oscilloscope Training, Introducing errors intentionally 1

[DM2]

I want to train some technicians on the use of an Oscilloscope for trouble shooting electrical interference on communication circuits. I'd like them to see real world examples of what noise looks like and what capacitance does to the communication signal. The communications circuit would be either RS232, RS485, possible some Ethernet or other communications protocol.

Normally we talk about the issue, what causes it, etc. We have equipment in our panel shop (PLC's) that represents what's in the field. We use this equipment to simulate the application in the real world and test programs before shipping to the site. We use this equipment for training as well. For training, I'll set the equipment up so it's functioning properly, then send the students out of the room, break something, and then have them come back into the room to figure out what's wrong.

I'm looking for ideas on how I can:

Simulate noise on the circuit
I thought about wrapping an un-shielded circuit around a magnet, but I'm not sure if that's the best way, what size the magnet should be, etc.

Demonstrate what happens to the signal with increased capacitance
I thought about putting a capacitor in the circuit, but not sure what size of capacitor, or if this is the best method.

Any suggestions or advise would be appreciated.

Regards,
DM

"Real world Knowledge isn't dropped from a parachute in the sky but rather acquired in tiny increments from a variety of sources including panic and curiosity."

 

[catserveng]

I used to do some technician training, had the same problem. I used a combination of resistor and capacitor decade boxes, and used a Variac and wrapped wire around different parts of the circuit to generate noise. For higher frequencies I used a signal generator. I just would set up a training circuit and then mess with it to get undesirable results.

I found I actually learned a lot myself setting up for student labs. Best way is to just play around with it.

I also used to make sure to have the students make measurements with both the scope and a meter, so they could see the difference in how device actually measured to signals.

For communications I dealt with mostly Modbus RTU (both RS422 and RS485) and CanBus/J1939, so I would usually find a damaged or salvaged controls to use as the master and slaves in the circuits. Use a resistance decade box as a termination resistor in one part of the circuit, great way to show how improper circuit burden impacts the comms.

Hope that helps, MikeL.

 

[itsmoked]

Any kind of buzzer with the wires near it will bring in arcs-from-contacts type notice in wholesale amounts.

Keith Cress
kcress -

http://www.flaminsystems.com

 

[danw2]

Sounds like an outstanding idea. There's precious little in the way of 'scope' shots that I found for serial signal problems

1. Capacitance

The reason there are published RS-485 baud rate versus distance tables is that the increasing capacitance with longer cable lengths affects the waveform.

I suppose there's two ways to demo that
- add a capacitor at a working baud rate and and watch the before and after results on the waveform
- increase the baud rate for the same length of cable and watch the waveform change.

2. Termination:
Highly recommended for an understanding of the need for termination, this video uses a square wave signal generator, a scope and a coil of twisted pair cable.

[URL unfurl="true"]https://www.youtube.com/watch?v=zrDxSM91Jcg[/url]

Duplicate that exercise for an understanding of what a reflection is why termination is required.

Loading with too much terminating resistance or in the wrong location can be shown using several nodes and a collection of resistors.

3. Biasing
The problem with demonstrating biasing is finding convenient power supply and ground point to bias to. The concept of distributed versus end-of-link biasing is well covered in this video
[URL unfurl="true"]https://www.youtube.com/watch?v=_54h2flKi_Q[/url]

 

[VEBill]

A good lesson would be to have them use an oscilloscope to measure a non-isolated (center ground) balanced signal. The point would be to use two probes and the invert-add (difference) function.

Don't use 240 VAC as your non-isolated balanced test signal. The noobie error of applying the scope probe ground to one side* would certainly be lesson they'd never forget, but this point could be better explained without a high power demonstration and smoldering oscilloscope.

(* For thread lurkers: Don't try this at home, dangerous.)

 

[VEBill]

For the noise lesson, you might build a simple capacitor coupling circuit to inject HF noise from a signal generator onto the digital signal.

The point would be to recognize the added fuzziness of the trace.

A permanent magnet would probably have precisely zero effect.

 

[DM2]

First, thanks for everyone's input. I’m a mechanical engineer so forgive my ignorance on some of the comments. Most of what I Google on the Internet is about making communications circuits bullet proof, when what I want to do is break them.

Catserveng…
So you took the wire from the Varac and wrapped it around the signal circuit, and then adjusted Varac?
You don’t happen to have a sketch of your decade box do you?

Itsmoked,
So just taking a 24vdc buzzer, wrapping the buzzer wires around an unshielded RS485 circuit and turning it on will be easy to see on a scope?

Danw2
You say “…add a capacitor at a working baud rate…”
Based on the chart, at 19.2, I’m guessing a 100 Pico farad capacitor would be easy to see on the scope. Would I connect one capacitor in series with each leg of the RS485 (i.e. 2 Capacitors)?

VE1BLL,
Unfortunately we don’t have a signal generator


Regards,
DM

"Real world Knowledge isn't dropped from a parachute in the sky but rather acquired in tiny increments from a variety of sources including panic and curiosity."

 

[danw2]

1) The table might not be clear. It's maximum capacitance per meter of cable for a given baud rate, not the maximum capacitance per run.

You'll need a lot more than 100pF to load the lines

Put the cap across the A/B or (+)/(-) driver lines.

2) Untwist the twisted pair cable to induce more noise on the lines. You might try something like zip cord with parallel wires to get more normal mode noise.

 

[itsmoked]

So just taking a 24vdc buzzer, wrapping the buzzer wires around an unshielded RS485 circuit and turning it on will be easy to see on a scope?

Yes.

Keith Cress
kcress -

http://www.flaminsystems.com

 

[zappedagain]

Don't forget the simple test of leaving the ground lead disconnected when probing an isolated source, so you can see how long it takes them to identify the 50/60 Hz noise from the mains and fix it.

X