|Elliott Sound Products||Project 144|
Mains Power Sequencer
Rod Elliott (ESP)
Anyone who has tried to turn on several large power amps at once will know that it's pot luck whether the main circuit breaker trips or not. It's not just power amps - even large banks of lights or anything else that draws significant inrush current will have the same problem. It doesn't happen all the time, but it can make powering up your system a rather hair-raising experience.
The major problem is that any one large power amp (or motor, bank of lights, etc.) draws a significant inrush current. When you have several, that current is magnified until the point is reached where the circuit breaker simply cannot withstand the surge and trips. Even if you do have a regimented power-up scheme, that doesn't work if the power goes off for any reason, then comes back on again some time later.
Likewise, the equipment (and the power switches for each) may not even be within reach, so it then becomes a major issue to ensure that everything powers up and breakers don't keep tripping. For this very reason, several manufacturers make power sequencers - power is applied to all the connected equipment one after the other, rather than all at once. Commercial sequencers typically operate as a number of 'banks' - 3 is not uncommon, and each is powered up with a 1-3 second delay between each. Each bank may have 2-3 power outlets, but for many loads that may still cause problems.
The system described here uses a 4-stage sequencer, and it works both with power-up and power-down. It's also easy to add a low voltage remote switch and over/under-voltage protection. The latter might be useful for stage systems that are run from a 3-phase break-out box. Should the neutral fail (and it happens more often than you might imagine!), you can easily end up with several racks full of blown amplifiers. Those with conventional transformers will probably just blow a fuse, but many switchmode power supplies will self destruct with a severe over-voltage. Unfortunately, any voltage protection scheme is almost certainly unable to protect against this kind of fault, but over-voltage protection may still be considered useful by some constructors.
Bear in mind that it is actually unlikely that power can be disconnected fast enough to save everything under neutral fault conditions, but hopefully at least some of the gear will survive if powered off quickly enough. I consider the over-voltage protection to be worthwhile, but it is unrealistic to expect it to be able to save everything from damage (or even complete destruction).
In common with most commercial power sequencers, the last item to receive power is the first to be turned off when the sequencer is turned off. Sequencers are often used to ensure that any equipment that is likely to cause loud transients is turned on first, and before power amplifiers (for example), and likewise they are the last to be turned off, so the opportunity for nasty noises is minimised because the amps are off until everything has settled, and are turned off first so that they can't reproduce the switch-off transients caused by other equipment.
Before we continue, I must provide this warning and disclaimer ...
Much of the circuitry used in this project operates at mains potential, and is therefore potentially lethal. Do not attempt construction if you are not 100% confident of your abilities to safely work with and wire mains circuits. In some countries, it may be illegal for non-qualified persons to construct or work on mains powered equipment. ESP accepts no liability for death or injury if you choose to build the project. Do not ignore these warnings. The material presented in this article describes equipment that can kill or seriously injure anyone who builds it. Extreme caution is advised. NEVER work on the project with mains power applied.
It is very important to understand that in some countries (such as Australia!) this unit would probably be classified as a 'power distribution board', and as such requires mandatory approval. If you build one for yourself and you ensure it's safe there's probably not much to worry about, but it may not be technically legal to use it anywhere that has regulations prohibiting unapproved electrical items that fall into specific categories (known as 'prescribed articles' in Australia). You must check local regulations to verify that it is legal to use a system of this kind. This is the responsibility of the builder - I don't know and I'm not going to even try to find out where it can or can't be used legally. If approval is mandatory where you live, then do not build this unit!
For what it's worth, I'm fully aware that much of the basic circuitry could be eliminated by using a PIC or other microcontroller. However tempting this might seem, it also has certain risk factors. Should the PIC used go out of production there's nothing you can do if it fails. You literally end up with a box of bits and pieces that you can't use any more. A PIC would still need transistors to drive the relays, and it still needs a power supply. You also need 'real world proof' input circuits and the necessary code has to be written.
Compare that with a simple hardware design. In 20 years time, you'll still be able to fix it should something go wrong. The power supply is most easily fixed by replacement, and you might need to change the odd electrolytic cap that has lost capacitance. Other parts are so common and have been with us for so long that they are unlikely to go away any time soon. They are also cheap, but most can be expected to have an almost unlimited life. A PIC won't help you one bit if the relays fail, but you will be in serious difficulties if the PIC fails and you don't have a spare.
From the perspective of many DIY people, showing a design implemented in hardware also demonstrates general circuit principles and provides a great learning tool. Even if they don't build the project, just seeing how things can be done can provide valuable insight. You get very little of that while looking at the diagram of a PIC - a central inscrutable block with a few input and output lines. While PICs are wonderful tools, there are some things that really should be done with hardware - apart from anything else, it's a lot more fun.
A cost comparison between the two techniques would reveal that the PIC approach is cheaper, but probably not by very much. Only considering the sequencer and not the loss-of-AC detector, there are certainly savings to be made, but there are additional parts to be considered - the PIC alone can't do everything without help. It needs a 5V regulator, and suitable protection from the outside world is essential. Suffice to say that the hardware version exists here, and the PIC based version doesn't (and it's not likely either).
The sequencer is based on a series of opamp comparators, supplied with a linear ramp as a capacitor charges via a constant current source. While it is certainly possible to just charge a capacitor using a resistor, the resulting curve is exponential. This makes it harder to get equal timing between the units being powered up, and if power-up is made equal then power-down will be unequal. The linear ramp is easy to achieve, and for the small extra cost (a few cheap general purpose transistors) we get a far better overall result. A second current source (a current sink, actually) discharges the cap when the power sequence switch (local or remote) is turned off.
This means that the equipment is powered down with the opposite sequence as it was powered up. The design presented here uses a 4-stage sequence, with a nominal 2-9 second delay between each relay closing. The relays open again in the reverse order, and the timing for both can be made faster or slower. The on and off sequences will usually use equal timings.
R1 and C1 are included to guard against noise and/or voltage transients that may cause false triggering or other problems. In most cases they aren't strictly essential, but I recommend that they are used regardless. Exposing any electronic circuitry to the 'outside world' without protection is never a good idea. Note that the external (remote) switch and wiring must have a series resistance of no more than perhaps 20 ohms or so. That represents a vast length of even very ordinary cable, so should never be a problem. It's also easy to include a 12V trigger input if desired (see below for more details).
Figure 1 - Switching, Current Sources & Comparators
The schematic for the first section is shown in Figure 1. Although transistors are shown as BC549/ BC559, you can use any small signal NPN and PNP transistors you have handy. The transistors are not critical, as they all operate at 12V maximum and a few milliamps at most. C2 should ideally be a low-leakage cap, rated at 25V to ensure that leakage does not cause malfunctions. This part of the circuit works as follows ...
- Decide whether the unit will be operated locally (SW2 set for 'LOC') or remotely, and whether the remote will be a short to earth/ground or a +12V trigger. In some cases, you may decide to use a key operated switch for SW2 so that it can't be tampered with.
- When the switch is operated (either locally or via one of the two remote methods), the upper current source (a current mirror) is enabled, and C2 charges linearly until it reaches the maximum of a little under 12V. As the voltage exceeds each of the voltages shown at the comparator inputs, the respective comparator output goes high, which in turn switches on the appropriate relay (Figure 2). Turn-on current is determined by VR1A and R3.
- When the switch is turned off, the lower current mirror turns on (via R2, VR1B & R4), and discharges C2. Again, as the voltage falls below the comparator's reference voltage, the respective comparator turns off again. Turn-off current is set by VR1B and R4.
- All comparators have hysteresis (positive feedback) to ensure there is a clean switching signal with no unexpected on/off transitions that may cause equipment problems. Q5 is used as an emitter follower to isolate the voltage ramp across C2 from the switching artefacts caused by the hysteresis feedback.
- The 'SD' (shut-down) connection allows the power supply to discharge C2 very quickly if there is a loss of AC. See Power Supply section for more details. If the loss-of-AC detector is not used, connect 'SD' to '+12V'.
With the values shown, each output will turn on at ~2 second intervals with VR1 set for minimum, and ~9.5 second intervals with VR1 at maximum. This includes output #1, which is delayed to allow for cascading. By doing this, the output of the first unit is used to trigger the second (etc.). With two units, this provides an eight stage sequence, with the last circuit turning on after 8 time delays. There is no limit to the number of units that can be cascaded, but it could make the power-up sequence far longer than desirable.
|Note that cascaded units will be triggered to turn off when output #4 is de-energised,
so the power-down sequence for cascaded units is not the reverse of power-on! The power-down sequence for the second unit starts when the
output #4 is turned off, and not when output #1 switches off.|
Increasing the value of VR1 extends the time - the interval is around 9.5 seconds if the pots are at maximum resistance. I do not recommend using higher value pots. In general, there's little point providing very short or very long intervals, but long delays might be useful with some equipment that uses switchmode power supplies or for gear that insists on making loud noises shortly after power is applied.
Should power be interrupted, the unit will instantly reset via D2 and the loss-of-AC detector in the power supply (Figure 3), and the preset sequence will be followed when power is restored. This prevents circuit breakers from tripping with high power loads if there's a short interruption, because the equipment is forced to go through the power-up sequence that you have determined is safe. Likewise, gear that makes silly (but loud) noises still gets to be sequenced, so the power-up phase should be calm and peaceful.
Without a sequencer, if power is momentarily interrupted, when it returns all equipment is powered on simultaneously. If the load happens to be a rack full of 1kW amplifiers, multiple banks of electronically ballasted lamps or other high power loads, the chances of the breaker not tripping is rather remote. Things that make noises will be heard making noises, speakers may be damaged, etc.
If you wish to use (or simply include the provision for) a 12V trigger signal, simply wire an NPN transistor as shown in Figure 1. The base needs a 10k series resistor to limit the current, and the 2.2k resistor between the base and emitter prevents leakage from causing problem. This is shown in Figure 1 and is indicated as optional. In most installations it won't be needed, but for the minimal cost of a transistor and a couple of resistors it provides the facility if it's ever required.
Note that as described, the three different ways of turning the sequencer on or off can all be used, but the unit is ultimately able to be controlled by the 2-position main 'LOC/REM' selector switch and SW1 ('Sequence ON/OFF') on the front panel. You can even use the 'LOC/REM' switch while the system is powered and nothing will happen, provided the front panel 'Sequence ON' switch is in the 'ON' position.
When switched to remote operation, the sequencer will not power-down until both external control signals are removed. For example, if the unit is turned on using the 'REM' port, then it is switched off again by removing the short between the 'REM' port and earth. If there is power applied to the '12V+' trigger input, the sequencer will not shut down! Only one type of remote control should be used, never both. Using both types will (not might) ultimately lead to problems, and rude words being used in abundance .
If powered on via the remote port, the front panel switch can be used to disconnect the remote and restore local control.
The outputs from the comparators are used to switch on the power relays. Each has a LED indicator in parallel so you can see the progress of the sequencer and that power is available on each of the outputs. Again, BC549 transistors are shown, but any small signal transistor that can handle the relay coil current is fine. The transistor base current is set by the 2.2k resistor in series, and will be about 4.5mA. Assuming transistor gain of 100 (the minimum we'd normally expect), that's enough for 100mA collector current, so the relay selection is easier because high sensitivity relays aren't needed. (Note that it is good engineering practice to always ensure that available base current is around 5 times the current theoretically needed.)
The 'TRIG' output is intended for cascading sequencers. If the 'TRIG' output is connected to the 'REM' (remote) input of the next sequencer, it will switch it on and so the second sequencer will power up the next 4 loads in the same way as the first. As noted earlier, there is no electrical limit to the number of cascaded sequencers - the limit is determined by how long you are prepared to wait for everything to be turned on, and/or by the amount of gear you have that needs a sequenced power-up. Also, remember that the second sequencer will start its power-down sequence as soon as output #4 is turned off (which happens first).
Remember that the total load on the sequencer(s) cannot exceed the rating of the power outlets you use. It doesn't matter how many items are sequenced on, provided the total load is less than 10A (typical 230V outlets) or 20A (120V outlets). High current outlets may be available that will permit more high powered equipment to be connected. All relays need to be rated to handle the worst case inrush current of the connected equipment, and 10A (continuous) is the minimum recommended.
Figure 2 - Power Switching Circuits
The circuit is entirely conventional, but be warned that mains voltages are present and proper clearance between mains and low voltage wiring is essential. Under no circumstances should the two sets of wiring be closer than 5mm unless additional insulation is used. The wire colours shown are the standard IEC colours - wire colours may be different where you live, although the IEC code is being adopted worldwide (albeit slowly in some places). In the US and Canada (and likely some other countries as well) the active is black and neutral white, with the earth/ground lead solid green. This is so important that I'm going to repeat the warning ...
|WARNING: This circuit requires experience with mains wiring. Do not attempt construction unless experienced and capable. Death or serious injury may result from incorrect wiring. In some locations it may be illegal to work on or modify mains powered equipment unless licensed. Ensure you know the regulations that apply where you live, and if equipment like this is not allowed, do not build this project.|
Under no circumstances should the mains relays be mounted on Veroboard or similar. This material is fine for everything else, but is completely unsuitable for mains voltages due to the insulation material (paper/phenolic) and close track spacings. In most cases it will be easier to secure the mains relays with a clamp (and glue), and simply hard-wire the mains connections. All relays must be rated for the maximum allowable current from the power outlet, and preferably with an additional safety margin.
For example, for a 10A mains outlet, relays should all be rated for a minimum of 10A, and ideally around 20A. Arc-quenching circuits may be used if desired, but will generally not be necessary if the relays are sufficiently rugged. Different countries may have specific rules about arc-quenching circuits, and local requirements and/or common practice should be followed.
The mains outlets naturally must be of a type that's normally used in your country. All must be 3-pin (including the earth/ ground), rated for the maximum current that can be drawn and wired according to the wiring standards that apply where you live. If desired, you can use a fuse for each outlet, but this isn't really necessary and just adds to the cost of the project. I strongly recommend that you use a fuse and/or circuit breaker for the incoming mains though. It is important to remember at all times that if your power outlets are rated for 10A (for example), then all equipment connected to the sequencer cannot draw more than 10A combined.
For example, a 2kW amplifier will draw close to 10A (~20A at 120V) if operated at continuous full power. The dynamics of music are such that the average current will be somewhat lower, but it is still quite possible for a single stereo power amplifier to exceed the nominal rating for standard power outlets. It is the user's (or installer's) responsibility to ensure that the power outlets are not overloaded when the equipment is used normally.
The circuit breaker is intended to protect against overloads caused by too much equipment being powered at once. If there are specific regulations that dictate the rating and type of such circuit breakers where you live, make sure that you follow the regulations to the letter.
You can also fit a HRC fuse (high rupturing capacity) to protect against serious equipment faults or short-circuit mains. The fuse value required may be determined by local regulations. It may not be required, or your local regulations may stipulate that both a fuse and a circuit breaker must be fitted.
I consider that a suitably rated thermal circuit breaker is mandatory and the fuse is optional. Check your regulations to determine what protective device you need to include to ensure compliance. The breaker rating has not been provided because it depends on local regulations and the available current from the outlet. The breaker would normally be rated for the maximum allowable current based on the type of mains lead and mains plug you fit to the unit.
Many countries allow for higher current loads and there may be dedicated outlets available that allow you to use the sequencer at a higher current than the normal household outlet provide. If these are available, make sure that the sequencer's mains lead and plug are the correct type and are rated for the full allowable current.
I strongly recommend that you use a 12V switchmode power supply. These usually have a rather short hold-up time when loaded, so if there is a power interruption the DC voltage will fall to zero very quickly. As always though, this can't be guaranteed. This applies equally to a linear supply, but the major difference is that a linear PSU will be more expensive and take up far more space than a suitable 12V SMPS. However, it will also be more reliable, and this is an important consideration. There are quite a few small SMPS available that are surprisingly cheap considering their performance. The supply needs to be able to provide up to 500mA - most of the current is drawn by the relays. A higher current supply will not cause any problems and may be easier and cheaper to get.
Naturally, you can also use a conventional transformer based supply, with a bridge rectifier, filter capacitor and regulator (e.g. 7812 with heatsink, or similar). The voltage needs to be fixed at 12V, but it doesn't have to be especially well regulated. As noted above though, it will generally be cheaper to use a switchmode supply. I have shown the complete supply, including the loss-of-AC detector, but if you don't feel that it's necessary for your application you can omit everything other than the power supply itself. The 'SD' connection must be joined to the +12V line. The 'loss-of-AC' detector ensures that equipment is turned off fast, and that the circuit is ready to sequence power back on after quite short interruptions (a few AC cycles only). Without the detector, when power is restored after a very brief interruption everything may turn on at once.
Figure 3 - Power Supply & Loss-Of-AC Detector
I must leave it to the reader to figure out the best way to mount the power supply itself - there are too many possibilities to try to cover them all here. The simplest is one of the small 12V supplies intended for 'netbook' computers, or you can also use a modified plug-pack (aka 'wall-wart') supply, by either removing the insides or terminating the mains to the existing pins - you could also provide an internal mains socket to allow it to be plugged in. Again, this is up to the individual constructor, but make sure that whatever you do is safe and complies with any regulations that may exist. Each relay will draw perhaps 60mA (240mA for all four), the rest of the circuit draws less than 50mA, so a 500mA power supply is more than enough.
One part of the circuit that may cause some consternation is the two 1nF capacitors from the mains input to the mains failure detector. These caps must be Y1 Class (reinforced insulation, certified safety type) components, and must maintain proper clearance and creepage¹ distances to satisfy any country-specific regulations. The value is considered low enough that it is harmless to humans (230V through 1nF is about 72uA, and roughly 45uA with 120V), and they are connected between the mains and low voltage output in almost all switchmode power supplies to suppress RF interference. The cap does not need to be changed for 120V 60Hz operation.
- The terms clearance and creepage refer to separation distances between hazardous voltage (mains) and SELV (safety extra low voltage). Clearance is the separation distance through the air, and creepage is the distance between the two voltages when located on a PCB or other insulating material. Depending on local regulations and materials, these distances may be the same or different. 5mm is an absolute minimum, more is better.
You may well ask why there are two caps and a switch, and that would be a good question . If a cap is only provided for the active (live) connection, the sequencer will not work at all if the active and neutral are transposed. The 'Power On' LED will show that mains is available, but the sequencer will be disabled. Many would consider this to be a good safety precaution (which it is), but it would prevent the unit from powering up at all, and the old rule that "the show must go on" would be broken. For this reason, an optional second Y-Class cap is included with a 'mains reversal override' switch so you can a) use the system, and b) advise the venue that there is a problem with their mains wiring.
The loss-of-AC detector is designed to remove power to the sequencer within less than 50ms after the last mains half cycle has vanished. This can be extended if desired, simply by increasing the value of R29 (22k). Increasing the value to 47k means that you will need a 100ms mains drop-out before everything is turned off. The sequencer will automatically power everything back on again as soon as mains is restored, using the same sequence times as set up in the sequencer itself.
The loss-of-AC detector senses the small current passed by the Y-Class caps, and keeps C6 discharged as long as AC is available. Should the mains be interrupted for more than a few cycles (you can select this), C6 charges and the comparator turns on Q11, which discharges C2 in the sequencer. Note that Q11 is a MOSFET for convenience and cheap high current capability. It has to be rated for a much higher current than any of the other transistors, because the peak discharge current is very high. Use any compatible MOSFET - it's not critical, and the cheapest TO-220 MOSFET you can get will almost certainly be fine. R35 limits the current to a maximum (theoretical) peak of 4.4A, which discharges C2 to less than 2V almost instantly. 5W rating for R35 might seem like overkill (and it is), but the discharge current is high as is instantaneous power dissipation. D8 and C7 ensure that the supply to the reset circuit will remain available for a short period after the mains has disconnected.
In general, the circuitry will operate much faster than the relays. There is always a practical limit as to how fast power can be removed. While it might be tempting to use solid-state relays (SSLs), don't even think about it ! Since many power amplifiers use switchmode power supplies, some of these may be completely incompatible with 99% of SSLs, and can cause serious problems. If used, you run the risk of failure of the SSL, the amplifier's power supply or both. This point cannot be over-emphasised, but you won't find much (factual) information on the topic.
This part of the circuit (if needed) is described in Project 138 and will not be repeated here. The circuit should be built as described, including the off-line linear power supply. Be warned that every part of the circuit is at mains potential and is lethal. There is one major change though, and that's the relay wiring. Instead of switching the mains to an external outlet as described in the project, we want to switch off the mains applied to the sequencer.
The relay switched mains output becomes the input to the sequencer, but it only needs to switch the mains to the power supply section, so the relay doesn't have to carry the full load current.
Note that the 12V power supply for Project 138 absolutely, positively must not be used for this project as well! You will need two separate power supplies, one for P138 and one for this sequencer. Attempting to use the same supply will create much arcing and sparking, burnt wiring, blown circuit breakers and other wholesale destruction.
However, the P138 article does show you exactly how to use the intestines of a switchmode plug-pack power supply in a safe manner, so that part alone might be considered useful. Even if you decide not to include the additional protection (and to be perfectly honest, I wouldn't bother), there are still options to protect the attached system(s) from transient spikes.
Feel free to add a suitable number of MOVs (metal oxide varistors) such that the wiring complies with the wiring code used where you live. Make absolutely sure that the MOVs are rated for the normal mains voltage where you live - a MOV intended for 120V mains will explode mightily with 230V! It may be illegal to connect MOVs to protective earth, so they may only be connected between active and neutral conductors. Varistors are capable of fairly good spike suppression, but I suggest that you include a one-time thermal fuse in series with them. After a number of protection 'events', varistors may become partially conductive, and this can cause severe overheating and extensive damage when they finally explode.
MOVs should always be located on a separate board that allows easy replacement should a failure occur.
Many people (to some extent including myself) are not fond of using electrolytic caps in timing circuits. In this case, it is very unlikely to be a problem because of the relatively high charge and discharge current and because there is no need for great precision. The minimum is about 100µA, which might not sound like very much, but it's a great deal higher than an electrolytic cap's leakage current. Because the cap is charged using a constant current source, the charge current doesn't taper off to almost nothing as the cap nears full voltage so a small amount of leakage won't cause any issues.
However, after many years of use you can expect that the timing will change slightly as C2 ages and either becomes leaky or loses capacitance. Assuming that you don't wish to service the sequencer every 10 years or so, this is where we can use electronic techniques to make a variable high-value capacitance as shown in Figure 4. It's overkill, but the circuit is good fun to play with and you can learn something new in the process. Even if you don't build the sequencer, I hope that a few people will play with the capacitance multiplier .
Figure 4 - Timing Circuit Using Capacitance Multiplier
U3B (the other half of the loss-of-AC detector) and associated parts comprise a capacitance multiplier. The effective capacitance is determined by the actual capacitance (in this case a 1µF MKT polyester cap) and the values of resistance around the opamp. Capacitance is determined by (approximately) ...
C = C2 * ( VR1 + R5 ) / R6
C = 2µF * 10k / 100 = 200µF minimum
C = 2µF * 60k / 100 = 1,200µF maximum
In practice, the maximum capacitance calculated will not be achieved because of the finite gain of the opamp. Where we calculate 1,200µF, the real value is a little over 1,000µF. The effective capacitance has still been increased from the original 2µF (C2) by a factor of more than 5 hundred, simply by using some circuit trickery. However, because we are using an opamp with bipolar input transistors rather than FETs, the opamp's amplified input current has to be considered. The input current from the opamp (the input transistors are PNP and current flows out of the input pin) skews the timing slightly, and it will take a bit longer to discharge than to charge. This is why R4 is reduced to 82k, and this makes the charge and discharge cycles almost exactly equal. Ideally one would apply opamp bias current compensation, but that just makes the circuit needlessly complicated.
Although the overall circuit is a little more complex than the Figure 1 version, there's not much in it really. You only need a single-ganged 50k linear pot for VR1, and the polyester caps have leakage currents that are several orders of magnitude below the opamp input current, and they should last forever. You no longer need the MOSFET in the power supply - a small signal transistor will be perfectly ok. You need to change R34 (100 ohms) to 2.2k, and increase R35 from 2.7 ohms to 10 ohms 1/2W.
With the values shown above, the minimum period between outputs is about 3 seconds, and the maximum is 16 seconds. You can vary the timing range by changing R3 and R4, or simply use a single 1µF cap to get delay times from 1.5 to 8 seconds. Higher values increase the time and vice versa. Note that the input current from the opamp (the input transistors are PNP and current flows out of the input pin) will tend to skew the timing slightly, and it will take a bit longer to discharge than to charge. This is why R4 is reduced to 82k, and this makes the charge and discharge cycles almost exactly equal.
The basics of the circuit shown have been tested, and it works exactly as described. Sometimes, running a simulation doesn't prove that a circuit will work properly, and the only way to be certain is to build it. The bench-test results and simulations are so close that I have to accept that the simulation is accurate. With the opamps I used in my bench test, the bias offset current is about 0.11uA (in case you were wondering ).
Meanwhile, I also bench tested a basic circuit using electrolytic caps (not even low leakage types), and that also works perfectly. The charge and discharge currents are normally around 100uA (0.1mA) and this is far greater than the leakage current in a standard electrolytic capacitor. Note that you cannot use a tantalum cap for C2, because it will probably be destroyed the first time the loss-of-AC circuit operates. Tantalum caps do not tolerate high charge and discharge currents¹, and usually fail short-circuit if subjected to any 'abuse' that other caps will handle for decades.
1 - Reliability of Tantalum Capacitors (NASA)
If you are building this project, I fully expect that you are not a beginner, have excellent skills with mains wiring, and fully understand the risk of electric shock while working on a project of this type. If any of these criteria don't apply to you, then you must not attempt construction. It is all too easy to make a potentially fatal mistake if you do not fully appreciate the danger of mains electricity and/or do not understand safe wiring practices.
All diodes throughout the circuit (except D1) are 1N4004 or similar. D1 is a 16V zener (15V is also fine) and is simply to protect the remote input from possible nastiness. It can't protect against everything of course, but will be fine for most 'accidents'. LEDs can be any colour you prefer. Standard LEDs will be quite alright - they all set up for a forward current of ~5mA, so are bright enough to be seen, but not obtrusive. High brightness types will (of course) be extremely bright with 5mA, and the series resistors can be increased from 2.2k to perhaps 10k - I leave that to the constructor.
The sequencer itself is non-critical, and the circuit can be built on Veroboard or similar. Make sure that C2 (Figure 1 circuit) is nowhere near anything that gets warm or hot as that will increase leakage current. It is highly doubtful that there will ever be enough interest to warrant a PCB, so don't expect one to appear in the pricelist. While it might look complex, I expect that any reasonably competent electronics enthusiast will be able to wire up the board easily enough. The same applies to the power supply - except the Y-Class capacitors! These can be mounted onto a piece of blank fibreglass or similar, and they must be wired in such a way that it is impossible for the capacitor leads to be shorted together.
Use the pinouts shown here for the dual opamps. Note that the second half of U3 is not used unless you use the capacitance multiplier shown in Figure 4, so join pins 6 and 7, and connect pin 5 to earth/ground. Do not substitute opamps - the LM358 was selected because its output goes to zero volts with a single supply, and the inputs can be referenced to the negative supply (earth). Most other opamps do not, and the associated transistors will be permanently turned on.
These opamps are very common, and extremely cheap - you shouldn't have to pay more than 50c each for them (I've seen them for less than 30c - quite remarkable for a dual opamp). At some resellers they will cost more of course, but they are still inexpensive.
The power wiring must all be completed using wire with mains rated insulation, and proper creepage and clearance distances must be observed between mains and control wiring. The earth connection must be continued through from input (either a socket or fixed mains lead), to the chassis and then to all outlets. All mains wiring should be neatly bundled and use cable ties to keep the wiring neat and well away from low voltage circuitry.
The most difficult and expensive part of this project is the case, HRC fuse and/or circuit breaker (if used) and mains connectors. A standard 1-unit rack case should have more than enough room to mount everything. This part I must leave to the constructor, as the regulations in different countries will dictate how the unit must be constructed.
I suggest that the power switch (shown in Figure 3) should be fairly inconspicuous, and mounted at the rear of the case. The end user needs to see that power is available ('Power On' LED) and be able to turn the sequencer on and off ('ON' switch in Figure 1). If the main power switch is accessible, you know that it will be operated instead of the power sequence switch much of the time. If fitted, the mains reversal override switch should also be at the back of the case, and preferably with a protective cover. This switch is provided as a contingency only, and the end user should never have the opportunity to play with it.
If you don't need the loss-of-AC detector, the only items in the power supply that you should retain are the mains switch, the SMPS itself, the 'Power On' LED and its resistor. Everything else can be left out. The 'SD' terminal should be wired directly to the +12V supply in this case. I also suggest that you add a resistor (1k or thereabouts) from +12V to earth to ensure that C2 fully discharges when mains power is removed.
Connect the sequencer to a suitable 12V power supply with the negative side connected to earth/ground. A plug-pack type supply is ideal, and as noted you can use a switchmode type. Leave the switch in the 'off' position, and connect the 'SD' point to earth. All opamp outputs should be at (or near) zero volts, and the voltages at the negative signal inputs should be as shown in Figure 1.
Switch on and nothing should change, and in particular Q4 should not get even slightly warm. Remove the earth connection from 'SD', and each of the opamp outputs should go to about 10V or more in sequence. Switch off, and all outputs should return to zero volts, again in sequence. The last output to go high will be the first to go low again.
To test the relay switching, simply apply 12V to each of the inputs. The respective relay should click, and its associated LED illuminate. When both are working to your satisfaction, connect them together and re-test to ensure that the sequence is right. If you mix up the leads from the opamps to the relays, the relays will switch out-of-sequence.
To test the loss-of-AC detector, you need to connect the mains. When the power switch is turned on, the Power LED should light, and the output of U3 should be at or near zero volts. Short the test point ('TP') to earth, and the output of U3 should go high immediately. The power supply will easily hold enough voltage (thanks to D8 and C7) to ensure that C2 (Figure 1) is fully discharged before the voltage fails completely.
|Copyright Notice.This article, including but not limited to all text and diagrams, is the intellectual property of Rod Elliott, and is Copyright © 2013. Reproduction or re-publication by any means whatsoever, whether electronic, mechanical or electro-mechanical, is strictly prohibited under International Copyright laws. The author (Rod Elliott) grants the reader the right to use this information for personal use only, and further allows that one (1) copy may be made for reference while constructing the project. Commercial use is prohibited without express written authorisation from Rod Elliott.|