Staying in phase and measuring sags...

Other Hardware General Electronics
1

Hi everyone,

I have a question how to best measure the impact of short, sudden sags on input voltage and wonder what the best way is to measure such impacts in a reproducible and scientific manner. One instance of such sags is when the upstream network of your line power is perturbed and the T&D network switches your home to a different circuit.

An easy way to measure 'sag resistance' is to simply hook up the equipment to a traditional variac and play the 'how low can you go' limbo game re: reliable unit operation. However, that type of operation is more of a brownout scenario, not a sudden sag that only lasts a couple of ms. Presumably, a power supply should be able to tolerate a lower input voltage on a few ms basis than on a continuous basis.

But, assuming I would want to explore such lower voltages on the basis of a few ms at a time, I am somewhat stumped on how to do it. For instance, let's say I'd take two variacs, set one at reference and the other at a sag voltage. That way, both output voltages have the nearly same phase change due to the variac and hence should align almost perfectly. Then, how to flip flop between them?

One option is a reed switch like this one, that has a very fast release time and more than adequate contact rating. Said reeds release / switch on a 0.5ms basis. Simple enough, but even 0.5ms w/o power is likely enough to cause some sort of measurement error.

Doing the flip flop with Triacs also doesn't seem easy to this non-EE due to their habit of not 'turning off' until the current reverses. I suppose one could use zero-crossing detecting Triacs and only flip and flop on alternative wave cycles but that limits the time resolution to a mere 8.3ms. Not the end of the world and the folks at the power switch tail site offer a complete zero cross switching solution.

Then there is this charge-coupled MOSFET circuit that should allow very fast switching between sources on a somewhat arbitrary basis. You'd need four of these, with two sets attached to each variac. With some OR'ing, upstream the output would then be based on one MCU pin being high or low... But, unlike the Triac solution, there doesn't seem to be a fully-built solution.

Or should we invest in a $$$ arbitrary waveform generator like this one? Decisions, decisions...

2

Finding it hard to understand your request. Do you want to measure something? You say
"measure the impact of short, sudden sags" - what impact? on what?

Surely you just want to measure (rms?) mains voltage, which is done via a transformer
and signal conditioning circuitry normally. Transformer brings voltages down to safe
values, an opamps / whatever to present to the ADCs.

Presumably, a power supply should be able to tolerate a lower input voltage on a few ms basis than on a continuous basis.

Mains voltage does zero-crossings all the time - power supplies cope with those events
all the time and they are "a few ms". Or do you mean more than a few?

3

OK, so you are testing the impact of short sags on pieces of electrical/electronic equipment. You probably know what you are looking for as far as disturbances in the equipment.

What you want to know is how to generate arbitrarily short sags in AC line voltage. Yes?

How much power? What voltage? What about spikes?

I'm inclined to suggest building a PWM controlled inverter, if the power isn't too high. With a high enough PWM and carefully designed filtering, you should be able to filter out effectively all of the PWM switching artifacts while leaving the filtering able to allow dropouts and sags through.

4

Hi and thanks to you both for the answers. The equipment we're looking to testing is consumer electronica like DVRs that have a fairly moderate power draw (<50W) but which can react somewhat sensitively to very short power outages or even sags. You can imagine how happy home users might be if their DVR omits recording the latest Game of Thrones episode just because there was a short hiccup in the line power supply.

However, I am not aware of any industry standards re: how long a power supply is supposed to last (i.e. be able to ride through an incoming hiccup, high or low). Based on published specifications, I presume that most power supplies are tested on a continuous +/- 20% nominal VAC basis but I would really appreciate any resources you can recommend.

I'd like to explore three scenarios, i.e. testing on a

  1. Continuous brownout basis (i.e. at what sag voltage does the DVR stop recording)
  2. Intermittent brownout (i.e. how low can you go with a 0.5ms, 1ms, 1.5ms, etc. brownout without affecting the recording)
  3. Complete dropout basis (i.e. all the way to 0VAC) at 0.5ms, 1ms, etc.

I haven't considered spikes, and I would be very appreciative of any industry norms you could point me to. I saw an interesting circuit that used a 18V transformer as a buck or boost converter to explore the behavior of the attached equipment at +/- 15% of nominal voltage using a DPDT switch to switch back and forth.

5

However, I am not aware of any industry standards re: how long a power supply is supposed to last (i.e. be able to ride through an incoming hiccup, high or low).

These do exist.
It depends on the class of the device but for consumer electronics, set top boxes and the like a device is supposed to be able to withstand three missing mains cycles without resetting or malfunctioning.

6

Hi Grumpy_Mike!

Thank you for your insights, I really appreciate them. So it seems that using a Triac-based, zero-crossing power tail may work very well in terms of keeping humans isolated insulated from mains voltages while also affording us the ability to measure impacts on the basis of 1/2 wave (i.e. 8.3ms) increments. I imagine that the sags experienced by the equipment must be longer than 3 cycles to cause the reported problems. I'll circle back when I have more data. Thank you again.

7

Triac-based, zero-crossing power tail may work very well in terms of keeping humans isolated from mains voltages

No they are rubbish for that, the leakage from the triac or SCR mans that you still get an electric shock from a device that is turned off. I know from experience a long time ago. Ouch! :fearful:

8

Oh dear!

So, do you think a reed switch (or parallel banks thereof) listed at 0.5A capacity would be able to handle the kickbacks and inrushes that a 50W switchmode power supply can generate? Or is the Mosfet approach the most likely to produce clean breaks regardless of where in the cycle the power is switched off?

9

So, do you think a reed switch (or parallel banks thereof) listed at 0.5A capacity would be able to handle the kickbacks and inrushes that a 50W switchmode power supply can generate?

It depends what these actually are, but a reed switch sounds unlikely.
Why do you want to isolate it anyway? People should not be touching live stuff, or potentially live stuff, anyway, even if it is off.

10

No no, I meant insulated, not isolated. my bad. I will go back and correct that. I had meant to use this power switch tail that uses a triac to turn on and off the power supply to the DVR. Does that make more sense? Apologies again for the typo.

11

I don't think insulated is the correct word either.
If you get hold of the mains input of a device that is controlled to be off with a triac or SCR you are not isolated nor are you insulated, so I am not sure what you mean.

12

Constantin:
No no, I meant insulated, not isolated. my bad. I will go back and correct that. I had meant to use this power switch tail that uses a triac to turn on and off the power supply to the DVR. Does that make more sense? Apologies again for the typo.

Those 'power tail' products generally do use isolated input via a opto-isolator input device. I base this on the spec for it's controlling input requirement, ( DC input: 3-12vdc (3-30ma), terminal block accepts #14-30 AWG wire) which is common for opto-isolated device that use an led as the interface connection to the microcontroller. However the standard power tail does automatic zero crossing turn-on and you have no control to say turn the tail off for just one AC (or one half) cycle. They do sell a product called the Zero cross tail ( http://www.powerswitchtail.com/Pages/ZeroCrossTail.aspx ) which can provide a timing signal that one could use to control the timing of the power tail such that one could probably create single cycle drop outs.

As far as the how any given device will respond to short AC drop outs really depends on it's internal DC power supply design the device is using , how much main DC filtering capacitance reserve the supply has and of course the current draw requirements of the device.

All in all this seems to be a demanding test/analysis effort and you really have to have a detailed specification requirement to test against.

13

Thank you all again. The test is challenging and I really need to get my hands on some data first to see how one would be test the condition. If this turns into a project, I'll be back with better information!

14

I really need to get my hands on some data first to see how one would be test the condition.

See:-
http://www.compliance-club.com/archive/old_archive/020122.htm
Look at section 6.5 AC mains supply dips, dropouts and interruptions

15

Thank you, thank you, thank you.

Needless to say, you are an amazing resource and thank you again for sharing that information. More reading!