I think you are getting to the point where 1 ounce (28g) of experiment wil outweigh a pound (453g) of theory
If your are going to do some level of multiple unit production (your posts seems to indicate this) then I definitly would buy components for trying several circuits. Especially as you do not yet (as I understand it) hold a stepper in your hand.
My favorite references for the steppers : here and here (the last is by GrumpyMike a helpful & productive user of this forum)
The L293D you refer to is described as "Quad Half Bridge" on the datasheets which means two full H-bridges in the circuit shown. I have used the L298N (the D/N suffix is a package difference) or plain transistors in my circuits.
Both uni- or bi- steppers are "simple" for their circuits, but twice the component count for a bipolar, unless you use a "good" chip then it is just a count of 1 (well, they usually require som external capacitors...). You can connect a unipolar as a bipolar, some people say that gives more torque. YMMV.
Cheap ... and you're using an Arduino? Well for protyping. (that is with a or two)
Anyhow, your question is so openended at the moment I can think of too many answers. And I played a bit with the concept (cheap Stepper control) in my mind, too, some time ago. And my conclusion (for what it is worth) is that all I need externally is the driver transistor/MOSFET, the rest I want(ed) to do in software. Keep in mind though, as "robtillaart" (another forum users) has as his motto: In theory there is no difference between theory and practice, however in practice there is ...
A unipolar stepper only requires four drivers (ie transistor/MOSFET, possibly with a resistor) - cant get much cheaper than that. Oh and the flyback/backpulse Schottky diodes if the stepper is "big". And 4 pins on the Arduino.
Software wise it is "easy" to drive the lines for full and half step. Quarter and other microsteps requires either a more sophisticated driver circuit to supply half-current or a PWM-like. Ah- the Arduino has that, too. I just have not tried it. Well, I did a quicky-test of a unipolar stepper with full&half and the software worked.
A Bipolar stepper must use two H-bridges. Again, it can be made with 2x4 transistor/MOSFETs (and diodes) if you trust your software skills never ever to send a signal that make a short from PWR to GND. Otherwise a simple inverter in the circuit can guarantee that, and then you only need two pins to control each H-bridge. Same considerations for the quarter/micro steps as for Unipolar.
Lastly, there is the concern of power. To get the steppers to react faster/stronger you want a larger voltage, but to avoid sending too much current you need to current limit. That is where the fancy motorcontrol chips start earning their worth. Maybe the PWM of the Arduino can be good enough.
and a note on mechanics (for your plotter): Anything that is not driven directly from the stepper is liable to mechanical slack or backlash. Your mechanical design is more important then the electronic/software if you want the plotter to draw accuratly. Doing that cheap is the real challenge. I have dissambled a few scanner/kopiers and the ingeniuity in the cheap designs to overcome this is interesting. F.ex. like a scanner only needs to be accurate when moving in one direction (it wont scan on the return stroke), or by including more feedback measurments of actual position (seen in most inkjet printers)
If you just want continuity, then you can join them all at one end, and then you only need a one-to-50 at the other end. But you probably want to know about shorts, too.
type of shift registers or multiplexers
So just use the little search box (top right on every page) on this forum for "shift registers" or "muliplexers" and you get plenty of ideas from others who have done it before and already written about it. If you have specific question ask them in the LED and Multiplexing section
As for the project in general : Yes, enough 1-to-8 multiplexers to get your 100 lines, and a simple program with two nested loops (to test every-one-against-every-other wire should do it. Hint: You set all pins to HIGH on the one side of the cable, except one to LOW. Then you probe for the LOW one on the input side. By using the internal pullup resistor no more parts are needed. (I did a 15 minute program for working out a weird thumbwheel switch I picked up as junk. It only had 8 leads, so I did not need a shiftregister). Your program only needs to write the unexpected result, and you can save the LCD by using the Serial/USB/computer (unless you wnat a standalone test box)
Oh dear. I bought something a few weeks ago and forgot to post here.... well, better late than never. Wanted to play with servos, found the cheapest possible ones in town (avoiding shipping costs, thus total price was cheaper) bought 3 different sizes.
After making one jerk with the standard tutorials I made this (see comments on youtube)
Now, when I am bored I continue the inverted kinematics problem (ie making it move in a straight (cartesian) line between two points at an even speed).
Lets see .... 9V makes the Servo go HOT ... is it wise to use 9V ?
On the other hand, if you got more time/money, you could put in a 6V regulator. OR You could power the Servos of the 5V pin on the Arduino, the onboard regulator has lmited the 9V to that. It can not take unlimited power, so depending on your size of the servos it mau fail occasionally as a servo takes too much power making the Arduino reset or "go funny". OR You do two batterypacks - one that does 6V which only goes to the servo + and -, and another that does at least 7v and goes to the Arduino power jack (and which only needs a small current, ie the 9V battery will do). Remember to connect the Gnd(-) of the two battery packs together.
Subject is just instead of the "small matter of programming" and "simple yet powerful" phrases. Anyhow, to the actual subject:
On some (big expensive professional) sites, when you enter a new question or request for assistance, the system returns a search list of 10 "hits", either of previous posts or of a FAQ. It derives the search terms from the submitted question (presumably, though I do not know for sure, detecting popular search term/words are in the question).
The idea of course is that there is no need to submit the question, saving everybody's time (the would-be-poster gets an instant answer) (And thinking about another related thread - it should always include "BlinkWithoutDelay" )