Over-engineering

[DKS]

You know everything you guys are saying sounds about right for the MR forums, but the one time I posted there I most have hit the jackpot I guess.

I wanted to make a loop to loop layout using tortoise switch machines. When the locomotive hit the loop I wanted a photo cell to throw the turnout, then I wanted the contacts on the machine to swap the polarity of the mainline.  The 2 machines had to work together to make it totally automatic.  Probably easy, but I didn't know...

Anyways one guy replies how it can be done with some 555 timers  ??? and long story short he completely went out of his way to draw up a circuit, then he breadboard tested it his self to make sure it would really work, then he sent me 10 pages of step-by-step hand drawn instructions with pictures and sent some harder to find components.

I was wading through the MR Forum topic [http://therailwire.net/smf/index.php?topic=14456.30] and came across this. 555 timers? Breadboarding? Hard-to-find components? I did this very thing (loop-to-loop, Tortoise switch machines, automated with sensors) for my WR&N with just a few relays. I find it interesting how some electronics guys tend to over-engineer stuff. Rick Spano has a double-ended yard powered with Tortoise switch machines that he wanted to control with a single rotary switch. One EE told him it couldn't be done; another said he could do it with a bank of logic circuits that would push ~$100 in cost. I did it with just a bunch of resistors and a home-grown power supply, no electronics at all. When he showed the finished system to a couple of EE's who were visiting his layout, they couldn't believe it. And I have zero background in electronics.

I guess I just had to toot my horn a little, here. Sorry, guys.

So, anyone else have over-engineering stories?

[asciibaron]

So, anyone else have over-engineering stories?

i worked for the federal government as an electronics tech - one of my projects was to design an enclosure for a device that was waterproof.  of course i had no clue what was going in the enclosure or what kind depth the water would be, i purchased hermetically sealed connectors for the hookup lines, devised a lip seal, and had the enclosure milled from a solid block of aluminum. 

when it came time for the EE to put the components in the enclosure, i was on vacation.  when i came back, the $10,000.00 worth of connectors and $5,000.00 block of aluminum had been rendered pointless - the EE drilled holes in the enclosure for the hookup lines and used RTF to seal the holes.  when i left a few years later, the connectors were still in the box on my top shelf.  your tax dollars...

-bob

[Chris333]

Where were you back then!     

I just looked for 10 minutes and couldn't find the papers he sent me. I know I kept them though. The circuit had a relay, 2 555 timers, some other bigger IC, various resistors and capacitors, and 2 pots to adjust the sensitivity of the photo cells.

[DKS]

Where were you back then!     

I just looked for 10 minutes and couldn't find the papers he sent me. I know I kept them though. The circuit had a relay, 2 555 timers, some other bigger IC, various resistors and capacitors, and 2 pots to adjust the sensitivity of the photo cells.

On the WR&N I used proximity detectors, which sense the presence of metal, such as a loco chassis. They were kind of expensive (~$20 each); now I use IR detectors from DigiKey, less than two bucks apiece. They come with built-in discriminators, amplifiers and output drivers that can handle small relays directly, all in one compact package that can fit between the rails. They also require no sensitivity adjustment. To make a loop-to-loop control system, you'd need:

2 IR LEDs
2 IR detectors
1 latching SPDT relay
2-3 standard DPDT relays
1 5VDC power supply

The circuit is really simple. The detectors fire the latching relay, which sets the mainline polarity and also throws the switch machines via the standard DPDT relays. Both turnouts are thrown at the same time. What's cool about it is that the latching relay always remembers which way the line was running, even after powering down the layout.

To give you an idea of what the circuit might look like, check this out: http://whiteriverandnorthern.net/detail_39.htm

Note that the whole layout control panel circuit is shown, which includes a lot of extra stuff you wouldn't need.

[SquirrelHollow]

I over-engineer far too often.  In search of the 'perfect' device, I ignore all the simple stuff. 
I always double, triple, or quadruple my component count.  When it's finished, it works exactly as I wanted it to.

Then a few weeks (or maybe just a few hours) later.... I realise the shortcomings of the simpler circuit would have been so minimal, it wasn't even worth the extra effort I had put forth.


One of my favorite over-engineering experiences relates to armor flooring for our helos. (also, plenty of old fashioned ignorance)
In 2005, the Kevlar blanket system we were using was determined to be ineffective to calibers larger than .30".  Because of the work we were doing in a certain less-than-friendly country; A $2mil+ contract was awarded to an unnamed company, under some kind of emergency acquisition program that ignores pricing and standard contracting practices.

Their first prototype was a tongue and groove, sectional armor system.  Each 'section' of 3 pieces would be held down with 1 bolt in each panel and 2 brackets on the outter edges.  Well engineered, tough as hell, light weight, easy to install, except....   Every aircraft is different.  They had designed it for one specific tail number.  As soon as they tried installing the 3 panels running from one side to the other, at the back of the aircraft, they ran into their first revision.  (The panels were 1/2" too wide)  They failed, however, to check the rest of the system.

So...  Phase II armor was tested...  This time, they had determined it was best to start at the front of the aircraft, due to design changes to the tongue and groove placement.  Three pieces went in fine.  The next 2, not so well.   The idiot engineers had decided to base this revision on the Marine HH-53D, instead of the recipient helo; MH-53M/Js.  J and M models have a 24"x42" avionics rack on the forward left hand side of the cabin, from floor to ceiling.  The floor panels required removing the rack.  That is not an option.  (Even if you don't need the avionics, it is considered a structural component.)

Phase III came along a few weeks later.  Now they had completely removed the tongue and groove style installation.  The panels were supposed to be installed with a single bolt, and remain free-floating.  They even added a 'sealant and non skid coating' to help avoid deterioration due to oil leaks.  The panels went in like a breeze.  Kudos to the engineer that took measurements, prepped the aircraft, and finalised that part.  Then of course... in flight testing....  5 people went to the hospital after slipping on the 'non-skid' coating.  The floor's free floating aspect also created so many vibrations that it was completely rejected by our unit.

Phase IV...  This is the golden donut.  One of the best days of my military career. 
The only changes for phase IV were a new MIL-spec non-skid coating, and a correction to the vibration issues.  Now, instead of the panels being bolted down to the floor...  48" sheetmetal panels were bolted to the floor with the armor panels velcro'd on top of them.  So, the bolts holding the metal panels were covered by the armor sections.  Not only that, BUT, the ENTIRE contact surface of the armor and sheetmetal pieces was velcro.  I warned them....  I warned them again....  They said it would be okay.

Two days later, after REPEATED, futile attempts to get the panels back up....  We had to take a gas-powered circular saw to the floor and cut a 12" square hole in 3 of the panels to get them up.  During the removal of the rest of the panels, every single section of the actual aircraft floor was ruined.

The engineers didn't understand why I kept laughing at their situation.

[SquirrelHollow]

FINALLY!!

I've been working on this damn thing for far too long now.  I can't say it is finished, but all the vital circuitry has been tested.  Everything operates as planned, and has timing within 0.03 seconds every 5 minutes.

So, how many ICs does it take to flash some LEDs?

7 - minimum, to test the circuit.
13 - for the finished product.




Unpictured, but included in the finished product:  44 more LEDs, and 6 more ICs to control them.  Adding them to the circuit is a piece of cake.  They just cascade into each other.

Incase you were wondering, the final result will be a clock. 

It gets its timing from the 60Hz signal in the 115VAC to 12.6VAC power supply.  Three of the ICs are dedicated to decoding the signal, and passing a single clock signal to the 8 bit counters controlling the LEDs.


This was my first experience with a ripple counter, and I really had no idea what I was getting into.  I still don't understand many of the signal principles (inputs, and outputs both), but I managed to get it to work with the other chips with some advice from electronics geeks.

The fun part is still ahead of me, though...
I get to plan the PCB, get it printed, and solder all 137 components to it.  I can't half-a$$ this project since it's going inside a permanent enclosure, and I'd like it to keep fairly accurate time.

Over-engineered... I think so... a bit.

[DKS]

Rick Spano has a double-ended yard powered with Tortoise switch machines that he wanted to control with a single rotary switch. One EE told him it couldn't be done; another said he could do it with a bank of logic circuits that would push ~$100 in cost. I did it with just a bunch of resistors and a home-grown power supply, no electronics at all. When he showed the finished system to a couple of EE's who were visiting his layout, they couldn't believe it. And I have zero background in electronics.

Here is that Tortoise switch machine yard ladder control circuit, in case anyone was interested: http://whiteriverandnorthern.net/clinic_38.htm

[Sokramiketes]

You know everything you guys are saying sounds about right for the MR forums, but the one time I posted there I most have hit the jackpot I guess.

I wanted to make a loop to loop layout using tortoise switch machines. When the locomotive hit the loop I wanted a photo cell to throw the turnout, then I wanted the contacts on the machine to swap the polarity of the mainline.  The 2 machines had to work together to make it totally automatic.  Probably easy, but I didn't know...

Anyways one guy replies how it can be done with some 555 timers  ??? and long story short he completely went out of his way to draw up a circuit, then he breadboard tested it his self to make sure it would really work, then he sent me 10 pages of step-by-step hand drawn instructions with pictures and sent some harder to find components.

I was wading through the MR Forum topic [http://therailwire.net/smf/index.php?topic=14456.30] and came across this. 555 timers? Breadboarding? Hard-to-find components? I did this very thing (loop-to-loop, Tortoise switch machines, automated with sensors) for my WR&N with just a few relays. I find it interesting how some electronics guys tend to over-engineer stuff. Rick Spano has a double-ended yard powered with Tortoise switch machines that he wanted to control with a single rotary switch. One EE told him it couldn't be done; another said he could do it with a bank of logic circuits that would push ~$100 in cost. I did it with just a bunch of resistors and a home-grown power supply, no electronics at all. When he showed the finished system to a couple of EE's who were visiting his layout, they couldn't believe it. And I have zero background in electronics.

I guess I just had to toot my horn a little, here. Sorry, guys.

So, anyone else have over-engineering stories?

Not to rain on your parade... .but I'm sure you got the idea for resistors and susequent grounding of pins 1 or 8 on the tortoises from an article in the railroading press at some point.  So did you "engineer" it yourself, or just happen to have the proper prior knowledge? 

But, I don't trust many EE's either.  Overall that whole profession seems to over engineer, or pack 10 pounds of features in a 5 pound sack, when you asked for 1 pound of usefulness.  An ME can design a mechanical solution to anything an EE might create... but an EE can't do much other than move electrons.   

[Chris333]

Didn't David have an article in MR about how to do this?

Maybe he learned it from reading his write up?   

The MR Mag index can't seem to find the issue, but I know I have it. Around 1993ish.

Edit:

Found it Nov, 2000 MR "Route control for slow-motion turnout motors"

[DKS]

Not to rain on your parade... .but I'm sure you got the idea for resistors and susequent grounding of pins 1 or 8 on the tortoises from an article in the railroading press at some point.  So did you "engineer" it yourself, or just happen to have the proper prior knowledge?  

Hey, thanks for your vote of confidence, Mike. Actually, the idea occurred to me based on work I'd done some years before on the first control system for Rick Spano's Hyde Yard (which is now on its third incarnation and hopefully will be the subject of an article soon). I was using a logic circuit that switched output states between high and low; the output required a pull-up resistor, because the high output of the IC chip was not sufficient to reliably be read as high. But when the chip switched to low, it was a solid low, and the resistor provided a load so there was no short. I picked up this very basic and widely-used principle from a Popular Science article at the time (it was the 70s or 80s) on a DIY IC alarm clock project, and that one article provided me with a gold mine of practical education in digital logic principles.

The pull-up resistor concept was lurking in the back of my mind when the challenge for controlling Field Yard was raised by Rick. I had no idea if it would work, so I developed a quickie test circuit as a proof-of-concept. I did not derive the idea from any prior modeling article, nor did it come from any circuit I encountered that was designed for similar applications. So, it was adapted from a standard principle, but it was not based on any design expressly created for this purpose.

So, did I engineer it myself? Since all electronics circuits are based on known principles, one might argue that no one engineers anything; they just adapt and re-use. But I contend that, since it is not based on any existing circuit to my knowledge that does the same task in the same way, I would tend to think that perhaps I did engineer it myself, at least to some degree. Of course, someone else might have done the same thing already; simultaneous development happens all of the time. But I don't know of anyone who has. (I've also never seen an asymmetrical dual voltage power supply like mine, either, but that doesn't mean I invented it; it just means I designed one for myself through creative adaptation.)

Oh, and the color illustration in MR is wrong; I made a mistake in the artwork I sent them. The schematic, however, is correct, as is the color illustration of my online version.

[SAH]

 An ME can design a mechanical solution to anything an EE might create... but an EE can't do much other than move electrons.   

Hmmmm.  In my business a project invariably gets to the point when increases in production efficiency are no longer possible until programming changes are made to overcome mechanical design deficiencies.  Just moving electrons around?  I think not.   

Steve

[SquirrelHollow]

I always think the best compromise is having an EE on hand for speed bumps the ME runs into.

Working around the problem often only serves as a bandaid, in my experience.  I'd much rather have a knee with a new joint, than a knee with an ace bandage and no cartilage. 


Holy Sardine! 
Pass .. me ... down the ... Electrician .. repellant Bat-spray!

[Chris333]

speed bumps the ME runs

Are you thinking of Atlas c-55, ME doesn't have speed bumps 

[wm3798]

How about some civil engineering?  When I started my layout, I was relegated to the garage for the early stages of construction.  My plan includes a lower level staging yard with a one-turn helix to rise about 3", then a lift bridge spanning the aisle, which starts 45" out from the door due to the location of the chimney, the main then wraps around the room, folding back onto that same first section, then diverging into two separate routes that end up rising or falling, depending on which route you take.  All this had to be tucked under a sloping attic ceiling, stay within a 36" wide table top, and rise to approximately 42"...


At the time I was working with little more than a level and a tape measure, and I was building the framework 75 feet and three storeys away from where it would be installed...  Plus, to make it even more exciting, the early measurements were taken in a room that was under renovation, so in addition to setting baselines and levels, I had to calculate in the additional thickness of drywall that wasn't there, door and baseboard trims that weren't there, and a bunch of other variables.


When I finally had the room done, and lugged the first section up for installation, everything was right where it should be, within about an 8th of an inch.  Using leveling screws on the legs (a 'la Ntrak) I recaptured that within a minute.

You electrical and mechanical guys want to take all the credit, but if you're not working in a solidly built environment, you can't very well work all your magic. 

Lee

[SAH]

I always think the best compromise is having an EE on hand for speed bumps the ME runs into.

Fortunately that's the way it usually works in the real, for profit world.  And the best case real world scenario unfolds when the EE and ME don't snipe at one another. 

Steve