Last update: 6:23 PM 2/20/2008

Home Page.

Back to Projects.

Demultiplexor Page.

I'm building a prototype of a possible Integrated Circuit, one version seems to be like a demultiplexor ("demux"), and the other a decoder, but I'm going to call it a demultiplexor because it sounds more high-tech. Oh, and it takes longer to type. Big plus (note the sarcasm).

(their functionality is decoder-ish-ing, but I'm going to use it for demultiplexing, if it makes sense.)

The Wikipedia articles have lists of chips with demultiplexors built-in, so why reinvent the wheel?

Because I can improve it, that's why!

Oh, and don't forget the fact that I haven't found them at any electronics store near me.

MY GOAL FOR THIS PROJECT: To make a special-purpose chip into a more general purpose one, via the publication of non-standard uses for it, thus increasing its market demand, so that MY specially-built-for-general-pupose-use-chip can fill that market niche, thereby replacing the other, harder to find chips.

I've had to put this project into cold storage, because of the BEST competition. See the BEST page for a little introduction to it. Some actually useful information can be found at the BEST website.

I recently compiled a general, but very technical overview about the TTL DU-1B prototype I'm trying to get working as a proof-of-concept. As usual, I had ulterior motives for this article, so I slapped it here and added some stuff, so that I can have something more specific to show, instead of "I'm working on it".

DEMUX Intro:

I am developing a prototype demutiplexor using discreet components. Once I get the bugs worked out, this will hopefully be presented as a possible IC (Integrated Circuit ) to various semiconductor manufacturers so that I can start my own business selling part designs.

After full mass-production, designation standards will be as follows:
Du-XXL (Substitute XX with either single or double numerical digit(s) indicating the particular version number, and substitute L with buffer arrangement type). Du stands for �DUstin's demultiplexor�, which is pronounced �Duh�. No, I'm serious. Really.
This system has a specified number of inputs, along with an �S�-input (S for �special�, indicating a separate function from the other inputs), and a large number of outputs (which can be found using custom liner equations, shown later).

Currently, there are two types of buffer arrangements, which are actually single-letter acronyms:
B: �Buffered� for greater output source/sink current. Possesses PP�s (explained later)
A: �Ack, no buffers!� A humorous acronym, signifying that it lacks PP�s for indirectly powered operation.

Elaboration of designation standards:
[A PP is an acronym for a Power Pin, a pin on the IC package specifically dedicated to receiving power from an external source.]
If the device possesses PP�s, current source and sinking capabilities are increased dramatically (this variation is designated with a B for Buffered outputs).
If the device lacks such PP�s, power is supplied via input, output, and S-input excess, allowing (largely) non externally powered operation, which can be advantageous if there is no room for extra power supply routes on PCB�s (Printed Circuit Boards). The disadvantage to this approach is that current source/sink capabilities are limited to what is supplied, which, in turn, may be problematic in situations where available power is miniscule, leading the user to require extra support circuitry. Device function description:
Depending upon the binary pattern on the inputs, S-input is connected exclusively to a single output with the matching binary address. During normal operation, there are never more or less than one output connected to the S-input, while all others are disconnected.
Usually, demultiplexors are used to split an already multiplexed data stream into it�s regular, demultiplexed form, hence the name.
For more information on multiplexing, see: www.en.wikipedia.org/demultiplexor
The reason for developing such a system is that it will extend the possible number of outputs available to common reprogrammable microcontrollers. This may prove to be its biggest target market.

Construction Story:

Last year, I researched prototype construction because I found it to be fascinating.
This was originally intended to be a punch card belt reader, (reminiscent of later versions of the British �Colossus�, the 1940�s forerunners of the modern computers) with eight Cadmium Sulfide photoresistors reading the position of holes in the punch card, sending out an eight-bit binary code, where a series of sub-circuits looked for their own binary combination, signaling a high when they encounter it.
This would require close to a thousand parts, way out of my budget, so I did some research into the binary system with all of it�s mathematical precision, drawing chart after chart, discovering ways to reduce the necessary number of components to a feasible level.

Due to the highly ordered nature of math with real numbers, I was able to find shortcuts that allow me to use fewer parts while maintaining, even gaining, functionality.

At this point, I went to a smaller scale, reducing the number of inputs from 8 to 3, more than dramatically reducing circuit construction complexity while still retaining proof-of-conceptability (very important) and ran a few computer simulations (YAY for Falstad.com!). The results were encouraging. After this, I set out to construct the first prototype to see if this system will remain plausible in real life.

Presently, my first prototype is constructed out of transistors, resistors, and LED�s (Light-Emitting-Diodes).

FET�s or MOSFET�s (Metal-Oxide-Semiconductor-Field-Effect-Transistor) would have yielded the desired results immediately, but, considering my current budget, this was not an option.
I chose transistors for the sole reason that I was able to acquire them at a very acceptable price, as opposed to FET�s or MOSFET�s.

I managed to get the logic core to work, however, the TTL (Transistor-Transistor Logic) input buffers became problematic, causing input cross-talk that spanned across all stages of logic control, preventing proper operation. This problem is easily fixed.
The next step that remains is to take the input buffers out of the loop, and replace them with CMOS (Complimentary-Metal-Oxide-Semiconductor) drivers.
This will prevent any transistor from having direct, or indirect contact with each other.

Another, thankfully minor, unforeseeable problem that occurred, was that the combined Vdrop (Voltage-Drop) of the stacked transistors in series added up to the point that it required 9v (Nine-volts) to operate the LED�s properly (5v was the target operating voltage). This Vdrop problem would not have arisen had I used MOSFET�s

My choice of CMOS IC�s are limited, due to the raised operating voltage (Most IC�s operate at 5v.) I am tempted to start again, using another approach: pure logic-gates, but this seems to be risky, involving many IC�s instead of discreet components, as was the goal. Perhaps I will construct both versions and compare them side by side.
Nevertheless, the first action that I will take next will involve getting the TTL prototype functional. While this is not essential, with the fact being that the final product will be CMOS, I have already spent a good deal of money and effort into the TTL prototype. Once functional, I will evaluate the overall behavior of the device, and determine more uses for it, so that I can sell it to a wider market.