I'm working on a circuit that uses an MCU and I've run out of I/O pins. I'm therefore trying to multiplex candidate signals in order to make the circuit conform to the maximum pinout limitation.

Here's exactly what I want:

• I need a circuit that will redirect a clock output (working voltage 3.3VDC) to either Out1 or Out2 depending on the state of CS.
• The circuit in question must draw zero, zilch, nada current when in the idle state.

​

The simplest circuit I've found so far that does what I want is a little too complicated for my taste. This because I'm thinking that the only way for it to work with zero current draw while in the idle state would be to use an impractical amount of discrete components:

​

Any suggestions?

 

link
SN74AUC1G19 1-of-2 Decoder/Demultiplexer
Sorry the link is for a 2.7 volt version. I know there is a 3.3V version.
---edited---
link2 There are many versions. Try this one.

 

link
SN74AUC1G19 1-of-2 Decoder/Demultiplexer
Sorry the link is for a 2.7 volt version. I know there is a 3.3V version.
---edited---
link2 There are many versions. Try this one.

Thanks for the recommendation, Ron. I'm looking into it, but I'm not sure I fully understand the device's power specs. It seems that it draws 1 µA while idle (which is acceptable for my application) and 10 µA while at work ... would you say that is correct?

 

Supply current is hard to understand from that data sheet. You might also look at other Companie's data.

The part should have very low current. It draws current on the clock edges. So a fast clock, 50mhz, will pull power and a 50hz clock pulls much much less.

Also if the inputs are at about 1/2 supply it will draw too much current. Keep inputs near 0 and Supply.

 

There are a number of "1G19" parts that functionally are the same thing. There are parts for 5V or 3V or 2.5V ....... and there are parts for fast/high power and parts that are slower and lower power. You might search for other versions that are lower power.
Need help, ask.
I was trying to find a small package version for you. There are 5 to 8 pin logic gates that I use when space is important, and I do not want to use a 16 pin version.
RonSimpson

 

What is the clock frequency?

 

What is the clock frequency?

Rather low, less than 100 KHz

 

It looks like CS is asynchronous with respect to the clock signal, and the clock is shown as 50/50. If so, I don't see any harm in the clock signal being inverted through the switch. This reduces the number of gates from 5 to 3, a good number for a quad package.

Note that the current values on a CMOS logic datasheet are not normal, and are difficult to confirm. The vast majority of the circuit's operational current draw happens at the signal edges. The rest of the time, from microseconds to years, the part really does draw less than 1 uA.

If you can find one with a low enough threshold voltage, this is doable with three p-channel FETs. What are your requirements (not goals) for idle and operating current in both CS states?

ak

 

Thanks for chiming in, AK. Your help is always appreciated

this is doable with three p-channel FETs

Mind elaborating? I have plenty of pFets laying around with gate thresholds that perfectly match the circuit I'm working on.

What are your requirements (not goals) for idle and operating current in both CS states?

Preferably zero power draw during idle (although 1µA is perfectly acceptable) and 10µA during clocking is also acceptable.

 

Here's another possible option, TMUX1247.
It specs a supply current of 4nA, and since it is an analog switch, the only additional power consumed when carrying the 100kHz CLK is the power required to charge and discharge the stray IC signal path capacitances.

 

Here's another possible option, TMUX1247.
It specs a supply current of 4nA, and since it is an analog switch, the only additional power consumed when carrying the 100kHz CLK is the power required to charge and discharge the stray IC signal path capacitances.

Very, very nice... thanks!

As a side note, I'd like to clarify that I AM NOT a lazy googler, it's just that sometimes the intricacy of technical english escapes me, and I am unaware of the keywords I should use for highly specialized searches. In this case, one of such words would be "multiplexer". Thank you all for your help.

 

This one looks promising. Mouser has stock, and it's .9μA quiescent.

 

If the worst part draws 1uA current and they won't guess at the average part. Then it is most likely the average part draws much less than 1uA. The 1uA is where they call the part bad and toss it.

 

Mind elaborating?

Why? After the two parts suggested so far, I got nothing.

I have no idea what they cost, but with guaranteed performance and zero development time, they're worth at least $10 each, if not $20.

ak

 

I have no idea what they cost,

Digikey.com has parts for $0.49 to $1.07 @1 and down to $0.05 at 5000.

 

Here's another possible option, TMUX1247.
It specs a supply current of 4nA, and since it is an analog switch, the only additional power consumed when carrying the 100kHz CLK is the power required to charge and discharge the stray IC signal path capacitances.

Dumb question. The TMUX1247, being an analog switch, can work bidirectionally, right?

 

The TMUX1247, being an analog switch, can work bidirectionally, right?

From my reference link:

Do you want bidirectional operation or is that a problem?

 

From my reference link:
View attachment 262305
Do you want bidirectional operation or is that a problem?

Not a problem ... bidirectional is good, thank you. In fact, I'm going to draw another diagram and show the exact details of what I'm trying to accomplish. BRB

 

It's basically just a CMOS analog switch that conducts equally well in either direction when ON.