solid state relay??

You could use a H-bridge, rather than build it from discrete transistors/FETs they come as a single or dual bridge in an IC, possibly with built-in snubber network.

Absolutely

What you need is a switching MOSFET. And a couple of resistors to control the gate and maybe a diode to absorb spikes.

I am using LB8721s.

Ground the gate to negative with 150k and feed it from your Pi with something like 150 ohms. and connect the source to negative and the drain becomes the switched negative output. You may want to put a 1N5187 diode across the load to calm any switch off spikes .

On a bit of metal to absorb heat these will do about 20A continuous - they have very low 'ON' resistance so they don't get too hot.

These are only single pole and only able to switch the negative of course.

If you want to switch the positive (as well) you will need a P channel device and I don't know what to recommend

Good thought. If you want to *reverse* the voltage . I am not familiar with how todays model railways actually work. I thought they had electronics in them to control speed and direction.

Actually re reading the question he wants a *polarity neutral* switch.

That is not easy. And is not exactly what an H bridge will do.

I'd probably use a P channell and N channel MOSFET in parallel

But realistically I'd probably use a 3.3 or 5V Songle relay. They are uber cheap in bulk from Aliexpress.

I use this sort of relay in my central heating controller to switch the zones.

If you want single relays then they are less than 50p in bulk.

The 5V coils are around 70ohms which was too low for my application direct from a PI so I added transistor buffering - the relays draw about 70mA

This is the sort of thing its not hard to design and get PCBs made up for at minimal cost. I use old style components as my eyesight is not good enough for surface mount goodies.

Mike,

Carefull there : any kind of sparking (like can happen with a bad contact between the loc and the rails) could cause quite a power spike on the solid-state relais and destroy it. I would suggest keeping the relais you now have.

Use Andys' suggestion*. Just make sure the snubbing network is upto the task.

• Take a look at the H- or T-bridges used for stepper motors.

Regards, Rudy Wieser

I was wondering about using back to back IGBTs. See 'four quadrant switching' on p23 here:

Never actually used an IGBT for real so not sure how 'small scale' they can go, but I'm seeing 6.5A versions so they're not just for kiloamp-scale locomotives:

Agreed, I think that's the simplest solution here.

Theo

There are various solutions, each with its own advantages and disadvantages.

One is to use two MOSFETS with the sources and drains parallelled, the switched path being that between the two drains. You'll see a drop of about 0.6V across it when it's on. (If this is a problem you can connect a Schottky diode in parallel with each MOSFET, which will reduce the drop to nearer 0.3V.) The principal snag is that the control voltage, which is applied between gates and sources, has to float.

If you can drive relays momentarily but not continuously, consider bistable relays.

David

Some people are still using the traditional DC systems, others use DCC, which requires electronics in each loco.

Adrian

plus ££££££

I was trying to get something simple done on the cheap. The expense of DCC would barely be justifiable.

I do have some, bought for another part of the project. But at IIRC ~£7 a pop, they're not cheap: I need 8.

I'm beginning to agree. I've been hunting round for prebuilt modules to do the trick, and nothing provides isolation like a good old-fashioned relay. Maybe I just have to put up with the current drain.

Many thanks to all who've responded with ideas. I'll look around a bit more, then press on with the relay bank.

A 20p transistor driving a 40p relay is probably best then

Well its only 70mA a relay When its on

On 01/08/2025 17:52, The Natural Philosopher wrote: ,,,

x8 though, worst case. Over half an amp. Seemed a lot just to let some current through to the tracks.

I'm seeing things like this for searches of 'magnetic latching relay' for £1 a relay or £2 on a board with GPIO inputs:

You have to hold the latching inputs for a certain length of time - not sure what happens if you exceed the rated time, I don't know if there's a danger of burning something out (railway modellers will perhaps be all too familiar with this idea from motorised points).

Theo

There are also "high sensitivity" type relays which can use less.

I repaired a modern model train controller a while ago and it had a bank of mechanical relays inside. Variable speed control was done with a very beefy set of MOSFETs in a H-bridge circuit, which had nevertheless been blown apart when the owner hooked the power supply up backwards.

A Pi pico draws up to 2.5A

So you are faced with having a little 5V/2.5A PSU block on board anyway.

:-)

-- There is nothing a fleet of dispatchable nuclear power plants cannot do that cannot be done worse and more expensively and with higher carbon emissions and more adverse environmental impact by adding intermittent renewable energy.

So you're looking for an AC SSR. Forget that the polarity is switched before the relay as to all intents and purposes it's effectively AC.

The main issue I had when doing this on a Pi some years back was that all the SSRs I used had a 5v input voltage with built-in limiting resistors on the opto isolator part. Saying that, I didn't find one that wouldn't work with the Pi's 3.3v output. I've no experience of the Picos outputs though.

I was switching 240v AC not 12v AC - and there is another issue - a lot of them are aimed at higher voltages - say 36-240v AC

Searching ebay find this for example:

Personally, I think that if I already had a working bank of mechanical relays (with opto isolated inputs such as the multitude sold for Adruino and Pi) I'd just suck it up and give it a separate 5v PSU capable of driving the coils. (Or take the main system 5v and give it a good decoupling from the electronics side to minimise any interferance).

Sure, it may well be a waste of energy but half an amp at 5 volts is a mere 2.5 watts. Run that continually for a year and that's 22 KWh which should be under a fivers worth of energy, but the chances of running it like that are probably remote... (how much was the last coffee and cake you bought?)

And mechanical relays will allow the passing of any DCC signals if you're using that.

-Gordon

Remember that although a PI can only drive 3.3V it can happily (and usually does) run off 5V.

I bought a load of these

I've found that drawing up a PCB for these sorts of things is so simple that I dont even prototype them

You need a PSU and best practice is a suppression toroid, and suppression cap and fuse on the mains side and a small smoothing cap on the output side. and a diode to feed the PICO so that you can program it with the power off. Picos pins feed NPN transistor bases via resistive droppers and collectors go to the relay coils and then to 5V.

The whole thing is a single neat circuit board . The pico solders onto it using pins

If I don't need wifi I use the RP2040 Zero unless I need more pins to be accessible. Its cheap as chips