Measuring dropouts on a waveform

[RTSLVU]

I need to automate measuring the transfer time of a UPS.

The customer specification is less than one cycle drop out of the UUT output voltage.

What would be an effective way to measure the dropout time of an AC voltage given a transfer waveform captured using NI-Scope?

[santo_13]

Could you please share the waveform snippet of a typical transfer event?

[Hooovahh]

I also think we might need more information.  But in the mean time I had to do something that sounds similar years ago.  I had a PWM waveform that I needed to look at and make sure it didn't have any issues with dropping out.  Eventually Crossrulz came up with a pretty decent solution using a two level threshold looking for the values to go above, and then below specific values.

https://forums.ni.com/t5/LabVIEW/Two-Threshold-Analog-to-Digital/td-p/2738762

[RTSLVU]

Honestly I don't have anything at the moment. I was hoping for more high level ideas on how to approach it I at this point.

I could capture a waveform with the scopes we use when we do it manually to look at but it wouldn't be a LabVIEW waveform data type though.

Let me see if I can come up with some sample data...

[santo_13]

Even a scope capture snippet would be a good starting point just to come up with a test method.

[RTSLVU]

This is the best I can come up with at the moment...

The CSV waveform files are too big to upload to the board...

[Hooovahh]

Okay that is a bit more of a pain.  I guess as a first try I'd just have a range of values in the coerce in range.  Then have it track how long it has been in that range.  If the time it was in that range was greater than X ms then you fail, or maybe capture the waveform at that time.  If you have a consistent sample rate for the AI maybe just a shift register tracking how many samples were between say -10V and 10V.

[RTSLVU]

---

@Hooovahh wrote:

Okay that is a bit more of a pain.  I guess as a first try I'd just have a range of values in the coerce in range.  Then have it track how long it has been in that range.  If the time it was in that range was greater than X ms then you fail, or maybe capture the waveform at that time.  If you have a consistent sample rate for the AI maybe just a shift register tracking how many samples were between say -10V and 10V.

---

This gives me something to think about... I think tracking the number of consecutive samples within a window might just be the ticket.

[santo_13]

Here is my thought process,

• Assuming you're able to trigger at 50% timespan, exactly at the recovery pulse rising edge
• Consider only the first half of the total capture
• Convert the values to only positive by taking the absolute value
• Compute power (assuming you get both voltage and current waveforms)
  • Look for the end region which is below a threshold power, which may be below 0.1W?