Looking for TVS recommendation for Motor driver PCB output to "15V" Pitman

The reason I'm asking this question is because, I only have a vague notion of how the recommendation came about for using a 1.5KE39CA at track voltage levels.  What I think I understand about this choice (please correct me if mistaken) is the 1.5KE39CA has a Reverse Voltage Standoff rating of 33.3V which is about 18% above the 28.3V Peak voltage of the maximum 20V RMS output of a typical O gauge transformer.  Reverse Standoff Voltage is the point below which a TVS diode conducts very little current.

Since the DCDS board is a bipolar bidirectional DC power source, there should be no RMS consideration.  So my thinking is that a TVS diode with a lower Reverse Standoff Voltage than the 1.5KE39CA could be used to provide better protection of the DCDS.

Would the voltage in the 15.1 VDC motor driver circuit (excluding transients) ever significantly exceed this voltage under normal operating conditions?

Are there any thoughts about using something in the range of a  1.5KE20A (RSV = 17.1V) or  1.5KE22A (RSV = 18.8V) for this application?

Any help would be greatly appreciated.  Thank you, Steve

I have no knowledge about the motor drive circuitry in "expensive" Lionel engines.  But if it's using modern electronics, the motor would be driven by FET type transistors that are pulse-width-modulated for speed control.  Assuming this is the case, the current to the motor is being interrupted thousands of times per second!

Additionally, assuming this is a DC can motor, motor brushes are notorious for interrupting the continuous flow of current into the motor windings.

I suggest any transients on the track voltage will be filtered by the AC-to-DC conversion in the engine - that is, the designers would have built protection into the circuit to prevent large voltages from reaching the transistors that drive the motor.

Also, for modern transistor motor drive circuits, you will see protection diodes either built-in (so called body diodes) to a FET type transistor or external diodes for older bipolar type transistors.  These serve to protect the transistor from seeing large voltage excursions by essentially shorting the transient into the capacitance of the DC supply or to DC ground as the case may be.

But I see no negative effect of installing a TVS - it's a no harm no foul.  Though I don't quite understand what you mean by a 15.1V DC motor.  I find it hard to believe Lionel circuit apply a steady 15.1V DC to the motor?!  I am of the school of thought that there is really no such thing as, say, a 15V motor.  It may be most efficient at 15V DC in terms of Watts in vs. Horsepower out...but you can apply 12V DC to it or 18V DC to it and as long as you stay within its power envelope (Watts of dissipation), then you're OK.   That said, the most cost effective motor drive circuitry for O-gauge would be to use a bridge rectifier to create a DC motor supply.  This would peak at ~1.4 times the RMS AC track voltage for a sine-type supply so that with 18V AC on the track, the motor bridge circuit would be working with over 20V DC and the motor would see voltage pulses in that range.  So using either of the two parts you list would short the motor drive on each pulse!  Again, if Lionel actually regulates the motor power supply to 15.1V DC and only switches that voltage to the motor, then I retract this entire paragraph...though I'd like someone to explain to me why they'd do it that way!

Stan, the motor is rated at 15 volts, but being a plain can motor, it will obviously run on any reasonable voltage.  I agree that a TVS across the motor wouldn't pose any issue AFAIK.

Stan and John, thank you for your replies.  I really appreciate your insights on this.

I'm going to take some measurements and report back when I have more details about the DCDS output.

@SteveH posted:

Are there any thoughts about using something in the range of a  1.5KE20A (RSV = 17.1V) or  1.5KE22A (RSV = 18.8V) for this application?

If you decide to go ahead with this project, I'd err on the high side, maybe the 1.5KE22A part.  The last thing you want happening is the motor drive FET's banging against a TVS and drawing high currents.   I don't know that this will have any effect, but I suspect it probably won't do any harm.  The reverse voltage spec for the FET's used on the DCDS is much higher than the 1.5KE22A part.

Maybe I'm missing something.

Seems to me like the DC bus, and hence the peak of the PWM waveform, is going to be 1.4 X track voltage regardless of the motor nameplate, at least with the low-cost controls we are discussing. I have worked on industrial stuff with SCRs or transistor rectifiers that in fact controlled the DC bus voltage but I doubt that any of our stuff is that sophisticated. Max track voltage of 22 gives a PWM peak of around 31 regardless of how the motor is rated, or what the effective value of the PWM output is. Yes, you can argue about a couple diode drops in the rectifier.

Is there a substantial capacitor across the H-bridge output to average the PWM waveform? That would make my argument faulty, but it would force the H-Bridge to supply very high pulses of current, which I suspect they aim to avoid. And, if there is such a capacitor, that already provides transient protection.

Valid points, I guess we should actually scope the motor drive to see what the real situation is.

@PLCProf posted:

...

Is there a substantial capacitor across the H-bridge output to average the PWM waveform? That would make my argument faulty, but it would force the H-Bridge to supply very high pulses of current, which I suspect they aim to avoid. And, if there is such a capacitor, that already provides transient protection.

I guess it depends on what you mean by "substantial." A full H-bridge motor drive by definition can reverse the voltage polarity to the motor.  So that capacitor would need to be bipolar.  A "substantial" capacitor that can smooth the motor pulses would be hard to come by and most likely quite expensive.  Those small capacitors you frequently see across DC can motors are bipolar but are for suppression of motor noise so a nearby AM radio does not go nuts.

@stan2004 posted:

I guess it depends on what you mean by "substantial." A full H-bridge motor drive by definition can reverse the voltage polarity to the motor.  So that capacitor would need to be bipolar.  A "substantial" capacitor that can smooth the motor pulses would be hard to come by and most likely quite expensive.  Those small capacitors you frequently see across DC can motors and bipolar but are for suppression of motor noise so a nearby AM radio does not go nuts.

Yes- Exactly - And the key to PWM device selection is the fastest possible switching time to reduce device dissipation, but fast rise-time increases the charging current through any capacitance on the output.

In industrial motor controls, inductors are often/usually fitted to PWM outputs to isolate the drive from the ill effects of cable capacitance.

If you think about it, the "flyback" (reverse polarity) diodes connected across the power devices limit the voltage excursion of the output to the value of the bus voltage.

OK-

Had a few minutes this afternoon, and needed an excuse to play with my new differential scope probe (trying to make GRJ jealous...)

Took some scope shots from a WBB 2 motor diesel into which I had installed a DC Cruise Commander. Power is nominal 18 VAC from a 180W brick through a Legacy Powermaster.

See image "Full speed overall" -PWM locked out, motor tied to bus, bus ripple in evidence between about 16 and 23 volts. The "scratchy stuff" between bursts is the back EMF from the motor used in the speed feedback loop. Back EMF about 15 volts. The scope says that the P-P value is 28.4, but I suspect the negative blip is too short to be measured accurately.

Image "Mid speed detail" shows the overall waveform at a somewhat lower speed, you can start to see the PWM pulses. Bus ripple still evident, back EMF about 12 volts. P-P output 25.4

Mid Speed fine detail - Expanded view of above, detailing PMW waveform. P-P output 24.6

Slow speed overall - motor just crawling, back EMF maybe 1 volt, back EMF measuring periods very long. P-P output 25.2

Slow speed detail - expanded view of above, skinny PWM pulses. P-P  output 25.0

I don't see anything that looks like a transient that I would worry about, the back EMF voltage is noisy but it is limited to the supply rail voltages as previously discussed.

Note that in all cases the P-P motor voltage is right around the track voltage x 1.414.

I have no idea what the rated motor voltage is.

In reality you have a dual-layer PWM system, you have the 25 kHz chopping frequency, but you also have the measurement interval for the back EMF. At slow speed that interval is wider than the power-on interval!

@PLCProf posted:

Had a few minutes this afternoon, and needed an excuse to play with my new differential scope probe (trying to make GRJ jealous...)

OK, I"m suitably jealous.   What differential probe are you using?  For that matter, what model Regol 'scope is that?  Jealousy may run in both directions.

One curiosity about what you're seeing.  The driver board being discussed is the Lionel DCDS, no back-EMF is used with that board, it's a tach driven speed control.  I don't know what you'd see on the board Steve is speaking of.

I have a few tach-controlled Legacy and TMCC locos but I don't know what driver card is in them. Let me check.

Steve's locomotive would be Odyssey I with the classic TMCC Magnet ring on the flywheel and a hall effect sensor.

I don't think I have any Odyssey 1 anymore. Maybe I'll look at an Odyssey 2 just for giggles.

Boy, you've excised all the TMCC with cruise from your stable, that's pretty picky!

Only ancient thing I have is a K-4 that was sold as TMCC convertible, you needed to pull out the E104 card and install the R2LC, which I of course did. Don't know where that fits into the hierarchy.

Well, it doesn't have Odyssey, so that one is out.

@gunrunnerjohn posted:

OK, I"m suitably jealous.   What differential probe are you using?  ...

For most if not all train stuff looking at differential signals, I use a 2-channel DIY differential probe - $0 out-of-pocket using parts from the stash.  I see I fabricated a PCB for it but this was sometime in the last century so I'm sure the documentation is in the bit bucket.  But it's just a dual-channel op-amp configured with 10x scale-down so as to be compatible with a scope 10x probe setting for properly scaled on-screen parameter measurements.  The other reason for a 10x scale down is to get a common-mode voltage range to allow the same scope to probe O-gauge track voltages with its "ground" (outer-rail) and  the circuit board signals after a bridge rectifier with its "ground."  Only a few MHz bandwidth but even for, say, the signals shown above - 25 kHz PWM motor drive - a 1 MHz bandwidth probe is 40 times the pulse rate which is adequate for seeing what's going on.

I can't off the top of my head think of a train application where I needed a high-speed (i.e., >100 MHz bandwidth) differential probe. Those probes are VERY spendy - no doubt more than the cost of a Regal scope.

@stan2004 posted:

Those probes are VERY spendy - no doubt more than the cost of a Regal scope.

This is the one I use for train projects, but train projects are a small part of my electronic activities.

Differential Probe 1:10/100, 25 MHz, Model 4232|Probe Master

It is designed for power circuits rather than fussy RF stuff, the test leads are stout and unshielded. For my purposes I need the higher voltage rating. Not really that expensive, but not free either.

Too bad the documentation for the budget differential probe is down the drain.

@PLCProf, @gunrunnerjohn, and @stan2004 Thank you all for your significant insights, contributions and comments. I really appreciate your willingness to help me on this quest and to further the collective knowledge base here on the forum. I've learned a lot with your guidance working on this project.

My initial measurements reveal that the maximum peak to peak voltage across the motor in my set-up is around 29 Volts when the transformer is briefly pushed to full throttle (19.6VAC under load). I wouldn't normally run at this voltage, but thought it wise to test there to know where the limit is. I checked voltages in 3 modes of operation: Conventional with Odyssey off and on, and using Cab2 TMCC control. The latter produced a max. voltage reading about ½ volt higher than both conventional modes. Throughout the testing I took detailed voltage measurements of input and output voltages for each of these 3 operational modes at minimum, medium and maximum speeds using the three measuring tools described below.  I may post those measurements later (or sooner if anyone is interested).

Before testing I installed a 1.5KE39CA across the track power at the DCDS input.

Testing apparatus set-up is as follows:

• Power to the locomotive is provided using a postwar ZW transformer through a 5A Airpax Instant circuit breaker.
• 20MHz dual trace analog oscilloscope in X+Y Subtract mode to facilitate differential measurements, using two standard probes (ground leads removed) set at 10x, tips connected across the intact motor leads. The display was set to show 5 volts and 5mSec. per division.
• Triplett 630-APL Type 5 RMS VOM and an inexpensive DMM.

The 3 videos (attached at the bottom of this reply) show the oscilloscope wave forms at start-up, accelerating to full speed, and decelerating back to a stop.  This video shows TMCC operation:

With a second 1.5KE39CA temporarily connected across the motor leads, there was no noticeable change in the output voltage, waveform nor motor operation. Based on these findings, I'm thinking that a 1.5KE39CA (with its 33.3 reverse standoff voltage rating) may be a good choice for this application.

Without knowing the time constants of and operational charge on the capacitors in the DCDS output and consequently how much of any voltage spikes they may instantaneously absorb, I would still like the added assurance that a TVS diode in the circuit would provide against back EMF spikes from the motor due to sudden power interruptions.

Given this information, what are your opinions about which TVS would be appropriate to install across the DCDS output?

Thanks again, Steve

I sort of like your altitude on the pick, the 1.5KE39CA seems like a reasonable choice based on your measurements.

I will offer up one caution based on real world experience!

The most prevalent failure mode of a TVS protection diode is a short.  However, that is also the failure mode that will very likely kill your motor driver, be it TMCC or Legacy.  I know that a shorted motor on the DCDS or even the newer Legacy RCMC will take out the motor drivers in a flash, been there done that more than once!   Personally, I don't think I want a TVS in position to zap my board, since that was the whole purpose of installing it in the first place.  Of the three RCMC boards killed by the crappy Canon motor shorting, I managed to rescue only one by installing new FET driver parts.  I actually have better luck with the TMCC or early Legacy DCDS in repairing them, but that's no reason to roll the dice.

I'm pretty sure that a dead short ed TVS across the motor leads would have a similar effect on the motor drive circuit.

My advice is this is one "addition" you may want to rethink.

@SteveH, I got a notification of a post from you in this thread, but when I came here the post was missing.  Did you delete it?

Just curious if there are any documented instances of a Pittman taking out a DCDS driver? I have quite a few and never lost any either at home or on the club layout.

One TVS from pickup rollers to ground should be more than sufficient. Older Lionel electronics are fairly rugged. MTH not as much.

The ones to be concerned about are RCMCs driving Canon motor’s. Those do have a history of taking out the drivers but a TVS won’t help here as its due to the motor shorting rather than an inductive pulse.

Pete

@Norton Thank you for letting me know about your positive experiences with DCDS and Pittmans.  I don't know of any specific instances of issues.

John, last night you made a very convincing case against the original notion of connecting a TVS in parallel with the motor if the TVS were to fail shorted.  The FETs would definitely not be happy with a direct short across them.

I did have an idea last night, but after initially posting it, I decided to reconsider,  then had a second idea that may be better.  I deleted that reply and decided to sleep on it before replying again.

  So per your suggestion, I'm rethinking this.  I think I like option A better and I'm not sure option B (the one I posted last night) would even work.  Here are a couple more ideas for consideration:

Option A

Resistor of a suitable value to prevent short circuiting DCDS FET outputs if the TVS fails in the shorted condition, but high enough for minimal voltage drop.  Shorted diode failure would blow FA fuse causing the lamp, conspicuously located below the locomotive, to illuminate.

Option B

Thinking that any TVS failure in this option B circuit would not present a direct short to the DCDS output.  Any motor coil spikes may (or may not) be dissipated back to the track power circuit where additional TVS diodes would be present.  But, I'm uncertain if there would be sufficient voltage potential to cause these side chained diodes to turn on.

The resistors would hypothetically be of a suitable low value to limit current (no dead short across the DCDS outputs) but not high enough to present a significant voltage drop to a spike.  Any motor coil spikes may (or may not) be dissipated back to the track power circuit where additional TVS diodes would be present.

Thoughts on these options?

@Norton posted:

Just curious if there are any documented instances of a Pittman taking out a DCDS driver? I have quite a few and never lost any either at home or on the club layout.

One TVS from pickup rollers to ground should be more than sufficient. Older Lionel electronics are fairly rugged. MTH not as much.

The ones to be concerned about are RCMCs driving Canon motor’s. Those do have a history of taking out the drivers but a TVS won’t help here as its due to the motor shorting rather than an inductive pulse.

Pete

I don't know of any documented case of such a Pittman failure.  My concern is the TVS failing shorted across the motor outputs and doing the deed.   Truthfully, this is simply a place I don't see the utility of a TVS.

OTOH, I do see the utility of a PTC in series with the motor leads.  I used a PTC for all the ERR Cruise-Lite installations I did as I had a motor stall that very quickly took out the FET drivers on the board, it only took seconds!  I haven't used a PTC for the full Cruise Commander or any of the Lionel DCDS boards, but I have had a few DCDS boards come in that needed new drivers from a short, normally a chassis short to a motor lead.

@SteveH posted:

Thoughts on these options?

Truthfully Steve, I think this is a solution in search of a problem.  I'm not seeing how this is warranted.  I've looked at a ton of motor drive circuits, and I've yet to see one where the designer's saw the need to incorporate a TVS into the outputs.  I suspect the reason is, the issue you visualize probably simply isn't happening.

John,  PTCs in series with the motor leads sounds like a good precaution.  What hold current value would you recommend?

Well, it would depend on the size of the locomotive.  However, I figure that a 2A hold is probably the biggest I'd go with, they don't trip until they get to around 4A, and you really shouldn't see 4 amps on any motor unless you have a stall  My Vision Line Big Boy pulls about 4 amps with all the smoke on pulling 30 cars, and about 1.5 amps max if I turn the smoke all off.

I use use a 1.1A hold one on the ERR CC-Lite boards.  I still have a couple of the CC-Lite boards, but when I use them, I don't want them cooking as there are no more.

For maximum protection, put a PTC in each motor lead, that way if either lead gets shorted to the frame, you have protection for the drivers.

I might have missed it or perhaps it's common knowledge amongst Lionel operators, but do you have a diagram, schematic, or close-up photo of the motor drive circuit for the board under discussion?  Does it use a FET H-bridge?  Or a bipolar bridge?  Or maybe an integrated motor drive IC chip?  Can you see or do you know the part numbers of the key components?

I don't know the exact configuration Stan, I'm assuming it's a FET bridge as the drivers are FET's.  Whatever they're doing, they're using the onboard uP for the controls, it's the only LSI chip on the board.  The driver FET's are the IRF540 and the IRF9540.

If that photo is on your bench, can you see the cap value between the FETs, and the bridge rectifier marking presumably telling us the Amp capacity?

@stan2004 posted:

If that photo is on your bench, can you see the cap value between the FETs, and the bridge rectifier marking presumably telling us the Amp capacity?

It's an GBU1002 10 amp bridge, and the large cap is 100uf 35V.

John, your current recommendation for PTCs sounds good, thank you.

I plan to take some current readings of the motor draw under load and choose an appropriate pair of PTCs for this application.

With the addition of PTCs in series with the motor leads, this would seem to allow a TVS paralleled across the motor leads back into consideration, since the PTCs should protect against a shorted TVS.

Other than being maybe overly cautious, does anyone foresee any technical issues with this arrangement?

@SteveH posted:

John, your current recommendation for PTCs sounds good, thank you.

I plan to take some current readings of the motor draw under load and choose an appropriate pair of PTCs for this application.

With the addition of PTCs in series with the motor leads, this would seem to allow a TVS paralleled across the motor leads back into consideration, since the PTCs should protect against a shorted TVS.

Other than being maybe overly cautious, does anyone foresee any technical issues with this arrangement?

Pic does not display.

@SteveH posted:

Thank you for letting me know.  How about now?

Bingo!

Electronics ignoramus here...  What is DCDS and what does it do?  (Tried googling it; best I found was https://www.spiedigitallibrary...2.2296146.full?SSO=1 which sounds like way over the top for trains)  Search also found Detroit County Day School - I'm guessing that's not it either.

@Mallard4468 Good question. In short it's the Lionel 691DCDSFP6 Odyssey 1 Single Motor driver board for a Pittman motor.

Without an actual schematic, there's some guessing involved.  Basically it is a AC to DC variable power converter and motor driver circuit board.  The DC voltage is a high frequency pulsing signal, positive for forward travel, negative in reverse.  This type of signal is called Pulse Width Modulation.  Longer spaces between pulses equate to the motor getting less power to run at slower speeds.

@SteveH posted:

John, your current recommendation for PTCs sounds good, thank you.

I plan to take some current readings of the motor draw under load and choose an appropriate pair of PTCs for this application.

With the addition of PTCs in series with the motor leads, this would seem to allow a TVS paralleled across the motor leads back into consideration, since the PTCs should protect against a shorted TVS.

Other than being maybe overly cautious, does anyone foresee any technical issues with this arrangement?

Steve, just be forewarned, seems to me a lot of hoop jumping for what is basically the Oddesy driver, …but that’s your gig, and have fun doing what you’re doing…..Oddesy can go flaky with out rhyme or reason, doubt your protection on top of protection would propel it to going flaky, but so you know, cracked magnets, flaky drivers, the lovely Oddesy lurch are all things Oddesy loves to do…..just giving you the heads up in case you box yourself into a corner, and then you’re looking at an upgraded driver …..and if you gotta go that route, might as well go all the way and get four chuffs, etc…..the J you bought is one of the best IMO as far as mechanics of it is concerned, and with some upgrades, can fairly easily be brought up to today’s standards…..I doubt I’d spend a lot of time trying to overly protect what’s considered by some an archaic system…that locomotive is completely modular, so board swaps aren’t life altering,…..😉

Pat