Many years ago I measured spikes on my DC layout over 20 volts from older locomotive motors. Back then I used 18 volt Zeners paralleled back to back on the motors that kept them below 20 volts. That was way before DCC and I didn’t have anything the spikes would damage but I just didn’t like them being there. Mel My Model Railroad http://melvineperry.blogspot.com/ Bakersfield, California I'm beginning to realize that aging is not for wimps.
Those that have published monitoring information on DCC show spikes around 30 volts on poorly run or too long bus runs, so I doubt very much you would see more than 50V on DCC, in fact you shouldn't even see those 30V spikes, at least not of any notable duration. It's not just LEDs that don't like 30V spikes, that's well beyond the voltage any decoder is required to withstand per NMRA standards.
--Randy
Modeling the Reading Railroad in the 1950's
Visit my web site at www.readingeastpenn.com for construction updates, DCC Info, and more.
A more direct question is: how frequently do we see spike or impulse noise on the DCC voltage? Is it more observable in the + or - swing, or equally probable either way? I surmise from other information (ease of burnout) that there is little or no voltage-clamping protection in decoders (although it occurs to me that varistor protection ought to be possible, perhaps with 'replaceable' devices in sockets accessible without removing shells so protection can be re-established if small varistors stop clamping, as so often happens on cheap surge protectors).
I was about to note 'any diode becomes an avalanche diode at sufficient voltage' and then thought 'has anyone used actual avalanche diodes to protect LEDs against spikes'? That ought to be something you could almost integrate into higher-power LED modules like those used for vehicle headlights... and protecting higher-value decoders seems to me to logically follow.
I note the reference provided above specifically mentions Zeners as protection devices against DCC reverse current:
The zener diodes will be connected cathode to cathode, and this assembly is connected in parallel with the LED. The zener voltages are selected to be greater than the forward voltage of the LED, but less than or equal to the maximum reverse voltage. During normal operation the current flows through the LED, whereas during the reverse phase current can flow through the zener pair, limiting the reverse voltage seen by the LED. Data sheets for LED often have these protection circuit schematics.
Yes, a diode with a PIV of 50 doesn't help much when the transient spikes are 200+ volts. I'm willing to bet most any electronic failure in a modern car after you get past infant mortality is a shortcut taken somewhere to save a few cents, or something simply overlooked by the design engineers.
Negative spikes on a great many types of diode -- not just LEDs -- are a substantial concern, made much worse by smaller device or die size and the push to lower and lower nominal voltages for the various reasons. I could cry you a river over all the impulse noise and spike fun I've had over the years, and cumulative damage from improper clamping in an automotive environment doesn't really surprise me ... although I'm highly grateful you formally documented it because I hadn't carefully thought the implications for LED devices through before.
Andy
I use AC on my Atlas #65 switch machines and Peco PL-10s with latching relays for position indication. They have been in service for about 10 years with no problems.I happened on a good deal on eBay, a box of 24 relays for $25 free shipping. Mel My Model Railroad http://melvineperry.blogspot.com/ Bakersfield, California I'm beginning to realize that aging is not for wimps.
And negatiove voltage hurting an LED isn't a myth. Here's a paper done on it with respect to transient spikes in automotive applications (warning - it's a pretty technical paper, probably not for everyone)
https://docs.broadcom.com/doc/5980-1504E
While a current limted (witht he resistor you need anyway for normal operation) LED usually can handle a high reverse voltage without instantly going poof, there seems to be an accumulation of damage over time. An ordinary silicon diode like a 1N4001 might handle 50V reverse, but most LEDs are only rated for 4 or 5 volts reverse voltage. White LEDs may be even lower.
Will it work? Yes. Is it best design practices to just feed AC to an LED through a resistor? No. Often the "track on" indicator on DCC systems uses a red/greed bicolor LED - there you have two LED 'chips' back to back, so the one that lights on the positive half of the signal protects the reversed one because the voltage drop across it is whatever the lit LED's forward voltage is - typically 1.8-2.1 volts. Well below the 4 or 5 volts reverse voltage the LED can stand. For a single red LED as an ac power indicator, a simple silicon diode (even those low current glass types you buy in packs of 100 will work here) will limit the reverse voltage to less than 1 volt.
You need either some sort of control circuit that includes LEDs (there are several on Rob Paisley's electronics site) or you need, as Greg said, some sort of latching relay, because there is only momentarily any power in the wires running to an Atlas switch machine (if there's continuous power, it will burn out the machine in short order). An Atlas Snap Relay would work, that gives you two sets of contacts to use, one for the indicators and another to power the frogs, if you are using Custom Line turnouts. Since the contacts on the relay are independent of the power source, you don't even have to worry about using AC with LEDs.
betamaxyou don't want to reverse bias an LED. They don't like that.
More to the point: there is little reason to use AC in this application in preference to one of the old red/green 'reversible' LEDs (which used two opposite-polarity 'cores' in one housing, so that if you reversed polarity only the one "forward-biased" would light. This would automatically give you either of the semantic possibilities folks use with turnouts: "green" meaning lined as desired, or 'green' corresponding to switch normal position as with some valve coding schemes.
What AC (or, more interestingly, asymmetrically-biased AC, as with 'pulse power' without rectification) does with such a diode is light both cores alternately, in step with the frequency of the AC. Persistence of vision produces an orange color, and applying asymmetry shifts this toward 'red' or 'green' as the case may be, giving you a range of 'oranges' for caution searchlight aspects. This of course has about as much use in switch control as maybe gates do in finite-state-machine computers (which is basically what betamax said about the use of 'having both lit at the same time') ... but that's another discussion for another day.
i think you're asking how to latch the state the turnout was last switched to?
if so, you could use a latching relay that would be switched along with the turnout. the relay coil can be powered with AC. the relay contacts would power one of 2 LEDs powered either by AC or DC
greg - Philadelphia & Reading / Reading
Not too difficult.
You need a series resistor to limit the current and provide the required voltage to the LED, and another diode wired in parallel, but in reverse, across the LED to protect it from the reverse voltage. Examples of how to do it are at the DCCWiki: LED Lighting.
Basically it is the same with DCC, as you don't want to reverse bias an LED. They don't like that.
If you want to indicate positions, another LED in place of the protection diode will allow you to see which position the turnout is in, based on the direction current flow. Will not work on AC as both will light up.
Gang: am using Atlas under table switch machines and controllers. Is there a way to use LED's to indicate turnout indications, using AC? Seems I remember a simple circut in MR, years ago. Now where did I put that issue?? .......Thanks Andy