Relay chattering problem

[Quantised]

Is this circuit very close to being unstable as the relay chatters or clicks open/shut about once or twice a second at certain times.

I have measured the pins on the LM339 and 8 is 3.4v and 9 is 4.8v.
Pin 14 shows 6.45v

I assume the relay is about 300 ohms and so expect around 40ma through the TIP47 but could be more. The relay is a DPST that switches 240v mains.

It seems the base of the TIP47 is quite close to the limit of turn on/off and if the 7.8v supply fluctuates (seen just a drop of 0.1v at times) by just a little could it cause instability. Not sure how the 5.6v zener effects way it works?

Any ideas how to fix this or is this circuit good and therefore I may have to look elsewhere?

 

[ronv]

Not exactly sure what you are trying to do, but..... Try making the resistor 8.2k. The 339 can not drive so much current. Then I think you can remove the capacitor and the zener and it will work.

 

[Quantised]

ronv - this is an old circuit on a board that has recently started to chatter the relay and as this is the controlling part of how the relay is turned on and off I can only assume it is fragile or as components age it gets close to a problem!

What do think a LM339 can handle, when pin 14 goes low it just falls to 1.3v and I thought 9.5ma would be OK for it to sink so is the 339 overloaded?

Also not sure what the 10u does - is it a sort of soft turn on. It is a pretty strange cicuit and trying to figure it out.

 

[ChrisP58]

The 10u cap will ad a small delay between when the output of the 339 goes high, and when the relay trips. The cap has to charge up to about 6.3 volts to trip the cap. The problem is that the BE junction of the bjt also acts like a zener such that the cap voltage stays right at the threshold of the Vbe + Vzener. As such, the transistor is always just barely turned on. Any variations in the Vbe and/or Vzener due to thermal (or anything else) will cause the transistor to turn off until the cap voltage adjusts.

If the circuit needs the delay caused by that structure, leave the parts as they are with one addition.

Insert a small resistance (10 to 100 ohms) between the zener and the base of the transistor. This will allow the cap voltage to go a bit higher and the base of the transistor will see a current source instead of a voltage that is barely enough to turn it on.

That should clear up your chattering problem.

 

[Quantised]

Chris - this is interesting as it seems it is unstable.

I do not think the cap is important for the delay of the on time as the relay just clicks once when started up and then the control chip slowly brings in the power. Maybe 5 or 10 secs later.

No idea why the zener is even there and the logic behind the 339 is simple, if there is AC volts available then the 339 goes high and if the AC is not present the 339 goes low. That is all there is to it so the cap and the zener seem redundant as there is nothing else connected to that part of the circuit.

Even so I am still a bit confused as if a 10 to 100 ohm resister is placed between the zener and the base would it not just reduce the current to the base?

Really need the Janet and John version.

 

[alec_t]

Part of the problem may be due to lack of a freewheel diode across the relay coil. High back-emf spikes could be causing interference which affects the comparator.

if there is AC volts available then the 339 goes high and if the AC is not present the 339 goes low

Since you are detecting AC, the cap functions as a reservoir. The zener provides a switching threshold.

Edit: No, not a reservoir. As I understand the intended operation of the circuit, the cap discharges every mains cycle. Providing its voltage can't ramp up enough to exceed the zener threshold during the cycle the relay won't be energised. When mains fails the threshold will be exceeded so the relay is energised. Simulation shows that a low cap value (< ~7uF) will just cause relay energisation, so is likely to cause relay chatter.My money would be on a duff cap being the culprit.

 

[Quantised]

This part of the circuit is all DC. The AC feeds a full bridge rectifier and then to a voltage divider, the DC output is fed through an opto isolator and then feeds through 2 339 one which is invert mode. It is a really tortuous circuit that seems to have been put together and then bits added here and there to make it work!

The normal mode of operation is in the OFF position for the relay, ie this means AC is detected. It stays in this mode all the time so Pin 14 is pulled low and sits at 1.3 volts. This seems strange as one would expect it to sit at near 0v
For the relay to chatter it must be turned ON and that means pin 14 will float HI to 6.4 v but immediately (1/2 sec) go low again. The only thing that controls the relay is this circuit????

I have attached a an updated schematic which also shows the fly back diode alluded to by Alec as I found it a way away from the rest of the circuit.

So I measured the inputs to the 339 and the output and the volts are shown in the schematic, These readings are all very stationary and I do not see small fluctuations.

Maybe I should be looking outside the box, poor connections, Knackered relay, temp breakdown etc as this only happens when the system reaches equilibrium.

It is a simple battery charger circuit that detects the AC and turns on the charge mode, when the batteries are charged it drops into a float mode and holds like 13.4v on the batteries, whilst in this float mode of just supplying enough current to keep the batteries at 13.4v does the relay chatter.

Getting quite lost in the 339 logic that has been implemented.

 

[eTech]

Hi

Do you have a part number for the relay so we can look up specs?

eT

 

[alec_t]

whilst in this float mode of just supplying enough current to keep the batteries at 13.4v does the relay chatter.

Perhaps there's little hysteresis around the 13.4 trip point and something is switching heavy current which could cause interference spikes?
Ok, so it's only DC at the 339 inputs in theory. Any brief output jitter on pin 14, due to spikes on that DC, should be caught by the cap, but it looks as though it's getting through. I'd replace the cap.

Pin 14 is pulled low and sits at 1.3 volts. This seems strange as one would expect it to sit at near 0v

As per the datasheet the 339's sink current may be as low as 6mA, so it's not strange. 6mA through 820Ω would drop only 4.9V below the 7.8V supply, i.e pulling pin 14 down to 2.1V.

 

[ronv]

I really do need a new calculator. The decimal doesn't always work. Can't imagine why - it's only 30 years old.
I think alec has it right with the bad cap. I think the circuit looks something like below where it is AC going to the comparator. But since it is rectified your meter reads it as RMS. Hard to tell how it measures the output wave shape.

 

[Quantised]

eTech - the relay is a W389CX7 by Magnacraft and the spec sheet I found for it shows the coil is 100Ω for the 12v model

Alec - The ref voltages are supplied by a 317 set to 8v and that also drives the 339 and the other circuits except for the relay which is driven from the 12v line. So the mains passes thro a bridge rectifier and then a voltage divider, the result is opto isolated via a 4N35 to further voltage dividers and is then input to another 339 - the output of that 339 via more voltage dividers drives the pin 8 of the 339 in question. So there is a lot of components upwind that could influence the circuit.

So my thinking is the only way for the TIP47 to energise the relay is for pin 14 to go high and the only way it can go high is if the pin 8 drops low as pin 9 is fixed at 5v.

As Ron says the meter will see the RMS of these pins and so measuring what is going on is a transient is difficult.

Very difficult to get at the circuit when it is running but I could put a clip on the quad 339 with pigtails and then connect up to a scope and see if it has any spikes and chase it back.

If I can decode the circuit further is it worth drawing up the circuit and posting it?

 

[alec_t]

So the mains passes thro a bridge rectifier and then a voltage divider, the result is opto isolated via a 4N35 to further voltage dividers and is then input to another 339 - the output of that 339 via more voltage dividers drives the pin 8 of the 339 in question.

Unless the rectified mains is smoothed, the input to pin 8 will have a strong 100Hz (or 120Hz) component. So not fixed DC (although your DMM may be telling you otherwise).

pin 9 is fixed at 5v.

Or should be? Could a bad solder-joint or interference cause a variation?
Have you replaced that cap yet?

 

[moffy]

Replace the cap. Check your 7.8v supply for ripple.

 

[Quantised]

OK I have had a scope on pin 14 for a day or so and waiting for the chatter, when it happens all I see is a very quick transient about 6-8ms and maybe a drop of 0.3 volts, from the stable 1.3v, at most but the scope is really simple - no trigger, no capture etc. I do not see any rise in voltage that could trigger the relay!!!!!!

When the unit is idle there is no ripple at all on the pin 14 or 8 or power supply even though the AC is on but when the unit starts to fire a triac BTA26 there is a 500 to 1000mv ripple on almost everything. The ripple is very strong when the triac fires hard but as it shortens its pulse the ripple follows suit and is a short sharp spike - from 1.6 down to 0.3v on the steady state 1.3v DC like 1 or 2 ms.

The chatter only happens at the point when the triac pulse has diminished to almost zero, the transformer hum is almost inaudible and then the chatter, sometimes for a few cycles and other times for minutes on end. As the triac starts to fire up again the chatter stops.

Looking at the circuit attached below it seems there is some filtering as the AC on signal travels to the pin 8

Putting the scope on the 7.8v supply now and wait for the next incident.....

 

[Quantised]

This just goes from bad to worse. The relay circuit, the 339 driver, the TIP47 et al are all at the correct voltages and do not fluctuate when the 'noise' is there.

So I listened with a kind of stethoscope on the relay and elsewhere and it is not coming from the relay. Absurd as the only moving part is the relay and rest is solid state and one automatically thinks it must be the relay chattering - in sa sort of soft way.

So now my question is what makes a sort of soft cluck cluck sound - the relay goes CLICK pretty loud and it is a totally different sound.

The main components are a large transformer driven by a triac from AC and controlled by a board with 1 chip, 2 339's and a bunch of diodes and resistors.
The output side has like 32 Fet devices acting as diodes to get a DC signal. There are some large caps like 16,000u and that is it so what can make the cluck noise. I have never come across this before so any ideas will be welcomed and checked.

 

[Janis1279]

Hi
Are you checking a voltage levels on the relay coil? May be it is so low to fully turn on state.
A low value capacitors are frequent cause of problems on the older designs.

 

[alec_t]

So that would seem to rule out the relay. We didn't catch any wild geese on that chase .
Transformer laminations can cause noise.
Have you checked the transformer mounting for tightness? A slightly loose mount could allow a transformer on a metallic base to click when pulsed.
Have you tried the stethoscope on the transformer?

 

[Quantised]

Transformer is tight and bolted and looks new, stethoscope - sounds like it is coming from everywhere and I think it is not a mechanical noise (all noise is mechanical I know but it is not a rattle or similar) and so reckon it has to be electrical.

Looked at the +12v main supply line that connects to the batteries and also right next to the 16,000u cap and it shows a hefty spike on the scope trace - attached. Cannot get the whole trace as the video takes it section by section.

The GND is set 3 divisions below the centre so each div is about 3v, as the signal is around 13v looks right and the pulse is about 3 v on top so quite a powerful pulse, the small pulses are from the triac firing which is probably 100Hz so the pulse duration looks like 4 to 5 ms. The end of the pulse always coincides with a triac pulse on all of the 30 odd pulse the video caught.

Do you think that the power of this pulse is enough to make the xfrmr and all the power electronics make that noise. I did record the noise and maybe I could upload it - big file!!!

A 5 ms pulse would sound like a cluck - the obvious concern here is what is the nasty pulse going to do to the rest of the components?

urrent thinking is towards the triac and its firing circuit, tough to get new triacs here though.

 

[eTech]

What is providing the 7.8 v and 12v DC? Is it a regulated supply? A battery and charger?
Thinking maybe a DC choke in line with the source DC supply(s) might help?

eT

 

[Quantised]

The 7.8 comes from a LM317 and the 12 comes from the main battery DC terminals - the one which is shown on the scope.

I will check the 7.8 again when the board is replaced - it is out for components to be renewed like the triac. I do not see a choke but see very large smoothing RC network in the 12 supply line.