Current limiting MOSFETs....

There are plenty of solenoid drive ICs (Search on ti.com for some examples), which can do initial high drive + PWM hold. The older analog way with resistor limit is not used much anymore due to efficiency.

You could also use 74HC123 or 74HC/CD4538 series monostable to limit the drive pulse. If you need controller pulse, you might need AND/OR with the output and the control signal and have the monostable to the limiting only. If you do the pulsing in firmware instead of timer, you can use retriggerable setup. With timer your firmware might lock up and timer could keep on working.

-- mikko

Nothing wrong with a capacitor in the gate drive, simple and effective.

Here's one solution:

Since a fried diode will end the game anyway, I guess it's alright to cut the power. So you don't need to monitor the current on all MOSFETs individually.

Since you use a 4-to-16 decoder, at most one solenoid will be active, so you could just put a current shunt or Hall sensor in the ground connection, to monitor the current of whatever solenoid is on.

Then you can use two comparators:

One with a high threshold, which is not normally reached with "normal" current, to detect if the MOSFET turns on and the diode in the solenoid is shorted.

To catch a melted MOSFET being on continuously, or a software bug keeping the solenoids on all the time, another comparator with a lower threshold, which detects a normal amount of current for the solenoid. This should go to a retriggerable monostable which is reset every time the current drops to zero. If the current does not drop to zero for long enough, the monostable will expire and signal an error.

Then the output of these should go to something like a RS flop that latches the error and deasserts the output enable on your decoder, and also disables the power supply, if it has an enable input. You could also use a high side MOSFET switch.

If the power supply has a lot of capacitance, and the diode shorts, a huge current will flow, so you don't have a lot of time to turn off the MOSFETs by deasserting the decoder's output enable. The gate resistors will probably have to be tuned. A fast comparator can be useful here. The comparator can be a simple BJT if voltage on the shunt in case of a short circuit exceeds 0.6V.

Danke,

PTC resettable fuses in series with each solenoid minimizes the number of current control components for maximum reliability. PTC doesn't trip during normal operation, so the lengthy cool down of several minutes is irrelevant. The thermal mass of PTC is much less than that of solenoid for reliable protection.

That would be my immediate thought.

Capacitively couple the drive so that the drive from the CPU can stick high or low forever without causing damage.

The simplest way is usually the best way.

Otherwise sense the current with a low value resistor and arrange for that to turn off the drive as necessary. Preferably in hardware only.

Are you seeing instances where this has (or is likely to have) happened?

As you can't predict haow the processor will fail if the watchdog isn't around to keep it in check (i.e., a double failure), the CPU could just as easily enter into a loop where it is turning the FET on and off, repeatedly.

I assume (past posts) you are looking for a drop-in replacement for a BJT or similar. I.e., a single three-terminal circuit for each device to be replaced?

At one point (in the distant past), SCRs were used and "forcefully" turned off by bleeding gate current away (now it's nominally a 4-terminal circuit)

Is there a reason you can't replace the existing devices with their exact replacements? (I assume you are dealing with small numbers/retrofits)

S*it rolls downhill. Your goal is to ensure the actual drive to the switch (*or*, the power available to it) is constrained, regardless of how you choose to drive it. Any part can fail so you want to have the most faith in where the rubber actually meets the road.

I don't see where John claimed to be driving them from a decoder. I suspect any combination -- including all and none -- could be active at any given time and for differing amounts of time, depending on their application.

Who's going to *see* the error if the processor has lost its mind?

Any sort of watchdog should bring whatever it is "watching" to a safe state that can persist indefinitely. E.g., restarting the CPU just means it can loop between "restart" and "fail", forever.

When I detect a fault, I remove power to the field so nothing can be actively driven "wrong". If you have designed so these things "can't happen", then it is easy to rationalize shutting down when it *does*.

In the applications I suspect John is addressing (based on his past posts), there's no real guarantee as to what will be available to address a problem, besides the "FET-replacement-circuit" itself. I.e., the CPU can be on a completely different board, interconnect cables can be "open", etc.

Scat shat on slight slopes stays grounded. Lessin' you're a scat kicker or something similar.

Danke,

--

Most solenoids and relays need a lot of current to pull in, and much less to keep them locked once they are seated. So one can drop the current after a tenth of a second or something.

A second mosfet and a resistor will work. Or PWM.

Old timey solenoids had two coils and a switch contact to do that. Like the notorious Lucas overdrive solenoid on MGs.

AC relays and solenoids do the current step down automatically; the seated magnetic loop has a higher impedance than the open one.

John Larkin Highland Tech Glen Canyon Design Center Lunatic Fringe Electronics

Would this work? fixed gate drive:

• | solenoid | |----

------| | |<---------- | | R [ ] === | --- C few uF | | /// ///

Use fixed size font

Input a fixed voltage square wave pulse This will give a big peak and when C charges drop to a lower limited current Works with transistors too.

If you have two different power supply voltages, slowly charge a large capacitor through a resistor from the higher voltage but connect it to the lower voltage supply through a diode. When the circuit is completed, the solenoid will receive a pulse of high current as the capacitor discharges, then a steady sustaining current as the diode conducts.

I used this method in a telephone answering machine. It had +50v and

The Laycock de Normanville overdrives were notorious on lots of different makes and models for failing to make contact. One day I decided to investigate these repeated failures in my Standard Vanguard and discovered that an 'O'-ring seal (shown in the original design drawings of the overdrive gearbox) had been omitted in production.

Hypoy gear oil is designed to keep metal surfaces apart under pressure and that was exactly what it was doing to the contact faces. I put a seal on the operating shaft, cleaned the contacts and never had a single failure after that.

It ought to work. The classic "capacitor in series" pick and hold, (more-or-less mentioned by the Original Poster and endorsed by Piglet), connects directly to the solenoid. Another alternative is a capacitor parallel to a solenoid. A summary of both classic circuits appears at:

Danke,

Yes, if the C is big enough. It will need a lot of gate swing too.

One could put a parallel RC in the drain lead too, to reduce the required gate drive.

The fet could be an SSR.

Spice it!

John Larkin Highland Tech Glen Canyon Design Center Lunatic Fringe Electronics

Interesting, uses 2 voltage sources.. Bit complicated.... Many componenst...

Yes, PWM should work too I use a lot of Microchip 18F14K22 micro controllers, those have a build in PWM generator, but you can do it in software too of course.

Some secret sauce is needed to complete the complicated circuit's cornucopia of components:

The circuit schematic is incomplete and is offered as a basic reference. Contact a Lee Sales Engineer to obtain drawing LFIX1002200A; it includes additional notes and operating instructions.

Danke,

You need to be more specific than just saying "momentary." There are thermal relays for this purpose but they're sort of large, compared to a component level solution, and they're pricey.

Assuming by momentary you mean a few seconds, a consultation with AI turns up this:

The following standard market options match this exact time-to-trip profiles from standard manufacturer log curves: Bourns MF-R050:

Vmax: 60 V

Ihold/Itrip: 0.5 A / 1.0 A

Trip Time Behavior: Its standard rating is 4 seconds at 2.5 A. Plotted onto the log-log time-to-trip curve, a 2 A current (which is 4 times Ihold) generates a trip action at roughly 6 to 10 seconds, securely staying under your 30-second requirement.

This is the sub-50 cent gem here:

and here

Datasheet:

Alternatives are:

Littelfuse PolySwitch RXEF050 is about the same.

The Trip Current Paradox: PTCs do not act like sharp digital switches. A part rated for a 2 A trip current will take an exceptionally long time (often 100+ seconds) to break if the fault is exactly 2 A. Sizing the trip current lower (~ 1 A) guarantees the thermal expansion triggers well before the 30-second deadline.

Ambient Thermal Derating: PTC fuses are highly sensitive to surrounding air temperatures. If your device operates inside a warm enclosure (≥ 40°C), the hold current drops. You may need to scale your base Ihold parameter slightly upward to prevent nuisance tripping during ordinary circuit operations.

According to the Bourns MF-R series datasheet, the Bourns MF-R050 does not have a single fixed "leakage current" value. Instead, it has a specified Tripped Power Dissipation (Pd) of 0.75 Watts typical at 23°C.

Applying this equation to your circuit parameters highlights how the device behaves in its tripped state:At 28 V (Your Circuit Voltage): The leakage current drops to ~26.8 mA. This is the steady state trip-state leakage.

The ~27 mA leakage current is necessary to dissipate the 0.75 W required to keep the internal core of the fuse at its active phase-transformation temperature. The surface temperature of the component will stay locked at roughly 125°C until the power supply is completely switched off or disconnected. And that is toasty.

Not this one. I've worked with this type before. They're so-call impedance limited types made with microscopic 40-ga wire. The impedance is virtually all resistive. It probably takes a full minute to burn it up.

Are your solenoids Bally 27-1300? These come in at 14.1 ohms and intended for 28VDC system. Part code means 1300 turns 27ga enamel wire. And "A Bally 27-1300 coil can sustain continuous 28VDC power for only 3 to 5 seconds before irreversible thermal damage begins, and it will completely melt or short out within 10 to 15 seconds." Reference from Museum of the Game.

1. The 3 to 5 Second Window (Thermal Degradation)Pinball coils are wound with Class 130 (B) or Class 155 (F) polyurethane/polyamide magnet wire insulation. This thin enamel film is rated to withstand a maximum temperature of 130°C to 155°C (266°F to 311°F).At a continuous 55.6W, the internal core of the coil reaches this breakdown temperature within 3 to 5 seconds. The Result: The enamel begins to soften and degrade. While the coil might not be visibly smoking yet, its lifespan is permanently compromised because the insulation has broken down structurally.
2. The 10 to 15 Second Window (Catastrophic Melting)Once the insulation fails, the physical geometry of the coil causes a runaway chain reaction: The Short-Circuit Cascade: As the enamel melts, adjacent bare copper wires touch. This bypasses whole sections of the coil, dropping the resistance significantly below 14.1 ohms. Current Spike: When resistance drops, current increases (I = V/R). If the resistance drops to 5 ohms, the current spikes from 2 Amps to 5.6 Amps, shoving the wattage up to 156 Watts.The Meltdown: At this point, the temperature instantly rockets past 200°C (392°F). This is the melting point of the nylon/plastic coil sleeve inside the core. The sleeve warps, melts, and fuses to the metal plunger, completely seizing the mechanism within 10 to 15 seconds. Concurrently, the outer paper wrapper chars and smokes.

Repair Community Verification. This reality is well-documented across repair archives like the Pinrepair Guides and community forums such as Pinside. Technicians frequently document that if a driver transistor shorts out upon powering a machine up, a failure to cut power within roughly 10 seconds will result in a "crispy," ruined coil that must be cut out and replaced.

Based on this info, the appropriate Bourns PTC FOR THIS COIL is the MF-R030, which, according to datasheet, trips in 1 second timeframe, well away from a damaging duration. The part does have a 1 ohm initial resistance, but because the current is near 95% design value, there should be no problem there. And since your driver pulsewidths are 25-30 millisecond range, an activation repetition frequency of up to 5Hz ( times per second) , keeps the average DC current below the MF-R030 Hold Current rating of 0.3A, which is 0% chance of nuisance trip.

Most of this is AI, except for their PTC part selection, where it failed to put two and two together.

Most amusement devices are controlled by a board that is designed for a variety of different applications.

The hammer drivers aren't tailored to specific loads; the software "knows" which coils it can keep energized and for how long. The SOA is enforced in the firmware to allow the hardware to be more "universal"

The processor knows to release the drive to a pop-bumper driver (for example) much quicker than for the "lockout coil" (prevents the machine from accepting coins when not ready to do so... as such, it is typically energized for very high duty cycles but doesn't "melt" as the coil design accommodates that high duty cycle)

And, even if the kinematics have been altered (e.g., the machine isn't leveled properly -- or, is intentionally being operated in an abnormal deployment), the CPU is smart enough to know not to let a coil remain energized even if the current state of play

In commercial settings (not home arcades), repairs are incredibly inconvenient. And, time consuming. The machine may be sited in a bar, prison, hospital, bowling alley, private club, etc. The "operator" typically only plans on VISITING the machine to empty the cash drawer and make small/quick fixes (replacing rubbers, etc.)

Encountering "out of order" signage (or a complain from the location of some misbehavior) means his schedule has been disrupted and will likely be disrupted tomorrow or later as he makes time to actually fix the machine -- using tools that he will transport TO the location.

Note that there are other machines with similar issues -- juke boxes, vending machines, etc. -- that have similar repair constraints. And, the folks making the repairs aren't always particularly skilled...