# Help with Telephone Circuit

Discussion in 'Electronic Basics' started by Trudeau, Feb 14, 2005.

1. ### TrudeauGuest

I have build the following project.
http://electronicsworld.tripod.com/telephonesimages/12.gif

The circuit works for the most part.

There is one problem however that I have.
The problem seems to show its self in two different ways.

1. The light turns on at erratic times, eg when the furnace turns off. Not
all the time but every once and a while I hear the forced air fan of the
furnace turn off and click the relay in the smart light turns on and the
light turns on. I'm sure the furnace is on a separate circuit.

2. When someone calls in at the end of the first ring when there is darkness
the light turns on. Fine exactly what I want. However sometimes the light
keeps clicking on, remaining on for as long at the capacitor allows, clicks
off and then clicks back on then remains on for as long as the capacitor
allows etc. several times. Some times its 3 times sometimes up to 10 times.
It's what I call a bounce. The light turns off, you hear the relay click OFF
and then it seems to bounces back ON and remains on for its scheduled time.

The relay is a 5 vdc relay with the right voltage rating to switch 120 Volts
AC. I have it connected to a outlet in a box that I plug the 40 watt lamp
into.

I assume the problem is that some how the a 120VAC surge in power eg the
furnace and the lamp turning off is tripping the relay. The tell tail sign
of this is that if I pull the plug from the lamp that is plugged into the
"triggered" outlet and sort of push the plug in and out a few times I can
make the relay "CLICK" and it keeps the light on for the scheduled time
limit. In other words when I make my own "Power Surge" it trips the circuit.

It seems odd to me because, the 120 volts is suppose to be completely
separate from the phone circuit. How can the 120 volts AC cause the relay to
"bounce" as described. And more to the point how do I prevent it?

I hope Ive made it relatively clear on what my problem is.

Do you think it is simply the relay, is perhaps a bad choice or even could a
different Diode than the 1N4007 that is used over the relay be used to solve
the problem??

Thank you.

2. ### peterkenGuest

Looks to me as if a glitch comes in from the phone line itself (the "ring"
terminal) which triggers the first '123.
Anything of size say about 2V *will* trigger IC2 and thus IC3 and so on,
meaning the circuit is built way too sensitive to just detect something like
a ringtone.

As far as I see it, I'd increase the level at which IC2 'passes' anything.
Might do it by connecting 2 zener of say 27V in series with R1, making sure
the zeners are "anti-series" (cathodes connected together).
Motivation : Since a 'ringtone' is at least about 60Vac it will pass, but
anything smaller will be blocked by the zeners.

Another way would be to "slow down" any glitches at the output of IC2.
Motivation : Since a "ring" always lasts for at least say 0.5 sec, removing
any glitches shorter as say 0.2 sec might do the trick there, so no more
false triggering occurs.

3. ### John FieldsGuest

---
The first thing I'd do would be to put about 1000µF across the 5V
supply where it comes into your box (I'm assuming you're using a
wall-wart of some sort) and the second thing I'd do would be to put a
0.1µF ceramic directly across the Vcc and ground pins of the chip.

One other thing you might consider is replacing the chip with HC,
since the HC versions usually come with Schmitt trigger inputs (check
the manufacturer's data sheets) and your day/night detector has _no_
snap to it at all. Also, the HC will provide you with substantially
more base current for the relay driver than the 123 can. Lastly,
since the 123's low output voltage will never go above 0.4V, I'd get
rid of the 2.2k resistor on the base of the relay driver.

4. ### TrudeauGuest

The power supply is a 9 Volt AC/DC adaptor and then I use a 7805 to bring it
down to 5 Volts (although when I put my meter on it it appears to be more
like 4.8 VDC)

I think ( and I have the whole thing in a junction box right now
functioning) if I recall the caps I used with this are 10uf ??
Eg

(12-9v) [7805]
+Vin --------1 2 3--------Vout (5v)
| | |
+10uF | +10uF
| | |
- ---------------------

Are you suggesting that I change these Capacitors ??

5. ### Anthony FremontGuest

I wouldn't recommend that. I've been advised in the past to use a
smaller cap on the output side. The reason being that when power is
removed, the output cap can hold the output pin voltage above the input
pin voltage since the input cap will discharge faster. I haven't
sacrificially killed any 7805's yet to prove that out, but it seems
reasonable.

I've seen quite a few schematics showing a diode connected from the
output pin to the input pin. It remains reverse biased as long as power
is applied. When power is removed, the diode only allows the output pin
to rise a maximum of .6V(or whatever Vf is for the diode) above the
input pin. This should allow any size capacitance to be used on the
output side.

I would also recommend adding a .1uF cap to ground on the output pin to
help prevent oscillation.

6. ### peterkenGuest

your use of the diac is about identical as the two zeners,
only difference is the diac voltage decreases to about zero if triggered
while the zener-voltage remains at its level

Increasing R1 to 100k makes the internal led of IC2 "glow" less for lower
voltages,
might also reduce small glitches however is less reliable

7. ### peterkenGuest

Whenever *any* circuit gets powered there are *always* transient effects,
even in integrated circuits
Increasing/decreasing capacitances will not cure those effects, increasing
power supply caps might even get things worse since longer transients exist
Designing circuits so that transient effects cannot harm anything does the
trick
Just look at C2/C8 timing caps, and imagine a transient on the power
supply...

Avoiding "power-on-light-on" might be done for example by slowing down the
base of T1

One other thing : I'd reduce R8 to say 560...680 Ohm, since a TTL out
*never* reaches 0V at "low"
(usually somewhere between 0.4 and 0.8V, bringing Vbe of T1 at say 0.5V
worst case thus *possibly* conducting T1 somewhat unwillingly if IC3out is
"low")

8. ### TrudeauGuest

Thanks for your advice. In trying to understand this a little bit, so that I
can apply this to other circuits I make.

Why do you think a slight spike or oversight (if those two words are
synonyms in this case) would cause the relay to trigger. An increase in
voltage "shouldn't" cause the relay to trigger. If the voltage dropped I
could see this. Eg. when I plug in the DC adaptor it understandably turns
the light on for its period of time because its charging the circuit, if I
can put it that way. Perhaps I'm confused, I assume that when the Furnace
for example turns off the voltage surges for a moment. Maybe it does drop.
As if its turned off then on. I suppose lights sometimes flash when motors
are turned on indicating that there is a drop. But as well shouldn't the AC
adaptor which is 9 volts hold some Inductance keeping the circuit from
dropping. Unfortunately I don't have a scope to test this, and would not
know how to use one if I did.

Sorry for writing like a stream of consciousness .

Appreciate your help in understanding this phenomenon. For future projects.

9. ### TrudeauGuest

I should say, (I didnt before for brevity and I didnt think it was the
issue) I increased the R1 to 100K and I added a diac to the ring line
inseries.
The reason I did this was because when ever someone picked up a phone the
light came on. The diac and 100K solved this problem. The break over voltage
interestingly enough I think was 27-32 Volts. About what you wanted. Your
use of the two diodes together are an interesting way of doing this. I think
(I'm strickly amature so I may be wrong) but the diac does the same or
similar as the two diodes together head to head. I have sometimes wondered
why in some circuits people had diodes anti-series I see now. Your using the
break down voltage in your favor. IF IM RIGHT.

Regards

10. ### TrudeauGuest

For full disclosure (I didn't put all the details in my original post for
brevity)

1. I put a diac in the ring line with a break over voltage of about 27-32. I
did this because when people picked up the phone the light came on . I also
put a 100k Resistor at R1. This seemed to solve the problem of the light
coming on because of phone line issues. eg people picking up phone or
turning phone on in the case of wireless handsets. I didn't mention these
changes because I didn't think they were the problem. or that the phone line
was the issue.
I think based on what you have said you would agree.

2. I actually did change the LS123 to a 74HC123. I did this because an
electrical engineer acquaintance was nice enough to look at the schematic
more resilient to voltage issues. It appears you two are singing from the
same song page.

I am using a 9 Volt AC/DC adaptor ( Wall Wart as you call it)
I then am using the following

(12-9v) [7805]
+Vin --------1 2 3--------Vout (5v)
| | |
+10uF | +10uF
| | |
- ---------------------

One poster has already suggested that I
"i would put like a 200 min on the input side and
a 500 min on the output side."
I think (and verbal descriptions are always hard to understand) Your
suggesting the same kind of thing.
Using a 1000uf on the in line from the + to the -.
I'm not sure if the mis leading info I gave you regarding the HC / LC would
change your advice on the 0.1uF from Vcc to ground on the chip.

Regards

11. ### TrudeauGuest

PS. I did ground Pins 6 and 14 as per datasheet for HC123.

12. ### JamieGuest

i don't see what your 5 Volt supply is, but its common to have
electronics perform resets and do funny things when powering up.
its called oversight.
any ways, you could try using a large capacitor on the 5+ line to
help hold the charge for those short minor pulses in the line.

13. ### JamieGuest

yup!
10 is very small, try something in the
line of 200 uf and up.

i would put like a 200 min on the input side and
a 500 min on the output side.

14. ### John FieldsGuest

---
The reason the light was coming on before was because of the pulse
created by going off-hook. When the phone is on-hook, the phone line
sits at about 45VDC and then when you go off-hook that voltage drops
down to about 6V, so you were coupling that ~ 40V level change through
C1 into the opto's LED, causing it to flash momentarily, but long
enough to trigger the first one-shot. What you've done with the diac
and the resistor change is to limit the amplitude of the pulse into
the opto to the point where it no longer generates enough light out of
the LED to turn on the transistor and fire the one-shot.

However, the phone line could still be part of the problem if what's
happening is that the fan cutting off is generating a magnetic pulse
which is being radiated by the power wiring, picked up by the phone
wiring and being coupled into your first one-shot. Not likely, but it
_could_ happen, depending on how close the phone line is to the fan
wiring... The easiest way to find out if that's the problem would be
to disconnect the phone line and exercise the fan a bunch of times to
see if the problem goes away.
---
No change; with the HC123 in there there's even less need for R8 since
Q will, essentially, go to the ground rail when it goes low.
---
---
Yes, but thinking about it a little bit more makes me think that this
would be a better way of doing it:

WALL-WART +9-12V>----+-----[82R]------------->RELAY +5
|
+----[78L05]---+-------->LOGIC +5
+| | |
[1000µF] | [0.1]
| | |
WALL-WART GND>-------+-------+------+-------->LOGIC GND
|
+----------------------->RELAY GND

While there's nothing wrong with the 7805 you're presently using, the
78L05 can easily supply all the current the logic needs out of a nice
little TO-92 package and, this way, the relay's operating current is
taken from the dirty side of the regulator and any transients or junk
it creates will be sucked up by the 1000µF cap and also blocked by the
regulator. Also, a separate ground line for the relay connected right
at the cap to keep the relay's ground current out of the rest of the
circuit couldn't hurt

The 82 ohm (1/4 watt) resistor is based on your having a 5V relay with
a 100 ohm coil drawing 50mA from a 9V source, the resistor wasting the
other 4 volts from the [9V] wall-wart. _But_ since you say "9-12V" I
suspect that what you have is a 9V wall-wart, but that it's being so
lightly loaded that its putting out 12V. If that's the case, then the
resistor would have to be 150 ohms, 1/2 watt in order to drop the
other three volts.
---

No, that 0.1µF is always a good idea since it heads troubles off at
the pass...

15. ### TrudeauGuest

Actually, and again because I'm learning the the >(12-9v) may have been
miss leading. I copied it from the little diagram that I had in how to use
the 7805 and it meant to me anyway that I could start with anything from 9
to 12 volts. and the regulator would bring it down to 5 volts. I know that
you could actually start with a higher input voltage but for us home bodies
who has a 30v AC/DC adaptor. Most everything is 12 or under. (my experience
anyway) I have not or don't recall anyway actually checking the voltage on
the AC/DC adaptor accept what its rated for which is 9Volts.

Again thank you for the advice I must now think it through. I have not seen
a 78L05 but looked it up and sure enough there it is. It does look like a
better regulator. Or at least more suitable for this application. I can't
imagine anything I would do that would actually need 5v, 1 amp that can come
from the 7805.

Thanks again I will digest it all.

16. ### w_tomGuest

Some of the usual suspects not seen in responses. Do you
have a bypass capacitor adjacent to the 74HC123? Bypass cap
is more critical with the HC version compared to the LS
version. And it must make a short connection to that 74HC123
chip. Schematic does not show it.

You did connect power and ground to pins 16 and 8. HC chips
can be powered, unstable, via other pins if the Vcc and Gnd
pins are not connected.

Another has recommended the relay be grounded separate from
HC ground. Remember, wire is an electronic component and not
a perfect conductor.

Many responses assume the phone line is electrically
isolated from the rest of that circuit. Maybe. But then
separation between phone line components and wires is
important for proper operation.

You are suffering a transient problem. But too many post as
if a transient enters on one wire, causes the problem, and
terminates. View transient problems in terms of both incoming
and outgoing path. For example, the transient may enter on
one AC power wire and leave on the other. Or it may enter on
one AC wire and leave via a not so well separated wire to
phone line.

Remember, there is no complete isolation between relay wiper
and relay solenoid. We have seen where a transient on the HC
output will actually trigger that IC to change state.

An oscilloscope using two probes (in differential mode)
would go a long way to finding the problem in but minutes.
Currently we all can only speculate. Provides above are some
examples of the usual suspects.

17. ### TrudeauGuest

Thanks for you tips. I keep adding data to my knowledge.
The idea of grounding the relay etc is an interesting one suggested by you
and another poster. Im a little unclear on this exactly. I would not have
thought that on such a little circuit it would be much resistance. If I
understand what people are saying I asume what people really mean there
should be a separate negative - connection put closer to the emitter side of
the T1 close to it to prevent the electricity traveling as far to the relay.
And absorb some of the voltage fluctionations. If you put it on the
collector side it would trip the relay. And not work properly. If people
mean someone else I would apreaciate knowing.
Another respectable answer was to "separately" power the + side of the relay
of a sort.
see "John Fields February 14, 2005 1:42 PM" in this thread for details. I
would have never thought of that and find it a facinating way of doing it.

Its a lively news group with some thoughtfull answers.

Regards

18. ### w_tomGuest

In electronics, noise on the ground pin will cause more
noise problems as compared to noise on the Vcc line or other
IC pins. Therefore we may create separate digital ground,
analog ground, and periperhal driver ground that only join at
one point. This longer distance in copper between the
different family of components creates reactive electronic
components - the copper traces - to separate the different
grounds.

In your case, digital electronic ground and the relay
peripheral ground would be separate until both meet where a
power supply ground connects. Copper traces are more
electronic components. Resistance is not the problem. Copper
impedance is the reason we do these things. This electronic
nature of copper is also why the bypass capacitor (typically
ceramic) must be located adjacent to the 74HC123.

19. ### ryan wiehleGuest

looked at your schematic. I suggest you replace the opto-coupler
with a schmitt-triger type opto-coupler to avoid
multiply spikes during the ring-cycle.

20. ### peterkenGuest

Since ringtone is about 60Vac there'a always multiple triggering during the
rincycle...
Replacing the opto with a schmitt trigger type would only be helpfull if a
slow changing dc voltage would be applied at the 'ring' line