demultiplexer design troubles

[krivan]

Hi folks,

I have an application where I have to have an analog 4:1 multiplexer and an analog 1:4 demultiplexer. Basically I have to design some switching logic to connect 4 transistors to two amplifiers.

So I designed the multiplexer with a conventional transfer-gate logic, it works fine.

However, when I want to use this circuit in reverse mode [see pic1] somehow I can't select one specific output. The input voltage appears on the right output but lower voltages appear on the other outputs as well. These voltages are lower than the input one.

I tried also to have 4 transfer gates and switch on/off them using a 2-to-4 decoder but the result is the same, although the control voltages of the transfer-gates are provided fine[from the decoder].

So I don't understand. There is one 4:1 MUX which works OK. I wanna use a slightly modified copy of it [rotated the transfer-gate blocks because the input/output ports] in reverse direction and it does not work however I use exactly the same transfer-gate blocks.



Does anyone have idea how to design then a proper analog demultiplexer? 1:4

Thx,
Krivan

 

[manson]

Hi fellow,

I would be able to make better judgement if you had sent your waveforms, but
whatever I surmise that the problem lies in the output capacitances of your TG switches. It is quite normal, when the switch turns off pevious voltage remains in the abovementioned capacitances.

 

[saro_k_82]

There is a problem in trying to use the mux as demux in the reverse direction. When it works as a mux, all the voltages in the circuit are well defined, driven by low source impedances (TOP and DOWN)., but when it works as demux, the nodes gets driven or left floating. So they hold whatever they held sometime back or what they acquired during transition.
So if you want the other outputs to be quiet, you need to expicitly drive zero volts when they are not needed. You cant expect them to behave nicely when they are floating.
You could chuck the first two TGs., tie input side of all the 4 TGs to the input and use a one-hot decoder to do the job., in addition to providing a ground path to all the outputs when not in use.

 

[krivan]

Thanks manson and saro_k_82 for the replies, they really helped.

I attached the simulation results of the DEMUX.

Input : MUX_OUT [as I want to use the MUX as DEMUX]
Code: 00 [PHI1, PHI2]
Selected output: 0 [at the bottom in the picture]

So here you can see what I talked about.

Saro_k_82: yes sure but why? I mean that in the DEMUX case I use the TG blocks the same as I used them in MUX case. These are just simple transmission-gates, they should behave the same in both cases. They have an input voltage at their input and should produce the input voltage, or tri-state at thier output depending on the control voltage.

So, at the moment I don't see the physical reason why they float in one direction and why they don't in the other. Can you please explain?

 

[saro_k_82]

In the mux case., all the input side of the TGs will be driven by low source impedances (there are 4 inputs). The other side of the TGs are driven by the TG's that are ON. The output will always be driven by the TGs. So no node will be floating.
In your demux., 3 of the 4 outputs will always float. One of TOP and DOWN nodes will float.
Even though a TG is bilateral., you don't use them here symmetrically.., the outputs should drive a load and input gets driven.