50hz detector needs improvement

Thread Starter

erifyab

Joined Apr 30, 2009
5
Hi

I'm trying to detect 240v 50hz without direct contact. The circuit needs to be quite reliable, and very selective (ie only pick up one cable from a bundle).

I have developed the following circuit which simulates OK in ltspice, but I would really welcome comments / suggestions before I order the bits and build it. I don't have access to a scope so fault finding once I've built it is tricky!

The idea is to wrap a piece of thin single strand wire (like a cat5 core) about 5 or 6 times around the cable I'm trying to monitor, which I figure should induce some sort of 50hz signal in it, then send that through a 50hz band-pass filter (IC1) with quite high gain, followed by a comparator (IC2) with a preset to set the reference voltage to give some sensitivity adjustment. Then a quite crude diode/cap to flatten it out and a buffer (IC3) to drive the output. The zener probably isn't necessary but it gives a nice steady flat output regardless of current, and allows me to drive the next stage which will be an AND gate to combine several of these circuits.

I'm only using the 1013 op-amp because I know where I can buy them and ltspice has a model for it!

improvements
==========
I'd love to loose the 3rd op-amp as the 1013 is a dual package and that would mean only one IC. Is this even the best way to do it? If anyone has a different idea I'd love to hear it.

I would really appreciate any help/suggestions/ideas/insults!!

Thanks.
 

Attachments

Last edited:

PRS

Joined Aug 24, 2008
989
If your similation works the design should be do-able. May I ask why you want to make this instrument?
 

Thread Starter

erifyab

Joined Apr 30, 2009
5
I have to switch on some emergency lighting in a block of flats when the mains supply to the normal lighting fails. Because the existing system is so old and doesn't meet regs I cannot connect directly into the existing circuits without effectively 'modifying' the circuits and thus taking on responsibility for them.

The emergency lighting is fed from a new separate mains supply, and normally in this situation I would use 'maintained' emergency lights which are always on, so it wouldn't matter which supply was on or off as there would always be light. In this case for long boring reasons that wasn't possible, and wasn't done, so I find myself at this point!

There are 6 feeds leaving the fuse board to various lighting in various areas, and I need to know if any 1 of those supplies fails, in order to switch the emergency lighting (which is only one circuit covering all areas).

Admittedly this is supposed to be 'work' and I should by now have given up and conceded defeat - but it has become more of an obsession and a determination to solve the problem in an elegant way, and as often happens is no longer part of the 'day job'!!

The one bit of the design I couldn't simulate was the antenna / sensor bit. I have assumed 0.1v 50hz in the input which seems reasonable? I was trying to limit this to about 0.6v with the 2 diode shunts, but thinking about it that would only work for dc I think?!

Thanks for the interest

Matt
 

SgtWookie

Joined Jul 17, 2007
22,230
Matt,
You posted the schematic, but not the .asc file (the LTSpice source file).

Many of us have LTSpice installed, and it would help to speed up getting feedback if you'd provide your .asc source.

One observation is that you're running U2 with no feedback (open loop); this will cause the opamp's output to be in constant saturation, which can lead to heating and low response times. LM1013's have a minimum open-loop gain of 8,000,000.
 

PRS

Joined Aug 24, 2008
989
Ok, erifyab, I got it now. Your golden fleece is a worthy pursuit, I think. And Sgt Wookie will probably give you the solution you need. Your idea of wrapping a few turns of wire around each hot wire is good, but we are talking here of current detection, not voltage detection. This is because the current in the hot wire will induce a current in the small coil. Do you really want to turn on the emergency lighting just because no current is present? What about the case where no one is using electricity? It would be much simpler if you tapped the sources physically.
 

Thread Starter

erifyab

Joined Apr 30, 2009
5
OK thanks, and here's the asc file.

PRS, good point. Actually very good point. I do need to pick up voltage, not current. Some of the light circuits are on a time clock and are switched off at night, so I wouldn't want mine switching on just because the main lights were intentionally switched off - I could just use a light sensor to do that. How do I pick up voltage, not current? And yes trust me I know how easy it would be to tap the sources directly - that's what I normally do. You'll just have to trust me that on this occasion I can't!:(

Matt
 

Attachments

Thread Starter

erifyab

Joined Apr 30, 2009
5
If you do not "terminate" the sensor coil, i.e. leave the other end of the coil of CAT5 wire unconnected, then you will only get voltage signals, via capacitive coupling.
That's what i was doing anyway, so that's lucky! Thanks. Has anyone got any idea what sort of signal level I could expect in my 'antenna'?
 

PRS

Joined Aug 24, 2008
989
erifyab, I think eblc has a good point and your analogy to an antenna is good because that is really what it is. I should think you could use a very high resistance, say a FET, to detect the presence of a signal.

As I understand it, you're not worried about the value of the voltage, you already know it. So amplify the signal at the antenae after input through a follower to the necessary voltage for a detector to feed a comparator which switches the emergency supply???
 

jvjtech

Joined Jan 26, 2008
23
I would suggest that the current detection circuit use a windowed type of current transformer. One wire, e.g. the line side of the mains supply, of the circuit being monitored can be passed through the window of the CT. (Power off - of course.) The CT can then be tied into an op amp and comparator circuit, etc. I though I would add that one is not limited to the device models available with LTspice. Models available from TI, NatSemi, etc. can be used in LTspice. Regards. JJ
 

Thread Starter

erifyab

Joined Apr 30, 2009
5
Thanks jvj for your suggestions, I'm trying to pick up voltage, not current though. And thanks for the tip about spice models too.

PRS, can you explain some more about how to use a FET on the end of my 'antenna'? My limited understanding of them is that although they have _very_ high input impedance you need 3 ish volts at the gate to turn them on. I realise I can bias the input, but would it still pick up the potentially tiny signal I could be getting? Or are you saying the picked up voltage would be much higher than I'm imagining?

The idea of the op-amp as an input was to use it as a 50hz band pass filter as well as an amp - to try and reduce false positives from any other transmitting source. I'm assuming cordless phones, wireless internet etc is prob too high a frequency to bother me, and most other things are too low power, but I know how many false positives you get from those 'volt-stick' indicator pens which are supposed to detect mains voltage but also light up for a plethora of other things.

SgtWookie, thanks for your observation about the second opamp being open-loop. I am trying to use it as a comparator, which I thought meant driving it to saturation - but I definitely don't need a gain of 8million! If I set up feedback to give a gain of say 5 or 6 then surely it would still just drive to saturation? I need to do my homework again, methinks.

Thank you all for your help, please keep it coming!

Matt.
 
Top