Multiplexing

[Richardcavell]

Hi all,

I am a novice self-taught electronics enthusiast. I have been considering how to send x number of digital data lines into a microcontroller that has y data lines, where x > y. The way I've chosen to do this is to use an output line to turn on one of two transistors that are mutually exclusive. That will connect either M or N to the relevant data line. This idea can be propagated across all the data lines, thereby effectively doubling the number of possible inputs at the expense of one output data line.

1. Is this the basic method by which multiplexing is done?
2. There will be a propagation delay through the NOT gate, but not on the M side of the circuit. Should I be worried about this?
3. There will be a current from base to emitter through the transistor that is active, even if the current-emitter junction has no current. This will mean that the data line will always read as true unless I account for it. How do I account for it?

Thanks in advance,

Richard

 

[Richardcavell]

Hi all,

This is following from my 'Multiplexing' post in which I try to figure out how to select one of two pathways to be placed into a circuit. I was imagining that the X-Y part of this schematic might be part of a resistor ladder, where a resistor is either swapped in or out of a series of circuit-parts. I have attempted to sketch, using resistor-transistor logic, what this circuit might look like.

1. Is this the correct RTL implementation of my previous schematic?
2. Is R5 necessary?
3. What do I do about the fact that from R4 to Y there will be a considerable base-emitter current, that can add significantly to the current flowing through R1?

Thanks in advance,

Richard

 

[Les Jones]

Hi Richard,
It would be easier to use a data selector IC such as a 74HC157 rather than build it from transistors. This would combine 8 inputs into 4 outputs (Inputs to the micro.) Yet another alternative would be a CD4051 which would combine 8 inputs into 1 output. This would only use 4 pins on the micro to read 8 lines rather than the 5 your method would use.

Les

 

[Pommie]

You need to google (and read about) multiplexing.

Mike.

 

[dr pepper]

Your idea might well work, however have a look at something like a 4094, a serial in parallel oit shift register, you can daisy chain 4 of these to get 32 o/p's from only 2 pins on the micro.

 

[granddad]

Cmos 4067 will do 16 to 1, both directions, analogue or digital...

 

[AnalogKid]

You are on the right track, working through the various issues that came up as analog and digital multiplexing were developed over many years (the first analog mux patent is from the 1890's). Do you have any interest in the many mux ICs that already exist, or do you want to pursue your circuit development?

On you schematic you have an upward-pointing ground symbol (strongly not recommended) labeled as +5V (totally not recommended).

Be more clear about what X and Y are.

ak

 

[Richardcavell]

Hi, AnalogKid.

I definitely want to pursue this circuit, as reinventing the wheel is how I learn. If I solve it with ICs without understanding how to do it using discrete transistors, then I feel like I'm cheating. Sorry about the +5V/GND debacle.

X and Y are simply two parts of a circuit which are connected via either R1 or R2, depending on the input.

Richard

 

[AnalogKid]

1. Is this the correct RTL implementation of my previous schematic?
2. Is R5 necessary?
3. What do I do about the fact that from R4 to Y there will be a considerable base-emitter current, that can add significantly to the current flowing through R1?

1. No. https://en.wikipedia.org/wiki/Resistor–transistor_logic
You show Q1 and Q2 as transmission gates. In RTL, all transistors are NPN and all emitters are grounded.
2. No.
3. That's just part of the fun of designing with bipolar transistors.

In general, there are two ways of multiplexing signals, transmission gates and digital gates. In its most simple form, a transmission gate is a MOSFET in series with the signal, like a light switch. Actually, like a series resistor that changes from around 5 ohms to 100 megohm. For high isolation there is a second gate from the output of the first one to GND, driven by the opposite logic polarity so one is on when the other is off, forming an L-pad. This works with both analog and digital signals.

The other way uses digital logic gates and multiplexes digital signals only. The incoming signals are not passed through to the output; they are reconstructed at the output based on the control signals. For the inputs that are not selected, at some point they are shorted to GND by the control signals. The multi-transistor NOR gate example shows a version of this. If you call A a data signal, and B and C control signals, you can see that when either B or C are ones, changes in A do not appear at the output.

ak