Inverter IC for model signals

Have you considered using something like a few 2N2222A transistors?  They will handle tons of current and you can build your inverter.

If that doesn't float your boat, how about one from this family?

SN5406, SN5416, SN7406, SN7416

If you're looking for more current drive than they provide, I'd do the transistor trick.  Really high current IC's are probably going to also be a bit more expensive.

The ULN2803A is used in the MTH AIU to drive the relays.  It is an octal array of Darlington transistors.

Looks like a good choice Dale, didn't find that one.

Here's the ULN2803A at Digikey, less than a buck.

So how much current and voltage do you really need?  Some logic families like 74ACxx can source and sink 24mA.  A hex inverting 74AC04 from DigiKey is 52 cents qty 1.  I'm assuming you don't want to deal with surface mount parts - otherwise there are newer logic families with even higher drive.

Note that while the 555 is spec'd as 100mA symmetrical drive, the voltage drop can be several volts as you drive higher currents. So playing the same specmanship game, 74xx logic families can deliver more current than you'd think.  And for applications like flashing LEDs, you can parallel logic gates where, say, an octal inverting buffer could be a better value per channel.

If the L293 (which is not even hex) is exactly what you want except without inversion, then considering its cost (over $2 at DigiKey), why not just put any 25 cent hex inverter logic chip in front of it? 

I'm curious about your need for bipolar/symmetrical drive. As your searching shows, multi-output buffers with sink-only capability are much cheaper and more readily available.  It also seems that while you might eventually find that nostalgic Sprague device from the last century, I'd guess it will be well over $1. Depending on what you're doing how about using a 555 chip strictly as a bipolar buffer.  At less than 15 cents a piece, a "hex" buffer would be about $1.

Well, the 7406 will dish out 40ma, surely that would do for driving several LED's at full brightness and probably 4-5 at very nice brightness.  It's not necessary to drive all your LED's at 20ma, and many times they're way too bright with that much current.  I also don't understand why you need to source and sink current, what's that about?

Your posts are so much fun to read.

Too bad I don't understand any of it.

I picked up a pair of MTH grade crossing flashers.  It has a string of four LEDs and two 1K current limiting resistors wired from supply to ground.  The 555 astable timer is wired so pin 3 output is connected at the LEDs midpoint.  As the timer output goes high, it effectively shunts the upper two LEDs and its resistor from supply to pin 3 midpoint.  When pin 3 goes low, it shunts the lower two LEDs and its resistor from midpoint pin 3 to ground.  This is same way that the prototype RR operates their flashers with a mechanical relay contact shorting out one lamp set or other to give flashing effect.  

The MTH signals have this 555 circuit in each base so when activated they flash, but due to time differences they quickly get out of synchronization with each other.   If I connect the LEDs mid-point from its 555 pin 3 as a slave and connect to other flasher 555 pin 3 as master, both sets flash in unison.  If each flasher is oriented with their cross bucks facing the roadway, they flash BUT instead of both sides flashing side to side across roadway like the big boys, the lights are 180 degrees out of synch again.  So therefore the mid-point on slave needs an inverter than can sink and source probably 10mA to effectively shunt their respective LEDs.  I thought of simply turning the crossbucks 180 degrees on pole and turning signal 180 degrees, but sign is glued and I don’t want to cut it off and re-glue to opposite side.

So this is why I was looking and searching for an inexpensive inverting source/sink driver.  An LM324 op-amp might work.  There is a way to make a 555 work as a driver, but studying the 555 timer data sheets and looking at circuit sheets that driver hookup is not obvious.

Well, here's a solution.

LM675 High Power Op-Amp

I think I understand the situation.  I'd think you don't want to mess with the supplied LED wiring/harness so I retract the idea of 5V logic which would work (5V can drive 2 red LEDs in series) but would require altering the 1K resistors and adding a 5V regulator.

A simple inverting buffer using 5 cent switching transistors (whatever you have in your junk bin) should do the trick. When pin 3 is "high" driving the master LEDs "high", the NPN is forward-biased (PNP turned off) and the slave LEDs are driven "low".  And when pin 3 is "low" driving the master LEDs "low", the PNP is forward-biased (NPN turned off) and the slave LEDs are driven "high". 

The buffer is powered by the existing supply voltage so no additional voltage regulator is needed.

Or if you already have a LM324, a single pair of resistors could create a V+/2 reference for all the inverters. You would be massively over-driving the op-amps but so what at crossing-gate flashing speeds.  It has asymmetrical source/sink capability but at 10mA, it looks like the voltage drop is about the same so you may luck out.  Otherwise, LED brightness would vary between sides and/or this brightness imbalance would change with supply voltage.   

Yes, it's not obvious, but no external parts required.  How it works is described here.

As an aside, while you undoubtedly have gobs of accessory power, the 555 itself draws as much current (about 10 mA) as the LEDs!  A LM324 draws a fraction of that, and the complementary NPN/PNP pair even less.

Wow! Thanks to everybody for their input.  I think I will use the 555 inverting configuration as I have gobs of them in my bin (I am cheap ya know!).

Will let everyone know how this turns out.

Don't feel bad Denny.  When I lurk on the layout and scenery forums and read how the masters get beautiful layouts in small spaces, and how masters turn a huge layout into a masterpiece in just 8 hours worthy of blue ribbon, while I stand there with my oval of track, paint and plaster dripping on the carpet with moutains looking like melting ice cream sundaes.  :-)

Just a note of caution:

The 2-transistor buffer posted above has one important flaw.  If you don't connect an input, BOTH transistors will turn ON, creating a direct short from the plus rail to ground, and probably wiping out the transistors.  If a control voltage at +V or ground is applied, it works as advertised.  Just don't try any voltages in between!

Yes Dale what we called 'shoot through' that ocurs in switching supplies if the high and low rail transistors or FETs are not allowed dead time for them to turn off before the other turns on (whole chapters and papers explaining this, which is left as an exercise for the student to peruse).  Thats why switching supplies and other configurations have sophisticated driver chips that can be programmed to add delay.  Of course with this flasher circuit application, shoot through is unlikely at slow speed IF the input is immediately high or low at power up.

And after all the finagling with 555's and transistors you'll still have a flasher with instant on-off that looks toylike not prototypical. With a 40 cent PIC I can get prototypical slow on-off, 2 control inputs to give bidirectional control all without any resistors or caps. 

However, you do have to invest the time to learn to program the PIC!

As far as the slow on/off, you can do that with a large cap across the LED and a series resistor to the switch.  This presumes the switch sinks and sources current, not an open collector.

When I fully retire AND have the time and money to invest in an expensive PIC programmer that might be used once, AND learn PIC and assembler/compiler languages over several years, AND build the board that might/not work as expected etc, then I guess I will investigate this route.    I can assume Gary, that you have already been down this route as a computer/software engineer and have the equipment invested so 40 cents PIC is a no brainer.  Where I work the owner has someone working who writes programs PICS.  I can tell you there is a frustration that comes into play especially when it turns out the PIC won't do what they thought and then have to redesign for different device with more pins or memory or a function.  You might say well they didn't read specs closely or understand limitations etc.

Besides I enjoy designing with discrete parts.   I know there are ready to go flasher boards out there for alot of money that has a few parts on them that will do the slow on off and detection, but whats the fun of that?  Each to his/her own I guess.

Sam, I agree with you, especially in light of not being a programmer.

I confess I've used a bunch of different uP products in my career, and I'm seriously considering the PIC for such projects.  However, I've programmed in assembler and C for most of my life, so it's a bit different than someone starting from scratch!  I have a good friend that is up to speed on the PIC, so I'm going to consult with him first.

The programmer and debug system for the PIC is pretty cheap, and you can get a free assembler and C compiler, so those aren't really stumbling blocks.

I am so ancient that I cut my teeth on 8008 and first 8080s, hand coding the little suckers!  Then there were PDP8 which you hand loaded the bootstrap in by front keys to get them to go.  And I learned C and C++ and assembly language (and shhhhh! BASIC) but have forgotten most of it now.  Nowaday we just turn on our appliance computer with its 10 trillion bytes bloatware and we are happy. 

Don't forget that many RR are converting to LEDs which are instant on/off.  So slow on off would be apropriate for 1950s era signals.

Excellent point about appliance computers, I have one here.

I actually cut my teeth on IBM mainframes in the 60's, I worked for IBM.  There was no stinkin' PC's, they came much later!   Funny thing, the room sized mainframes we worked with pale in comparison to the machine that sits under my desk!  In the 70's, I worked on a joint project for Univac and Decision Data where we were the first volume users of the 8080 in the world, were were building the BC/7 business computer.  I built my first computer with a wirewrap gun before the IBM PC was even dreamed of.

Indeed.

5 cent transistors in a paired emitter-follower configuration with a single R-C time constant can achieve a suitable effect. One after the master 555, another after the inverting buffer slave 555.

For some reason the ramping effect does not show up that well in this video but I think you can see the difference when the cap is removed.

Good solution and I can easily incorporate this into my board with the junk trannys in my drawers.

A larger cap and a slower blink rate will also make the fade effect more pronounced if desired.

Great stuff guys. Now let's keep going. We need something to turn the circuit on and off at the proper time. Turn on say 5-10 feet before the grade crossing. Off as the end car passes the crossing. Make that bidirectional and, oh yeah,  for a 2 track crossing. I did it in my PIC with about 20 lines of code. How about crossing gate motor control? Just some more lines of code.

Since the OP clearly stated that the PIC was not an option for several reasons, why do you keep beating on this option?  I think we know that it's possible to do it with a programmable part.

Sam, John, et al,

Could you use a 555/556 Bistable (flip-flop) - a memory circuit (opposite of the

555/556 Inverting Buffer (Schmitt trigger) or NOT gate) as a triggering device to the input of the NOT gate?

If so, the "SET" and "RESET" connections of the flip flop could be magnetic reed switches along side the track with magnets on the engine and caboose or observation to activate or de-activate the crossing signal.  ie: left side of train to activate, right side to de-activate.

You guys are the experts here... these are just my thoughts and/or suggestions

Best,

Dave

Yes, Dave reed switches can be used and are used in HO work though no reason can't be used for O, just need to get reed positioned close enough to reliably trip and be protected if engine derails and runs over and breaks reed capsule (ask why I know this?)

I am using Hall effect latching switches to trigger signals on the approach with one magnet polarity, then turns off bell when reaching island  roadwaysensor.  Caboose has opposite polarity magnet that turns off approach sensor but island sensor still on so lights flash bell silent.  When caboose passes island sensor it turns off sensor and signals.  So put a senor on one side of track for approach and island.  Rails effectively shield opposite sensor from tripping.

Have this working in my workshop now.  Got some powerfull magnets from minitronics that reliably trips sensors up to 1 inch.

Do you have MFG part # and source URL or other means of acquisition that you care to share?

I'm interested in trying to make this work for a crossing signal that would trip and reset for trains going either direction thru the crossing.

Cost is a factor; as always, that's why I thought that reed switches triggering a flip flop 555 which in turn would control a NOT gate 555 might not be too expensive a way around this.  Your idea may be even more cost effective.  Care to share your schematic and approximate cost?

Best,

Dave

Nope, can't share, if I told you I would have to run you over with my MTH Y6b engine!!

I am using Honeywell sensor SS461A which is latching type.  Came from Mouser as PN 785-SS461A about $1.50 each.  Go to Mouser website and put in PN then on that page click the data sheet link that will take you to this device specs, rather than fussing with Honeywells horrible website.

For your application, you might need the non-latching type sensor that signals when magnet went by to trip the set and reset flip-flop lines.  All these sensors are open collector requiring a pull up resistor to supply voltage.  No magnet is off and high output.  Magnet near is on and output goes low until magnet removed.

Here is the schematic which is a work in progress.  Maybe not elegant and sure gurus here will see simplier ways to do this (besides using a PIC).  But this was using stuff from my junk box except for the Hall sensors and MTH flasher/bell set.  If not readable email me and I will send it to you.

Nope, can't share, if I told you I would have to run you over with my MTH Y6b engine!!

OK, well, let's see - Y6B, two sets of 8 drivers... I'll just triple-head a few Hudsons... Three SIXES beats a pair of EIGHTS!!! 

Thanks for all the info and the schematic!!!!!!!!!!!!!!!!!!!!

Best,

Dave

I could probably save a resistor or two from your design, looks good to me.

OK John, I'll bite; which resistors would you eliminate and why?

Best,

Dave

Sorry guys, I was joking, it looks fine.  I guess I didn't make it clear enough.   I'm sure there are other ways to do the job, but that looks fine the way it is.

Peripheral Interface Controller , my son at UofI is a third year electronics software geek has learned to program these.  Sheesh! When I was in engineering we were just learning about ICs and when his grandfather was in EE he just missed transistors.

No point here just an observation of engineering education that America should not be left behind by calling us snobs!!

Hooked board up to lights and sensors.  Behold it worked as designed, sort of.  Per the Honeywell spec sheets, I oriented the senors so S magnet turns on lights, N magnet turns off.  Well turns out the MTH speakers have their N magnet facing down to ties.  So front of engine S magnet turned on lights, and diesel engine low riding speaker N magnet turned them right back off!!

However if the sensor was flipped over, then an N magnet turns on and S magnet turns off.  Luckily all my magnets were on sticky foam (one of Murphy Law axiom here) to locate the sweet spots, so I just flipped magnet polarities and volia no more shut off until caboose or observation car S magnet rolled by.   

That's great.   Sounds like it worked out as intended.

That must have been a head-scratcher. Out of curiousity I measured the flux density at the rail height which was about 8mm below the grille of the low-rider I measured. It was highest directly under the center of the speaker but you obviously wouldn't place the Hall sensor on the center rail.  At half-way between the center and outer rail, it was still 3 times the minimum trip point of a SS461A.  An interesting consideration that also applies to reed switches.