Relay for 74HC-controlled outlet strip

(Almost certainly overthinking this one, but another case where the
wrong decision means more waiting for another mail order.)

A couple of projects I have in mind would benefit from
CMOS/microcontroller-controlled outlet strips, and I'm trying to make
some decisions about what sorts of relays this might require. I'm
assuming first that the contacts need to be rated for 125VAC (or more)
and for well over the current I intend to drive (does P=IV apply to AC;
e.g., is 10A right for 12500W @ 125VAC?). And I'm looking for either
12V or 5V coils, whichever comes easier.

The drive circuit I often see laid out is something like this:

+------+------> +12VDC
| |
- @
1N4001 ^ @ coil
| |
+------+
|
R |/ 2N2222
signal >--------/\/\-----|
|>---| GND

I would assume that I should select R to drive the transistor to
saturation. A relay I'm looking at has a coil rated for 120mA, 100
ohms. As far as I know, that ought to work, since even the TO-92
work-alike of the 2N2222 is rated for 500mA or so typically.

Any flaws in my logic yet?

Another matter concerns switch configuration. Would it be a better idea
to leave one blade always connected (single pole) or do I need to
connect and disconnect both at once (double pole)? I'm assuming that
the third grounding prong is not disconnected under any circumstance.

The IDEC RU2S-C-D12
(http://www.alliedelec.com/Search/ProductDetail.asp?SKU=814-0064) seems
to do what I want. Am I totally off?

One more thing: Say I'm running the microcontroller circuit and the
relay from the 5V and 12V pins, respectively, of an ATX-style SMPS. Are
there any special precautions that need to be taken other than the
typical reverse diode across the coil, decoupling caps on ICs, and so forth?

Thankses
PSM

 

Why wouldn't you use a solid-state relay, which will work directly
from a logic signal?

 

If you didn't want to mess with all those discretes, what about a
ULN2003 or equiv.?

 

So far it appears that they're a lot more expensive for the same contact
rating (about $4 vs. about $12) and high frequency response isn't so
important.

 

If I were driving more than one, I'd probably use one of the ULN2803As I
already have, but we're only talking one relay per device, so the splay
of discretes really isn't an issue.

Thanks
PSM

 

Instead of using your regulated power to run a relay coil, consider
using a phototriac-output optocoupler to drive a 120VAC relay.
It has most of the good features of a solid-state relay without
the complication of sucking half an amp from a power
supply intended for CMOS.

 

An interesting idea...one that could easily remove the complexity on the
DC side of things.

A 5VDC solid-state phototriac relay I'm looking at has the following
phraseology in the data sheet: "No overvoltage absorption element is
built in. Therefore, if the G3MC is connected to an inductive load, be
sure to connect the overvoltage absorption element." I would guess that
a coil relay is an inductive load. What would one use as an overvoltage
absorption element?

Thanks
PSM

 

---
No. The contacts will be rated for volt-amperes, not watts, and at
125VAC, 12500VA, the contacts will need to be rated for at least
100A if your load is resistive. (But, I suspect that 12500 was a
trypo and you really meant 1250, yes?)
---

---
For the same ampere-turn coil, a 12V relay will draw 5/12 the
current of a 5V relay.
---

---
A PN2222 has a beta of between 100 and 300 with an Ic of 150mA, so
that would seem to be a good choice. However, in order to be sure
that the transistor is saturated, it's customary to "force" the beta
(Hfe) to 10. That's done by dividing the collector current by 10
and pushing that current into the base.

For a 120mA relay that would require 12mA into the base, which your
micro might not be able to supply.

However, there's a possible out. Let's say your micro can supply
2mA into the base. In that case the transistor's beta will be
forced to:


Ic 120mA
Hfe = ---- = ------- = 60
Ib 2mA

so it should still work. Try it.
---

 

Use X10. The life/home you save may be your own.

 

I did in fact mean 1250, not 12500, but I was talking about watts. No
lightbulb, appliance, or computer PSU has ever been rated in VA. (This
is odd to me, since from what I've read it's VA that applies to AC. Why
are these things rated in watts?)

I'm looking at using the relay-switched strip with a few combinations of
small items. For example, a pair of wall transformers marked 25W and a
toy lamp rated 17W. What's my current per watt at 125VAC?


The single-pole configuration would be

mains
+---(~)---+
| |
| O
| \
| O
| |
| |
+--[] []--+
outlet
strip

while the double-pole would be

mains
+---(~)---+
| |
O O
\- - - - -\
O O
| |
| |
+--[] []--+
outlet
strip

Is there any reason to prefer one over the other? Is one more likely to
arc than the other?

Thanks
PSM

 

Because that's what you pay the power company for.

If the load is a resistor, watts = V*A

But consider a capicator or inductor. It takes current, but it
gives it back on the other half cycle. No net power.

Light bulbs have a horrible (10x) starting current when they are cold.
That's nasty to relays.

I'd add up all the power, divide by 125, round up generously, then
look in the catalogs and probably round up some more.

 

Agreed. Bringing X10 into the picture was a complexity and expense I
wanted to avoid. A hard CMOS line is protocol-free, isn't as
susceptible to radio interference (intentional or accidental), and would
cost well under $10 to implement. As for safety, I'm not ignorant to
any electrical and/or fire hazards inherent in this mod. The whole
reason this idea seems reasonable is that outlet strips already have a
big power switch, so what's the harm in inlining and/or replacing that
switch with a relay of the same contact rating? Nothing significant, as
long as I'm not careless.

Thanks
PSM

 

---
Yes, I know, but relay contacts aren't rated in watts, they're rated
in volt-amperes. Or horsepower, or tungsten loads in some
instances.
---

---
Generally, they aren't. Look on their nameplates and you'll see
that their operating voltage and current are listed separately. The
reason for that is that they're reactive loads except for toasters,
irons, and other appliances which present resistive loads to the
mains.
---

---
If the transformers are truly just transformers, (as opposed to
being AC to DC converters) then they're mismarked and should be
rated in VA.

If your mains are 125VAC and your loads are resistive, then:

P 1W
I = --- = ------ = 0.008A = 8mA
E 125V
---

I prefer the DPDT configuration since mains HOT and NEUTRAL are both
switched. That way, even if the outlet strip or a load with an
exposed ground and a non-polarized plug is plugged in backwards
it'll be safe to touch when the relay isn't energized. Of course it
can still kill you when it's on, but that's a different story.

BTW, you showed the outlet strip with outlets in series. I'm sure
you meant something like this:

.. mains
..+---(~)---+
..| |
..O O
.. \- - - - -\
..O O
..| |
..| |
..+--[] []--+
..| |
..| |
..+--[] []--+
..| |
..| |
... .
... .
... .
..| |
..+--[] []--+
.. outlet
.. strip

 

I'm only calling them "transformers" because "AC-to-DC adapter" is a bit
long-winded and I've never been partial to the term "wall wart". But,
in any case, that's what I'm talking about--black boxes with
transformers, bridge rectifiers, and probably a couple of other things
(never actually broken one open).

What general rules are to be applied when determining whether a load is
resistive or inductive? Having no formal training in such details, I
tend to assume that something with a coil in it is inductive. By this
logic, the AC-DC "transformer" (itself loaded with a small electronic
device such as a router, cable modem, laptop, ...) is inductive, the
motor part of the toy lamp is inductive, and the incandescent bulb part
of the lamp is resistive. I get the feeling that I'm entirely off on at
least one of these.

That is not only what I meant but in fact what I thought I had said
(abbreviated to a single "out" socket). My prediction of ambiguity in
my illustrations leaves something to be desired...

Thanks
PSM

 

X10 isn't only for "remote" control. You needn't drive the protocol
or push buttons to use it.

 

No, you buy one of their interfaces for $5 or perhaps less (perhaps
free).

 

Don't have one now, but I get ads for them all the time. So many
that I throw them away (modules too). This stuff is *cheap*, if you
don't buy it from RadioShaft (or Leviton) and there is a module to do
just about everything.