Using SSR to switch transformer

o t d

Yes, don't waste your time on it.

R f

to

an

ant

n

ced

l

Transformer cores do not hold any significant magnetic flux when there is no current in the windings.

However,

Can't ever happen whatever you try.

Results may vary with transformer type.

Nope, can't be done.

and always start at opposite phase.- Hide quoted text -

Mine does.

First picture shows current through 3 half cycles, starting and ending with positive cycle.

Second picture shows same waveform, after a minute of inactivity. Notice increased current in first half cycle due to core saturation. The effect is also audible in the transformer.

This happens every time if the waveform starts with same polarity as the previous one ended, even with considerable time inbetween. I also ran it starting and ending with negative cycle, and the traces are the same (but inverted).

Alternating positive and negative bursts shows no extra current.

The experiment has a 230V->27V toroidal transformer controlled by SSR+MCU, and a small series resistor to limit the current. Secondary side is permanently attached to 1 Ohm load.

Without the voltage waveform as well as the current your trace is of no value. I expect your not triggering quite how you think you are.

I've added a voltage trace (measured across the primary) to the second waveform:

During the current spike, the voltage drops due to current limiting resistor in series with primary.

e t e t

You seem to be triggering at the zero crossing point and not on the voltage peak. How big is your current limiting resistor? How are you measureing the current waveform?

Yes, for this test I wanted to demonstrate worst case saturation effect, so I'm triggering on the zero crossing point.

If I trigger the SSR on voltage peak, I get similar results, but somewhat lower peak current, as expected.

The series resistor is 33 Ohms. Current trace represents voltage across this resistor. I'm using 3 voltage probes (one side of resistor, between resistor and transformer, and other side of transformer), and tracing the differences between them. Initially, I was using a 0.1 Ohm/10W resistor, but it died with a flash and a bang during a current surge. :) So I decided to pick something bigger.

re

g

ice

ect

the

it

33 ohms will seriously alter your results. What do you mean by "tracing the differences"? does your scope have 3 channels? Or are you not measureing them all at the same time. You should get almost no surge if you fire at the top of the voltage waveform, if you are then you are doing something wrong.

I sure that the 33 Ohm resistor alters the results, but the fact that there is no current surge when using bursts of alternating polarity does imply that the core is still magnetized to some degree. Also, with an even number of half cycles, there are no current surges with repeated bursts.

Without the series resistor, the effect is similar. Unfortunately, I don't have a real current probe, so I can't measure the current without the resistor.

Here are the voltage waveforms across the transformer primary, switching

3 half cycles at peak voltage _without_ series resistor. First burst, followed by exactly the same burst a minute later.

You can see the voltage drop across transformer as current increases. The lights in my room also briefly dim during second burst.

The scope has 4 channels, so everything is measured at the same time.

e:

here

ing

otice

ffect

s the

an it

e d , n W )

second-no-R.png

e quoted text -

You only fire the first wave at the voltage peak not the rest, they should be at the zero crossing point so the voltage is has no breaks. When I was making tin can seam resistance welders I had exactly the same set up as you, problems only occured when the transforme didn't recieve alternating half cycles.

On a sunny day (Tue, 25 Mar 2008 10:52:34 +0100) it happened Arlet Ottens wrote in :

Question: did you have the 1 Ohm load connected in teh experiments? You transformer ratio is 230 / 27 = 8.51, so the 1 Ohm is seen as the square of that in the primary, so as 72 Ohm. I would expect the magnetic filed to collaps sooner with 1 Ohm load.

33 Ohm series with 72 Ohm load seems not a pleasant ratio. Better use a lower shunt if the load is connected?

This effect is caused by a net DC component from an unequal number of positive and negative excursions. In the test sets I design, we set up a burst of 5 cycles for instantaneous trip testing of circuit breakers. While the current is increasing, we make sure there are actually five full cycles, and each subsequent pulse is clean and without a high current surge. But if the timing is set so that there are 4.5 or 5.5 cycles, we get large inrush currents.

This is, essentially, magnetic core memory, similar to that used in ancient computers for data storage.

If the voltage waveform were reduced gradually, the net DC component would be lower, until the magnetization becomes insignificant.

We tried reducing inrush current using large thermistors, but they would eventually heat up and lose their effect, and we could not wait for them to cool between pulses. Careful control of the number of pulses helped a lot, but when the breaker trips, there are very often odd numbers of half-cycles and the next test would have a large surge.

One way we found to minimize it was to apply a short burst of about 1/2 the voltage, with the proper even number of half-cycles, and then the next full-voltage pulses were OK.

Paul

Yes, the 1 Ohm load is always present.

The 33 Ohm series resistor was just for safe testing, and to prevent more things from blowing up. I didn't have anything smaller. During normal use, and with correct SSR trigger, it's not used.

Yes. In my case, I can keep an eye on the cycle count, and make sure there are always an integral number of cycles during normal operation. The only case where this isn't possible is when the power is suddenly removed, for instance when the machine is unplugged while in use, or a circuit breaker trips. This is expected to be very rare.

To cope with those cases, I use a soft start at power up. See:

The waveforms I referenced were created on purpose to test the effects of bad timing, and to demonstrate the permanent magnetization of the core. They don't reflect typical timing of the finished product.

I think you can do the same thing by just remembering the polarity of the last half cycle that was triggered on. But it is a bit hard on an EEPROM cell to keep reversing that one bit, every time you turn the SSR on. If the DC supply lasts long enough after the plug is pulled, you can save it to ram and just store it nonvolatily after the power is turned off (or the plug pulled).

Congratulations on performing a classic and very educational lab exercise. Your direct experience with this puts you in a small minority of people who understand this problem this well.

--
Regards,

John Popelish

On a sunny day (Tue, 25 Mar 2008 12:17:33 +0100) it happened Arlet Ottens wrote in :

Right, so evaluating 33 + 72 = 105 Ohm. Total power at 230 V is 230 x 230 / 105 = 503 W. From that 33/105 goes into your resistor, so 158 W. Did you use an electric heater as resistor? Because 230 x 230 / 33 = 1.6 kW ;-) Just curious to the setup...

Yes, it's an old 1.5 kW electric hotplate that I use for stuff like this.

One or more incandescent light bulbs is very useful for performing tests where high currents might occur. There is about a 15:1 ratio of resistance from cold to hot, and you get a very visual indication when something isn't right.

I use an old car headlight for 12 VDC stuff.

Paul