Post by Greg Elmassian on Apr 11, 2017 11:29:31 GMT -5
You know the 4007 is probably good enough, I would probably pick a Schottky... I talked to my head of hardware engineering and he said that the 1n4007 is pretty typical.
(how fast are you sweeping your scope and what'sit's bandwidth?)
I was more commenting on the fact that you have agonized over this, so I would have expected a "cut above" on the BEMF diodes, but what you have is sufficient. I would have picked a signal diode like a 1n4148.
To Greg : The bandwith is 70MHz and I made now the tests with 500us or 1ms/division in "one shot" mode ; the probe was connected between Gnd and the "common" connection of the switch (i.e. the + terminal at the capacitor). During the first 5~6 ms I saw transients between -6 and +16v, sometimes +21v but very rare, probably caused by "glitches" at the switch ; nothing like the former +/-50v. Because many relays (2PDT and 4PDT) are rated for "mains" switching, I think these rather low voltage peaks (even those with 50v) are harmless, but even 21v is better than 50v. I will try the 1N4148 ... as soon as I can order some of them (the disadvantage of living in a small village) : ordering only 10 items of 1N4148 would be rather silly and the delivery costs much higher than the costs for the components themseves !
I must add that I made at least 40 back and forth tests and that, with 7.5v and a 1500uF capacitor, no false response was noticed from the turnout ; so I think they are VERY reliable, using optimum control circuitry like the indeed very simple Kato system (thanks again to BAZman), beside which the Rokuhan looks again like a gas...(etc)
Last Edit: Apr 12, 2017 0:14:46 GMT -5 by alberich
I could mail your the 1n4148/1n914's. They're only $.01 here in the U.S. I know what you mean but doesn't matter small village or a few km away. Sucks spending $5-10 for something in a $1 padded envelop.
Thank You, but in about 7 days I will go to the next electronics store (30km) and I hope I will find them. I am waiting at the end of the month for the base structure (2.9 x 1.45m) for the layout, made by my local "carpenter" ; I can't build such big structures alone, planks are full length and there will be only one "big" plywood plate for the whole structure ; for my former layout I had my father helping. And in the present time I can't do anything else than experimenting on the kitchen's table (Fortunately there is no wife ... ). And then I am waiting for my rolling stock (about 4 weeks) and some tracks to build a testing oval, and test the PWM and distance for deceleration : this will define the place for track-insulators near the stations, a detector in the negative power line will trigger the decelerating and a second one on a section in the stations themselves the power off, so I can run my RDC's ; as You see the problem with diodes is not urgent !
Post by Greg Elmassian on Apr 12, 2017 0:35:46 GMT -5
Just for reference, you will see that the 1n4007 has about 4 times the capacitance of the 1n914/1n4148 at 1MHz, and of course higher at higher frequencies... so that is the reasoning for using a switching diode instead of a general purpose rectifier diode.
Power supply for turnouts : the problem is that I don't know how many amperes I will need for my turnouts ... I wanna use the Kato system. There will be a maximum of 4 turnouts connected to the same switch ( = relay), and flipping together. Or should I use a double switch, 2 turnouts connected so the capacitor goes to Vcc, while the other 2 are "inverted" so that the capacitor goes to Gnd (I hope my explanations can be understood) : the current would be the same for both positions but divided by 2 ? I used a 0.2 Ohm shunt in series with one coil, but I got absolutely HORRIFIC values on my scope : +/- 1V peaks, and because the shunt is a pure resistive device, one can say that there are transients of 5A ... in both directions (I didn't yet replace the 1N4007 but I think this will not change anything) ! As I can remember from my time at school, it's called "damped oscillations". As I saw this I gave up the idea of controlling a turnout with 1/4 L293D, I think the driver would not survive to the transients. I planned to use a readymade 5A switching power supply but I think the transients will make trouble : activation of a breaker (if present) or damage(s) at the supply's electronics ! I think they are intended for use with "conventional" appliances, not inductive loads. Or are they protected against such issues ? It is not possible to get the information : "buy and don't grumble if You have problems ... " Or does anybody know which system Kato uses ? I saw a "KM-1" unit, 12v and 2A but I don't know if it's a conventional or switching power supply. All the way 12v (or even 15v on some descriptions) would be too much, but I only saw "regulated and filtered" ... I think it's "analog" technology. I thought of simply add a BIG filtering capacitor (with a limiting resistor to charge it, and I could decrease the "amperes" available at the supply), I know it should provide charge to several capacitors, so for 4x1500uF I am thinking about 10000 to 22000uF, because charge transfer quickly (exponentially) reduces the voltage available. But in this case recharging current will recharge (slowly) the capacitors at the turnouts, too ... and flow through the coils until the voltage reaches the nominal value. Not good ... Perhaps the simplest way would be a transformer, 6v and 1A per turnout switched at once, with a bridge rectifier or a simple P600 diode (6A) and no filtering at all ?
Last Edit: Apr 24, 2017 0:16:08 GMT -5 by alberich
Post by Greg Elmassian on Apr 24, 2017 12:48:38 GMT -5
Funny, you seem to have gotten yourself into the predicament like the loco power. You want to protect one loco (or turnout), but you also want the same circuitry to protect multiple locos (up to 4 turnouts).
When I am presented with issues like this, I normally move the protection methodology from the power supply to the individual units themselves.
Clearly this also gives you a "scaleable" solution, i.e. adding more locos/turnouts does not change the basic system (power supply).
One alternative that is available for your turnouts is to not fire all 4 off at the same time, but sequence them one at a time, less of a problem.
Thank You for the answer, but I only wanna know what kind of power supply would reach to "fire" 4 turnouts simultaneously, without taking damage itself ... So I could use a 4PDT relay, one switch for the 4 turnouts (the contacts are for 250V 7A and the peak is very short, the freewheel diodes would help, too, to avoid sparks) and the 3 remaining for power to the tracks (with my detection system for stopping the MU Budds, each track in the station has 4 "poles", there is a supplementary insulator in each rail, so 3 "poles" must be switched ! ). I didn't want to protect anything ... only the power supply itself, so I thought that perhaps a basic (transformer + rectifier) 6V 4A supply could reach. KISS : Keep It Simple and Stupid !
Last Edit: Apr 24, 2017 14:01:38 GMT -5 by alberich
Post by Greg Elmassian on Apr 24, 2017 16:57:10 GMT -5
You actually gave part of the answer yourself, you are uncertain about the supply because you don't know if the turnouts will trip the breaker.
And of course this depends on the breaker and also if you are using a capacitor to discharge, etc.
You have analyzed this so much with many different scenarios, and honestly, shifting requirements that honestly, an answer is impossible.
Now before you get upset, I cannot see how a moderately filtered supply made of a transformer, bridge rectifier and mild filtering would need more than one amp per turnout.
But if you are making a capacitor discharge system, then it's a different animal, i.e. those systems do not rely on constant current capability of a power supply, they go the other way and make sure the constant current the system can supply is much lower.
So, I don't know where you are at, seems the pendulum has swung away from a capacitive discharge system, where very little steady state current is required, so I gave you my recommendation on a "full time" supply with moderate filtering. (The filtering is really not needed at all).
p.s. I'm trying to help, really, but you have to really grab the reality of your "moving target".
One though is how Rokuhan trips the turnouts. First their throttles can only supply 1A maximum. They consistently power two turnouts and generally will do 4. However their turnout switch discharges a 1000uFd capacitor through 1K resistor. So you really only need to supply 1mA to the turnout.
Post by Greg Elmassian on Apr 25, 2017 19:13:43 GMT -5
Well, it's really not that simple, because if you supply 1ma continuous to the turnout it most likely will not move. (try it, hook a 10v power supply through a 10,000 ohm resistor)
The issue here is that you are mixing continuous power ratings with pulses of power, which need to be described in duration and intensity.
If the output of the capacitor is indeed always going through the 1k resistor, then the maximum current would be using Ohms law and considering full voltage of the power supply through the 1k resistor into a dead short (when first dumping current the resistance will be close to zero of the turnout motor):
V=IR or I = V/R or current = 10v / 1000 ohms = 10 ma, not 1 ma... but are you sure the discharge path includes a 1k resistor? Normally it's the charge path only.
In any case it's not simple since we are dealing with reactive elements and the current is not constant.
I'll check the switch one more time, but I've twice checked the PCB as I was surprised that the resistor is in the output. The input side is just two polarity steering diodes for the capacitors. And yeah, I should never do math without a calculator. I have a schematic of the C002 switch, but at the moment I'm not confident enough to publish it.
The Kato system (which is the simplest I have seen) works fine with 7.5v and a 1500uF capacitor for one turnout. My lab power supply is limited to 3 Amps. So I think that I will : 1) Build the Kato system as drawn by BAZman, without Led's but with four 1500uF capacitors, and increase the number of connected turnouts from 1 to 4 : I already bought 2 Rokuhan "Y cables" (for the 4 connectors), it's easier for the turnouts during test on the workbench ; so I can see if 3 Amps are enough ; and if not, I can use a simple proportionnality calculation ... Sometimes trial and error is the best method. On the final layout the turnouts will be "hard wired" ( = soldered). 2) Then I will buy a 6v transformer with enough VA and a margin, according to 1) ... And I will add a bridge "full" rectifier and no filtering. 3) But, as I have already seen, voltage peak goes up to 10~11v (minus 1.2v due to the loss in the diodes). So I will try again with only one turnout if I can't reduce the capacitor's value. 4) And finally I will build the new system with 4 capacitors and turnouts, and then hope it will work.
Last Edit: Apr 26, 2017 1:03:06 GMT -5 by alberich
Post by Greg Elmassian on Apr 26, 2017 10:35:56 GMT -5
OK, wow this is going all over the place.
Mark, thanks I'll be interested to see what you find. A 1k resistor in series with the "output" of the cap of course would limit the max current it could put out, and that makes sense technically, but it's kind of unusual. In any case, it makes it simple to calculate the max current it could dump, by assuming zero resistance in the turnout "motor".
Alberich: clearly one idea behind a capacitive discharge system is to give a short, high current "punch" to the device being operated, but also that the short "punch" is indeed that, in fact the peak current would be way higher than just connecting to an ordinary supply.
But the other half of this is that most systems are set up so that after the "punch" the steady state current is very low, so low that it cannot damage the device being operated.
Thus 3 amps would most likely be too high for a good design. The perfect design would dump the capacitor to the turnout quickly so the turnout action is very positve, and it would be able to recharge the capacitor in some reasonable amount of time, like 2 seconds or longer, AND that if you left the motor connected to power, after the capacitor discharged, the steady state current to the turnout would not damage or overheat it.
That's the ideal design.
In many cases, to achieve the last goal, the recharge time of the capacitor is too long and a compromise is made.
The whole thing again depends on what the turnout motor really needs, and how long you can wait until you trip a second switch.
Now you have also stated you wanted to operate as many as 4 turnouts simultaneously. Well, that blows the whole thing up, since enough current to do that and recharge the cap in a reasonable time is most likely too much for a single turnout.
Again, you want all these features that really conflict with each other. If I had all your requirements, I would forget the capacitive discharge, use a big supply, and have a one-shot controller (like a 555 timer) to control the power to the switch. If I had to protect from stupid operators, I would add in something that would not let them operate the switch so often the turnout would overheat. I would have a hard time with that though, I just would not let that kind of person anywhere near my trains!!
The principle of the Kato system (look at it on page #1, post by Bazman) is to both discharge AND charge the capacitor through the coil, so there is no delay to allow the capacitor to recharge. I must admit that this system is VERY simple and ingenious ... and certainly more than the original Rokuhan system ; it could even be built with a bridge of two transistors ... but then there is no supplementary power switching to the tracks. Due to the topology of my layout I could even use power routing ; too long to explain in detail : simply said, there are 2 stations with the tracks in "parallel", two turnouts (among the 4) would switch one rail and the two others the second rail. But I wonder if power routing will STAY reliable ; on the other side most people who experienced issues used DCC with much higher currents than my "little" DC for the two Budd's. So, because I already bought 2 L293D's, which I don't need for other purpose, I will "sacrifice" one and try to adapt the Kato system with one driver per turnout ; if the system survives to about 25~40 back and forth switching, I think it could work ; otherwise I will "meet" the problem very quickly when the IC will go to "paradise for circuits" !
Last Edit: Apr 26, 2017 12:48:38 GMT -5 by alberich