Question about insulated signal blocks with DCS

Danr posted:

My layout is not especially large (14x14) but I have 4 isolated rail trigger sections in each of two loops.  These isolated sections are between 2 and 4 feet long.  It happens that the isolated section are all on the same rail, but I do not believe that has any real effect since both rails are tied together.  These sections trigger relays controlling my signals.  I have both Legacy and DCS and have zero issues.

Thanks,

I wonder if any clubs are having issues. The club sees a lot of activity so there could be many more spikes than on a home layout. That increase is all I am worried about at this point. Yesterday I took some measurements of a relay's EMF with a digital oscilloscope and found some pretty big spikes when the connection to the relay is broken.

This is the track side EMF spike I get off a 12Vac relay when I lift the common wire to the relay coil off the track.

this one spike alone has a V-max of 184 volts and a Vpp of 272 volts!! I zoomed in on one of these spices and found they are very short in duration 7 nano seconds. But that is quite a jolt, I feel like I should be trying to clamp this at the source with some fast acting diodes. Something like the 1N4148 or 1N5818? maybe @Adrian! or @gunrunnerjohn have some info here.

Do you have TVS Diodes installed? (I can't recall if there are some in the MTH control units or not). Regardless, it can't hurt to have some added. 1.5KE36Ca is the standard recommendation.

That's usually why many relays have a kickback diode across them to swamp out that spike.  The TVS is also a good idea.

bmoran4 posted:

Do you have TVS Diodes installed? (I can't recall if there are some in the MTH control units or not). Regardless, it can't hurt to have some added. 1.5KE36Ca is the standard recommendation.

The above reading was certainly taken without any TVS diode on the relay. I initially wasn't hesitant to just say that a TVS diode is the solution. I thought @Adrian! on another thread mentioned that the TVS diodes are a bit slower to react (micro seconds not nano seconds). Slower than this 7 nano second spike. That's why I figured I may have switch to dc relays and use high speed diodes to stop the EMF issue. The 1N4148 mentions a 4 ns (maximum reverse-recovery time).

Is't there also some consensus that the TVS diodes are more likely to fail on a layout that has many trains thus having more than occasional derailments.

gunrunnerjohn posted:

That's usually why many relays have a kickback diode across them to swamp out that spike.  The TVS is also a good idea.

agreed, now I can't exactly run the relays on ac unless I use a rectifier diode in series before adding the kickback diode across the coil. Plus I guess I am still unclear on the clamping time of a I had the impression they were slow.(relatively speaking.)

I don't know how you got the idea that the TVS is slow.  One picosecond response time, or .001 nanoseconds.  The whole point of the TVS is to catch fast transients, it wouldn't be of much use if it took nanoseconds to clamp!

From the datasheet of the Littlefuse 1.5KE series TVS.

gunrunnerjohn posted:

I don't know how you got the idea that the TVS is slow.  One picosecond response time, or .001 nanoseconds.  The whole point of the TVS is to catch fast transients, it wouldn't be of much use if it took nanoseconds to clamp!

From the datasheet of the Littlefuse 1.5KE series TVS.

I realized what got mixed up in my head now. The subject that came up previously is that TVS diodes break down if subjected to long duration spikes. (longer than 200ns or so I think). I was just thinking of what that "Beefy Clamp" board of Adrian's was doing. Those slower to act Schottky diodes taking over the clamping effort after the earliest part of the spike. Relieving the TVS diode of it's duty and extending it's life. I just realized I was swapping the clamping times of TVS and Schottky diodes in my head.

Thank you GRJ

Ryaninspiron posted:

Danr posted:

My layout is not especially large (14x14) but I have 4 isolated rail trigger sections in each of two loops.  These isolated sections are between 2 and 4 feet long.  It happens that the isolated section are all on the same rail, but I do not believe that has any real effect since both rails are tied together.  These sections trigger relays controlling my signals.  I have both Legacy and DCS and have zero issues.

Thanks,

I wonder if any clubs are having issues. The club sees a lot of activity so there could be many more spikes than on a home layout. That increase is all I am worried about at this point. Yesterday I took some measurements of a relay's EMF with a digital oscilloscope and found some pretty big spikes when the connection to the relay is broken.

This is the track side EMF spike I get off a 12Vac relay when I lift the common wire to the relay coil off the track.

this one spike alone has a V-max of 184 volts and a Vpp of 272 volts!! I zoomed in on one of these spices and found they are very short in duration 7 nano seconds. But that is quite a jolt, I feel like I should be trying to clamp this at the source with some fast acting diodes. Something like the 1N4148 or 1N5818? maybe @Adrian! or @gunrunnerjohn have some info here.

As you start trying to clamp this you may find it turns into the whack-a-mole game.  7ns right?... so on the order of 140 MHz (possibly even higher). That's a 6 ft wavelength (and a 3 ft half-wavelength right there). So you can't stop all of them completely unless you put a TVS every 3 feet along the entire section (at which point there will be a lot of junction cap added up, and your DCS bits will look like the shark fin waveform). Or you can put the TVS in the engines and at the TIU ports (the things you want to protect) and let the impulses exist on the track. Or you can goto an SSR (recommendation) which isn't inductive. Or just go through the 5 stages of grief and make everything optical (like we did).

Ryaninspiron posted:

Do insulated outer rails (used for signal blocks) cause issues with DCS on very large layouts?

I only ask because I have heard it can help legacy signal strength if you bridge insulated signal blocks with a capacitor at the gap between blocks. I am wondering if those same concepts apply to DCS at all.

Also can relays activated by an outer rail cause any issue for the TIU? especially in terms of EMF? Are opto isloators the best solution(or even recommended) for signal activation? (large layout several hundred feet per line.)

I write answers to this all over, but I can put it here too. 

So conceptually your layout is floating with respect to like all of the planet earth. The TMCC/Legacy voltage is a potential between the big planet earth, and the ground of your layout. Lets call it "VL". The DCS signal sits on top of this in series lets call it "VD". So at the outside rails of this wonderful layout has potential "VL" with respect to planet Earth and the center rail has potential "VL+VD" with respect to planet Earth. Now your TIU and DCS Trains are with respect to the the outer rail (VL) so at the PS3 board or TIU board it's going to be Vin = (VL+VD)-VL  = just VD so conceptually the Legacy signal is invisible.

That was the concept. Now in practice:

1. The sources VD and VL have Thevnin impedances, lets call them ZD and ZL keeping our convention.

2. The center rail has capacitance to ground Cin and the outside rails have capacitance to ground Cout. One has 1 rail and one has 2 rail so even considering these to be similar is a bit silly. So from basics Zin is going to be 1/[JwCin] and Zout is going to be 1/[jwCout].

Okay so by good old linear superposition we can compute the contribution of each VL and VD to the voltage at the locomotive Vtrain.

These results have some interesting implications.  (1) even balancing the inside and outside rail capacitance perfectly won't stop the DCS signal from collecting some superimposed legacy signal, because there is another path to ground through the legacy base (ZL) that more strongly influences the outer vs the inside rail.  (2) If ZL goes to infinity (that is saying if the legacy base goes away) there is no cross talk. That's a no brainer. (3) The term (ZL||Zout + Zin) which describes the parallel capacitance of the layout and legacy base directly eats away at your DCS signal at it gets smaller (cap gets bigger) since... Vtraind = Vd/(1+Zd/[ZL||Zout + Zin]).

So in general anything thats in parallel with ZL makes it a smaller number, (IE having more paths to planet earth through power supplies, relays, capacitance inside control devices), makes the situation worse. If you like math you can breakup Zd into unit sections of track and put other reactance in series to consider the little capacitors, but given the minor role Zd plays in the transfer function (since Zd<<Zin) , the real key is to manage the capacitances to planet Earth.

ADRIAN!,

Thanks for all your work clarifying the what's REALLY happening under the hood with both DCS and Legacy.  It's nice to finally be able to put some real numbers to the problems.

Ryaninspiron,

Here's a little applied experience that might add to the math.

Ryaninspiron posted:

Do insulated outer rails (used for signal blocks) cause issues with DCS on very large layouts?

Usually not.  You need to be careful around turnouts and where the insulation moves from one outside rail to the other, but that's to ensure reliable track power rather than DCS signal.  I've built layouts up to 3200 sq. ft. with 4000+ feet of track with occupancy detection via insulated rail throughout that have excellent DCS signal (strong 10's).

I only ask because I have heard it can help legacy signal strength if you bridge insulated signal blocks with a capacitor at the gap between blocks. I am wondering if those same concepts apply to DCS at all.

No.

Also can relays activated by an outer rail cause any issue for the TIU? especially in terms of EMF? Are opto isloators the best solution(or even recommended) for signal activation? (large layout several hundred feet per line.)

Yes, using insulated rails to drive the coils in relays will more often than not cause noticeable interference with the DCS signal.  Opto isolation is an excellent solution.  On most layouts I use RR-CirKits OIB-8 opto isolation boards for occupancy detection.  They have a variety of other products that allow you use the occupancy detection to drive signals and accessories

Ryaninspiron posted:

The best way to prevent this kind of issue is not to let it happen in the first place.  If you totally isolate your accessory power and use a soft close and release circuit for the relay, there will be no coil kickback to bite you.

This is the circuit I use for insulated rail triggering of signals.  The choke provides DCS signal protection and the resistor provides inrush limiting, there is no issue with arcing at the rails.  The relay has a soft power-on and power-off to eliminate any kickback spikes from the coil.  AAMOF, I don't actually use the diode at D2 as it wasn't needed, it's a knee jerk reaction for me to put a diode across any DC relay even though it's not always required.

With this kind of circuit, you can power the accessories from separate isolated power using the isolated relay contacts.  Any spikes generated by the accessories are not directly transmitted to the tracks.

Ryaninspiron posted:

...

This is the track side EMF spike I get off a 12Vac relay when I lift the common wire to the relay coil off the track.

To be clear, the scope is showing the voltage between the center-rail and outer-rail of a DCS layout?  That is, the track is power by 18V AC and this is driving a 12V AC relay coil via the insulated-rail section?

It's just that at 200 nsec per division, those spikes seem remarkably evenly-spaced as opposed to what would presumably be random interruptions in contact that I'd expect would be separated by more than ~100 nsec if a mechanically induced event.  Stated differently, are you sure you're not measuring some kind of high-speed digital circuit signal somehow coupling into your measurement?

Separately, as to your question about EMF and driving a relay.  I believe you're addressing the issue of the so-called back-EMF spike when a relay coil quickly turns off such as intermittent wheel contact straddling an insulated-rail section.  Seems to me though that if this EMF spike is from the relay coil, then that spike is not on the track side and the TIU is "safe".  OTOH, if the EMF spike is from the inductance in the track (and the instantaneous interruption in current), then it seems to me you have far bigger problems.  That is, a relay coil probably draws 1/2 Watt of power or, say, tens of mA at 12V.  But an engine or even just a bulb lighted passenger car can draw tens or hundreds of times more current.  In which case an intermittent loss of wheel contact would generate a spike in the thousands of Volts thru the same inductance!  

Dave Hikel posted:

...

Ryaninspiron posted:

….

Also can relays activated by an outer rail cause any issue for the TIU? especially in terms of EMF? Are opto isloators the best solution(or even recommended) for signal activation? (large layout several hundred feet per line.)

Yes, using insulated rails to drive the coils in relays will more often than not cause noticeable interference with the DCS signal.  Opto isolation is an excellent solution.  On most layouts I use RR-CirKits OIB-8 opto isolation boards for occupancy detection.  They have a variety of other products that allow you use the occupancy detection to drive signals and accessories

 

As Dave says, opto isolation can be a remarkably effective and inexpensive solution to noise issues.  From my previous post I'm still not clear on exactly what problem is being solved.  In any case, refer to this recent thread from which I copied the following:

The surrounding discussion gets a bit nerdy but all can be explained if this is determined to be the matter at hand.  In the other thread I was lamenting on the lack of an AC-input optocoupler modules for O-gauge applications.  But Dave found one I was not aware of.  Now it's not clear that the OIB-8 is actually available (all the price options are crossed out?) but clearly it was purpose-built for O-gauge AC.

In the other thread I was proposing changes to a $2 (4-channel) DC-input optocoupler module to make it compatible with AC O-gauge.  I suggested a bridge rectifier and a capacitor.  And there they are in the OIB-8!  Member rtr12 might be making a circuit board that would bring the cost per channel way down relative to the OIB-8 price.  In any case, unless you already have a box of AC coil relays looking for gainful employment, you simply can't beat the price/value of the eBay DC relay modules (about $1 per relay).  They have trigger buffers so that you don't directly drive the relay coil.  Hence you only need a fraction of the ground trigger current flowing thru the insulated-rail section - just a few mA rather than tens of mA.  In general, reducing the trigger current requirement gives you more options.

If optocoupler isolation for relays is something you're considering, you may want to tack on to the other thread if the focus of this thread is EMF.

Adrian! posted:

Ryaninspiron posted:

This is the track side EMF spike I get off a 12Vac relay when I lift the common wire to the relay coil off the track.

one spike alone has a V-max of 184 volts and a Vpp of 272 volts!! I zoomed in on one of these spikes and found they are very short in duration 7 nano seconds. 

As you start trying to clamp this you may find it turns into the whack-a-mole game.  7ns right?... so on the order of 140 MHz (possibly even higher). That's a 6 ft wavelength (and a 3 ft half-wavelength right there). So you can't stop all of them completely unless you put a TVS every 3 feet along the entire section (at which point there will be a lot of junction cap added up, and your DCS bits will look like the shark fin waveform). Or you can put the TVS in the engines and at the TIU ports (the things you want to protect) and let the impulses exist on the track. Or you can goto an SSR (recommendation) which isn't inductive. Or just go through the 5 stages of grief and make everything optical (like we did).

Very interesting, I am not afraid to go optical (AC input DC output Optocoupler) such as the H11AA1 (which I have on hand). Wouldn't I need to worry about keeping spikes (from other sources) off the LEDs inside the Optocoupler in that scenario? TVS again?

I am curious about what your optical detection circuit looks like. or are you talking optical as in Photoresistor or IR object detection style?

Adrian! posted:

Ryaninspiron posted:

Do insulated outer rails (used for signal blocks) cause issues with DCS on very large layouts?

I only ask because I have heard it can help legacy signal strength if you bridge insulated signal blocks with a capacitor at the gap between blocks. I am wondering if those same concepts apply to DCS at all.

Also can relays activated by an outer rail cause any issue for the TIU? especially in terms of EMF? Are opto isloators the best solution(or even recommended) for signal activation? (large layout several hundred feet per line.)

I write answers to this all over, but I can put it here too. 

So conceptually your layout is floating with respect to like all of the planet earth. The TMCC/Legacy voltage is a potential between the big planet earth, and the ground of your layout. Lets call it "VL". The DCS signal sits on top of this in series lets call it "VD". So at the outside rails of this wonderful layout has potential "VL" with respect to planet Earth and the center rail has potential "VL+VD" with respect to planet Earth. Now your TIU and DCS Trains are with respect to the the outer rail (VL) so at the PS3 board or TIU board it's going to be Vin = (VL+VD)-VL  = just VD so conceptually the Legacy signal is invisible.

That was the concept. Now in practice:

1. The sources VD and VL have Thevnin impedances, lets call them ZD and ZL keeping our convention.

2. The center rail has capacitance to ground Cin and the outside rails have capacitance to ground Cout. One has 1 rail and one has 2 rail so even considering these to be similar is a bit silly. So from basics Zin is going to be 1/[JwCin] and Zout is going to be 1/[jwCout].

 

Okay so by good old linear superposition we can compute the contribution of each VL and VD to the voltage at the locomotive Vtrain.

These results have some interesting implications.  (1) even balancing the inside and outside rail capacitance perfectly won't stop the DCS signal from collecting some superimposed legacy signal, because there is another path to ground through the legacy base (ZL) that more strongly influences the outer vs the inside rail.  (2) If ZL goes to infinity (that is saying if the legacy base goes away) there is no cross talk. That's a no brainer. (3) The term (ZL||Zout + Zin) which describes the parallel capacitance of the layout and legacy base directly eats away at your DCS signal at it gets smaller (cap gets bigger) since... Vtraind = Vd/(1+Zd/[ZL||Zout + Zin]).

So in general anything thats in parallel with ZL makes it a smaller number, (IE having more paths to planet earth through power supplies, relays, capacitance inside control devices), makes the situation worse. If you like math you can breakup Zd into unit sections of track and put other reactance in series to consider the little capacitors, but given the minor role Zd plays in the transfer function (since Zd<<Zin) , the real key is to manage the capacitances to planet Earth.

Thanks a ton for this, I took enough electrical engineering classes in college (before I switched to computer programming) to follow most of the math behind this. One of the biggest takeaways I have from your post is to try managing the capacitance to earth.

Across the layout there are a few dc transformers being used for control tower block indicators. One of the club's two main branch lines has 48 block indicators that are all wired down 10ft down from the control tower and then spread far and wide across the 4000sq.ft layout. Plus the pretty much the whole layout has some variation of ground plane wiring nearby to help the legacy signal. I imagine all this wiring is contributing to the issues.

I have learned that the TIUs are all early revision Ls without the TVS diodes. The main issue with the club is that the signal seems to have random weakspots across the layout. some places you hit the horn and it gets stuck on. other places you can control the speed but not sounds. Sometimes the built in signal test reads 3 sometimes it reads 10. It has been a mystery for years I am told. All the Z4000 transformers have chokes on the output.

Dave Hikel posted:

Ryaninspiron,

Here's a little applied experience that might add to the math.

Ryaninspiron posted:

Do insulated outer rails (used for signal blocks) cause issues with DCS on very large layouts?

Usually not.  You need to be careful around turnouts and where the insulation moves from one outside rail to the other, but that's to ensure reliable track power rather than DCS signal.  I've built layouts up to 3200 sq. ft. with 4000+ feet of track with occupancy detection via insulated rail throughout that have excellent DCS signal (strong 10's).

I only ask because I have heard it can help legacy signal strength if you bridge insulated signal blocks with a capacitor at the gap between blocks. I am wondering if those same concepts apply to DCS at all.

No.

Also can relays activated by an outer rail cause any issue for the TIU? especially in terms of EMF? Are opto isloators the best solution(or even recommended) for signal activation? (large layout several hundred feet per line.)

Yes, using insulated rails to drive the coils in relays will more often than not cause noticeable interference with the DCS signal.  Opto isolation is an excellent solution.  On most layouts I use RR-CirKits OIB-8 opto isolation boards for occupancy detection.  They have a variety of other products that allow you use the occupancy detection to drive signals and accessories

 

Thanks,The club I am in has a 4000 sq. ft. layout room so it is good to hear about your experience with such layouts. That IOB-8 board is basically what I was considering building from scratch at about half the cost though. Good to see that solution working well.

gunrunnerjohn posted:

Ryaninspiron posted:

The best way to prevent this kind of issue is not to let it happen in the first place.  If you totally isolate your accessory power and use a soft close and release circuit for the relay, there will be no coil kickback to bite you.

This is the circuit I use for insulated rail triggering of signals.  The choke provides DCS signal protection and the resistor provides inrush limiting, there is no issue with arcing at the rails.  The relay has a soft power-on and power-off to eliminate any kickback spikes from the coil.  AAMOF, I don't actually use the diode at D2 as it wasn't needed, it's a knee jerk reaction for me to put a diode across any DC relay even though it's not always required.

With this kind of circuit, you can power the accessories from separate isolated power using the isolated relay contacts.  Any spikes generated by the accessories are not directly transmitted to the tracks.

Certainly an interesting circuit, nice work. Also aside from the inrush resistor limiting current, isn't D1 also going to do the job even if the emf spike somehow backs out of the voltage regulator input? (not sure that's even possible.) Just scientifically curious why it wasn't needed.

Ryaninspiron posted:

Very interesting, I am not afraid to go optical (AC input DC output Optocoupler) such as the H11AA1 (which I have on hand). Wouldn't I need to worry about keeping spikes (from other sources) off the LEDs inside the Optocoupler in that scenario? TVS again?

I am curious about what your optical detection circuit looks like. or are you talking optical as in Photoresistor or IR object detection style?

The circuit I posted gives you the isolation that optical would using an insulated rail.

stan2004 posted:

Ryaninspiron posted:

...

This is the track side EMF spike I get off a 12Vac relay when I lift the common wire to the relay coil off the track.

To be clear, the scope is showing the voltage between the center-rail and outer-rail of a DCS layout?  That is, the track is power by 18V AC and this is driving a 12V AC relay coil via the insulated-rail section? I set the track power to 12v for testing this.  Besides that the scope is hooked to the rails, the relay was driven using a pin wire jumper that I touched to the same point as the scope probe. The EMF we see is on the track power after I lift the relay power wire off the track. 

It's just that at 200 nsec per division, those spikes seem remarkably evenly-spaced as opposed to what would presumably be random interruptions in contact that I'd expect would be separated by more than ~100 nsec if a mechanically induced event.  Stated differently, are you sure you're not measuring some kind of high-speed digital circuit signal somehow coupling into your measurement? It's my first time measuring EMF and it is certainly not a high speed digital circuit. Just the track connection point to a passive TIU channel.

Separately, as to your question about EMF and driving a relay.  I believe you're addressing the issue of the so-called back-EMF spike when a relay coil quickly turns off such as intermittent wheel contact straddling an insulated-rail section.  Exactly. Seems to me though that if this EMF spike is from the relay coil, then that spike is not on the track side and the TIU is "safe". This spike is on the track side.  OTOH, if the EMF spike is from the inductance in the track (and the instantaneous interruption in current), then it seems to me you have far bigger problems.  That is, a relay coil probably draws 1/2 Watt of power or, say, tens of mA at 12V.  But an engine or even just a bulb lighted passenger car can draw tens or hundreds of times more current.  In which case an intermittent loss of wheel contact would generate a spike in the thousands of Volts thru the same inductance!    

 

Responses above, Thanks.

On your last point about why a light bulb does not create an equivalent spike, I thought a light bulb would be considered an almost purely resistive load and is quite a bit different from a relay coil in terms electron momentum. Not sure about locomotives though. Never did look for spikes on the scope wile they are running.

Ryaninspiron posted:

 

 

gunrunnerjohn posted:

The best way to prevent this kind of issue is not to let it happen in the first place.  If you totally isolate your accessory power and use a soft close and release circuit for the relay, there will be no coil kickback to bite you.

This is the circuit I use for insulated rail triggering of signals.  The choke provides DCS signal protection and the resistor provides inrush limiting, there is no issue with arcing at the rails.  The relay has a soft power-on and power-off to eliminate any kickback spikes from the coil.  AAMOF, I don't actually use the diode at D2 as it wasn't needed, it's a knee jerk reaction for me to put a diode across any DC relay even though it's not always required.

With this kind of circuit, you can power the accessories from separate isolated power using the isolated relay contacts.  Any spikes generated by the accessories are not directly transmitted to the tracks.

Certainly an interesting circuit, nice work. Also aside from the inrush resistor limiting current, isn't D1 also going to do the job even if the emf spike somehow backs out of the voltage regulator input? (not sure that's even possible.) Just scientifically curious why it wasn't needed.

There is no spike from the coil in this circuit.  The voltage smoothly drops as the capacitor discharges, there is no sudden disconnection to cause a fast collapsing magnetic field for the relay coil.

The inrush limiting isn't about spike suppression, it's to limit the sudden current through the circuit so there's no arcing at the wheels when the initial contact is made.  L1 is to prevent DCS signal degradation.

gunrunnerjohn posted:

Ryaninspiron posted:

 

 

gunrunnerjohn posted:

The best way to prevent this kind of issue is not to let it happen in the first place.  If you totally isolate your accessory power and use a soft close and release circuit for the relay, there will be no coil kickback to bite you.

This is the circuit I use for insulated rail triggering of signals.  The choke provides DCS signal protection and the resistor provides inrush limiting, there is no issue with arcing at the rails.  The relay has a soft power-on and power-off to eliminate any kickback spikes from the coil.  AAMOF, I don't actually use the diode at D2 as it wasn't needed, it's a knee jerk reaction for me to put a diode across any DC relay even though it's not always required.

With this kind of circuit, you can power the accessories from separate isolated power using the isolated relay contacts.  Any spikes generated by the accessories are not directly transmitted to the tracks.

Certainly an interesting circuit, nice work. Also aside from the inrush resistor limiting current, isn't D1 also going to do the job even if the emf spike somehow backs out of the voltage regulator input? (not sure that's even possible.) Just scientifically curious why it wasn't needed.

There is no spike from the coil in this circuit.  The voltage smoothly drops as the capacitor discharges, there is no sudden disconnection to cause a fast collapsing magnetic field for the relay coil.

The inrush limiting isn't about spike suppression, it's to limit the sudden current through the circuit so there's no arcing at the wheels when the initial contact is made.  L1 is to prevent DCS signal degradation.

Ah yes, I see. No dropouts other than when the regulator drops below it's minimum input. In terms of component cost, Couldn't the regulator also be replaced with with the correct resistors (of appropriate wattage) to bring the relay to an appropriate voltage?

The circuit was intended to run on anything from 10 VAC to 22 VAC, the regulator insures consistent operation of the relay over the entire voltage range.  Given that I buy the 12V regulator for 7 cents, it didn't seem too ostentatious to have the better and more reliable circuit.

gunrunnerjohn posted:

The circuit was intended to run on anything from 10 VAC to 22 VAC, the regulator insures consistent operation of the relay over the entire voltage range.  Given that I buy the 12V regulator for 7 cents, it didn't seem too ostentatious to have the better and more reliable circuit.

I see, good point. Not a bad deal for 7 cents then.

Ryaninspiron posted:

stan2004 posted:

...To be clear, the scope is showing the voltage between the center-rail and outer-rail of a DCS layout?  That is, the track is power by 18V AC and this is driving a 12V AC relay coil via the insulated-rail section? I set the track power to 12v for testing this.  Besides that the scope is hooked to the rails, the relay was driven using a pin wire jumper that I touched to the same point as the scope probe. The EMF we see is on the track power after I lift the relay power wire off the track. 

... 

...

On your last point about why a light bulb does not create an equivalent spike, I thought a light bulb would be considered an almost purely resistive load and is quite a bit different from a relay coil in terms electron momentum.

...

But that's exactly the point!  I think we are in agreement that the EMF spike originates at the relay coil due to the sudden interruption of "electron momentum" when it is disconnected from the track.  But since the relay coil is disconnected from the track at that point in time, how can that EMF spike be measured on the track-side?  For example, if you have a chance, why not measure the EMF of the relay coil itself when you lift the power wire to the relay from the track.  I believe you will see dynamics that are much slower than the nanosecond pulses you're seeing.

That's why I'm curious about EMF spikes originating from the interruption of "electron momentum" in the inductance of the track itself.  In other words a straight, non-coiled, piece of wire or track has some inductance - assemble a few hundred feet of track and I can imagine the inductance being large enough that a sudden power interruption to any type of load (bulb, engine, etc.) can unleash the stored energy in that inductance.

Bottom line #1: I don't believe the EMF from a relay coil powered by an insulated rail section is an issue wrt damaging a TIU channel.  Instead, it is the EMF from large engine loads that are suddenly interrupted from derailment or intermittent power interruptions going over turnouts, crossovers, etc.

Bottom line #2: Since the EMF from a coil or long length of wire/track is proportional to amount of current being interrupted, reducing current levels is a good thing.  Optocouplers can be a key component in reducing the triggering current in an insulated rail circuit to a few mA rather than tens of mA...so at minimum a 10x reduction.  Likewise, since it is the suddenness/abruptness of current interruption that directly affects the size of the EMF spike from an inductor, strategically placed capacitors can slow down interruption and drastically reduce the EMF. 

stan2004 posted:

As Dave says, opto isolation can be a remarkably effective and inexpensive solution to noise issues.  From my previous post I'm still not clear on exactly what problem is being solved.  In any case, refer to this recent thread from which I copied the following:

In the other thread I was proposing changes to a $2 (4-channel) DC-input optocoupler module to make it compatible with AC O-gauge.  I suggested a bridge rectifier and a capacitor.  And there they are in the OIB-8!  Member rtr12 might be making a circuit board that would bring the cost per channel way down relative to the OIB-8 price.  In any case, unless you already have a box of AC coil relays looking for gainful employment, you simply can't beat the price/value of the eBay DC relay modules (about $1 per relay).  They have trigger buffers so that you don't directly drive the relay coil.  Hence you only need a fraction of the ground trigger current flowing thru the insulated-rail section - just a few mA rather than tens of mA.  In general, reducing the trigger current requirement gives you more options.

If optocoupler isolation for relays is something you're considering, you may want to tack on to the other thread if the focus of this thread is EMF.

There is an almost complete but "never tested" 4 channel version of a PCB using the PS2505-4 IC chip which accepts an AC input. Parts are ordered, but I don't have them and it will probably be a while (Aliexpress). Needless to say, no testing until parts get here and no sample PCBs have been ordered. 

The circuit used for the PCB was the one discussed in the other thread that Stan linked to above and is similar to the 'ready made board' in the center of the top picture above. It doesn't have the bridge rectifier like the OIB-8 in the second picture above. 

If anyone is interested I can post the Diptrace files of what I have so far. The PCB layout file needs some additional labeling and possibly some further arrangement of the components. And, as stated earlier it is 'completely untested'. Posting the files to the other thread would be more appropriate, so that's where I would put them. That's where I was going to post them after they are tested and working.