Hi,

Have a switching application for a model railroad...I'm making an H bridge that's controlled by two 555's to throw a track turnout (8 of them, actually! Each with its own H bridge and timers, tripped by reed switches/magnets on train). My father in law is the train enthusiast...

The basic work is done, and works great. Good up to 20mA at 12V.

Now he tells me he runs his signal LEDs off the switch machine (Tortoise)...to each control station - all five. So, we have to up the beefiness of my H bridge to carry the load of them, as well. Which takes me over the ratings for 2N3904/6. I'm now somewhere in the realm of 720mW, and would like the safety of having a 1W rating.

I've looked at MPSW42G (NPN) and MPS6726G (PNP) at Mouser, and they seem to fit the bill, at the right price. Just don't know if they "complement" each other. I haven't picked transistors for this kind of application before, but imagine it's not too complicated, right?

Any thoughts? Thanks!

 

If they are truly acting as switches, then the wattage is the voltage drop across the transistor x current. If you are saturating the transistor your voltage drop should be 0.6 volts x current. How many amps are you running?

 

Tortoise switch machines have separate contacts which are often used to control the signals. Signal power is separate from switch machine power, so I wouldn't expect that your circuit would have to carry the current for the signals.

That is unless your FIL has his signals wired differently…

 

Hey guys, thanks for the replies...

Gopher: I anticipate only having to carry about 60mA, so let's just call that 100mA because FIL isn't so cognizant of power and that blue smoke, ha ha...from what you wrote above, maybe i don't fully understand transistor dissipation? It's not simply I^2*R for BJT's?

dj: What's going on is that the tortoise is powered as it is working...that is running a series of LEDs to indicate that it's in motion (his 'indicators'). 5 LEDs (5 different control 'stations'), I'm allowing 8mA each, so 40mA for them. Tortoise can draw 15mA in a stall, so 20mA safety for that = 60mA total. At 12V, 720mW.

In essence, anything that comes out of the tortoise in his setup has to go THRU the tortoise, and due to that, thru my H bridge as well, since it's the only way I have to reverse polarity (always "in-circuit"). My thought is "hey, just make the H able to handle 1W, so we're bomb-proof". That will work.

All I need is a type of NPN and PNP that can do 1W and general switching duties. Just not sure about 'complementary' transistors - will *anything* work, or should they be from a 'family'? The prototype works great, altho gets warm, and I want to increase its power handling.

 

For a DC signal, transistor dissipation is V x I. V is Vce, the voltage from the collector to the emitter, and I is the current through the transistor and the external circuit. If the transistor is acting as a saturated switch, the Vce is actually Vcesat. This is on the datasheet as a table entry, a plot, or both.

For a small signal general purpose critter like the 3904/06, Vcesat can be below 0.1 V. For a signal current of 0.1 A, that is a steady state power dissipation of 10 mW, not very much. It is important that the transistor be turned on and off quickly, because to get from on to off it passes through the linear active region where power dissipation in the transistor soars. For example, switching a 12 V 100 mA circuit slowly means the transistor spent some time with 6 V across it and 50 mA through it (assuming a resistive load for this example). That's 300 mW, a whole lot more than 10.

To make sure the transistor is firmly in the saturation region, the base current should be around 10% of the collector current. Check the data sheet to make sure this does not exceed the max base current.

If you still want to go to bigger parts to be safe, try the TIP series of TO-220 power transistors. Old, friendly, rugged, reliable, cheap.

ak

 

Nice, thanks, Analog and Gopher.

I guess this is where I'm at: I'm triggering the H bridge (saturating the transistors) with a 555, so the 'climb up' is pretty fast.

I did think that I needed to use V*I, and I have, which is why I think I'm going to need 1W. I come up with 720mW dissipation, and the 3904/6 can only do like 650mW.....

Do you think *any old NPN and PNP rated 1W* would work? My question was really 'is there any scheme for matching them', or no? Mouser has some good stuff (my original post) for cheap, that looks appropriate...

Looks like I need to provide about 10mA base current, for 100mA....back to the drawing board!

 

.1A x .1V = 10mW. Mike, can you explain where your 700+ mW figure is coming from? Not that using bigger transistors (say, TIP-31 and its complement, TIP-32) will hurt anything, but...

 

I didn't search for particulars on the latest transistors but the 2N4401 and 2N4403 come to mind at 34 milliwatts for 100 ma and they have nice fat ratings for current and voltage.

Basically, I have a drawer full of them, so that's where I'd look. They are old, so they're probably cheap. Try something you have in your junk box. I think you're over rating this job, as to3 said.

 

A nice little beast for this application is the ZTX450 and ZTX550 - these are two complimentary (NPN/PNP) devices that are well suited for this application.

Decent beta at higher current levels, fast, 1W max power dissipation.

 

I would not bother to use 2x 3904 / 3906 in parallel.

Yes I can think of your answers, but I can test them with component tester for hFE.

Besides I have other transistors too. for instance http://en.wikipedia.org/wiki/2N2222

have a bag with 500 here since recently.

Why do you need to match them for the P and N types?

you just saturate them, you dont neccessarily need 1/10 base current, check the datasheet, or just try some resistors and see how the thing works.

If you are unsure breadboard it so if the transistors burn out, you can replace them easily.

 

BD139 and BD140 are cheap

 

Hello,

Perhaps the following post from the Components selection guide might help:
Transistor Selection Guide

Bertus

 

Great, thanks for the ideas, and the link Bertus.

Ok - I am going to have to pass ALL current into this device (tortoise) thru my 4-transistor H bridge. The bridge is necessary because this servo reverses polarity. It normally draws only 4mA, or 15mA in a stall.

FIL has indicator LEDs running off the 'accessory' terminals on the tortoise. They would be 'on' when the tortoise is moving, and are powered by it. Each draws, oh 10mA perhaps, being overly conservative. He's doing 5 in parallel, each w/dropping resistor. So my load is more like 60mA. I'm sure in reality, i can trim that back to like 45mA, but...

At 12V, that is 720mW. This flows thru the transistors of my H bridge. Not something trivial, but nearly 1W! Unless I don't see how the bridge operates? Current flows into it, thru the devices, and to the load....measuring the current flow confirms this. If I switch the thing often enough, my 1/4W base (load reversing) resistors get fairly hot (going to make them 1W also).

So, using 1W devices seems to be a way I can ensure i'm not going back in and replacing 4 transistors on 8 boards, down the road, ha ha. Cheap enough insurance, I think.

 

Can you attach a schematic sketch of how the Tortoise is wired? I suspect it is the circuit shown on the right of the wiring examples from the Circuitron Switch Machine instructions.

I am confused by your description. The accessory terminals of the switch machine are independent of the switch machine power. They are simply a pair of SPDT switches. FIL must have them cross wired to the power. Also, Tortoise switch machines are always powered, so it is not clear to me how the indicators only display when the machine is in motion.

 

Currently he has it kind of bass ackward...it sort of works for him, but it's not right by any stretch. No power unless active (pins 1 + 8). I'm trying to get him a bit "sorted out", lol. Yes, he's only got one power source to the tortoises, so aux. outputs are based on if it is "on" or not. He wants to indicate "in motion". He uses snap switches to reverse the tortoises, which is nuts...they pull over 1A just to switch, and at times he fries them! My idea was to go to H bridges...low draw, no fuss....

So ok - the H works beautifully, but a momentary pulse won't throw the tortoise all the way (even with a 'helper cap' on the tort.). Some of his rig will have reed switches w/magnets on the train = a fast pulse. So part II: I came up with using a 556 for each tortoise - a reed in each direction can 'auto control' the tortoise ahead, and you can over-ride that with a toggle. That is working really well on the test bench.

Trying to get him out of a spot - he's bought all this stuff (20+ toggles MOM only in ONE direction, jeez....), and the table is already wired. BIG train, 15 yrs. in the making and all that. So wiring the tortoises 'right' is off the table for him.

My job as he's defined it: come up with a system to pulse each tortoise for oh, 5 seconds or so, AND indicate that it's in motion. As an aside, I'd like to have dedicated LEDs that tell you the position of each crossing all the time! But he's intimidated by rewiring 10 tortoises + miles of already-set-up wiring.

I'll draw this up in a few mins and post it...

 

So, DJ - you're saying the tortoise pins 1+8 are always kept energized? I checked one and it doesn't cut out (draws 4mA all the time), that seems like quite a waste - UNLESS the idea is that you hold the MOM toggle til it has made its run?

The real issue is that he wanted to make part of this automated, using reed switches under the track. When the train hits the reed, the pulse is too fast to throw the tortoise all the way, hence my getting involved with 555s etc. To make the pulse long enough to go all the way (adding a cap across the pins didn't work well enough).

Anyway, here's a diagram of where we're at now. I'm using the H bridge to reverse the polarity since you can't do it with the reeds.....

 

Yes. The Tortoise design assumes that pins 1 & 8 always supply power. The stall current is designed to be small, so as to minimize power waste. It also is a method of ensuring that the switch points do not move under train movement.

 

So I need to design for power to be on at all times, gotcha. This way, at least the indicators will show which way the points are facing at all times, instead of only when moving!! Thanks.

 

Great, thanks for the ideas, and the link Bertus.

Ok - I am going to have to pass ALL current into this device (tortoise) thru my 4-transistor H bridge. The bridge is necessary because this servo reverses polarity. It normally draws only 4mA, or 15mA in a stall.

FIL has indicator LEDs running off the 'accessory' terminals on the tortoise. They would be 'on' when the tortoise is moving, and are powered by it. Each draws, oh 10mA perhaps, being overly conservative. He's doing 5 in parallel, each w/dropping resistor. So my load is more like 60mA. I'm sure in reality, i can trim that back to like 45mA, but...

At 12V, that is 720mW. This flows thru the transistors of my H bridge.

Please read carefully, as 2 others have tried to point this out and it has gone completely over you head.

The 720mW is the power dissipated by the transistor + the load. The transistor only dissipates power in the amount of the current through it times the voltage across it. In saturation the voltage is as most 0.5V and probably lower.

So the dissipation in the transistor is 60 mA x 0.5V = 30mW at the very most. Your 2n3904 and 2n3906 wont even get warm.

If they do, they are not in saturation, and you need more base current.

Bob

 

.....................................
At 12V, that is 720mW. This flows thru the transistors of my H bridge. Not something trivial, but nearly 1W! Unless I don't see how the bridge operates? Current flows into it, thru the devices, and to the load....measuring the current flow confirms this. If I switch the thing often enough, my 1/4W base (load reversing) resistors get fairly hot (going to make them 1W also).
..............

The power that flows through the transistors is not the power dissipated by the transistors. Most of that power is dissipated in the load. As others have noted, the transistor dissipation is just the volt drop across the saturated transistor voltage drop (normally one or two tenth's of a volt).

Note that he transistor is acting as a switch so dissipates little power. A wall switch can control 1500W of power, for example, but certainly doesn't dissipate 1500W or it would melt instantly.