and especially low Rds-on values as well.

I'm sort of deliberately challenging myself on voltage tolerance of this circuit, both to help the person I'm designing it for and as a good learning experience for me, which it has already been!

So... about that MOSFET, did I manage to wire it right? Drain and source aren't reversed? Gate resistor ok? I wasn't really sure how to size that - at the moment it's basically just limiting current demand on the 555. I don't expect switching speed to be an issue with low frequency and low power all around, but I could be overlooking something.

Thanks!

 

So... about that MOSFET, did I manage to wire it right? Drain and source aren't reversed? Gate resistor ok? I wasn't really sure how to size that - at the moment it's basically just limiting current demand on the 555.

Looks right to me.

I'm not sure the 100 ohm gate resistor is really necessary, but anything between 100 and 1000 ohms would probably be fine given that switching speed is not an issue.

Looks good!

 

Thanks so much for all your help! Think it's time to finalize the BOM and send it off.

Thanks!

 

Thanks so much for all your help! Think it's time to finalize the BOM and send it off.

Are you going to breadboard this gadget first?

I always do, simply because it brings good luck: almost every time I've tested a design before committing to PC board or sending it to a client, the thing turns out to work exactly as expected; and the few times I've skipped that step (mostly because I thought the design was too simple to bother breadboarding it), most often something doesn't work right and I end up red-faced.

Call me superstitious if you will, but that's the way it's been for me.

 

It is not being superstitious. It is practicing due diligence.

 

Why have you switched from an N-channel MOSFET in the original circuit to a P-channel MOSFET?
The results will be reversed.

 

Q1 is a P-channel MOSFET. If you want to replace it with a BJT you need a PNP like a 2N3906, not an NPN.

It is only simply switching an relay. I say NPN because I would modify the circuit to SINK current to the relay coil, not source it.

 

Precisely the reasons I ended up with this MOSFET.
Even looking at logic level MOSFETs, a lot of them looked dubious with Vgs of around 3V (555 output may not reach full 3.3V supply voltage,) so I tried to choose options with especially low Vgs requirements for the Vds and current expected, and especially low Rds-on values as well.

Use an pull-up resistor on the output of the 555 timer. You do not need an mosfet to simply switch an relay ON.

 

Please consider the following for switching an relay:

 

Please consider the following for switching an relay:

Have you read any of the other posts here? This circuit needs to run on as low of an input voltage as possible. With 3.3V or less available, there are concerns (very valid as far as I can tell) that the relay wouldn't get enough voltage to work reliably when switched with a bjt.

 

Have you read any of the other posts here? This circuit needs to run on as low of an input voltage as possible. With 3.3V or less available, there are concerns (very valid as far as I can tell) that the relay wouldn't get enough voltage to work reliably when switched with a bjt.

Yes well I must have missed something then. Operating a 5 volt relay off 3V3? Well the whole schema seems wrong then.