Mains Gate: Programmable Relay & Energy Monitor [120709-I]

HELP! I'd appreciate any assistance with this project particularly from someone with power experience.Proposal

HELP! I'd appreciate any assistance with this project particularly from someone with power experience.

Proposal

Rather than some US political scandal, this is an idea for "mashing up" five of the circuits from the July & August 2012 double issue to produce a really flexible "mains gatekeeper" circuit combining a multitude of control, measurement, protection and logging functions at minimal cost.

The sources are:

"A Zero Current Switch" by Matthias Haselberger.
"MOSFET Circuit Breaker" by Georges Treels.
"Equipment 'ON' Counter for 68 Years Max" by Vladimir Mitrovic.
"16 Ways to Switch your AC Power On" also by Vladimir.
"Soft AC Line Start" by Jurgen Kruger.

The new circuit starts with the mains switching element used in (1), an IGBT or high voltage MOSFET in the DC arm of a diode bridge. However Matthias's discrete control circuit is replaced by a microcontroller. Georges' idea (2), measuring current by using the voltage drop across the on resistance of the MOSFET, is adapted by using an ADC channel in the microcontroller. The Microchip PIC range has some controllers with in-built op amps which would minimise the BOM and cost of this circuit. Voltage will be measured on another ADC channel using a simple potential divider. We are now in a position to duplicate and improve all the functionality of these first two source circuits using firmware in the microcontroller. For example, we can measure power factor and apply any compensation automatically, so eliminating the need for the manual link and capacitor in Matthias's circuit. We can easily implement the ideal AC switch: zero volt switch on, zero current switch-off.

Next I will pinch two ideas from Vladimir (3). A separate programming circuit enables settings to be written to EEPROM in the microcontroller as well as logged values to be read. I would probably use a simple BOB style USB connection and a PC UI for this function. Following Vladimir again, we can use the mains voltage measurement already described as a timebase for logging the on time.

It is now trivial to embrace and extend Vladimir's other circuit (4). By using the mains timebase, we can improve the timing accuracy over the internal RC oscillator used in the original. We can eliminate the hardware programming switches and the corresponding microcontroller GPIO pins using the external EEPROM programmer. This will enable us to offer a much wider range of programming options including ones which respond to an external optically isolated input to provide a true solid state relay. I would also add an output which could be used to gang a number of the circuits for power saver arrangements (e.g. turn your peripherals off when the power drawn by your PC drops, indicating it has gone into standby).

By adding an optional power resistor switched by a second MOSFET or IGBT in parallel to the first, we can have a soft-start as implemented by Jurgen (5), with the extra benefit of programmability.

Finally we can push the firmware into completely new territory without needing any further hardware. We can measure and log mains voltage, power draw, power factor and accumulated electricity consumption. We can implement protection trips on any of these parameters. We could even implement timer functionality by building a firmware real-time clock on the mains timebase. A light level switch or thermostat would also be easy to add as an option using a few extra components.

Design Process

Schematic 1 is the first draft of a circuit for the mains part of the circuit. I have settled on the PIC16HV785 20-Pin chip as this has many of the analog components needed built in with the digital microcontroller. I have added a thermistor thermally coupled to the substrate of the main switching MOSFET as this should make it possible to thermally compensate the current measurement. I decided on a fully galvanic isolated design by using a photodiode input and IR LED output. This will work either as a direct simple switch or for a low data rate serial link (which will have to be bit-banged in the microcontroller as it lacks serial peripherals).

Model 1 illustrates the mechanical design of the above circuit. It is designed to fit in the bottom part of the DIN rail enclosure mounted on the mains in and out screw terminals as shown. The main components will be on the underside with the available envelope shown as a white box. The IR input and output will be mounted at centre on top of the board with a shield between to prevent optical crosstalk.

Model 2 shows how an interface circuit is mounted in the top part of the enclosure with matching optical transducers on the bottom of the unit. Again the available envelope for component mounting is shown as a white box. A number of interchangeable interface circuits may be designed. The simplest would be a connection from one of the pairs of screw terminals to the IR transmitter for a relay input. More complex would be a USB or Elektorbus serial interface with programming and reading capabilities.

Model 3 shows a concept for ganging several mains gates together. The interface circuit is split into two “bookends”. One bookend has an IR transmitter optically coupled to the input of the first mains gate, which has its IR transmitter bent through ninety degrees to fire sideways. Intermediate mains gates have both receiver and transmitter bent over to couple with the adjacent unit. The other bookend has an IR receiver coupled with the transmitter of the last mains gate which has its receiver bent over to couple with its neighbour. The firmware will allow messages to be routed from receiver to transmitter so that they get passed on ‘bucket brigade’ style and seen by all units.

Update 28-Nov 2012

Schematic 2 shows several improvements in the light of bench testing. I found that MOSFETS rated for mains voltages and currents (I am testing using a STP14NK50Z) need a higher gate voltage than the PIC microcontroller can switch, so I added a MAX4427 dual buffer. I was uncomfortable with the inefficiency of the control-side power supply concept of Schematic 1 and experimented with the capacitor-based solution shown in this schematic. Sadly, while the power supply and the switching cell both work well in isolation, connect them together and the sour smell of burning plastic ensues! Anyone have any clever ideas about this? Will I just have to bow to convention and use an isolating transformer?

Update 01-Sept 2013

I am reluctantly retiring this as a dead end because we have been unable to come up with a control power scheme which meets the original vision of an economical self-contained unit. Without this, a more conventional multi-channel design will be better. If anyone wishes to revive this under their own name, or reuse some of the ideas in their own design (which after all is exactly what I was attempting to do), please do so with my blessing!

Project Elements

Model 1 (PNG, 36.79 KB)

Model 2 (PNG, 18.36 KB)

Model 3 (PNG, 29.61 KB)

Schematic 1 (PNG, 30.92 KB)

Schematic 2 (PNG, 31.55 KB)

Discussion (9 comments)

JohnHusband 11 years ago

In power supply design it is customary to use an optoisolator to isolate control electronics from power switches, there are many types to choose from, so try to match parameters.

JohnHind 11 years ago

The original schematic does have an opto isolator on the control lines to the microcontroller, but the microcontroller is on the mains side. The concept was that the controller is regarded as expendable in the case of a fault and the whole circuit would be double insulated. As you can see from the other contributions, our problem is deriving a low-cost, reasonably efficient power converter for the microcontroller.

1 Comment(s)

TonGiesberts 11 years ago

The other solution for the power supply I mentioned is a resistive divider using a single extra diode instead of a bridge. See first accompanying schematic. This is the worst solution ever. But for testing purposes it will do and it may give some insight. The RMS value of a half wave rectifies sine wave is 1.57 times greater than the average value. So the power loss will be high, more than 11 W (hence the two paralleled 10 W resistors R1 and R2). The average value is about 43 mA, which is the DC current that can be drawn from C2. D2 is only a protection, should the current consumption be considerably less. But it resembles your first schematic. And this worked? In that case the resistor can be replaced by a buck converter.

If this indeed works without any problems (except for the large power loss) than maybe you can try an alternative capacitive divider. Downside is a twice as large capacitor as the one with a bridge will be needed (see second accompanying schematic). Connecting the anode of D2 to the cathode of D7 is maybe even better? Half the current through the capacitor C1will be used for the power supply so it will have to be in the order of 1.5 µF or simply put two of 680 nF in parallel (whatever takes up less space).

120709-mains-gate-advise2-schematic.jpg (164kb)
120709-mains-gate-advise3-schematic.jpg (145kb)

Arne.Hinz 11 years ago

Try the IR2183. This is a half bridge driver with an internal level shifter for the high side; thus it can withstand up to 600V. If you connect the MOSFET to the high side like in the example in the datasheet, the circuit should work.

PeterPugwash 11 years ago

I'm not sure if this will help, but there was a recent news item about a chip from Linear Technology that drives a bridge of 4 Power FETs to form a super efficient bridge rectifier, here are the links :- http://www.elektor.com/news/ideal-diode-bridge-controller-minimizes-rectifier.2494929.lynkx http://www.linear.com/product/LT4320 The downside is that the maximum input voltage for this chip is 80 V, but it does raise some interesting ideas about using the body diode to self power the output. I hope this might stimulate some new ideas, unfortunately I can't provide any suggestions myself.

JohnHind 11 years ago

I wanted to use a MOSFET rather than a triac for several reasons notably the ability to use its on resistance as a shunt for current measurement and the ability to switch off at zero current (as well as on at zero voltage). Perhaps you could post a schematic showing live side switching as I do not see how this would address the auxiliary power problem that is the blockage on this project. I would be happy to share authorship on this project to anyone who can come up with a workable power circuit (preferably bench tested).

JohnHusband 11 years ago

I did mention in another comment that it would be better to switch the line rather than the neutral for safety reasons, you have added another reason for this to be preferable to switching the neutral. Are you aware of the Triac as a mains switching device that is well tried and tested in various configurations?

JohnHind 11 years ago

Thanks for your efforts Ton, but I do not think these are workable. The first schematic does work, but it is really more of a space heater than an efficient switch! Can you elaborate on the buck converter idea? I do not see how this would work since it would itself require an auxiliary power supply for its control circuit. Or could it maybe be boot-strapped with a resistive divider and then self powered? Either way sounds expensive in components. The second schematic has the same problem as my original double-bridge arrangement: as you point out D2 could be dropped and the function of D7 doubled up. But if you draw the circuit this way the problem immediately becomes apparent: the auxiliary supply is effectively just the DC arm of the switching bridge, so it is at either rectified mains potential when the MOSFET is off and zero potential (give-or-take on resistance) when the MOSFET is on.

2 Attachment(s) 5 Comment(s)

TonGiesberts 11 years ago

In the way the two bridges are interconnected in the second schematic puts two diodes from both bridges directly in parallel. Simply leave one diode out in the smaller rectifier bridge as in accompanying schematic is shown. From the information so far it is not clear when the smell of burning plastic happens. Does it already occur without a load? What exactly gets to hot? Did the processor and the driver survive this? More complete schematics with component values would be appreciated, also from model 2.

If this is not helpful there is another way of supplying power to the processor, by way of a resistive divider. But this would mean a power loss of probably more than 6 W. Instead of the resistor a small buck converter could then be a solution.

120709-mains-gate-advise1-schematic.jpg (169kb)

JohnHusband 11 years ago

From the point of view of safety switching the neutral wire is seen as unsafe, because if the circuit fails the live mains is still available. I don't know if this difference affects the result (exchanging Line for Neutral) but I think you will understand the point of view (better an o/c live than an o/c neutral). I admit it is traditional.

JohnHind 11 years ago

Thanks for input Ton. From memory I believe I did try this configuration with no change. The two diodes in parallel are the same way round, so at most I'd expect one of them to burn out if it was underrated for the mains load. I never connected the ICs - basically my test circuit was as you've shown plus a dummy load resistor and zener diode (standing in for the one in the PIC) parallel with C2 and the gate of T1 grounded. When a load is attached to the output (I used a low wattage light bulb) it switches on and the components in the DC arm of the power supply bridge overheat. So the power supply bridge is shorting out T1. Values: C1 - 0.68uF (305v AC X2 Rated); R1 680R; R2 500K; C2 100uF. So Reactance of C1 at 50Hz is 4K7. Removing the link from D4/D5 to ground then the switch and the power supply both work OK, but the power cannot be used to drive the gate of T1 because they lack a common ground reference (and of course the current measurement would not work for the same reason). I think a buck converter would have the same problem - it would still need an isolation transformer.

1 Attachment(s) 2 Comment(s)

Raymond_Vermeulen 11 years ago

First off, you might try to use a small isolation transformer to isolate your circuit. It isn't exactly safe to touch as it is now.

Okay, to drive your mosfets you could use the SI8220 from Silicon Labs . But you would need to to power it's isolated side.

To sense current you could use the SI85xx series.

Further, a lot of the functionality you discribed is also found in completely integrated AFEs. Like the Maxim 78M6610+PSU and the Analog Devices ADE7953 for instance. Have a look.

JohnHind 12 years ago

Thanks Raymond. My original concept was to produce something similar in form factor and cost to a mains SSR, but with added features - the microcontroller would be on the mains side of the isolation barrier with the whole thing double-insulated or potted. So only a serial control channel to the outside world needs to have an isolation barrier. Problem is I cannot come up with a workable compact, low cost mains-side power topology for this. The problem is the need to have a common ground for the switching element and for the current sensing element - easy for DC, less so for AC! If I have to go for a proper isolated control circuit with its own PSU, the thing becomes a much more conventional design as you point out. In this case it would probably better to make a multi-outlet design so there is one control unit with multiple mains-side switch/sensing blocks.

1 Comment(s)

PeterWashington 11 years ago

Had you considered the PSoC 3 or PSoC 5 device families from Cycpress, I confess I haven't used them myself so I'm not certain that they are what you want, but they programmeable Digital AND Analogue function blocks, (similar concept to a FPGA), built in and there available from Farnell starting at around 10 GBP for the PSoc 5LP based an ARM Cortex M3 ?

See http://www.cypress.com/?id=4562

JohnHind 12 years ago

I've been evaluating PSoC quite deeply for other projects. I am seriously impressed particularly with the ARM based PSoC 5 and the fact that the development system is free, which is very unusual in the ARM market. I especially like that it has programmable digital logic as well as analogue and full pin-mapping. I do not know why all microcontrollers do not have full pin-mapping as it will simplify board layout and flexibility of prototyping boards greatly. However, having said all that, I think it is overkill for this project as it would not reduce the external components over the Microchip controller which has lower cost and power.

1 Comment(s)

Swieter 11 years ago

I like this project. I'm interested in seeing it happen.

As for the power supply, many moons ago I researched some similar ideas on transformerless designs. I didn't take my work much further than Googling around. That is I didn't get to create anything. Microchip has a couple Application Notes, and I am sure there are more that could be of assistance in the design.

AN236

AN954

If creating the power circuit isn't something you want to tackle, then maybe a module would work like MPL-03SE series from MicroPower Direct. Again, I haven't used these, but have a scribble in my notebook to try them out.

JohnHind 12 years ago

Thanks for this. The app notes both provide useful information, though essentially the same transformerless concept I was already attempting. This concept does not work because of the incompatable grounding requirements between the control and switching circuits. Unless we can find a way round this, the economics of the design change fundamentally because too much of the circuit has to be duplicated for multi-channel units. I could not find an online shop selling the MPL-03SE or find a price for it. From modules I'd guess around $20. This again pushes us to a multi-channel unit to make it more economic.

1 Comment(s)

Hooligan0 11 years ago

Can you give us the values of R1,R2,C1,C2 you have used ? And what is (approximately) the current needed by IC1,IC2 and transistors control ? The smell of 'burning plastic' can be joule loss into R1 when C1 is empty.

JohnHind 12 years ago

Working at 230v AC, C1 = 0.68uF (305v X2 rated); R1 = 680R; R2 = 560K; C2 = 100uF. I calculate this should give 40mA through IC2 and the bandgap regulator in IC1. I estimate a maximum of 10mA will be required - 40mA is the rating of the bandgap regulator. I believe the problem with this circuit is the need to connect the control circuit ground to the -DC corner of the switching bridge. Without this link, both switching circuit and power supply circuit work fine (although of course the mosfet cannot be switched and the sensing circuits will not work). Once the link is inserted, the load comes on (with the mosfet still off) and the components in the DC arm of the power supply bridge burn - so the load current is being shorted through this path. At the moment I cannot see a workable circuit topology that does not require an isolating transformer.

1 Comment(s)

JohnHind 11 years ago

I would like to build this project in a DIN rail enclosure so it will fit in a consumer unit box for safe mains wiring. The one pictured in the project image is from CamdenBoss and available from CPC-Farnel. It is a bit expensive, hopefully I can fine a lower cost / better alternative.

0 Comment(s)

JohnHind 11 years ago

I'm looking at the PIC16HV785. This has a built-in zener bandgap regulator and two op-amps as well as a 12-channel 10-bit ADC and 256 bytes of high endurance EEPROM. It's a 20-pin package, a bit bigger than ideal for this application, and you need a high voltage programmer and a special 28-pin version for debugging.

An alternative is the PIC16LF1782. This is even bigger at 28-pins and lacks the internal regulator. But it has internal debugging and self-programming capability. It also has serial peripherals which would facilitate implementation of the Elektor bus, for example (with the HV785 the in-circuit programming link would be the only channel for configuration / reading - or bit-banging).

I am wondering if the mains-derived power could be made safe enough to avoid the need for isolation on the configuration link. I'm thinking about a fuse with a secondary shunt zener at slightly higher voltage than the primary shunt. If the primary shunt fails, the secondary conducts and blows the fuse. External IO lines could then be protected with diodes to VCC and GND.

0 Comment(s)