Power Latching IC

[ktomecek]

Hello folks,
I have a circuit that I would like to power when a button is pressed. It draws under 50 mA's. I would use the uP to toggle a control line to turn the circuit off again. And I would like very low current drain while the circuit is in low power mode awaiting wake up. So I need some sort of latching capabilities. I do not want to use a relay, since this is powered by battery. Is there an IC out there that will do something like this?

As usual, thanks in advance.

Regards,
Karl Tomecek

[Chris Smith]

SCRs and Triacs

The SCR will latch on for a DC power through, and stay on with DC going through it, and a tiny DC on to turn it on.

To shut it off you can drop the power, or other methods.

Ac power through it will make it collapse with out holding the gate on.



The triac will turn on with Full AC or DC going through it, using a small signal.

However, the Triac fails if the gate goes to Zero.

If you need to control AC and for it to remain on, use the SCR to latch on the Triacs gate so it stays on.

[philba]

use a logic level P channel MOSFET in series with the V+ feed. The gate on the P-MOSFET should be pulled high via a resistor to the hot side of the P-MOSFET (10K or higher, I'd guess). In addition, the gate is connected to a pin on your uC. The uC, when running, normally pulls the pin low, thus keeping the P-MOSFET on. Finally, your switch shorts the the gate to ground. Leakage current is dependent on the FET you use but is usually quite small.

So the scenario is
a) you push the button which pulls the gate voltage to 0 which causes the P-MOSFET to conduct and power the uC,
b) the uC asserts the gate pin low which keeps the P-MOSFET on
c) when it's time to shut it off, the uC asserts the gate pin high. The P-MOSFET turns off. uC loses power
d) the resistor pulling the gate high keeps the gate off until a) is repeated.

If you need to control a power supply, I'd use a SMPS chip with shut down capability and do something similar.

The SCR approach doesn't meet the stated needs as the power to the SCR must be removed for it to revert to the non-conducting state.

[rshayes]

You may want to add an inverting stage between the microcontroller pin and the gate of the P-channel FET.

When the power supply pin on the microcontroller goes to ground, it is almost certain that the output pins will follow, due to the isolation diodes in the integrated circuit. In general, the pins on an IC can not go higher than one diode drop above the positive supply pin without forward biasing the isolation diodes.

Adding a bipolar or FET inverter stage between the output pin on the microcontroller and the gate of the FET will prevent this problem. The output port of the microcontroller is now high to turn on power. When the output port goes low, power is shut off. As the power supply to the microcontroller falls, the output power will stay off.

You want to use a discrete part rather than an integrated circuit for the inverter. Practically any integrated circuit will have the same problem with isolation diodes as the microcontroller does.

[Chris Smith]

I would use the uP to toggle a control line to turn the circuit off again.

What can this do, change states, add in AC or DC,... + - what?

The SCR/ Triac can do many things, but what can your uP do for the gate?

The SCR can easily toggle on, and be canceled.

No multiple parts needed.

[Bob Scott]

The SCR will latch on for a DC power through, and stay on with DC going through it, and a tiny DC on to turn it on.

To shut it off you can drop the power, or other methods.

Good start Chris. Now show EXACTLY a simple way to remove DC power once the SCR is triggered ON using a uP. C'mon. I want to see your simple diagram.

Betcha can't!

[Chris Smith]

Are you still desperate for that good education?

Did you ever try to find out about the “Plasmonâ€

[Bob Scott]

Just answer the inquiry. What are you babbling about "Plasmons"? I didn't ask you to explain "Plasmons", just how to turn off the DC power with the uP after you trigger the SCR.

[haklesup]

Since you didn't describe the circuit subject to standby, we can rule out solutions that have to di with its inherent design. What you have left is some sort of switch in series with the power source.

Philbas FET solution is suitable and may have the fewest parts but you need to be careful about proper bias of the FET and still need some way to latch it on or off.

If you don't like working with discrete transistors, I might suggest the MAX312/MAX313 or similar analog switch in a 16 pin DIP package. That part has 4 switches (4x SPST) each with ~5ohms series resistance and rail to rail operation. Wired in parallel and activated with the same enable pin, it makes a very nice low resistance solid state relay (SPST). It uses only a few uA on its own and has less than 0.5nA leakage through the switches. This is a very versitile part family based off the older DG411 devices. You can even use it to switch multiple voltage sources (3.3V and 5V for example) or series analog signals (like a current loop for a sensor) just power it to the highest supply voltage.

http://datasheets.maxim-ic.com/en/ds/MAX312-MAX314.pdf

The other part to the design requires latching of the power enable circuit. You can do this directly with no additional circuitry if there is a dedicated pin from the uProc that you can hold high or low for this purpose only.

However if you are addressing this feature by decoding it with a register, that circuit needs to be able to latch that power up or power down signal so that it can serve as an enable signal for your switch. A flip flop will work, whether that be a D type or JK or another depends on the other parts of the design. For example will there be seperate commands for power up and power down or will a single command just toggle the switch. Will there be more than one module to stand by, the addressable method is expandable.

[dyarker]

Put a transistor in parallel with the SCR. The uProcessor turns on the transistor, the current goes through the transistor instead of the SCR which turns-off the SCR. When the uP turns-off the transistor power is off.

Cheers,

[rshayes]

You might prefer the P-channel FET solution. A SCR needs a certain minimum current to remain latched, which could be higher than the load current. Also, the voltage drop across the SCR is higher than a diode, probably in the 1 to 2 volt range, and may be unacceptable with a battery supply.

[philba]

how does the transistor across the SCR work? Would it drop the voltage enough? Basically most of the current would have to go through the transistor for the voltage across the SCR to fall to zero. right?

Interesting solution though for higher current situations, it might be fairly expensive.

[Chris Smith]

You might prefer the P-channel FET solution. A SCR needs a certain minimum current to remain latched, which could be higher than the load current. Also, the voltage drop across the SCR is higher than a diode, probably in the 1 to 2 volt range, and may be unacceptable with a battery supply.

5 ma hold current, or one tenth your requirement.

A parallel transistor can short out [ or cancell] the feed voltage, causing the SCR to fail, or the micro can reverse feed the gate voltage to also cause a failure?

[dyarker]

Actually, I like the P-MOSFET idea better if battery voltage is close to Vcc.

If battery voltage will be high enough that forward drop of SCR isn't important, then Chris' SCR suggestion is okay too.

Bob asked Chris twice how to turn off the SCR. Chris doesn't know so he went with Alice to Wonderland, babbling about plasmons and our education.

So, I answered the question.

The voltage across the SCR doesn't have to go to zero. The transistor only has shunt enough load current so that the current through the SCR is less than it's holding current. If satuation voltage of transistor is less than forward on-state voltage of SCR it will happen. An SCR is a 4 layer device, so it's forward on-state voltage is about the same as two diodes in series.

The transistor only needs to be on for a few microseconds, so you can select transistor by peak current rating instead of continous, and a separate heaksink usually isn't needed either.

Chris is right about 5mA hold current (of some SCRs).

"A parallel transistor can short out [ or cancell] the feed voltage, causing the SCR to fail, or the micro can reverse feed the gate voltage to also cause a failure?" is nonsense. The SCR and transistor are in series with the supply and the load.

---------------------------------------
ktomecek,

What is the battery voltage you plan to use?
And Vcc of the uP?a


Cheers,

[ktomecek]

Dale,
They are both 3.3v dc.

Thanks,
Karl