Convert From Common Cathode LED wiring to Common Anode for 3 Light Aspect Signals?

Replace the LEDs with load resistors and drive three NPN transistors with the LEDs plus resistors in circuit from collector to the positive power supply. IE, all their anodes will be common. The transistors will then pull the individual cathodes low to light them.

cjack posted:

Replace the LEDs with load resistors and drive three NPN transistors with the LEDs plus resistors in circuit from collector to the positive power supply. IE, all their anodes will be common. The transistors will then pull the individual cathodes low to light them.

Also...you could invert the signal to the red LED with a transistor and eliminate the two PNP transistors, driving the green and yellow directly with the 556 since it's outputs are capable. 

Yep, I'd just drive the LED's directly from the timer chip.

Great, thanks for all the suggestions!

After considerable effort RTR12 came up with the attached common anode revision, which after we have both tested it out, we are quite happy with. it seems to work as we had hoped, and now makes the wehonest and other common anode aspect lights useable! Yahoo.  My hat is off to Tom for all his dedication and hard work!

Posting the schematic here for any feedback or improvement comments anyone would like to add. Thanks in advance! We will likely delete the second diode D4 from the green led return wiring; it does not appear to be needed.

After all is said and done I will probably post the final versions with gerbers here, so that anyone that wants can order boards.

Rod

It doesn’t look like the green LED is off when the red LED is on.

What am I missing?

 I’m thinking the green LED should be driven with an OR gate keeping it off when either of the 556 timers are high.

The transistors are inhibiting current to the green LED whenever the yellow or red LED is lit.  Basically, I guess that's your OR gate.

My question is on the Block Occupancy Detection.  Since you have an AC opto isolator, why do you need an alternative trigger with a switch?  Can't you connect pretty much any trigger to the opto isolator?

That’s true on the first schematic but I don’t see that on the revised schematic.

cjack posted:

It doesn’t look like the green LED is off when the red LED is on.

What am I missing?

 I’m thinking the green LED should be driven with an OR gate keeping it off when either of the 556 timers are high.

cjack, I think grj nailed it. All I know is, it works quite well.

John, the NO switch was originally envisioned for conventional triggers such as a #153 mechanical switch, or insulated outer rail section, something like that. I think you are right though, the opto should handle most any powered trigger circuit. In the end I will probably leave both in because it's only a couple of small extra traces on the board, and you can just leave them bare if not needed.

Thanks for the feedback so far guys,

Rod

Oh I see that if the yellow or red LED is on, the resistor labeled R7 drops the voltage to below where the green LED will illuminate. 

Green is off when red is on because a Green LED has a higher turn-on voltage than a Red LED, say, ~3V vs. ~2V.  This "voltage starving" is sort of described in this previous OGR thread. 

The two diodes D3 and D4 effectively increase the Green turn on voltage to ~4.5V further insuring the distinction.  If you remove the 2nd diode D4 as proposed, I'm thinking you might see Green turn on every so faintly (turn down the room lights) when Red is on.

GRJ has the operation described above, just took me a LOT longer than it took him to figure out how to do it. I was trying to use the suggestions above, but I didn't quite grasp the proper circuit needed to follow the suggestions (you guys are all way ahead of me), so this could is be a variation of some of those above suggestions.

Stan, your are right about D4. On my bread board, with D3 only, by itself does allow the green LED to just barely flicker. I needed to add the extra diode to my circuit. When Rod tried it, he got no flicker (or light) from the green LED with only one diode (D3) and did not need D4.  When testing the previous common cathode circuit, we both needed the second diode during testing. The common cathode version was the starting point for this common anode version. 

I am guessing there is some difference in the LEDs we are using that is causing the two different results. I am using a 'bargain basement' (cheapest I could find with the most LEDs) LED assortment from ebay, short leads and all. Not sure what LEDs rod is using. 

Rod, I'll try just one of the we_honest signals separately to see if they work on D3 only or need to have D4 added. Not sure I thought of trying that before? Anyway I haven't tried it?

Diodes are cheap, just leave both of them.

I think that might be a good idea, since no one knows for sure what others will be using for LEDs. Also, as I'm sure you already know, it's difficult to predict what (or how) others might use the circuit. That is if they are even interested in using it at all? Definitely another very good exercise in learning though! 

I was just fiddling with the we_honest 3 aspect signals, I have and 2 of them and each worked just fine, individually, with only one diode. I was also using the signal's 1k resistor built in to the common anode wire. I was using a 470 ohm on the bread board when I needed to add the 2nd diode. I think Rod is also using that same 470 value. I suppose the 1k resistor vs 470 ohm could make a difference too.

So yesterday I played around with various values for R7 from 200 to 470R, and with/without inline 1K resistors to the leds, and with one D3 diode in the cicuit. At no time did I get a glimmer from the green led when either the red or the yellow were on. They were of course all very dim with the 1K and R7 from 200 to 470. I varied the supply voltage from 6-12 vdc.

But I wound up leaving D4 in the Diptrace files anyway, as shown in the schematic, just so we have that option. Then I ordered some boards so they should be here in a couple of weeks. Once they arrive I'll get a couple built and see how they work.

Many thanks for all the great input. Will keep ya all posted! 

Rod

PS; I just want to add a great big note of thanks here to RTR12 for doing all the heavy lifting on the right side of the schematic, and creating/testing a circuit that works like we wanted. If all goes according to plan we will wind up with a nice affordable little board that will operate all those common anode pre-wired 3 light aspect signals! Yahoo.

rtr12 posted:

...

I am guessing there is some difference in the LEDs we are using that is causing the two different results. I am using a 'bargain basement' (cheapest I could find with the most LEDs) LED assortment from ebay, short leads and all. 

Right.  In my experience, buying "cheap" Red LEDs verge on being orange as opposed to being "deep red".  Likewise, the "cheap" Green LEDs are almost yellow-green as opposed to being "deep green".

Without delving into semiconductor physics and such, note that the typical operating LED voltage varies with color.   Cheap red and green LEDs will have operating voltages closer together than Deep red and Deep green LEDs.  And in the circuit you guys are using, the function of D3 and D4 is to separate the operating voltages of Red and Green.  Hence, it could be that Rod is using darker LEDs that already start with a wider voltage separation and hence do not need the 2nd diode.

For the prototypical sticklers for detail, you can actually look up the FRS specification for the wavelength effectively specified in nanometers (nm) for Red and Green signals.  Some might find the following clip from 100 years ago amusing.  Given the technology of the day, rather than specifying in terms of nM wavelength as you might find today, note the definition of Green talks of a hint of "blue" meaning to the left of the spectrum as shown above.  In other words a deep green!

If you shop around, you will find interesting names for colors of LEDs such as "Emerald Green", "Pure Green", and so on.  Very few hobbyists, if any, can afford an optical spectrometer that can measure the wavelength of an LED.

Rod Stewart posted:

 

...Thanks for the feedback so far guys,

 

Have you considered layouts that don't have 6-12V DC readily available?  I'm thinking of guys that might want to use 14-16V Accessory AC.  I realize you can buy a 12V DC wall-wart or an AC-to-DC voltage converter module for a few bucks.  Just curious.

For peanuts, you could add a bridge rectifier, capacitor, and 8V regulator.  That would give you universal compatibility.  The 8V regulator is so it would still be compatible with 12VDC with no issues of low input power.

gunrunnerjohn posted:

For peanuts, you could add a bridge rectifier, capacitor, and 8V regulator.  That would give you universal compatibility.  The 8V regulator is so it would still be compatible with 12VDC with no issues of low input power.

Good plan John and Stan. We already have a bridge rectifier and 220 uF filter cap in the circuit, but a reg would be a good addition. A TO-92 LM7808 would fit with no issues and the 100 ma current limit is plenty for what this circuit is doing. It would be a good addon for the next revision. And we can use one value for R7 that fits all.  Darn, now I need to order some LM7808's.

Rod

Stan, that has to be the difference. My green LEDs are a very light green. The reds and yellows are not what I would call 'deep' colors either, also pretty light colored. I have some better (I think anyway) LEDs from Digikey around here somewhere that I am going to try today and see what happens. One of those may even be an Emerald green, that sounds kind of familiar.

The we_honest signals seem to have somewhat 'deeper' colored LEDs, more so than my cheap LED kit has. For the total signal price, I would expect the we_honest signals to also be using pretty inexpensive LEDs. They so look better than the ones in my kit though. I'm sure they get much better pricing than I do, so they could have the better quality LEDs in their signals and still keep pricing low I guess.

Thanks for the kind words, Rod! Not sure I am deserving as it was hit and miss for a while there, mostly miss. Probably a lot of luck too!

I have been watching the developing convert from common cathode to anode and am aware of the dual transistor OR function of the original schematic. It's solid with two PNP transistors in terms of various LED choices. I have to say that the voltage starved sensitivity to component choices leaves me somewhat anxious. I think a solid OR gate like a PNP with a couple of diodes on the base would eliminate all the component variations. Simple and devoid of a lot of testing. The idea is if either of the 556 timers outputs is positive, then the green LED is off. No problem. No issue with voltage either power supply or LED choices.

Rod Stewart posted:

 A TO-92 LM7808 would fit with no issues and the 100 ma current limit is plenty for what this circuit is doing. It would be a good addon for the next revision. And we can use one value for R7 that fits all.  Darn, now I need to order some LM7808's.

Rod

No-no!  You want the TO-220 package and likely also a small heat sink.  The TO-92 package, in spite of it's datasheet rating, is not suitable for 100ma!

cjack, I have had fairly reliable results using 2 diodes from the green LED cathode to Pin 9 of the 556. It also works for me with the green cathode to GND using 2 diodes. I had a couple of glitches (flickering green at times), but I think I now have all the gremlins removed from that part. Rod has had the same results (with no gremlins) using only one diode in his circuit. I think he has now made the PCB with 2 diodes and anyone using it can either add both or jumper one out as desired. I'm just planning on using both diodes for anything I do here. 

Are you thinking of a 3rd transistor somewhere in the circuit? I sort of fiddled with that earlier and never came up with a way to make it work properly. Then I stumbled on to the circuit as it is now, which has been the best one so far. I'm still pretty green with this stuff though, so what seems simple or obvious to you guys might be way over my head?

That TO-220 might be good for me.   I know I don't have any TO-92s, but do have assorted values in TO-220, just not sure what voltages they all are? Luck may run out with 8 volts though? 

It's probably a long shot to have the 8V one, it's not as common as other values like 5V, 12V, and 15V.

Exactly correct, the shot was way too long, no 8's!  How about a 5 or a 12 (or maybe a Rube Goldberg with a 317)?

LOL.  8 cents vs. 14 cents.  This high-finance is making my head spin! 

In any case, my concern is someone hooking it up to 14-16V AC Accessory voltage without the regulator.  That would release the magic smoke from the 556 which is spec'd to operate up to 16V DC. 

What may not be obvious is some train transformers that let you set the AC Accessory voltage can be problematic.  Even though the AC Accessory Voltage is set to, say, 12V AC the peak voltages that would reach the 556 IC chip (again, we're talking without a 780x regulator) would exceed 16V DC.  This devolves into a conversation about chopped-sine transformers which gets boring really quickly.  Best just to put in the 780x regulator and be done with it!

The reason I talked about the 8V regulator is it would work properly if you connected to 12VDC, even with a couple of diode drops for the bridge.  You could go with 5V, but you might want a bit more voltage headroom for the LED drive.

cjack posted:

I have been watching the developing convert from common cathode to anode and am aware of the dual transistor OR function of the original schematic. It's solid with two PNP transistors in terms of various LED choices. I have to say that the voltage starved sensitivity to component choices leaves me somewhat anxious. I think a solid OR gate like a PNP with a couple of diodes on the base would eliminate all the component variations. Simple and devoid of a lot of testing. The idea is if either of the 556 timers outputs is positive, then the green LED is off. No problem. No issue with voltage either power supply or LED choices.

cjack; you mean like this:

This is reproduced by permission from Rob Paisley's site; Circuitous. I had been in touch with him and asked if he had any ideas for modifying his original common cathode circuit to common anode, and this was one of three options he has since posted on his site. Note the date! Option #3 looks almost identical to the creation of our very own RTR12, which is the basis of the circuit I posted on Oct 23, above. The link to the Circuitous site is http://www.circuitous.ca/Timed3LightSignal555.html

Some subsequent suggestions from this thread are incorporated in the following revision to the basic circuit:

This has provision for a fixed regulator, which could be a 5, 8, 9 volt or other. I show an 8 volt version here as suggested by grj. There is also a selectable DuPont style  jumper block, depending on whether the regulator is installed or not. The 8 volt reg will need a DC supply of minimum 12 volts, if bridge diode D1 is left installed (though if the application is purely DC it would be best to just jumper this), OR minimum 8.5 volts AC. For the case of an 8 volt reg, R7 should be about 270 ohms, if the target led current is say 15 ma. Note that there is also provision for both D3 and D4 diodes in the circuit, since RTR12 needed both in his version of the circuit to keep the green led from lighting when either red or yellow were lit.

Good discussion here, let's keep it going!

Rod

Just bread boarded the 3 transistor #2 circuit above. It works quite well, though no better than that shown in R5.4 also above, from what I can tell. I had to change the resistor that grounds the base of the 2N3906 PNP transistor from 10K to 2.2K. The green led was not lighting after the yellow went out, so I reasoned that the base voltage was not close enough to ground with the 10K value to make it conduct. Anyway, it seems to work fine now. I also added a couple of diodes in line with the red and yellow leds, to better equalize the currents and reduce their brightness to more like the green. All good.

Rod

Rod, I can't find that circuit on Rob's site? I found some others, but that one seems to be hiding? I'm probably overlooking something obvious...    Anyway, I copied the one from your post above so at least I have a copy... but I'd be interested in reading what Rob posted about it though (that is if he posted anything).

Before ending up with the basic circuit we have now (the 10/23 version), this is what I was trying to do in prior attempts, but was unsuccessful. Rob's has figured it out and I also suspect this could very well be what cjack was saying above that I still didn't quite understand how to do. 

That's good news on the bread boarding. I must try that here too, just for the experience, I'll keep your resistor changes for the BB.

Also, didn't you mention a version Rob had with a 339 in it somewhere? If there is one I can't seem to locate that one either? Or am I getting mixed up with another circuit on Rob's site I was probably looking at?

Rod Stewart posted:

Just bread boarded the 3 transistor #2 circuit above. It works quite well, though no better than that shown in R5.4 also above, from what I can tell. I had to change the resistor that grounds the base of the 2N3906 PNP transistor from 10K to 2.2K. The green led was not lighting after the yellow went out, so I reasoned that the base voltage was not close enough to ground with the 10K value to make it conduct. Anyway, it seems to work fine now. I also added a couple of diodes in line with the red and yellow leds, to better equalize the currents and reduce their brightness to more like the green. All good.

Rod

That's it, #2. I wasn't sure that placing the PNP above the Green LED would work without more parts value sensitivity, but it looks good. The 2.2K draws more current from the 556 but it's got a lot available. Alternatively I thought about using one fourth of a CMOS 4001 driving an NPN transistor. The CMOS chip can use any of the voltages being suggested. But the above has less parts.

gunrunnerjohn posted:

Rod Stewart posted:

 A TO-92 LM7808 would fit with no issues and the 100 ma current limit is plenty for what this circuit is doing. It would be a good addon for the next revision. And we can use one value for R7 that fits all.  Darn, now I need to order some LM7808's.

Rod

No-no!  You want the TO-220 package and likely also a small heat sink.  The TO-92 package, in spite of it's datasheet rating, is not suitable for 100ma!

I hear you on this John. I plan to run a little breadboarding trial using a TO-92 package reg and see what she does. I'll stand by with the fire extinguisher handy! Mostly I use a TO-92-100 pattern for any TO-92 package lately, just to make the soldering easier. This includes the new board I have roughed out for this addition. And conveniently this pattern nicely fits the TO-220 pin spacing as well. Only hitch is the #1 and #3 pins are reversed between the two, so you would have to install the TO-220 backwards, but it could be done I think.

Tom, did you find that #1 circuit with the 339 comparators being used?

Rod

I can tell you that the TO-92 regulator used on the Lionel smoke unit frequently fails with just the 30-40ma load of the smoke motor on it.  I have built modules with the TO-92 regulator, and with about 40ma on them, feeding them with half wave filtered 18V track power, the regulator shoots past 100C in temperature.  My temperature guideline for these regulators is no more than 85C, and I prefer to keep them lower in the 70's.

For the few cents difference in price, let me save you some time.  Forget the TO-92 package and move right to the TO-220 model.

Point taken John. Thanks, I am going to convert the design over to TO-220 style and be done with it. Lots of them around the place anyway. 

For those following the Circuitous site #2 design above (using 3 transistors), I just got word from Rob Paisley that he is overhauling the circuit a bit. So there will be a new version along, likely later today. 

Rod

Rod Stewart posted:

.....

Tom, did you find that #1 circuit with the 339 comparators being used?

Rod

Yes, I sent you an email.  Found, copied and saved them all!  Problem was Chrome and the website being http: and not https: as Chrome was wanting. Guess Chrome thought it was protecting me from something? 

Got word from Rob that he has tweaked all 3 of the common anode circuits, particularly the #2 version which is now 4 transistor and 3 diodes. The link to the new circuits is here http://www.circuitous.ca/Timed...Signal555Others2.GIF

Take care,

Rod

Of the 3 options, I'd go with #2 if you are going to make an OSH Park (or whatever) PCB design.  Circuit #1 is limited by the drive capability of the LM339.  If you are driving more than one signal head in parallel, or if using high-power LEDs (100 mA instead of 20 mA nominal operating current), or even miniature incandescent bulbs then you're out of luck.  Circuit #3 depends on the different in drive voltage for different color LEDs.  This works of course of the Red-Yellow-Green LEDs but what about incandescent bulbs or a PRR style 7-LED signal head which uses all the same color LEDs?

Did you guys actually prototype circuit #2?  I think the "OR" circuit which turns off the Green LED when Red OR Yellow is on is skating on thin ice.

The Yellow and Red LEDs are turned on when one or the other 555 chip is ON (at pin 3 output).  This ON signal does NOT go to the positive supply voltage because there's an output transistor as shown in the 555 internal schematic above.  As the curve shows on the right, the "ON" voltage might be about 1V below the DC+ supply voltage.  Then each Yellow/Red output goes thru a diode which further lowers the voltage.

So this transistor + diode voltage drop now tries to turn off the 2N3906.  While a diode appears to have been added above the 2N3906, it's not clear to me that either the Yellow or Red output generates a high enough voltage to turn OFF the 2N3906.