light to uncoupler

I would add the light in parallel to the output of the AIU to the uncoupler.  The light should be of sufficient rating to handle the voltage you are applying.

George

Yes the  Uncoupler is the track section  of Fastrack.  And The AIU unit is a accessory interface unit. I have a little street light that I want to use but when I hook that up either with it in line with the AIU or power supply the uncoupler does not work.  I think it's because of two small of wire  for the voltage. 

Add the light in parallel with a resistor in series with the light.  Put the light with resistor in parallel, not in-line with the output of the AIU.

I would start with a 300 ohm resistor.  Is this an LED or incandescent light?

Also, if you have another light, like a lighted lockon, I would troubleshoot using that first, before trying your street light.

It is a street light with 3 incandescent lights in it from MTH #30-11034.  I will have to go to Radio Shack to get what i need to get this to work. I am in the dark when it comes to understanding electronics.

Before you go to Radio Shack, is this how you tried it?  I think I was wrong in saying put the light in parallel with the AIU output.  I would put it in parallel with the uncoupling track like the picture below, after the AIU.  (IN and 1 are for the AIU) (A and U are for the power supply).

George

When I first read the problem, I thought the task was to detect that the electromagnet was "ON".  Detecting voltage applied to the UCS terminals is undoubtedly good enough.  But if you want to detect that current is actually flowing thru the electromagnet a Current Transformer (about $1-2 on eBay, free shipping) plus a couple 5 cent LEDs is an option.

Here's a 1000:1 Current Transformer plus a couple back-to-back LEDs on the CT output.  A wire with the uncoupler current runs thru the donut hole of the CT.  The few Amps of uncoupler current when active is transformed to a few millamps (1000-to-1 ratio) of LED current on the CT output...plenty to light up LEDs.  The LEDs are placed back-to-back so that each lights up on the opposite half of the AC cycle and also serves to protect the other LED from excessive reverse voltage.

And here it is in action.

There are other ways to detect AC current like putting a series diode or resistor shunt to generate a voltage when current flows.  No matter the method, detecting current (rather than voltage) provides arguably more confidence that the electromagnet or load is actually active. 

What's the eBay item for that current loop Stan?

The one I happened grab out of the proverbial parts bin is eBay item number 381276598888

I don't know if ironic is the right word but it even comes with a resistor and circuit board that converts the CT output current to voltage so an Arduino can read it!  But since LEDs "want" to be driven by current (not voltage), the resistor is not needed.

I don't have the Arduino specs at my fingertips, but I am going to guess that .005A * 80 ohms = .4 volts is not going to lead to great results in the A/D section.....

Probably OK to sense on or off, though.

I'm not using it with an Arduino but it seems that if you're using a 3V Arduino, and you bias the CT output to A/D mid-voltage to sample both polarities, and you account for peak current of at least 1.4x RMS (more like 3x RMS with modern capacitive loads), etc. etc....80 ohms may not be so far off the mark?

Dunno - Only fooled around with an Arduino Uno, 0 - 5V was 1024 counts. Was not impressed, but I didn't put much effort into it. I will pick up a couple of those small CTs to fool with.

BTW, I have a few CTs if anybody wants them. They are 500:5 and 150:5. Great to fool with, you will never burn them out on a model railroad layout!

I looked around previously and found this one, so I ordered one to see how it works, I'd be using it with the PIC processor.  You can tweak the resistor to match the scale you want to measure.

gunrunnerjohn posted:

I looked around previously and found this one, so I ordered one to see how it works, I'd be using it with the PIC processor.  You can tweak the resistor to match the scale you want to measure.

 

I will need to test some of these small CTs.

Thing to keep in mind, if using a CT for measurement purposes, is that they are most accurate operating into a short-circuit., or looking at it differently, when there is practically zero flux in the core. If you increase the burden resistor, in an effort to get the output voltage higher, you will soon get to a point where the coil saturates, causing a marked drop-off in accuracy, as well as some other interesting effects. I am quite sure that Stan's application of LEDs relies on these other "interesting effects," but, it does light the LED, so who cares!

When feeding a resistive load, the waveform of the secondary voltage should match that of the primary current. If they differ, you are outside the rating limits of the CT, and accuracy will suffer.

Serious CTs have a specified "burden" rating, usually in VA, sometimes in ohms, below which specified accuracy is assured. These values are the great unknown in these tiny ebay units.

Good points, I'm more interested in having it trip at a given set point of current.  That could probably be done by an op-amp precision rectifier circuit feeding an A/D channel.  The precision rectifier circuit is to get rid of the diode drop.  Still in the thinking stages.

Never new about a Current Transformer...with this forum i swear you learn something new or more every day.

gunrunnerjohn posted:

...Still in the thinking stages. 

Yes, think twice, solder once.  If you're not trying to squeeze out that last bit of of A/D resolution, I think you're better off biasing the CT output to the mid-point of the A/D range and letting the PIC do the rectification (i.e., absolute value instruction).  I'd think the extremely low output impedance of the CT relative to the high input impedance of the A/D might make simple capacitive coupling possible...rather than op-amp rectifier or whatever.

Of course it depends on what you're doing, but there was a monumental reset in AC power meters when solar panels allowed consumers to put energy back into the grid.  Power meters now needed to detect/measure negative kW-Hrs so you got credited accordingly.  I figure you're not doing trying to make a residential power meter but a bridge or rectifier eliminates the polarity of the current before it gets to the processor.

Good point, with the 10 bit A/D/ let's assume I get maybe 8 bits of accuracy, still pretty decent for anything we'd need for model trains.