Odd Transformer Question

[bostonman]

I'm working on a WES51 soldering iron, and have a thread about it (https://www.eevblog.com/forum/repair/weller-soldering-iron-wes51/msg4026304/#msg4026304). Since this question is regarding a different aspect of the circuit (and mainly about transformers), I'm asking in a new topic.

Just to summarize, the iron wasn't heating, discovered the iron was the culprit, but, in my initial troubleshooting steps, somehow caused the circuit to lose its ability to blink the light when the iron reaches the desired temperature.

Having said this, and if anyone reads the thread, I've ruled out everything on this board except the transformer and RC on pin 5. The attached is the schematic and a wave form I captured showing the top of the 1M resistor going to pin 5, and pin 5.

First off, what I expected to see is pin 5 to look a bit less than a digital pulse. I expected it to have the same shape (only not going below 0 since it's probably being pulled to ground by internal diodes on the PIC) but clipped at the top. For some reason the rise looks much sharper than I expected, but, at the same time, the RC I calculated to be about 200us, however, I'm measuring about 480us.

The more oddball part is regarding the transformer. even with the oscilloscope referenced to a counter that's referenced to an atomic clock, I'm seeing both the top of the 1M and pin 5 moving in phase at about 2ms every 9s. Sometimes it moves one way, sometimes the other.

Can a bad transformer cause this? I know it seems like a crazy question, but I've done everything to rule out the phase shift being caused by my generator (or scope).

Since I'm measuring 60Hz, it seems I can't connect 10MHz from the atomic clock to an oscilloscope channel and use that as the reference channel because it causes my 60Hz channel signals to fly around the screen. I've tried everything to prove the frequency generator is causing the phase shift, but, I've tried two generators, and the same results.

I've taken the 10MHz from the atomic clock, plugged it into one channel of the scope, set the generator to 10MHz (which is referenced to the clock), connected that to the scope, and the two waveforms don't move. Makes sense since they are referenced to the same point, however, if the generator had a bad time loop constant between its internal electronics and the external reference, then maybe I'd see them moving in phase.

I then set the generator to 60Hz, connect my scope leads from the soldering iron circuit, and watch them move a the rate I mentioned above (approximately 2ms in 9s). It keeps moving in phase in one direction, stops, and then moves in the opposite direction. This seems to point to the generator, but it's referenced to an atomic clock.

[TheMG]

To make sure I'm understanding what you're doing... You have one scope channel connected to your frequency generator set to 60Hz, and the other channel connected to the DUT, which is connected to mains power?

If that's indeed correct, the phase shift you're seeing is simply a slight frequency drift of the AC mains supply, which would be expected.

[bostonman]

My initial setup was scope channel 1 connected to the top of the 1M resistor (which is basically the secondary of the transformer), channel 2 to pin 5 of the PIC (the RC).

I decided to reference the scope using a frequency generator set to 60Hz on channel 3, and saw the phase shift on channel 1 and 2.

Thinking it must be the generator, I connected the 10MHz output of the cesium atomic clock to channel 4 of the scope, but that caused channels 1, 2, and 3, to not trigger correctly when I set channel 4 as the trigger. I'm assuming it's trying to trigger on 10MHz but show 60Hz (I probably never dealt with trying to trigger off a different frequency on a separate channel as I've never had a four-channel scope).

I then removed channel 1 and 2, set the generator to 10MHz, and compared channel 3 (the generator) and 4 (the cesium), but this also caused whichever channel wasn't being triggered to show constant noise (more like just being unable to trigger).

After I connected the 10MHz from the cesium to the ref in of the generator and compared the output of the generator at 10MHz and the atomic clock at 10MHz, and they were locked together.

Finally, I set the generator back to 60Hz (still referenced by the atomic clock), connected channel 1 and 2, removed 4, and saw channel 1 and 2 moving side to side in phase.

I fully agree it could just be the line moving, but was curious if a transformer could cause this issue.

If you read through the troubleshooting thread I referenced, I've checked everything on this iron, and, as silly as it sounds, the only thing that hasn't been confirmed good is the transformer. This is why I asked such an oddball question as to whether the transformer could have some issue causing the PIC to not get the correct line frequency.

The only oddball stuff that aren't quite correct and haven't been confirmed good: the rise time of the RC (maybe this is causing the PIC to act odd), why the RC waveform looks more like a digital pulse than part of a cutoff sine wave, and seeing the phase shift (which again, I agree, it's probably line frequency shifting.

[Fred_47]

I think you'd need a 3ph primary and a moving (rotatable) secondary to cause a phase shift thru the transformer. And, the secondary would have to physically move wrt the primary (rotating) magnetic field.

Your signal generator is the almost certainly the cause of the phase shift.

Try setting the scope trigger to "line".

The power system frequency does vary from 60Hz. Prior to retirement in 1993, I saw shifts of ±0.005Hz cause by tripping large generators from the system at significant load (>5-600MW). Such shifts took many seconds to minutes as the energy stored in system inertia (speed of the rotating machinery) and turbine governors adjusted to a new stable system frequency where the generation matched the load.

Also note that a slip of 2ms in 9 seconds -> 60 ±0.0133333... Hz. If the power system is changing that fast, you'll be working in the dark shortly.

The puzzle to me is that your signal generator seems to be hovering around 60Hz without slipping a complete cycle.

A question for the metrology minded is how can you use the 10MHz clock to measure either the 60Hz± system frequency or your signal generator? There are 166,666 +2/3 cycles @ 10MHz in 1 cycle @ 60Hz.

I don't have a stable clock but if I did I would try triggering from the clock with a holdoff of 16.666666...ms. I think that would produce either 3 or 6 waveforms on the scope. Hmmm... a longer holdoff, say 50 or 100ms may show only 1 (60Hz) waveform.

[TheMG]

Also note that a slip of 2ms in 9 seconds -> 60 ±0.0133333... Hz. If the power system is changing that fast, you'll be working in the dark shortly.

Depends on location and the size of the grid. Where I live it is normal to see variations in frequency of up to +/- 0.3Hz on a regular basis, sometimes all within about 30 seconds.

[bostonman]

Also note that a slip of 2ms in 9 seconds -> 60 ±0.0133333... Hz. If the power system is changing that fast, you'll be working in the dark shortly.

That was hilarious.

From the responses, I'm going to end my thought about the transformer having a quirk causing the phase shifting. It sounded silly, but, from researching others who have repaired a WES51, they too have found quirks that caused the iron to stop/start working.

As for the phase shifting, I would think if it was the frequency generator, when I compared the two 10MHz, then the frequency generator would change phase because it's not keeping up with the time loop constant of the reference signal input from the atomic clock.

Since it didn't, I have to assume my generator is quite solid and stable enough to get a fairly accurate idea of the line frequency phase changes. Having said this, I have to believe my line is changing phase at 2ms/9s (timing was clocked by an iPhone, so don't take that 9s as an absolute).

Too bad I can't divide the atomic clock. It has a 1/5/10MHz output, but nothing is divisible by 60; assuming I can make a digital divider circuit easy enough.

[bostonman]

Tonight I became a bit more curious. I wanted to compare two transformers (one being the secondary on the Weller).

I connected a transformer output to channel 1, the Weller solder station to channel 2, and a 60Hz frequency generator referenced to an atomic clock.

Looking at the file named 'zoomed in', you can see the  two waveforms moved simultaneously at about 6ms in 33s.

If I select 'line' as my trigger, then channel 3 moves; which makes sense.