8M comm

[CWells]

I've been messing with serial communication for a while now...I have to have 2 outputs share an input line, making diode mixing necessary. I have some 1N914 diodes, and I was trying to find out how to orient them, and they seem to pass a logic signal regardless of which way I put it in the circuit. Huh?? Which end is the back stripe facing, and why am I getting a signal either way?

The input pin is grounded via a 10k resistor.

 

[eclectic]

http://www.fairchildsemi.com/ds/1N/1N914.pdf

Black stripe is Cathode.

What's your test procedure and circuit?

e

 

[hippy]

Or white stripe on some diodes, but stripe is universally ( unless someone can prove that wrong ) cathode, the pointy end -- as I could never remember my anode from cathode until someone ( can't remember who, but big thanks ) noted the Kathode looks like the |< end of the diode.

For Nxxxx baud rates you want the diodes pointing away from the transmitters, pointy-ends joined and into receiver, with possibly a 1K-10K pull-down at the join to 0V.

For Txxxx baud rates you want the diodes pointing towards the transmitters, anodes ( non pointy-ends ) joined and into receiver, with possibly a 1K-10K pull-up at the join to +V.

The 0V of all PICAXE have to be joined in either case, and both sides have to use the same Nxxx or Txxxx baud rates. I haven't found any diode which didn't work but low forward drop ones are probably best.

 

[slimplynth]

From the information given, my first guess would be a breadboard mistake, having them inline so the diode does nothing but appears to conducting in both directions. I'm guessing this is where a pebble cd could be handy, but I'm on my phone, in the bath.

 

[John West]

Or white stripe on some diodes, but stripe is universally ( unless someone can prove that wrong ) cathode,...

I once solved a production problem where every circuit bd being produced was failing test. It seems the manufacturer of the diodes used (a bag of 5 thousand of them) had put the stripe on the Anode end of the device. So it isn't always on the Cathode end - but it certainly should be.

 

[CWells]

Here's the setup I have now...

 

[CWells]

Quick update - I just found my flourescent worklight is sending EMF signals into the circuit...??? I'm trying to "clean up" the area now...

 

[CWells]

Another weird thing - my receiver pin is connected to ground through a 10k resistor. I have the connector plugged into nothing. I'm getting a stream of random signals through the serial line. If I reach up and touch my (off) worklight, the stream stops. I'm not touching the circuit, and neither is the lamp. Huh?

I found I can move the lamp into a certain position, and the noise stops - but the diode circuit still isn't working. A tester circuit works fine.

 

[Minifig666]

Check your diodes outside of the curcit. You may have broken diodes. I beleve diodes main failure mode isto conduct both ways.

 

[russbow]

I think the 10k should be moved to the left, and connect between the commoned cathodes and 0v, the 1k then from this common point to the input pin.

 

[Rickharris]

It depends how your light is wired. Unplug and see if it still does this. Possibly there is current in the ballast coil that is creating a magnetic field.

OR the electronics in the lamp are acting as an aerial and picking up RF and retransmitting it to your cct.

OR your lamp really should be earthed.

 

[CWells]

It's not just the lamp - I may have to rig up a Faraday cage or something, as I'm getting stray signals with it off too, if it's in the wrong position. No grounds in this house.

 

[Dippy]

Have you, as suggested by minifg, checked your components out-of-circuit?
When I'm prototyping I tediously test every discrete before use.

A lot of this is about signal:noise and how to separate them.
All sorts of things can have an effect.

Why don't you post your circuit? And a photo of layout too.
Values, layouts, proximities and boo-boos all have a part when things go pants.
A fresh pair (x No.of viewers) may be able to help - but only if they are given sufficient detals.

 

[BeanieBots]

I agree with russbow about the 10k position.
If it's that sensitive, then sounds like a floating input. Either the 10k or possibly even your download circuit.

Flourescents are notorious for noise. They require several kV to strike and then they produce a plasma to give off UV to flourese.

As mentioned by Dippy, post a picture and circuit.

 

[sierrasmith71]

I agree with russbow about the 10k position.
If it's that sensitive, then sounds like a floating input.....

I have seen this symptom--it really puzzed me for a bit---all I had to do to get a circuit response was to wave my hand near the prototyping board .... It turned out to be a floating ground on an input---10K was there but not connected to ground as I was sure it was!


David G.

Maryland USA

 

[Dippy]

Good point David. People should always treble-check. It's good practice and good practice.

This usually why something like; "Why don't you post your circuit? And a photo of layout too." is broadcast ....

 

[CWells]

Which is why I posted a pic...

No time until tomorrow to move the 10k...I tried swapping the direction of the diodes, no effect, and I tested them, they're fine. The problem doesn't seem to be with the code, or any bad connections, or any bad components...no low batteries, Picaxes are running fine, and a logic tester sees the signal before the diodes...

 

[BeanieBots]

Have you double checked that your 10k is 10k?
Diodes conducting both ways and erratic behaviour still indicate no pull-up/down. ie. a floating input.
This is tried and tested method of mixing inputs so however sure you are of correct circuit, somewhere, it's wrong.

Saw your schematic, but didn't see a pic.
Post a picture (photo), somebody might spot a silly.