Deleted
Deleted Member
Posts: 0
|
Post by Deleted on May 20, 2017 23:50:17 GMT -5
Hello again. Now I am waiting for delivery of my engines (2 Budd's to be coupled together and 2 Gp7/9's) from HRT (Jörg Erkel) Germany. I plan to use TCA0372's (2 power OP-Amps's, max 1A, 1€) for locomotive power. Now, looking at the datasheet ( www.onsemi.com/pub/Collateral/TCA0372-D.PDF , page 5), I saw that the IC can be used both for PWM and pure analog control ; I read once about MTL engines : "no pulse" . What's better for AZL, pure (variable) DC or PWM (7.5v or 9v, about 250 Hz, as found in a former topic) ? I am thinking first of reliability (and durability), and less of "slow motion". Thanks for any answer ...
|
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on May 21, 2017 1:27:10 GMT -5
Thanks for the answers. But to add a litle bit confusion : home.iprimus.com.au/nuggles58/Making%20A%20Basic%20DC%20Controller.pdf" rel="nofollow" target="_blank">ece.buap.mx/pub/JCid/Publicaciones/control%20de%20motores.pdf I am convinced that PWM is the better solution for "slow motion" ... unfortunately it seems to be BAD by overheating the motor in its basic version ; I read that every increase of motor temperature of 10 degrees Celsius (18 Fahrenheit) halves the motor's life cycle (?). I saw a test of several systems here : www.scottpages.net/ReviewOfControllers.htmlA good system seems to be a "phase control" via a SCR, like used for 220v appliances motors ; the schematics of a "Gaugemaster" design are shown ... unfortunately the components are obsolete ! Another system for railroads is shown here : books.google.fr/books?id=xll-DAAAQBAJ&pg=PA341&lpg=PA341&dq=model+railroad+phase+control&source=bl&ots=x9L83rSpvJ&sig=WHmO2d4K19Ad6NyLHchsRyu5YkA&hl=fr&sa=X&ved=0ahUKEwjbhOCajIHUAhXFrxoKHbrHA5wQ6AEIcTAN#v=onepage&q=model%20railroad%20phase%20control&f=false(figure 19.17) but it is running very quickly into a "gaseous plant" (perhaps due to the overload protection). I built an analog servo control (published in a magazine) with another "dual" OP-Amp L272M (now obsolete) in the 90's and it worked without heat sink ... but the motor got only power for short lapses of time. So perhaps the best solution would be a simple transistor, and a relay for reverse swtching ? And no PWM : I think ultimately that realism is good, but longevity is better ... On my former layout there were only 2 speeds, with pure "analog" control and an unfiltered supply, and the locomotives were modified with two Zener diodes, because a Roco runs MUCH slower than a Trix or Fleischmann at the same voltage, I hope this problem should not occur with "AZL only" locomotives. So I will see on a test/breaking in layout how the various locomotives behave at same voltage, and perhaps I will give up my too sophisticated solution. And, to start a train at "slow speed", perhaps a short "full speed" pulse ? A microcontroller does this very well ! As I wrote in a former topic, better solutions are the worst ennemies of good solutions .
|
|
|
Post by markm on May 21, 2017 17:29:14 GMT -5
Alberich,
If you have access to the Model Railroader magazine archives from the late 1960s to mid 1970s I can point you to a number of articles on throttle design and the "optimum" characteristics for powering the HO motors of that era. The basic conclusions from that era were: 1. Pulse power enhances low speed operation 2. Pulse power is desirable for acceleration 3. Pulse power provides no benefit for operation above 20-30 mph scale 4. Pulse power increases motor heat.
The throttle designs were DC with pulse power superimposed at low speed and during acceleration.
My opinion is that is work can be applied to the 3 & 5-poled motors use by Märklin and MTL F7s. I'm unaware of any serious work regarding more modern motors in the AZL units.
Being aware of the early throttle work, I have compared a Märklin DC throttle to the Snail PWM throttle running an MTL F7A, an MTL GP35 and an AZL GP38-2, Regarding the above conclusions I can say that: 1. definitely better low speed operation with pulse power, 5-10 mph lower starting speed. 3. On the GP38-2 performance actually degraded above 30 mph scale using pulse power by 5-10%. 4. Running the units for 4 hrs. each on pulse and DC, no noticeable difference in the temperature of the units. I suspect the current split chassis designs in Z and lower current draw of the motors makes heating less of an issue. Also several of the pulse designs of the era used an SCR.
Hope this helps,
Mark
|
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on May 21, 2017 23:24:56 GMT -5
regarding more modern motors in the AZL units. That's my main problem : PWM and coreless motors ... It seems ultimately that the optimum frequency is the main factor for overheating, or not (but not only : www.passmann.com/_fr-1e.html , first paragraph : the shape of pulses seems to be important, too) ! Frankly speaking, I would prefer simple (rectangular) PWM (or with a low pass filter, to "smoothen" pulses, between output and power transistor ? ), because it can be directly generated by the hardware in AVR's, and this would allow smooth acceleration and deceleration via software. With a clock frequency of 16MHz, a divide factor of 8 and a 8 bit resolution, it should be possible to go up to 7KHz (would it be sufficient ? ). And perhaps I could increase by setting a lower reset value ... and using a lower resolution : are 256 steps REALLY necessary ? Some people say that a higher frequency is better : forum.arduino.cc/index.php?topic=173147.0www.piclist.com/techref/postbot.asp?by=thread&id=%5BOT%5D+PWM+and+coreless+motors&w=body&tgt=_topBut I read from a french modeller that about 150Hz to 400Hz were optimal ! And another (not modeller) wrote that a solar motor didn't work properly over 120Hz. It would be interesting to know the OUTPUT frequency and shape ( from the power to the motor) of DCC decoders, or of the "Snail" controller. Or do they use a "mix" of PWM and another system, and perhaps no "sharp" pulses ? As Greg wrote, DCC seems to be working without any issues, and frankly I believe it ... Help, to all owners of a scope ! Or perhaps I should simply buy such a motor and make workbench tests, without any "risk" for my locos ? Where, in Europe ? I didn't find them at the "big" components resellers (Conrad, Reichelt Elektronik, Mouser Electronics). And : Reference for the motor used in AZL locos ? (Faulhaber sells many models ! ). As Faust said : "And here, poor fool, I stand once more, No wiser than I was before."
|
|
|
Post by markm on May 22, 2017 2:38:57 GMT -5
The data for the Snail is easy. The frequency is 60Hz (+-1Hz) and the waveform looks like: I'm still somewhat new to DCC myself so I haven't scoped it out yet. On the z scale forum on Yahoo there is a file with waveforms, I just can't find it. There is also a review of a MRC throttle with waveforms you might find of value. Going over to the DCCwiki.com site, they describe the output of "early" decoders to be 60Hz and that frequencies up to 10kHz are used, primarily to prevent an audible hum during operation. They also report that there is a trade-off with lower motor torque at higher frequencies. I can say that for the Snail, I noticed no hum for the 3 locomotives I mentioned. If you want to experiment, Rokuhan sells their motor as a standalone product. Hope this helps, Mark
|
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on May 22, 2017 4:06:50 GMT -5
Thank You Markm, the waveform looks like the voltage at a capacitor charged quickly at the beginning of the pulse and discharged slowly at the end ... I think a combination of 2 resistors and a diode could generate a very similar waveform at a capacitor (followed by the power transistor). And a simple low pass filter would provide both an exponential rise and falling. So I will make some workbench testing, first with the electronics "standalone" ; and then with a Rokuhan motor : I hope it's almost the same as in AZL locomotives (or perhaps Maxon ? ), at least it looks very similar. I think the 60Hz is the frequency of an internal oscillator (for battery power), perhaps a value coming from a former transformer-only powered version, where the positive (or negative) going waveform triggered an adjustable monostable circuit. As I wrote, in my case the "pure" PWM signal will be generated by a timer from the AVR.
|
|
|
Post by markm on May 22, 2017 10:29:24 GMT -5
Yes the Snail is a rather basic throttle. After about 80% duty cycle the output doesn't return to zero.
I don't think 60Hz is a magic number. In one of the Model Railroader articles it was stated that there was no improvement in performance after about 100Hz so 60Hz is mostly a matter of convenience. One consideration in frequency selection could be RFI. A good square wave can generate a lot of harmonics and model train tracks can make a great antenna.
It would be interesting to follow up on some of the claims from the DCCwiki. I can see where higher PWM frequencies could reduce motor torque, but I wonder if it's actually significant. They also claim that decoder manufacturers provide an unspecified means of varying the PWM frequency, but so far I can't find it in my decoder documents. If it really exists, it would be a great way to characterize frequency verses performance.
Mark
|
|
|
Post by Rob Albritton on May 22, 2017 10:47:00 GMT -5
One consideration in frequency selection could be RFI. A good square wave can generate a lot of harmonics and model train tracks can make a great antenna. Mark Correct. Capacitors are frequently added to the locomotives themselves for RFI (Radio Frequency Interference) certification. Another point to consider: So far you are talking about the circuitry that attaches to the rails. The locomotive may have additional circuitry between the rails and the motor that will change how a waveform looks before it goes to the motor. So this quest for some level of perfection on one end of the equation may be irrelevant if you do not have the other side of the equation in mind. While it is fun to tinker with the electronics - I should know, I have scratch built several throttles and even scratch built a DCC decoder! - the one element that has contributed most to good operation of Z scale is dirt. Keep you tracks clean. Wipe them with a good cloth. But also do not forget that when dealing with pulse power, dirt will also collect on the wheels of your locomotive (and rolling stock) so clean the wheels too. Good luck and have fun!
|
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on May 22, 2017 13:13:52 GMT -5
One consideration in frequency selection could be RFI. A good square wave can generate a lot of harmonics and model train tracks can make a great antenna. On my former N layout in the 80's, during building, I had a RFI problem too, and it was due to insufficient filtering on a 5v supply (I saw it with my former analog scope). And the power delivered by the supply was initially fully rectified sine ! So I think that, if one doesn't want have any trouble with his neighbours, he should use pure DC on the whole layout ; it sounds stupid, but nothing is harder than debugging RFI problems (I remember my first RC ship in the 70's, 3 months of testing and trying ! ). So I think that finally I go back to the basic solution, only 2 speeds for each train, DC ; and as I wrote, the "starting" problem at slow speed could be solved via a short "full" pulse. Thanks for all answers, but I come back to the basic principle : KISS, Keep It Simple and Stupid !
|
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on May 22, 2017 14:46:57 GMT -5
I don't know if the speed changes are so abrupt, most locomotives (I hope, this includes the Budd's) have flywheels on the motor axle. With my N locos I never had issues ... I don't wanna beat the absolute speed record for trains : A realistic speed for passenger trains like the Budd's seems to be 50 mph, and 25 for a freight train, about respectively 6.8 or 3.4 meters per minute for the models. Older throttles had no inertia simulation. Too much "momentum" (by an integrating capacitor) would give a longer "breaking distance" when the train stops in the station (and this could cause some other accidents, like running into a turnout). For example many modellers use a section with a diode when approaching a bumper, or a section without power in front of a signal, and this system gives a "sudden" stop, too. And I hope they don't damage their engines ... Automatisms before the "widely used microprocessor" era (i.e. beginning to middle of the 80's) on older layouts were controlled only via relays, with no progressivity. And Z scale exists for 45 years ! Or perhaps I should use more steps, the CD4051 would allow 8 (CD4067, 16 channels, is only available in SMD at most distributors, but I could find them in DIP, hooray ! ) and more "smootheness" for acceleration/deceleration. But all the way, I wanna stop the train "abruptly" from the lowest step that allows safe running. About PWM and RFI : sorry, my layout is not in a Faraday cage. But I don't know if the small currents used by Z locos at 60Hz (even with the harmonics) are enough to make significant RFI ; the power supply on my former layout, which caused issues, was about 5v-5A (standard TTL IC's) ! But I think, it was rather direct induction than real RFI. I confess, my "secret preference" would go to PWM (still with an "abrupt" stop, but at very low speed, otherwise the "stopping point" of the train will be unpredictable, if the loco is still "cold" or already "warmed up"). I think the best solution will be to build a test circuit (oval) and use a simple radio receiver near the layout. I hope that modern TV receivers are no more concerned by "QRM TV", like in the first years of CB. Perhaps the low pass filter that I suggested when I saw the waveform of the "Snail" (and which is never mentioned in any article about the L293 motor shield for Arduino ), smoothening the square waves, will be the ultimate solution, even if the voltage reaches the zero level at every period ? The main problem with square waves is their harmonic content. And if my locos "climb" 1.5% with a maximum of six 33' bay hoppers (by "Full Throttle") or four 40' box cars (by MTL), and in both cases a supplementary caboose (MTL), I will be happy ! Sometimes I wonder if some people would prefer me to give up my layout project ? Or are Z locos SO sensitive, and it was an error to try Z gauge ?
|
|
|
Post by Rob Albritton on May 23, 2017 8:54:51 GMT -5
are Z locos SO sensitive, and it was an error to try Z gauge ? If you want to think about the electronics because it is fun to do so, then by all means please continue. But if you are genuinely concerned about extending the life of your Z scale locomotives, then I suggest that you no longer worry and spend time on another aspect of your layout. We had one RDC that came back to us for service after 3 years. It was located in a display in a doctors' office in Canada. The RDC ran point to point with a hard wired 12v power supply. When the RDC reached the end of the line, a sensor would trigger a relay would snap reverse the polarity of the wires and send the RDC flying back in the opposite direction. It was run like this for 10 yours a day, 5 days a week. My simple math says this little RDC made about 1.8 million round trips before it needed service. And that was with the power being slammed full reverse at 12 volts. I don't think you need to worry so much about the locomotive. And no, I do not know the model number of the relay.
|
|
|
Post by dave on May 23, 2017 10:20:38 GMT -5
Alberich....it appears that you are way overthinking this. Get a Rokuhan controller and go play with the trains. If you are worried about the voltage, put a stop at 10.5 volts. I am not putting you down in any way but it sounds as though you are trying to talk yourself out of Z scale. If you want to run older Marklin stuff, get a Snail 9V controller.
Just my opinion and nothing more.
|
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on May 23, 2017 12:05:59 GMT -5
To Dave : 1) Unfortunately commercial DC controllers don't have a computer interface ; and MY layout will be automated via one or several AVR's. That's the reason for "scratch" building ... 2) Only AZL locomotives.
|
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on May 23, 2017 23:15:24 GMT -5
In this case, to avoid misunderstanding : computer = ATMega (embedded) with parallel ports (it has a calculation power better than my former C64) ; no Windows, Delphi, VB, USB adapter (although the IOWarrior by CodeMercs, Germany, works fine with another application), etc ... The full control over the system. As I wrote in a former topic, try programming an endless loop in Delphi (repeat until false) ... --> Crash : the routine must be called periodically via a timer interrupt ; and I don't want to dedicate a "full" computer to the layout. Another point : to say it immediately (and again), I DON'T wanna speak about Linux !
|
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on May 24, 2017 0:29:02 GMT -5
I want to switch several voltages via a CD4051 (or 4067).
|
|