Inline External Inrush Current Limiter

Inrush current can be a major problem if you have an amplifier with a transformer over 300VA, especially if it uses high-value filter capacitors (more than 10,000µF, or 10mF).  Traditional inrush limiters are internal, with ESP's Project 39 being an example of one of the earliest circuits published on the Net.  However, while this is fine if you're building your own amplifier and can accommodate the PCB and an auxiliary transformer, it's less satisfactory if you have commercial equipment that doesn't include the necessary circuitry.  You can likely fit a P39 board into the amp, but if it's under warranty or you don't feel comfortable making internal modifications, then it's not for you.

Inrush limiting can also be useful with some mains-powered power tools, such as angle grinders, circular saws and heavy-duty drills.  If you have a power tool that 'kicks' when it's turned on, that can be dangerous.  A soft-start will limit the kick, but not limit the power when it's needed.  This design is suitable for (almost) any load, but should not be used with air-compressors, refrigerators or air-conditioners (these need all the power they can muster, as they start under load).

For detailed analysis on inrush current, see Inrush Current Mitigation.  The article includes oscilloscope captures of transformer inrush current, both with and without a capacitor-input filter (the most common for many types of equipment).  It also covers many different types of equipment, but it's entirely up to the reader to decide if the system described here is suitable.

WARNING:  This circuit is directly connected to and controls household mains voltages, and must be built with extreme care to ensure the safety of you and your loved ones.  All mains wiring must be segregated from low voltage wiring, and in many countries, mains wiring must be performed only by suitably qualified persons.

There are a (small) few external inrush limiters, but they won't work unless you turn off the power at the wall outlet - assuming that your wall outlets have switches.  In many countries, they don't, and even where this is normal, the outlet isn't always accessible.  What's needed is an inrush limiter that operates when you turn on your equipment with its own power switch.  Having an external circuit that can do that may seem somewhat unlikely, but it's not especially difficult with a well designed circuit.

There is one essential component - a wide-range current sense circuit.  This is used to detect that the equipment has been turned on, and after a suitable delay, the current limiting element(s) are bypassed, providing the full current needed for normal operation.  An essential part of the circuit is a 'self-reset' function, so the current limiting devices are put back in series with the mains as quickly as possible after the equipment is turned off.  The same technique for monitoring current is used in Project 79, Current Sensing Slave Power Switch.  However, there's a major difference because a slave switch doesn't need to reset quickly - a few seconds is usually perfectly alright.

When a timer is involved, the circuitry becomes more complex.  A simple on/off switch can be delayed a little at power-on and power-off, but it's unpredictable.  An inrush limiter needs to be predictable, otherwise the delay will vary, possibly over quite a wide range.  The current waveform drawn by an amplifier (or a power tool) isn't predictable, and the timer circuit has to be able to provide a preset time delay with any waveform.  It may vary, but not so much that it will cause a problem.

There will always be some limitations with any circuit that measures the current, because it can fall to a fairly low value with light loading.  Most power amplifiers will draw sufficient current to make it easily detected, and having tested a couple of current transformers (AC-1005 and ZMCT103C) it was quickly apparent that they cannot be used 'normally' with very low current.  Even with two or three turns through the current transformer, the mains power transformer didn't draw enough current for it to be detected.  Extra primary turns increase the sensitivity, but a 1:1000 current transformer doesn't work well with less than 10W drawn from the power transformer.  If you want to know more about current transformers, see Section 17 in the transformers article.  Even with the detection scheme shown in Fig. 1, if your equipment uses less than 5W when idle (very unlikely I would think), you cannot use a current transformer.  Without some current drawn from the transformer secondary, the detector is only left with the equipment transformer's magnetising current, which can be a lot lower than you may imagine.  It's very hard to detect less than 50mA mains current, even with the burden resistor (in parallel with the current transformer's output winding) increased from the normal 100Ω.

Note that it's the responsibility of the user to determine if the power amp (or other equipment) used is suitable for use with a soft-start circuit.  Most equipment will be alright, but there will be exceptions.  The manufacturer or supplier should be able to tell you if a soft-start circuit can be used or not.  Please do not email me to ask, because I cannot answer the question.  I don't have information on every product ever made.

The most important thing is that the equipment must draw at least 20W (20VA) when idle, and preferably more.  If the minimum current draw is too low, the current-limiting resistors or thermistors will remain in circuit, and you may get the circuit cutting in and out while music is playing.  While this probably won't harm anything, it's not a good idea.  If possible, verify the power used by the amplifier under idle and operating conditions, using something like Wattmeter for AC Power Measurements.

Note Carefully:  The unit described must not be used with a power board or any mains distribution system.  It is designed specifically to power one (1) high-power load only!  If you were to use it with a power board or similar, the first device turned on will operate the soft-start circuit if it draws more than 30mA from the mains  Everything else will be turned on with no soft start at all.  This is almost certainly not what you want to achieve.  If other equipment has a 12V trigger input, a trigger output can be provided as shown in Fig. 3.

An electronics magazine recently published a design that does much the same as the one described here, except it uses a high-power MOSFET to control the inrush current using a principle similar to a leading-edge dimmer.  While you might assume that this has some advantages, it's largely an illusion.  The circuit is controlled by a PIC microcontroller, and is vastly more complex than the one shown here.  I know that a 'dimmer' type circuit works, as I've tested it, but there are no real advantages compared to a simple design as described here.  There are several disadvantages though, not the least being providing protection for the MOSFET (or TRIAC) switch from damage due to over-voltage or over-current (several extra parts are devoted to this).

The magazine circuit uses a MOSFET and PIC, a high-current bridge rectifier, plus two optoisolators, a 230V coil relay, an opto-coupled TRIAC driver IC and a high-frequency transformer that you have to wind.  Some parts are SMD, which is (IMO) an un-necessary pain, just to include a precision rectifier that isn't needed anyway.  Should any of these fail in a few years, the circuit is scrap if you can't get the parts any more (particularly the programmed PIC).  The version shown here uses very simple circuitry that can be repaired tomorrow or in 20 years.  All parts are deliberately generic, with nothing that can't be substituted for something similar.

There are several options for monitoring current in an AC circuit.  The first is a current transformer, which up until recently was the only low-cost option.  A CT provides excellent isolation, and all mains wiring through the transformer can easily be made very safe.  You can also use a diode string, with two or three diodes in series, with another equal string in reverse-parallel.  This provides a comparatively constant output voltage regardless of the load current.  The voltage developed across the diodes can be used to activate an optocoupler or can be coupled with a small transformer.  The transformer will be a mains type, typically used with the secondary across the diodes, and with the primary used for the output.  This combination is a lot harder to insulate properly to prevent accidental contact.

Another method is a Hall-effect current monitor IC, such as an ACS-712 or similar.  These are available as a small PCB designed to interface with an Arduino or similar.  Unfortunately, these are too noisy to be useful at low current (less than 50mA is buried in noise).  I have tested the ACS-712 and the measured noise is almost 20mV, and 30mA current only gives an output of 5.5mV.  Without additional processing it's impossible to separate the signal from the noise.

Finally, there's a current shunt - a low resistance in series with the load.  The voltage across the shunt is monitored, and it has a voltage proportional to the current.  A 'typical' shunt may be 0.1Ω (100mΩ) that will provide a voltage of 100mV at 1A.  Unfortunately, the shunt dissipates power, and with 1A it dissipates 100mW, rising to 10W at 10A (I²R).  Unlike the other techniques described, there is zero isolation, so all circuitry is at mains potential.  This option is strongly discouraged.

For a complete discussion of current detectors and measurement methods, see Current Detection and Measurement.  This covers all the available options, and will help to explain why I elected to use a current transformer.  They are readily available and low-cost, and provide the ultimate in electrical isolation (and therefore safety).  Reliability is unsurpassed - I've never heard of one failing.

Because all mains wiring is insulated (other than the relay connections), the CT is preferred, as it provides extremely good isolation between mains voltages and the rest of the circuit.  The detection threshold may be greater than a diode and optoisolator option, but that's not likely to cause any issues.  The only change that's needed depends on the current drawn by your equipment.  As a guide, the following table should help.  I_Min is the minimum current that can be reliably detected.  For most applications (other than really high power), a 3-turn CT primary is probably the best compromise with an AC-1005 current transformer.

The above is a guide, and is based on acceptable dissipation within the CT's winding.  For example, if you use 5 turns with a 10A continuous load, the output will be up to 50mA (1mA/A × 5 turns).  This will result in a current transformer dissipation of 100mW, assuming a 40Ω winding.  While this is acceptable for the current transformer, it subjects the base of the detector transistor to more current than it's designed for.  Ideally, the base current shouldn't exceed 1/3 of the collector current.  More can be tolerated for brief periods.  Most datasheets fail to mention the peak base current at all, other than for some power transistors.

The inherent nonlinearity of the CT is actually our friend in this role.  I ran tests and verified that the AC-1005 can provide 50mA with 50A (or 10A with 5 primary turns).  I tested the AC-1005 CT with 50A primary current (one turn), and it happily provided the full 50mA expected.  Including an input resistor (in series with the CT) limits the current into the base of Q1 to a reasonably sensible value, measured at around 30mA RMS with a 50A primary current.  The ZMCT103C is another contender, available from eBay for less than AU$2.00 each.

Based on tests I've run, you need 3-5 turns with the AC-1005, and 1 turn with ZMCT103C.  I'm unsure why the ZMCT103C performs well with only one turn when the AC-1005 requires 5 turns to get the same sensitivity with the detector circuit shown below.  The improved sensitivity of the ZMCT103C CT is a good thing, because it only has a very small centre hole (5mm) compared to the AC-1005 (9.5mm).  This makes it hard to get more than 3 turns, even with relatively thin cable.

The heart of the project is the current detector.  When the remote equipment is turned off, there should be zero current drawn from the mains.  In some cases, you may have an amplifier with an AC mains EMI filter, and this will draw a small (capacitive) current all the time.  If your equipment has an in-line filter before the mains on/ off switch, you will probably not be able to use this circuit.  It's difficult to make the detection threshold adjustable (although you could add a 100k pot in parallel with the CT's secondary), and the detector is designed to detect zero current and anything above about 30 milliamps, depending on the number of primary turns.  The detection circuit is very different from others I've used (e.g. Project 79).  Some equipment will draw very little idle current when turned on, but for most gear it's very unlikely that it will be much less than 100mA.

The circuit uses all low-cost parts.  Resistors can be carbon film (cheaper than metal film), and the two transistors can be almost anything you have to hand.  Only the opamp is critical, because it must have an output that goes close to the negative supply.  Most opamps can't, with a minimum output voltage of perhaps 1-2V.  The LM358 is one example that definitely works, but there are a few others.  Be aware that some have a supply limit of 5-6V, limiting their usefulness.  R13 (1k) can be added if you wish to use an opamp such as a MC1458 or similar.  The resistor ensures that Q2 can turn off when the output of U1B is low (~2V minimum).  The remainder of the circuit is unchanged.

Figure 1 - Detector, Timer And Relay Driver Circuit

With the circuit shown above, detection is 100% reliable with anything over 30mA RMS.  The timing is consistent, as it's determined only by the presence of current above the threshold.  The typical delay is around 300ms with any load current.  This may not sound like very much, but the vast majority of supplies will have settled to (close to) steady-state conditions before the timeout.  The exception is circuits with particularly large filter capacitors, and in this case, R7 may need to be increased.  With the values shown, the circuit resets in less than 50ms, but the relay cannot release instantly (about 6ms is typical).  You may wish to increase the time if the circuit is used with a power tool.

The opamp is an LM358, which was selected for a number of reasons.  It can pull the output to zero volts, so it's easy to drive the relay driver transistor.  The inputs can also operate down to zero volts (actually a little below zero), but that's not a requirement in this circuit.  Power consumption is truly miserly - around 500µA, independent of supply voltage.  Finally, they are cheap, and you should be able to get them for no more than AU$2.00 (or as low as 70c, depending on supplier).  I've used them in quite a few (non-audio) projects, and I don't think I've ever managed to blow one up, despite often rudimentary test lash-ups.  You can use almost any opamp you like if R13 is included.

The circuit has four stages.  The first is based on Q1, which discharges C1 with each current pulse detected.  If there's no current, C1 remains charged to 12V, and the second stage (a comparator) maintains its output at (close to) zero volts.  Current detection is either on or off, and there's no halfway point as the comparator has hysteresis provided by R6.  When current is detected and the voltage across C1 is less than 6V, the output of U1A goes high, and charges C3 via R6 - this is the delay timer.  Once the load current stops, C1 charges quite quickly, and when the output of U1A goes low, C3 is discharged via D2 in less than 50ms, the relay releases and the circuit is ready for another turn-on cycle.  The delay time can be altered by changing the value of C2 (10µF), with a larger cap providing a longer delay.  Alternatively, R7 can be increased (or decreased).  Don't go above 100k though (about 1.1 seconds).

U1B is a comparator for the timer circuit, with hysteresis provided by R11.  The output of U1B drives the final stage, the relay driver transistor (Q2).  When the relay activates, the inrush limiting resistor(s) are bypassed by the relay contacts.  You might be curious as to why I recommend a relay rather than a TRIAC.  There are two reasons, with the primary one being electrical safety.  The relay contacts are at mains potential, but the drive coil is fully isolated.  A TRIAC has both its input and output at mains voltage, although a TRIAC driver such as the MOC3021 could be used to provide isolation.  Secondly, relays have extremely low contact resistance, where a TRIAC dissipates roughly 1W/A, and almost always needs a heatsink.  TRIACs are also less likely to survive a major fault, while a relay is close to immune if the contacts are rated for 10A or more.  A more-or-less typical 12V, 10A relay has a coil resistance of 270Ω, and draws ~45mA.

The high sensitivity of the detector is needed because the current drawn can vary widely.  A low sensitivity could mean that the circuit would no longer detect current with the amp at idle, and the relay would release.  This is not acceptable, and that's why Table 1 has been included.  When the equipment is turned off, the relay will be deactivated in less than 60ms.  This is important, because if the soft-start circuit has bypassed the limiting devices (resistors or thermistors) and the gear is turned on again, there is no soft-start because the relay still bypasses the current limiter.

Ideally, you need to be certain that your equipment will never draw less than around 50mA from the mains (20VA @ 230V or 10VA @ 120V).  It's very unlikely that any equipment will draw less, and even a small power amp will almost certainly draw more than 150mA at idle.  Very low idle current could cause erratic operation, and while it won't hurt anything, it's undesirable (and the relay clicking on and off will be annoying).  The output from U1.7 can be used to provide a 12V trigger (positive only) as shown in Fig. 3.

Figure 2 - Power And Relay Wiring

The current limiting thermistors (e.g. N20SP010 20mm, 10Ω or similar) or resistors are wired as shown above.  The relay is wired with its normally open contacts across the limiting resistors, so that when it activates, the resistors are shorted.  This is exactly the same technique described in Project 39.  The fuse is recommended, and it can be part of an IEC mains input connector.  The value is determined by the relay contact rating, so with a 10A relay, you should use a 10A fuse.  Note that the 'COM' (common) is earthed (grounded) and must not be used if you decide on a transformerless supply.

The relay and other terminals carrying mains voltage should be protected by heatshrink tubing to prevent contact.  The line fuse will typically be part of the IEC mains input socket.  The mains output should ideally be a chassis mounting type as used for wall outlets where you live.  As an option (not shown), you can include a thermal fuse in contact with at least two resistors or thermistors.  This protects against a prolonged overload as may occur if the relay fails to operate for some reason.  Be warned - you cannot solder thermal fuses as you can other components, as the soldering heat will cause the fuse to become open-circuit.  They must be crimped or terminated with screw connectors, or use a heatsink on the leads (next to the fuse) to absorb the heat from soldering.

The MOV (metal oxide varistor) is optional, and needs to be selected for the mains voltage (230V or 120V).  If you include it, you'll need to make sure that you get one that's specifically intended for the mains voltage where you live.  The datasheets aren't always clear, but a reasonable 'rule' is that the rated RMS voltage should be 1.2 × the mains AC voltage.  For 230V use it should have a voltage rating of not less than 275V RMS, or 150V RMS for 120V mains.  If you're not sure, seek assistance from someone who knows how to use them properly.  MOVs are normally rated for the voltage where they pass 1mA, and I tested four 230V MOVs, and they passed 1mA (peak) at about 268V.  Different manufacturers have differing recommendations, and if you're unsure, don't fit the MOV at all.

You need to decide on the current limiting devices used.  Thermistors are a good option, but using the arrangement described for P39 (3 x 150Ω, 5W resistors in parallel, 50Ω) is a tried and proven technique.  For this project, I'd be more comfortable using 3 x 10Ω thermistors in series (30Ω total) for 230V systems, with two in series for 120V.  Either option will limit the worst-case current to under 10A at power-on.

The mounting for the resistors/ thermistors (and the MOV if used) is critical.  They have the full mains voltage on the leads, and must be securely mounted so they cannot fall from their mounting should they overheat (and melt the solder) for any reason.  Use fibreglass tubing over the connections if possible.  A separate enclosure or a cover is recommended, which must provide ventilation and keep stray fingers away from live connections.  I ran some tests on a 1kVA transformer using 3 x 10Ω 10W resistors in series (which I have set up in a high-power load box), and even after repeated operation they refused to get above 'slightly warm'.

If you use a 'conventional' power supply (mains transformer based) or a small SMPS, I suggest using a 12V supply.  These are easier to make with off-the-shelf parts, and you use a 12V relay.  You need a relay that can switch at least 8A, preferably 10A (this should be increased to 20A for high-power with 120V mains).  Most equipment won't draw anywhere near that much current, but it's always wise to use a relay that's capable of more current than you require.  This minimises contact erosion.

Figure 3 - Circuit Diagram Of A Transformer Based PSU

The supply doesn't need to be regulated, as the circuit behaviour changes very little with supply voltage.  Of course, you can use a 12V secondary rather than 9V, and add a 7812 regulator if you wish.  The 12V supply shown will be a bit over 12V before the relay operates, but the circuit will compensate easily.  The 12V trigger output is optional, and it gets its drive signal from U1.7 in Fig. 1.  This can also be used with a small switchmode supply.  It cannot and must not be used with a transformerless supply as shown in Fig. 5, because everything is mains voltage.

Figure 4 - Power Supply, Using A 12V Plug-Pack (Wall Transformer)

If you use a plug-pack, it can be external, which although easy to do is decidedly sub-optimal.  If it were to be disconnected (by accident or otherwise), the circuit has no power and the relay cannot close, so the limiting resistors will remain in-circuit.  A better option is to remove the 'innards', and mount the PSU internally, protected against accidental contact (half of the PCB is live).  The photo shows how the PCB can be mounted to a piece of acrylic or similar, which would have a cover and additional acrylic base-plate fitted to prevent contact with live parts.  Alternatively, re-mount the PSU in its original case, with the plug pins cut off and replaced by mains cable.

A switchmode supply will draw less idle current, but getting a good one can be a challenge.  My recommendation is to buy a 'plug-pack' ('wall wart') supply from a reputable supplier, and use the PCB, liberated from (and/ or returned to) its original enclosure.  Great care is required of course, and remember that you will be dealing with mains wiring, so it has to be as close to 100% safe as you can make it.  The advantage of the SMPS is that it's small, and has minimal losses.  The disadvantage is that it operates at mains voltage, and great care is required to ensure that no part of the mains circuitry can be touched when the cover of the complete unit is removed.

An alternative is a so-called 'transformerless' power supply.  This is something I normally don't recommend because it means that everything in the enclosure is at mains voltage.  You cannot work on the circuit other than by using an external power supply.  This is suggested for experienced constructors only.  There is also a change needed, as the series mains capacitance needed to power a 12V relay is excessive.  The relay needs to be a 24V type to minimise the capacitance needed.  No changes are required for the detector/ delay circuit.  The only capacitors you should use are X-Class (mains rated, fail-safe) types.  You'll see countless circuits using 'normal' capacitors rated for 400V DC, and these are guaranteed to fail at some point.  DC capacitors are not suitable for mains voltages, and their use with mains voltage is very dangerous.

Note that for 120V mains, you need two 470nF X2 capacitors in parallel, or there will be insufficient current to power the relay.  The power for the supply must be taken directly from the mains input.  If it's taken after the current transformer there's enough current drawn to activate the relay.  This would cause the whole circuit to by permanently bypassed.

Figure 5 - Transformerless Power Supply

With the values shown, the supply can provide 28mA (50Hz) or 32mA (60Hz), with the relay taking the lion's share when it operates.  Suitable 24V relays are easily found from all major suppliers, and should have a coil resistance of 1kΩ or more.  Any relay that draws more than this will cause problems with the circuit's operation.  Note that the 12V trigger option cannot (and must not) be used with this supply!

For use with 120V mains, you need two 470nF caps in parallel.  R1A/B are used to discharge the series capacitor when power is removed.  Without them the cap can store a dangerous voltage for a long time.  All resistors should be 1W to ensure they can withstand the voltage.  R1B is not required with 120V mains.  The capacitance for C2 is the minimum, feel free to use more if you prefer.  There are two 12V 1W zeners in series, because the dissipation is too high for a single zener (just under 1W for the pair when the circuit is idle).  Once the relay operates, it takes most of the current (around 24mA).  The start-up time for the supply is about 400ms for both 50Hz and 60Hz versions.  Idle power dissipation is around 1W for both versions.

These supplies are particularly dangerous if you need to take any measurements or perform diagnostics.  If you need to do any work on anything powered by a transformerless supply, it must be disconnected from the mains - do not rely on a switch on the wall outlet or elsewhere.  To work on other parts of the circuit (the current detector, relay driver, etc.) the circuitry must be powered from an external lab supply.  The circuit itself doesn't know the difference, and provided mains wiring to the power supply is disconnected, you can still test the circuit's operation.

The simulated waveforms below show the peak transformer primary current with/ without 30Ω limiting.  The green trace is for a 45V transformer, rectifier and 22mF filter cap, switched on at 100ms.  The load across the 64V DC output is only 22k (2.9mA), which is far lower than any 'real' circuit.  The peak current is 23A for the first half-cycle, diminishing rapidly with time.  When the limiter is active (same PSU circuit), the peak current is reduced to 7.3A.  The relay closes at 370ms, and you can see a very small increase in current as that happens.

Figure 6 - Inrush Current (Green, No Limit, Red Current Limited)

Unfortunately, the simulator doesn't show the transformer's inrush, which can be very high.  If power is applied at the zero-crossing, the transformer core will saturate, and the peak current could be as high as 45A (5Ω primary resistance was assumed for the simulation).  Worst-case transformer inrush current happens when power is applied at the zero-crossing point, but that gives the lowest capacitor inrush.  Turning on power at the peak of the AC waveform gives the lowest transformer inrush, but the highest capacitor inrush.  Normal power-on is effectively random with the vast majority of equipment.

As noted in the introduction, the article Inrush Current Mitigation is a useful reference.  The only real difference between this project and Project 39 is that this unit is designed to be external, in a separate sub-enclosure.  By sensing when the protected equipment is turned on, it bypasses the inrush protection resistors/ thermistors after the time delay, and they remain bypassed until the equipment is turned off again.

I checked a very large power supply that I use for testing, and it has a 1kVA transformer and provides ±95V DC.  With no inrush protection, the highest current peak I measured was 80A, and it will happily blow (i.e. vaporise) a 10A fuse (which it did - it's normally brought up slowly with a Variac so there is no inrush current to speak of).  With a 30Ω inrush limiter, it's impossible for it to draw more than 11A peak with 230/240V mains.

This project describes a technique that no-one else seems to have thought of.  It's likely that readers will see other possibilities as well, because the ability to sense that an appliance is drawing current and turn on something else is common.  Project 79 has been on-line since 2001, and shows another technique that isn't described here, namely using a small transformer (6V output or thereabouts) wired in reverse, so the primary becomes the secondary, to get a voltage boost from a low-voltage source.  When the project was published, current transformers were not readily available and that was the easiest.

You can now get a current transformer for less than $2.00 quite easily, but that was not the case in 2001.  Using a 'normal' transformer in reverse is still a good option, but it's likely to be more expensive than a CT.  As with most ESP projects, this one is geared towards providing information that you won't find elsewhere.  Feel free to let your imagination run wild, and make sure you read Current Detection and Measurement, making use of whichever detection technique you desire.

During the development of the circuits described, I made extensive use of Project 207 - High Current AC Source.  I tested current transformers to 50A (and beyond), something that would have been very difficult without the high-current transformer.  The small current transformer (ZMCT103C) I tested is too small to get ten turns through the centre hole, and I really needed to know what they would do when subjected to 50A.  Needless to say, the core saturated, but surprisingly not until the rated current was more than doubled (to over 10A).

This is a unique project, as most inrush limiters expect to receive power along with the device being powered.  I saw nothing that provided inrush limiting if the limiter is powered permanently and the gear is simply powered on/ off with its own mains switch.  I don't expect that it's something that many people will need, but if you have equipment that can benefit from inrush limiting but doesn't have it internally, then this lets you do it with no modifications to the gear itself.  It's especially useful for power tools that 'kick' when turned on (a very common problem).

Please note that if thermistors are used with equipment that starts and stops regularly, they may heat enough to reduce their effectiveness.  This is something that only the end-user will know, so if you build one of these circuits for a power saw (for example), high-power resistors are probably a safer option.  For circuits that draw very high current, you'll need to run tests to determine the maximum (and minimum) current, and re-evaluate the value (and rating) of the limiting resistors/ thermistors.

It's entirely up to the user to determine its suitability with his/ her gear.  I've provided a lot of information here, and adjustments will be necessary to ensure it's compatible with the expected current draw.  Predictably, ESP accepts no responsibility if the use or misuse of the circuitry provided here causes loss or damage to the powered equipment or any other gear.  It's expected that constructors will have the knowledge and skill to build the circuits in a workmanlike manner, and that all safety precautions are taken.

One thing that some readers will realise is that everything done by the circuits shown can be done using an Arduino or a PIC.  From my perspective, there are several things I don't like about that approach.  The first is that the way the circuit functions to others is inscrutable - it's just a small PCB or IC with some code in it that 'does things'.  If I were to go this way, I'd probably use an 8-pin PICAXE device, because I have them in my parts drawers and they are low-current and fairly flexible.  The input sensor and relay driver transistors are still needed, and the only parts saved would be a few very cheap parts - resistors, capacitors and a diode.  As I've discovered from past experience, if the IC dies after a few years, the program would have to be re-written as programming interface updates can break 'old' code.  There's nothing complex about it - the real 'smarts' are in the current detector anyway!

If done in software, it's a very simple state machine.  The input is from the current detector (active low), and the output drives the relay after a predetermined delay.  When the input goes high, that indicates that current is not being drawn, so the external device has to have been turned off.  The timer is reset, releasing the relay and the state machine waits in an endless loop for the input to go low again.  The state machine is so simple that even the lowliest PIC will still be overkill.  While it will save a few parts (and add others, such as a regulator), you have to ask if it's worth it.

The current drawn by an Arduino is a great deal higher than the circuits described (idle current is less than a couple of milliamps, mainly due to the 'power-on' LED), and when (not if) the processor platform of choice fails (this could be anything from a year to a decade), if you no longer have the code, the circuit is scrap.  The circuitry I've shown lets you see each step of the process, and anyone writing code would have to work out these details before starting.  If something doesn't work, then you have to analyse the code to see why.  I happily accept that the code is dead simple if you know what you're doing, but I still prefer the analogue approach.

With an analogue circuit, each stage can be tested and verified independently, and opamps like the LM358 have been around for a very long time, and aren't going away.  If one fails, it's easily replaced, and the circuit function isn't changed at all.  More to the point, the constructor learns about analogue electronics along the way.  I've lost count of the number of 'Arduino people' who post very basic questions about simple analogue functions - it's quite obvious that they don't know (and aren't even interested in) Ohm's law!

While a PIC can (at least in theory) dispense with most of the parts, it still needs the input detector (Q1 in Fig. 1) and a relay driver transistor (Q2).  It's possible to dispense with Q1 and implement the detection in code, but the PIC needs a good ADC (analogue to digital converter) and detecting low-level pulsating voltages accurately may be quite irksome.  The requirement for a 5V (or perhaps 3.3V) supply means that a regulator is needed, and the overall cost is almost certainly going to be greater than the solution described.  If you want to use an Arduino or other microcontroller, you're completely on your own.

As noted in the introduction, an electronics magazine has published a circuit using a PIC and high-current MOSFET.  I can only assume that the idea was prompted by this version (I know that my website is well known to the magazine publisher), but they obviously couldn't use my idea as that would be a breach of copyright. 

• Project 39 (ESP)
• Inrush Current Mitigation (ESP)
• Current Detection and Measurement (ESP)

Main Index

Projects Index