Re: (Beginner) Varistor or TVS

Thx Chris for your advice. Sad to say i'm currently on my own. About
the 240/120 isolation transformer i'll have to check with the lab. But
i'll definitely put my own safety as the first priority =)

I did some designs on my own, taking into account your advice and some
other resources that i've managed to find and i was wondering if you
could give your opinion on it:
http://geocities.com/Antonio_Magma/transient.jpg

I decided to use a 3-stage protection circuit with a gas discharge
tube, a resistor and a MOV (as stated in the Bel coupler data***).
For the MOV i followed the coupler data*** and i chose the LITTLEFUSE
Varistor 390V 130J (V250LA40B). The gas discharge tube, the EPCOS 350V
(EC350X). These parts i chose based on the availability of it in my
area.

However, for the series resistor i'm not sure how do i choose the
value, that's where i hope you can help me.

Next, i was thinking since my signal is transmitted at 4Mhz-21Mhz, i
could probably design a high-pass filter to filtering out everything
else. Thus, i designed a 4pF capacitor and 10kΩ resistor to form my
high-pass filter.

For the PSU, i decided to take your advice and i'll search for a linear
power supply of maybe 10V output then use a LM317 to bring it further
down to my operating voltage of 3.3V.

What do you think?

Chris wrote:
Ant_Magma wrote:
Thanks for you advice Chris. There were a few problems that had me
started with this project too late which i won't elaborate further.

If i follow your method, won't i have 2 plug heads (don't know what's
the proper term)? 1 plug brings in the signal to the transformer then
to the coupler and the other brings in the power supply from the wall
wart?


I have already ordered Bel's powerline signal coupler, can i try and
build it like the schematic drawn in the data***? The live and
neutral wires are connected parallelly to the coupler and the PSU
(step-down transformer). Since the coupler only passess the Homeplug
frequency of 4Mhz-20Mhz filtering out the 50/60Hz power freqeuncy i
should be safe right? For the step-down transformer i'm thinking of
using any encapsulated PCB transformer available on Farnell.


Please advice.


Btw, Chris i guessed you have read the data*** (however brief) it is
of the Bel's powerline signal coupler. At the last couple of pages with
the schematics, at the capacitor it writes 4.7n x2 class. What does x2
class mean? And the 2 resistors 200k, can't they just use a 400k
resistor?

At the Electrical specifications part of the Bel coupler, the
parameters of the coupler has this Hi Pot @60Hz 1mA and a value of
2000V. What does it mean? It rejects the 60Hz?

Hi, Ant_Magma. One thing at a time.

A real engineering prototype would have one line cord, with the split
to the powerline modem and the power supply occurring inside the box.
For a proof-of-concept prototype, and for initial work, you can have
two plugs. You can always add a small linear or switching power supply
later, after you have the thing working. Points off for
proof-of-concept as opposed to full engineering prototype, but no
points if it doesn't work. You need to get something going NOW.

Whenever you're working with line voltage in an engineering lab
environment, you should use an isolation transformer. This is first of
all for safety. As an extra benefit, you can realistically step down
from 240VAC to 120VAC, which should provide additional safety. I'd
rather work with 120VAC than 240VAC. The thing is, you apparently have
very little non-SPICE experience, and you're dealing with potentially
lethal voltages. You need an isolation transformer for your personal
safety. This is also why you need an advisor who has worked with line
voltage before, and can keep a watchful eye on this part of the circuit
(once you've got it done, you can enclose it and be done with it).

According to the coupler data***, it provides 2KV of isolation
between the line primary and the secondary. This is pretty much
standard for transformers. On paper, you should be fine. The X-10
looked good on paper, too. But I have a vision of your getting to the
last couple of days, and a line transient jumps up and smokes your
works. Pobre Ant_Magma. Line voltage accidents are usually explosive.
Apart from any electrical safety considerations, you don't and won't
have the time to fix this if a line transient smokes your prototype in
the last week. Do everything you can to keep your prototype safe, and
only use it and the other piece of comm equipment with an isolation
transformer.

By the way, here's what happens with a hipot transformer test. It
measures the coupling between the transformer primary and secondary,
and is required for proving isolation from line voltage. This is a
standard safety test which is used to get listing from regulatory
agencies like UL and CSA.

| .------. ,------.
| | )|( |
| | )|( |
| o------' '------o ^
| | DUT | |
| | | |I
| | | |
| | |
| / \ / \
| (2KV) ( A )
| \_/ \_/
| | |
| | |
| === ===
| GND GND
(created by AACircuit v1.28.5 beta 02/06/05 www.tech-chat.de)

The 4.7nF cap they're talking about is one rated for line voltage use.
In the event of overvoltage, it's made to be self-healing instead of
just shorting out. Don't use a cap which isn't rated for the line
voltage you're using. And 200K is a standard resistor value. Also,
resistors have a voltage rating as well as a wattage rating. If you
apply more than the rated voltage, it may arc over the surface of the
resistor, which would cause failure. Using two resistors means you can
halve the voltage across each resistor, which might be a good thing --
they're not self-healing. ;-)

On a personal note, I had the pleasure in days of yore to work in a
place where they liked to hire one or two undergrads and work them to
the bone part-time for peanuts. As part of the deal, they had
unofficial use of lab facilities after hours, and the use of all the
kibbitzing they could wheedle out of the staff for their senior
project. I tried to give them a hand when I could, and had quite a bit
of fun working with them. It's satisfying to see students finally "get
it".

I'm not qualified to help you here. I've never done a component-level,
"from-scratch" network interface. And in order to get an answer to two
or three basic questions here, you had to wait over 10 hours. If your
project is due in 9 weeks, you've only got about 1500 hours left. You
need to get that loop lag down, and get your questions answered now, in
quantity, by someone who knows what they're doing. If you wait for the
kindnesses of strangers, you'll still be talking at deadline. Get some
help, you're out of time.

Actually, the real lesson of the senior project is how to do projects.
By doing something enormous by yourself, you get a practical baptism of
fire in the gentle art of working under brutal pressure. It's a
required part of the engineering skillset.

Good luck
Chris

.