@Jerryrails posted:I don’t think I understand any of this but it seems that you are all suggesting I could now use the MTH WTIU and the Lionel Base 3 on my new layout and get them to work together. I just need this gizmo being discussed? Is it doable for someone like me? Is there a path to a hand holding list of directions? I would like to run both Legacy and DCS
Patience grasshopper, it's being fine tuned. I think it'll be soup soon. 
@Jerryrails posted:I don’t think I understand any of this but it seems that you are all suggesting I could now use the MTH WTIU and the Lionel Base 3 on my new layout and get them to work together. I just need this gizmo being discussed? Is it doable for someone like me? Is there a path to a hand holding list of directions? I would like to run both Legacy and DCS
So we already run Lionel and MTH/DCS in our club and in most cases with some fine tuning (GRJ's booster, some signal level adjustment) it works pretty well. I would imagine for a home layout that's enough to get you going.
The problem is this stuff is both locomotive and weather dependent so in a big club with lots of trains and lots of members there are isolated cases where a particular locomotive doesn't work well in a particular part of the layout (and often only in particular humidity levels). The work we're doing on this new suppressor idea is to try to solve those edge cases. We're sharing the progress as we test it out so the other folks with large layouts can share in what we find and learn
@Adrian! posted:So we already run Lionel and MTH/DCS in our club and in most cases with some fine tuning (GRJ's booster, some signal level adjustment) it works pretty well. I would imagine for a home layout that's enough to get you going.
The problem is this stuff is both locomotive and weather dependent so in a big club with lots of trains and lots of members there are isolated cases where a particular locomotive doesn't work well in a particular part of the layout (and often only in particular humidity levels). The work we're doing on this new suppressor idea is to try to solve those edge cases. We're sharing the progress as we test it out so the other folks with large layouts can share in what we find and learn
Where I think this is going to help me personally is with our modular club. Since the wiring is hardly "DCS inspired", the signal levels are not optimal in any case. If this can aid in some way improving the DCS signal quality, that would be a great help.
Cool, that works. Any thoughts on trying shielded inductors vs unshielded?
@gunrunnerjohn posted:Cool, that works. Any thoughts on trying shielded inductors vs unshielded?
I mean step one is to try this one out on the layout tomorrow and see how it goes. If it's helping the DCS performance then I can do a production run... if it totally sucks (as I expect) then I think my next step is hole thru scale components. I mean shorting out my 50 ohm sig gen is a lot easier than shorting out the ~10 ohms of source impedance I measured between the rails.
With a good ferrite core inductor you can get ~20uH or so with a series resistance like 0.01 ohms or so. Then use tantalum caps so the ESR is very low too. That drives the general look of it from the $1 to $15 regime, and the 0.3" scale to the 2" scale, but whatever, it needs to work, right?
@Adrian! posted:With a good ferrite core inductor you can get ~20uH or so with a series resistance like 0.01 ohms or so. Then use tantalum caps so the ESR is very low too. That drives the general look of it from the $1 to $15 regime, and the 0.3" scale to the 2" scale, but whatever, it needs to work, right?
I'm not as worried about the final price as I am about the effectiveness of the filter. 
We spend thousands on the layout and locomotives, etc., so spending a bit more to make them run right doesn't seem to be much of a sacrifice.
OTOH, regarding perfect being the enemy of good enough, it depends on how much more effective the additional cost makes the filter. If we get 90% of the benefit from a $2 parts cost that we could get from a $15-20 parts cost, then it might make sense to do a volume run of the $2 board. I am thinking that the boards would incorporate two sets of filters in parallel if the parts only cost a couple bucks, that would help with the performance it appears.
@gunrunnerjohn posted:I'm not as worried about the final price as I am about the effectiveness of the filter.
OTOH, regarding perfect being the enemy of good enough, it depends on how much more effective the additional cost makes the filter. If we get 90% of the benefit from a $2 parts cost that we could get from a $15-20 parts cost, then it might make sense to do a volume run of the $2 board. I am thinking that the boards would incorporate two sets of filters in parallel if the parts only cost a couple bucks, that would help with the performance it appears.
It seems to be uniquely the model train people in our club that pour $2900 into a locomotive, then immediately spend a month complaining about smoke fluid costing $19.
@Adrian! posted:It seems to be uniquely the model train people in our club that pour $2900 into a locomotive, then immediately spend a month complaining about smoke fluid costing $19.
Actually, that's pretty universal. I had a discussion with a guy about looking for smoke unit home brew recipes for his fleet of Legacy steam locomotives. He couldn't understand my stance of simply buying a couple big bottles of JT's Megasteam and using them until I needed more! He was sure he could save money by mixing up who know what to pour into his engines. 
I thought that the MLCC caps had a very low ESR, is there something I'm missing about their characteristics for this application?
@gunrunnerjohn posted:I thought that the MLCC caps had a very low ESR, is there something I'm missing about their characteristics for this application?
These ones were el cheapo… China film capacitors off Amazon. You’re right… a good ceramic or mlcc is low ESR but like real Murata ones with an actual QC department involved
@Adrian! posted:
These values are right on peaking 451 KHz.
Standard sizes don't do it. 6.8nF is a little too low freq, 5.0nF is a little too high... so split the difference and it's just right.
So Adrian, are you saying that something like 5.9 to 6.0 nF is about right and gives us the magic 455KHz notch?
@Adrian! posted:These ones were el cheapo… China film capacitors off Amazon. You’re right… a good ceramic or mlcc is low ESR but like real Murata ones with an actual QC department involved
I typically buy my stuff like that from Digikey or Mouser, they normally carry higher quality brands.
@Rod Stewart posted:So Adrian, are you saying that something like 5.9 to 6.0 nF is about right and gives us the magic 455KHz notch?
So the ideal value is lying between two standard values (5.0nF and 6.8nF) , so I split the difference like this:
(22uH)--(5nF)--(22uH)--(6.8nF)--(22uH)--(5nF)--(22uH)--(6.8nF)
The effective cap is then the midpoint between those values and yeah its very close to 450 KHz. As I posted earlier... the concept is "working kind of" we still have some issues with the design in terms of series resistance so I'm going to work out a rev 2 with less lossy components.
I'll of course share that when I have it done
I'd love to find some SMT inductors with a lower DC resistance and/or a higher Q. I think the better capacitors can be had in surface mount without too much issue.
I wondered when I'd get to the point where I would be probing my layout with either of my two RF network analyzers! Great work on this, Adrian!
I did a quickie layout, I'm shooting for a compact footprint so it can be easily installed in a locomotive. Since there's no chance I'd have an IR oven to build these, small quantities will be hand assembled, and a larger quantity would be sent to a PCB house. That being the case, I did a 2-sided design for compactness. Since it's assumed that these would likely be employed in parallel, I have a jumper between the front and back that can be installed to parallel two filters. Since this is a prototype, I could leave that out if it turns out that's the default configuration and just connected them always in copper.
I found inductors with a .3 ohm resistance, considerably better than the first ones used.
@gunrunnerjohn posted:I did a quickie layout, I'm shooting for a compact footprint so it can be easily installed in a locomotive. Since there's no chance I'd have an IR oven to build these, small quantities will be hand assembled, and a larger quantity would be sent to a PCB house. That being the case, I did a 2-sided design for compactness. Since it's assumed that these would likely be employed in parallel, I have a jumper between the front and back that can be installed to parallel two filters. Since this is a prototype, I could leave that out if it turns out that's the default configuration and just connected them always in copper.
Happy to report there was some initial testing at the AGHR club. This configuration of 2-parallel of the SMD ones like you have here works "pretty well" but not "perfect". I'm lazy and don't want to do it all again, so we might try just doubling up (4-parallel) what we have before we revisit.
I also tried discrete components, but the notch gets really pointy and hard to align. A tunable L or C may be the next step there, but like I said the SMD ones are getting us 95% of the way there already (plus the cost and difficulty of the installation is so much lower)
@Adrian! posted:Happy to report there was some initial testing at the AGHR club. This configuration of 2-parallel of the SMD ones like you have here works "pretty well" but not "perfect". I'm lazy and don't want to do it all again, so we might try just doubling up (4-parallel) what we have before we revisit.
I also tried discrete components, but the notch gets really pointy and hard to align. A tunable L or C may be the next step there, but like I said the SMD ones are getting us 95% of the way there already (plus the cost and difficulty of the installation is so much lower)
I think in general that SMD components tend are a little more consistent than thru-hole components, at least that's been my experience. I'm going to order some PCB's for this configuration and do a little bench testing. I dusted off my sweep generator, it's been in hibernation for several years. I wanted to make sure I had the ability to do the bandwidth testing.
As for the 90%-95% achievement, perhaps we don't want perfect to be the enemy of good enough. It sounds like trying to go that last little distance may be exceeding the point of diminishing returns.
@gunrunnerjohn posted:I think in general that SMD components tend are a little more consistent than thru-hole components, at least that's been my experience. I'm going to order some PCB's for this configuration and do a little bench testing. I dusted off my sweep generator, it's been in hibernation for several years. I wanted to make sure I had the ability to do the bandwidth testing.
As for the 90%-95% achievement, perhaps we don't want perfect to be the enemy of good enough.
I'm an academic and a researcher 
It needs to be perfect or it bugs me on the inside. Until DCS has the same responsiveness with the legacy base on as it does with the legacy base off, I'm always going to have a bad feeling on the inside when running trains 
@Adrian! posted:I'm an academic and a researcher
It needs to be perfect or it bugs me on the inside. Until DCS has the same responsiveness with the legacy base on as it does with the legacy base off, I'm always going to have a bad feeling on the inside when running trains
You're right, you're doomed to be bugged by most things in the outside world! 
I'm happy if I can solve a problem. If everything runs great, I'm not that concerned what the 'scope is telling me. 
==================
REMINDER
==================
Just to reiterate for everyone reading this (and those emailing). 99.9% of the people here will not need this device. It only solves one very specific problem:
If you find your DCS is unresponsive with the Legacy ON, but works fine with the Legacy OFF, especially in certain spots of a layout or with specific engines then this might help you. It will not be helpful in any other weak control signal situations. It doesn't make DCS stronger or Legacy stronger, it just helps keep them separated in the same layout.
Actually, I've personally seen this on both Harry Henning's large layout at his house and on our modular club layout, so I suspect the percentage that will benefit is a bit more then 0.1%! I do notice even on my fairly small 12 x 24 layout that I sometimes have trouble adding engines, but switching off the power to the Legacy base allow them to add. I usually isn't a problem, so I never chased it down, but this might be the solution.
I've been watching this thread with interest and am curious if there are any testing updates to report or circuit modifications in the works, maybe with a tune-able center frequency.
FWIW, I know if someone cranks up the TMCC/legacy signal booster on our club layout, (I built it with the adjustment feature) it will easily just overpower over the DCS signal. So yes, I'm following this for that situation- if you have a booster- this might give me a little more headroom to turn up TMCC/legacy and not drown out DCS.
I just got my prototype PCB's in, so I'll be putting a couple together and running some bench tests.
@gunrunnerjohn posted:I just got my prototype PCB's in, so I'll be putting a couple together and running some bench tests.
I too am quite interested in how this works for the common man's layout. Looking forward to your results John!
For those following we did put these in at AGHR. It's not *perfect* ... like there's still mode contamination voltage you can see on the scope as you move the ground clip around between building ground and layout ground but it is definitely "better" by a factor of 2-3 in voltage.
The DCS trains do feel more responsive, it takes us from like 1/10 commands getting through to like 8/10 commands getting through. The theory is sound and the testing agrees. Its not a magic solution though, just an improvement from awful. We're going to make up more modules so we can try stacking more than one per block and see if it improves more, or if we're reaching diminishing returns.
Well, we never expected "perfect", that's pretty hard to attain. Much better is a big improvement.
Well, mixed reviews.
When I do one simple filter, I get about a 2:1 reduction in the amplitude of the signal, but it centers a bit too high at closer to 500khz.
When I parallel both sides of the filter to get two filters running, I get waveform distortion and a more amplitude reduction. The distortion of the waveform bothers me a little...
It seems I'd have to jack the 5nf caps up to perhaps the 6.8nf value as well.
@gunrunnerjohn posted:Well, mixed reviews.
When I do one simple filter, I get about a 2:1 reduction in the amplitude of the signal, but it centers a bit too high at closer to 500khz.
When I parallel both sides of the filter to get two filters running, I get waveform distortion and a more amplitude reduction. The distortion of the waveform bothers me a little...
It really comes down to the output impedance of the sig gen you're using. This thing is like ~7 ohms series on resonance for mine. the distortion might be a result of the source driver not being able to deliver the demanded current across the full swing
Vo = Vin x Zfilter/(Zsource+Zfilter)
Very good point on the output impedance, I totally failed to consider that. I am, after all, shorting out the output at the notch frequency! I used inductors that had around .3 ohms DC resistance to try to improve the performance.
I may have to try connecting up my TMCC Buffer to beef up the signal, after all it's primary purpose is to lower the impedance of the output signal while amplifying it.
I do have to move the notch down closer to 455khz... I also discovered I used the wrong footprint for the inductors, so I'll get some PCB's with a larger footprint.
@gunrunnerjohn posted:Very good point on the output impedance, I totally failed to consider that. I am, after all, shorting out the output at the notch frequency! I used inductors that had around .3 ohms DC resistance to try to improve the performance.
I may have to try connecting up my TMCC Buffer to beef up the signal, after all it's primary purpose is to lower the impedance of the output signal while amplifying it.
I do have to move the notch down closer to 455khz... I also discovered I used the wrong footprint for the inductors, so I'll get some PCB's with a larger footprint.
Yeah, there's some subtly here. You're not trying to short out a sig gen at 455 KHz... and you're definitely not trying to short out your booster output at 455 KHz...
You're trying to short out two nodes that are both driven by the booster together (wrt to building ground), one driven directly (track outer rails), the other through a slightly higher impedance (center rail, driven via the TIU output stage and 60 Hz power supply).
Access to this requires an OGR Forum Supporting Membership
