I'm working on a hobby project and think I can harvest enough power from a slow, TTL level serial link. There are 5 data/control lines (all open collector) and a single ground. Signaling is the typical wired-or in open collector systems, each line relying on a 1k pullup to 5V.

Thinking through things, I figure I can harvest up to about 3mA from each line without affecting the operation of the link. With these being TLL level lines, pulling up to 3mA each, leads to a drop of 3V across the pullups, leaving the lines right about Vih for TTL. Realistically, I think I'd limit it to no more than 2.5mA, to keep things well above the 2V Vih level. With the 5 lines, plus sufficient capacitance, and a battery, I'm thinking I can easily ride through the transitions. They are all very slow. The maximum low time for any line is about 60us, and never are all of the lines low at the same time.

The idea then is to run each of the signal lines into a current limiter, then diode-or all of them into an LDO or SMPS to get down to the 1.8V I need for my microcontroller (uC). The uC subsystem sits about 10mA. With 5 lines, I can get 15mA when the bus is idle, and then ride through the transitions. I haven't completed all the math yet to determine how much backup power I need, but that will determine what kind of battery capacity I might need to meet long term operation.

So, here's the crux of my question. I need to limit each line to about 2.5mA. I've been researching current limiters. I'm thinking I need high-side limiters since I have multiple power sources. A low-side limit wouldn't work as I need to limit from the source, not the overall current. I think. But I'm not finding any design principles on what architecture or how to select transistors or resistor values. Or are there IC's that do this? The ones I've found don't go down as low as 3mA, so I think I have to go with a discrete solution.

Further, how do I simulate this uC load? I'm not really finding much there. I saw one page model it has some capacitance and a current source in parallel. I guess I can sum all the bypass caps on the ICs and then put the maximum current draw I expect in the current source. But when I did so with some of the current limiters I considered, I couldn't keep the voltage high enough or there was excessive inrush current when it turned on.

So, I guess two questions:

1. Any references or assistance on designing current limiters down to about 2.5mA? I'm ok with a voltage drop so long as it stays above 1.8V.
2. How do I model the uC subystem's power requirements?

 

I curious about where you think the power is coming from to power the TTL gates. Would it not be simpler and more straightforward to just have a 1.8-volt regulator?

 

I curious about where you think the power is coming from to power the TTL gates. Would it not be simpler and more straightforward to just have a 1.8-volt regulator?

The power comes through the 1k pull-ups on the link. But if I draw too much current on any line, the line will drop below VIH. The IV subsystem is at about 10mA, which is too much for any one line to supply. Hence the current limiter.

 

The power comes through the 1k pull-ups on the link. But if I draw too much current on any line, the line will drop below VIH. The IV subsystem is at about 10mA, which is too much for any one line to supply. Hence the current limiter.

Power doesn't normally come through pullups. The answer I was looking for is that power comes from a power supply. If you have a power supply of some capacity, why are you worried about harvesting power from pullup resistors? That is no way to establish a stable VCC required by most chips. What you want is a power supply that won't let the voltage drop when more current is required. IMHO, what you are trying to do is foolish in the extreme.

 

Sure. Normally a power supply. But I’m aiming for a standalone, using the comm links pull-ups to supply the bulk of the power. I’m shooting for the device to operate continuously for at least a month. I don’t want a separate power supply.

the devices on the ends of the link are open collector with 1k pull-ups to 5V. I want to leverage the power they have available to source without degrading the link comms.

I am aiming for *no* external connections other than the link itself. The battery provides the ability to ride through excessive low current cases. But generally there should be sufficient power to run without the battery. I’m considering the possibility of a supercap and avoiding recharging or battery replacement. The design isn’t purely powered parasiticly.

finally, there will be brownout detection to put the device is ultra low power when necessary, which is in the uA range.

You say foolish. But this is a challenge I’m trying to overcome. It is easy to use an external supply. But that’s not the goal. The goal is standalone.

 

Just because you see a windmill, doesn't mean you should go tilting at it. Good luck with your quest.

 

Just because you see a windmill, doesn't mean you should go tilting at it. Good luck with your quest.

Your help, such as it is, is appreciated.

 

Your help, such as it is, is appreciated.

The old X10 CM17A firecracker used Parasitic Power from the comm line but it needed at least a +-5v RS232 physical interface for reliable power with the ability to pull tens of mA without voltage drops, not 1K pull-up power designed for CMOS input loads.



Your idea might work but it likely will be like a dog walking only on hind legs.
Amazing but very limiting for anything useful.

 

Sure. Normally a power supply. But I’m aiming for a standalone, using the comm links pull-ups to supply the bulk of the power. I’m shooting for the device to operate continuously for at least a month. I don’t want a separate power supply.

the devices on the ends of the link are open collector with 1k pull-ups to 5V. I want to leverage the power they have available to source without degrading the link comms.

I am aiming for *no* external connections other than the link itself. The battery provides the ability to ride through excessive low current cases. But generally there should be sufficient power to run without the battery. I’m considering the possibility of a supercap and avoiding recharging or battery replacement. The design isn’t purely powered parasiticly.

finally, there will be brownout detection to put the device is ultra low power when necessary, which is in the uA range.

You say foolish. But this is a challenge I’m trying to overcome. It is easy to use an external supply. But that’s not the goal. The goal is standalone.

This is definitely possible but not sure if your application can do it or not because the way I did it worked because I had control over all the signals. The power is very small, but if your circuit draws microamps you can still do it.
What I would suggest for your application is set up some experiments and make some measurements.
My goal was to create an RS232 link that could power the external uC chip that was used for making measurements without the need for a wall wart to power it.

 

I think a 2.4V NiMH battery and silicon diodes would do what you want. When the diode is conducting it would drop about 0.6V, giving you 3V at the pull-up and the current is limited to 2mA. As long as the battery remained charged, it should work. And a 200 mAh could handle 10 mA if truckle charge.

But I do agree with others that it is puzzling why you want to do this instead of tapping power from the other electronics. Is the reason for this that your device is far from the two ends of the link?

 

I think a 2.4V NiMH battery and silicon diodes would do what you want. When the diode is conducting it would drop about 0.6V, giving you 3V at the pull-up and the current is limited to 2mA. As long as the battery remained charged, it should work. And a 200 mAh could handle 10 mA if truckle charge.

But I do agree with others that it is puzzling why you want to do this instead of tapping power from the other electronics. Is the reason for this that your device is far from the two ends of the link?

Hi Bob,

With one of my RS232 projects I did not want to have to use a wall wart or batteries to power it because if you only need a little power you can power the uC chip right from the RS232 port, which is very convenient.

The difference becomes obvious.
If you use a wall wart, you have to not only get a wall wart of a decent voltage (5v, 10v, 12v, etc.) you also have to provide a jack for it to plug into, or cut the plug off and wire it direct. However, now whenever you want to use the uC device you made you have to also find a plug to plug in the wall wart.
If you use batteries you have to provide a battery box, and again have a connector for that on the uC device you made, and either replace the batteries when they run down or have a charger on hand or a wall wart to charge them with.

When you can use the power from the RS232 connector on say the computer, you need none of that. You just plug the RS232 plug into the jack on the computer (which you have to do anyway) and you have your communications established as well as your power needs. Nothing else to do except use your uC device as intended.

I have to say though this may not always be possible. I was using a raw Microchip uC chip which uses very low power. One of the projects was a four channel voltage meter.

One big caveat is that it is much harder to get galvanic isolation using power from the RS232 port, because innately you need to connect the circuit right to the circuit of the RS232 port inside the computer, which means no isolation.
To get power with isolation that way, you'd have to use a transformer based DC to DC converter, but that would probably use more power than the RS232 port can put out. Maybe, if you designed a very special DC to DC converter that could do around 1ma output with around 1ma or so input you might be able to pull it off. I never tried that though.

 

I think a 2.4V NiMH battery and silicon diodes would do what you want. When the diode is conducting it would drop about 0.6V, giving you 3V at the pull-up and the current is limited to 2mA. As long as the battery remained charged, it should work. And a 200 mAh could handle 10 mA if truckle charge.

Why would the current be limited to 2mA? There is a 1k pullup to 5V. Putting a diode in series w/ the pullup doesn't limit the current. Or are you suggesting a different type of diode? I've looked for current limiting diodes, but have not found any as low as 2mA.

But I do agree with others that it is puzzling why you want to do this instead of tapping power from the other electronics. Is the reason for this that your device is far from the two ends of the link?

Twofold. First, there are solutions that already exist using external supplies. But they are large and bulky, usually relying upon a 12V source. Second, space is limited, and I want to make this solution smaller than the competition.

 

What I would suggest for your application is set up some experiments and make some measurements.

I've done so, mostly in simulation. I'm trying to figure out how to limit the current. My original post was about trying to design the current limiter. I can't seem to dial in the example circuits I have found of high-side PNP based current limiters. That's the help I'm seeking there.

Whether or not it will be possible--I think it is--the battery is the help. I just need help figuring out the details.

My goal was to create an RS232 link that could power the external uC chip that was used for making measurements without the need for a wall wart to power it.

Same here. There is a 12V DC supply available, but other devices do use that same supply. And it complicates the installation to tap into the existing power systems.

I gave the background to give context, not to seek overall design advice. I'm only looking for assistance on the current limiter.

 

Why would the current be limited to 2mA? There is a 1k pullup to 5V. Putting a diode in series w/ the pullup doesn't limit the current. Or are you suggesting a different type of diode? I've looked for current limiting diodes, but have not found any as low as 2mA.


Twofold. First, there are solutions that already exist using external supplies. But they are large and bulky, usually relying upon a 12V source. Second, space is limited, and I want to make this solution smaller than the competition.

The battery is 2.4V, the diode drops an additional 0.6V, so there is 2.0V left across the 1K resistor, which means the current is:

I = V / R = 2 / 1000

 

I've done so, mostly in simulation. I'm trying to figure out how to limit the current. My original post was about trying to design the current limiter. I can't seem to dial in the example circuits I have found of high-side PNP based current limiters. That's the help I'm seeking there.

Whether or not it will be possible--I think it is--the battery is the help. I just need help figuring out the details.


Same here. There is a 12V DC supply available, but other devices do use that same supply. And it complicates the installation to tap into the existing power systems.

I gave the background to give context, not to seek overall design advice. I'm only looking for assistance on the current limiter.

Hi,

Ok then let me ask you something.
Why do you need a current limiter?
I ask because if your max current is say 1ma and your circuit tires to draw 2ma, a current limiter will limit the current to 1ma and then the circuit won't work.
Does this sound right to you or is there something else to it?

 

The battery is 2.4V, the diode drops an additional 0.6V, so there is 2.0V left across the 1K resistor, which means the current is:

I = V / R = 2 / 1000

Right. For the battery, if the battery had a 1k series resistor. The case I'm trying to solve is at 1k pullup to a 5V rail, and limiting the current through that resistor to 2mA.

I'm trying to use something like this:

https://en.wikipedia.org/wiki/Current_limiting

Or:

https://www.circuitlab.com/circuit/sevhs5/pnp-current-limiter/

 

Ok then let me ask you something.
Why do you need a current limiter?

I want to keep the voltage drop across the 1k pullups to less than 2V. This keeps the lines above VIH under the cases where I'm drawing current from the communication links.

I ask because if your max current is say 1ma and your circuit tires to draw 2ma, a current limiter will limit the current to 1ma and then the circuit won't work.
Does this sound right to you or is there something else to it?

The uC subsystem is sitting under 10mA. I have 5 control lines. Under idle conditions, the circuit can be completely powered via the parasitic current through the control lines. However, when the lines are transitioning (pulled to ground--the link is open-collector, wired-or), I plan to use the battery to supply the current necessary to keep the circuit up and running.

The comm link is very slow and bursty in nature. There are very long spans of time between communication messages (every hour or so there is a burst of communications lasting about 30-90sec). I haven't completely characterized the link yet, but the goal is to have sufficient battery storage that the device can last a month or more. The parasitic power is to do two things: eliminate the need for an external supply (except for recharge) and keep the size small.

 

I want to keep the voltage drop across the 1k pullups to less than 2V. This keeps the lines above VIH under the cases where I'm drawing current from the communication links.


The uC subsystem is sitting under 10mA. I have 5 control lines. Under idle conditions, the circuit can be completely powered via the parasitic current through the control lines. However, when the lines are transitioning (pulled to ground--the link is open-collector, wired-or), I plan to use the battery to supply the current necessary to keep the circuit up and running.

The comm link is very slow and bursty in nature. There are very long spans of time between communication messages (every hour or so there is a burst of communications lasting about 30-90sec). I haven't completely characterized the link yet, but the goal is to have sufficient battery storage that the device can last a month or more. The parasitic power is to do two things: eliminate the need for an external supply (except for recharge) and keep the size small.

Well, I have given you a very simple solution.

 

Well, I have given you a very simple solution.

I’m not understanding then. What you suggested earlier is a battery only solution. But that is not my goal. My goal is a battery supplemented by parasitic power through the control lines. I know I can do a battery or an external power supply. What I am trying to figure out is how to design a current limiter of 2mA.

 

Why do you say the Ni-MH battery cells are only 1.2V? That is when they are half discharged with a high current or are almost dead with a low current. The voltage curve in the datasheet was with a 750mA load.
My Energizer Ni-MH AA and AAA cells are over 1.4V today and were charged a few weeks ago.