Archived Forum

If you have no experience of PCB layout and design for mains voltages, then you've got a lot of learning to do if you go with Plan A. Plan B gets around most if not all of the safety issues immediately.

On the minus side for Plan B, you're using the same transformer for voltage sampling and for power. If you use the typical bridge rectifier configuration, you need another transformer or an op.amp to isolate the transformer voltage that you're going to use for measurement from the one that goes to the bridge rectifier and regulators for the low voltage dc power supply rails. Even then, the current pulses as the smoothing capacitor charges will dent the waveform and introduce errors. If you want to avoid the extra transformer or op.amp, you can only use a half-wave rectifier, as the emonTx does.

Plan A is essentially the configuration that Robin Emley uses in his Mk2 Router (http://www.mk2pvrouter.co.uk/). Having two separate secondary windings (one for the voltage sample, one for power) largely overcomes the waveform distortion problems. Point 3 mentions isolation - apart from observing the necessary tracking distances, which is an absolute must, I'd make sure that there is earthed copper between the mains connections (both line & neutral) and the low voltages so that any leakage or flashover is to earth. Have a read of "GIVMon" to get some idea of the proper way to do it (that project was to obtain a 'better' voltage sample, but the safety precautions are relevant to you).

The main source of interference that we've encountered is from the digital circuits in the processor; so good grounding, shielding with earthed copper and filtering in the supply (there's plenty of advice on the Atmel website) are most important.

Those are the main points that come to mind.

Hi Robert - thanks for the input :-)

Happy enough with low voltage DC PCB design, but think option B sounds the best plan.  I had a look at the mk2 router, and can see the isolation gap you mention for leakage.

I did some circuit simulations last night and a transformer with a center tap and two diodes seems a good option.  I don't have a circuit schematic to hand, but will post later.

Is the 9V AC-AC in the shop (and that I have already) suitable for a supply?

James

Hi James, I'm always happy to supply parts for DIY constructors to experiment with, at their own risk. 

One of my original PCBs (the orange one) and a matching mains transformer (either 2 x 6V or 2 x 9V) including postage would be £10.  There's an email address on my Shop page for all such enquiries.

The original PCB layout has a tracking error in the vicinity of the display but it doesn't affect the operation of the processor, so will be fine for basic experimentation.  The error can be easily fixed if need be.  My latest (Rev 2.1a) PCB is listed as £7, but I'm always open to offers.

Robin Emley

If you use the shop ac adapter, you've got all the problems with separating the voltage monitor and the power supply that I mentioned above. Otherwise, there's enough power for a basic 'emonTx look-alike' but no more without distorting the voltage waveform that you're using for measurement.

(Hint: you need to add the transformer impedance / secondary winding resistance to your simulation, then check how the voltage is disturbed by the current pulse.)

Thanks for the posts both.

Would a multiple winding transforms solve the voltage/current issues with a single transformer?  One for power supply and one for measurement.  This is where my lack of knowledge starts to show....

i.e. http://uk.rs-online.com/web/p/pcb-transformers/1739737/

Robert -> I'll check the simulation, but I suspect the software isn't that clever.  Its LiveWire (http://www.new-wave-concepts.com/ed/livewire.html) mainly for home use.

James

As I wrote: "Having two separate secondary windings (one for the voltage sample, one for power) largely overcomes the waveform distortion problems." Largely, but not completely.

For simulations, I (and Glyn) use Linear Technology Corporation's LTspice IV (free, from their website).

I have occasionally used the LiveWire simulator from New Wave Circuits, but doubt whether it would be able to cope with this kind of application.  I also use their PCB Wizard 3 tool for PCB layouts.  Although the 'pro' version features double-sided auto-layout, I have always preferred to do my layouts manually.  Using this tool, creating Gerber files for PCB manufacture is straightforward enough.  These files can then be checked using the free GCpreview tool from graphicode.com.

Hi

So lots to ponder, and work out - thanks for the comments.

Robin - I prefer Eagle atm as I then use oshpark.com for cheap boards.  I've used expressPCB before which is very good and very quick, but also very expensive...

James