A/C causing clocks to reset and lights to flicker

[strugglinsparky]

Hi everyone. Usually just use this awesome site as a reference, but I couldn't find a post that delt with this.

A local heating company called me out to a house that the A/C inrush would cause all the lights to flicker and all the electronic clocks to reset. Lights flicker on inrush I have witnessed many times and I do not believe there is a fix or a issue with this, but every electronic clock in the house resets!!??. Microwave, electric range, sprinkler clock, digital garage door opener, everything resets to 12:00. They do not have any simple clocks House is out in the country and the only house on the local transformer, at the end of the POCO overhead line. The homeoners said this all started after they had a A/C installed. We did not wire the A/C so I was really hoping to blame the electrician.

Voltages all were "normal" at around 120v. There was only a max of 1v difference between A/B at any time. The lowest I witnessed was 118.5v at the house panel, when the A/C started up. The amperage on the house was 6 on A and about 15 on B. The voltage did increase a bit to the exterior disconnect (80' 4/0 SER) and a bit again to the pole disconnect (200' 250 MCM URD). 120/240 200 amp service. I use just a basic klein multi meter, so really quick fluctuations I cannot see. I checked for any hot connections with a temp gun and tested for torque. The highest inrush I measured was 22 amps and the running amps of the A/C was around 7.2. Nameplate on A/C was 18.2 min circuit with a 30 amp max overcurrent. I did not see the FLA/LRA. I was expecting much higher readings on the A/C. I texted the heating company tech but have not heard back yet.

I am perplexed.

 

[480sparky]

Possibilities:

Undersized branch circuit
Undersized service
Loose connection in circuit
Bad start capacitor in AC unit
Undersized POCO transformer
Loose connection in service

 

[synchro]

It would be good to know if this was a replacement for an existing central A/C, or if the homeowners did not have one before this one was installed. If they had a central A/C unit before and things were OK, then the problem is more likely to be from the new unit or its installation and less likely because of a service issue. If they never had A/C before then they might have an issue with the service, etc. that's gone unnoticed. This doesn't provide any definitive answer but it may help focus where to look further.

 

[zbang]

What you really need is a power quality monitor since you won't see the short dips on a regular multi-meter. But.... you can create your own sag with 2-4 portable heaters- connect them to branch circuits on the same supply leg, put your meter on than leg, switch on the heaters,and measure the voltage under load. Even with a 40-50 amp load, the supply shouldn't drop more than a few volts and certainly not 20-30. (Turn off other loads so if there is a neutral problem you're not increasing the voltage on the other leg.)

 

[StarCat]

Possibilities:

Undersized branch circuit
Undersized service
Loose connection in circuit
Bad start capacitor in AC unit
Undersized POCO transformer
Loose connection in service

The above are all correct in their direction, and need to be ruled out. Also what Synchro is saying is correct.
I would want to have a recording Ammeter such as Fluke 87 on the condensing unit to see what the START INRUSH is. All Single phase Compressors will have the inrush lowered by installing the correct
" Hard Start package." While it will not stop the dimming effect it will reduce it significantly. This is assuming all other matters are ok upstream of the unit.

 

[gar]

191003-0840 EDT

strugglinspaeky:

You failed to mentioned whether it is a 120 or 240 air conditioner.

A low cost test is two 120 V 15 W incandescent bulbs set side by side with one connected to each phase at the main panel. What do these lights do on start up of the air conditioner, and on turn off?

.

 

[strugglinsparky]

Thank you all for your help. The A/C is 240v max overcurrent of 30 amps with a 10g wire running to it. The heating company is reluctant to put the hard start package on the unit for whatever reason. This is a new A/C on a two year old house that has never had a A/C before this. The house is on a 200 amp service and the draw I read was 6 on A and 15 on B, so I dont think it could be a undersized service. I have contated the POCO and they are going to check their transformer in the next couple days. I'm hoping that fixes this. I will let you all know if I find something out.

 

[drcampbell]

Have you confirmed that the clocks resetting actually correlates with the air-conditioner compressor switching on? (or off) Can you duplicate the symptom by cranking the thermostat up & down? Most small electronic devices are going to need a lot more than a short-time voltage sag to completely reset.

 

[mopowr steve]

Even though it may have a 200 amp service, when ever that was installed or upgraded the power company may have only installed a transformer big enough to serve a house without A/C probably just a 10-15 KVA. Especially if they are at the end of a primary run. The POCO needs to be informed when a significant load increase is added so they can make adjustments as necessary.
Heck some of these A/C units have a hell of a LRA above 100 amps.

 

[hbiss]

The lowest I witnessed was 118.5v at the house panel, when the A/C started up... I use just a basic klein multi meter, so really quick fluctuations I cannot see.

I am perplexed.

It should be obvious that you just aren't seeing the dip on your meter but it is substantial enough to make the clocks reset. You are going to need a power quality analyzer to see it. To find the source connect it across the service and work back. If you see it at the service you know the POCO is to blame. Actually they should be able to provide a PQ analyses at the drop if no other problem is found.

-Hal

 

[gar]

191004-2156 EDT

I previously suggested using two 15 W incandescent bulbs to test for your problem.

These can be a very useful, inexpensive, and a readily available tool.

Ran an experiment with one 15 W bulb. Shortest off pulse I could generate easily at the moment was 60 mS, about 4 cycles.

Complete off for 60 mS produces a very noticeable flicker.

Changed to using the SPDT relay contact to switch between 124 V and 124 V with a 15 ohm resistor in series. The resistor drop is about 2 V. Still fairly large flicker.

Then used the relay to simply shunt the the 15 ohm resistor. Now a very small flicker. The relay is a P&B, and from memory has a transfer time of around 1 mS between NO and NC.

The two bulb approach gives you much additional information.

.

 

[hbiss]

Gar, I'm wondering how an LED would work instead of the incandescent lamps. Just an LED, a diode (maybe a FWB) and a resistor, not something like a LED light bulb with a driver. Thinking that the lag due to the heating and cooling of the filament makes the lamp less sensitive to voltage dips, whereas with an LED it's instantaneous.

-Hal

 

[gar]

191005-1424 EDT

hbiss:

The LED is very fast. I will take a look.

.

 

[Hv&Lv]

191005-1424 EDT

hbiss:

The LED is very fast. I will take a look.

.

Gar,
In reading here Iv’e noticed you like to experiment...
I have a question..
we have a Power line carrier system for our meter reads.
it basically works with the return signal traveling at the zero crossing or right at it.
On an oscilloscope there is a definite small spike during the signal injection to call on the meter right before (25-30 degrees?)
with that being said.. we have had some customers some time back that stated their clocks run fast. Some gain 8-10 minutes a week.

I understand clocks keep time by frequency. I notice mine get behind when we run my generator during a sustained outage.
(I’m not allowed to fix my own power lines, I live on a different system)

so, does the clocks count frequency at different timed intervals? I assume it’s like an on off switch during frequency peaks.
Wouldnt they all count at the same timed interval? Or are there others that count different points on the wave, like sampling?

 

[gar]

191005-1716 EDT

Hv&Lv:

If you have a synchronous motor driven clock, then, if power is not interrupted, meaning loss of a number of whole cycles, then I would expect the inertia of the motor to keep it in lock with the AC line. Power companies have for a very long time been required to keep the cumulative number of cycles per year quite close to a reference value.

Obviously if there is a power outage this is interrupted.

Watching my frequency meter the other day I was at 60.0 Hz except once in a while jumper +/-0.1 Hz. Switching to time period measurement, and calculating frequency, my frequency was much closer than 0.1 Hz most of the time.

A good electronic clock should do enough filtering of the AC line voltage to get good zero crossing information. I would not expect carrier line communication to interfere. A cheap clock might be a different story. I have several electronic clocks from the 1960s that keep very good time. Never had any problem from TED power monitor carrier signals, and they are strong.

I would expect a good clock to use a 60 Hz bandpass filter to get zero crossing information. This might have a filter bandwidth of a Hz or so. Otherwise they would at least need some sort of low pass filter, and possibly a threshold detector with a lot of hysteresis.

.

.

 

[synchro]

191005-1716 EDT

I would expect a good clock to use a 60 Hz bandpass filter to get zero crossing information. This might have a filter bandwidth of a Hz or so. Otherwise they would at least need some sort of low pass filter, and possibly a threshold detector with a lot of hysteresis.

.

Yes, a Schmitt trigger is commonly used to provide hysteresis to improve the noise immunity in zero crossing detection. They are very cheap to build by just introducing a little bit of positive feedback. The hysteresis creates a dead zone around where the voltage is small, and this makes it more immune to "noise" like the PLC signal.

Digital clocks just increment digital counters at each cycle of the AC wave form. Where this happens in the waveform doesn't matter on a long term basis, as long as it doesn't have false or missing detections that would speed up or slow down the clock, respectively . The counters "decode" when they reach a specific binary value to that indicates that an event like the next second, minute, or hour occurs, and then use this to increment other counters that hold the current time value which is displayed. The frequency is so low that this could be done in software on general purpose low cost real time processors if it made sense cost wise.

 

[Hv&Lv]

Yes, a Schmitt trigger is commonly used to provide hysteresis to improve the noise immunity in zero crossing detection. They are very cheap to build by just introducing a little bit of positive feedback. The hysteresis creates a dead zone around where the voltage is small, and this makes it more immune to "noise" like the PLC signal.

Digital clocks just increment digital counters at each cycle of the AC wave form. Where this happens in the waveform doesn't matter on a long term basis, as long as it doesn't have false or missing detections that would speed up or slow down the clock, respectively . The counters "decode" when they reach a specific binary value to that indicates that an event like the next second, minute, or hour occurs, and then use this to increment other counters that hold the current time value which is displayed. The frequency is so low that this could be done in software on general purpose low cost real time processors if it made sense cost wise.

That’s what I’m wondering. I think the digital clocks are counting the blips from the PLC signal, it would be a false detection.

 

[synchro]

That’s what I’m wondering. I think the digital clocks are counting the blips from the PLC signal, it would be a false detection.

It's certainly plausible that this could be happening.

Another thing that could be done to prevent false triggering in clocks that run off the AC waveform would be to insert a non-retriggerable monostable aka "one-shot" after the comparator that detects the zero-crossings of the waveform. This would block any re-triggering for a certain time period. For example, after the zero-crossing of the 60 Hz waveform is detected it could be made to block any detection that occurs before ~1/4 of a cycle elapses. That way it will keep it from triggering more than once before the voltage goes close to zero again on the next half-cycle
Of course doing this would cost a few pennies more, so it probably isn't done in consumer products.

https://slideplayer.com/slide/5778980/19/images/26/Non-Retriggerable+One-Shot+(OS).jpg


About a dozen years ago I got a few "atomic clocks" that synchronize to the WWVB transmitter in Fort Collins, CO. The clocks always keep very precise time and compensate for any daylight savings time changes automatically:

https://www.nist.gov/pml/time-and-f...ations/wwvb/help-wwvb-radio-controlled-clocks

 

[gar]

191006-2225 EDT

hbiss:

Ran an experiment on an LED. With a 2% change in excitation of current which is the necessary way with an LED. I did not visually see a change in intensity even with a long pulse. This was a red LED. A full wave rectified signal was used.

By comparison the 15 W incandescent was noticeable with the 2% voltage change at my 60 mS duration.

.

 

[George Stolz]

Jim, give me a call at the shop tomorrow if you get a minute.