Does anyone know if any commercially available decoders have built in wave-clipping (back to back zeners across the motor outputs and/or the power inputs) to take care of high voltage spiking?
Doubt it. It's not necessary. Most H-bridge driver chips have built in diodes to protect the drive transistors from motor spikes.
--Randy
Modeling the Reading Railroad in the 1950's
Visit my web site at www.readingeastpenn.com for construction updates, DCC Info, and more.
Not to my knowlwdge.....I add back-to-back zerer diodes to clip any high voltage on the rails.....the diodes on the MD in the decoder do not protect the decoder INPUTS.
The way the zeners work is sometimes overlooked:
The clipper using 2 zener diodes in series with the polarity opposing each other is more complex than it first appears. It works because of the properties of a zener diode. Normally, a zener is used in a reverse bias state and 'trips' at the rated voltage, BUT the zener acts like a regular diode when forward biased. So, connected with reverse polarity, the forward biased diode is just a diode, and the reverse biased diode acts like a zener, as it should. The result is that the reverse biased zener 'sees' the track voltage PLUS one diode drop from the forward biased zener diode, which results in the trip voltage being the zener's trip voltage PLUS 0.7V. When the polarity of the track reverses, the results are the same, because the two zener diodes swap functionality because the biasing was reversed. The best part of the configuration is that there is only leakage current drawn when the voltage is less than the trip value.....a win-win in any case. So, because there is no real current passing through the diodes unless tripped, and the nature of the DCC signal, ONLY excessive voltage spikes actually cause current flow. The ratings of the diodes can then be minimal, like 1-2 watts(the low end of what devices are available), and don't get warm like the snubbers, which draw current more or less continuously.
I hope this expalnation makes sense.
Is this a problem? I have dozens of DCC locos and have never lost a decoder yet, to ANY cause, let alone high voltage spikes coming in the track inputs. And they run on more than just my modest home layouts, the club modular layout is something like 27x160 fully deployed - there are some pretty long bus runs there, and no snubber circuits on the bus wire, either. To my knowledge, no one has ever had a decoder fried on the actual layout.
Yes I understand tha tthe MD protection diodes only protect the MD. WHat else is generating large BEMF voltage spikes?
Would this help?
sure does...thanks. At our club we have had (not often but periodically) all of the symptons listed ....loss of control, loss of programming, and even decoder failure not due to lock up and overcurrents.
Unforunately, we have had "symptons" (see reply below) that could be attributable to "spiking" ...hence my interest in "spike protection". I have had an MDC decoder burn out in normal (albeit protracted) usage, a Bachman Soundtraxx deoder fail in normal usage, and a BLI Paragon lose its brains due to repeated "spiking".
I think the track inputs can be sparky...we see it when we turn the room lights off and jack up the speed step setting..especially if the wheels and/or track are dirty.
I'm not sure about the MD inputs. The built in FED diodes come into play when the FETs are "off"...they provide a path for the collapsing current in the motor inductance and prevent excessive voltage build up. But Zeners across the motor would seem to me to have a slightly different function? Namely shunting transient voltage spikes ...?
Do you run lots of Athearn BB locos with the old sintered iron wheels? Sparks between the wheels and rails don;t necessarily mean high transients, the 15V is MORE than enough to arc over such a tiny gap. Clean track would really help - and NEVER use abrasive materials to clean the track - if you have in the past, burnishing with a hardened metal tool will go a long way to KEEPING it clean.
The only transients off the motor will be clipped by the protection diodes. Between the drivers and the bridge rectifier feeding them, there's little chance of motor spikes getting back into the rails from a decoder. If the decoder uses BEMF to control motor speed, you don;t want anything interfering with that - Bachmann locos typically have an LC circuit on the motor leads which makes a mess of the BEMF signal. Cutting off the caps is usually sufficient, the chokes are in series witht he motor leads and would need to be bypassed with jumpers, but alone without the C component they don't do much anyway.
The need and use of snubbers does seem to be somewhat brand specific. That probably has to do with the driver design in the booster.
Well there MAY be H-bridge driver chips without the built in diodes, I would think they are kind of rare. Any use would be to drive a reactive load, which menas diodes are needed, plus any MOSFET transistor design has parasitic diode across it as a matter of how a MOSFET works.
Still not bad design to add external diodes - for example, the source switcher I designed to control my test/program track and switch between 4 or more input sources (DC, DCC, PR3 programmer or LokProgrammer), the driver chip between the ATMega328 and the relays already has internal diodes, but I was putting one on each relay coil anyway. Belts and suspenders. Space is not an issue. Inside a loco - different story. The single biggest part of most motor-only decoders in the bridge rectifier, or the set of 4 diodes doing the job. The controller and the motor drive, and the function drivers, are all small parts compared to that. Only a sound decoder might have the main chip being bigger than the input rectifier, only because so much die space is needed for the sound memory. With a larger HO loco, especially a steamer, there's probbaly plenty of room to incorporate extra protection diodes, if it makes you feel more secure. More benefit would come with them on the track input though. In smaller scales, space becomes an factor.
Given that there is usually a filter cap after the full wave bridge on the decoder, I'd be curious to see how much ripple and noise from the rails actually makes it through. Something to throw my scope on one of these days and see. There is also a voltage regulator with some additional capacitance before the main processor. Most micros have some form of input clamping, so even the input lead that feeds track power in for the decoding of the DCC signal isn;t going to be able to go too far out of bounds.