[Amps] Solid state relays..again

Manfred Mornhinweg manfred at ludens.cl
Wed Feb 12 12:56:05 EST 2014


> Once in a while that thing turns on with a
> WHummpppp and a low frequency ringing that just trails off.

Let's see if I can explain this in a way everyone can understand.

When a power transformer is operating normally, the magnetic flux 
density in its core fluctuates in a sine wave function. The peaks of 
this sine wave are usually roughly where saturation of the core begins, 
but is not yet too bad. Typical values for power transformers would be 
between 1 and 1.3 tesla of flux density. Those who learned 
electromagnetics at least 40 or 50 years ago, probably are more used to 
the antiquated gauss unit - anyway, 10,000 gauss is 1 tesla.

The sine wave of the magnetic flux is in the same phase as the 
magnetizing current in the primary, which is 90 degrees lagging the line 
voltage. This means that when the line voltage crosses zero, flux 
density is maximum, and when the line voltage is maximum, flux density 
is zero.

Whenever the line voltage is positive, flux goes up, and whenever line 
voltage is negative, flux goes down (accepting some simplistic 
conventions on what is positive line voltage and increasing flux). So, 
at the time the voltage waveform crosses zero going from negative to 
positive, the flux density is at its negative peak. With the voltage 
being positive from there on, the flux density goes up - that is, it 
goes from -1 tesla to -0.9 tesla to -0.5 tesla, etc, reaches zero, then 
gets positive, keeps climbing, and by the end of the voltage semicycle 
the flux density has become +1 tesla. Then the voltage goes negative, 
making the flux density start going downwards again.

This should be clear enough, I hope. Now let's see what happens when you 
switch on a transformer.

If you do that precisely at the peak of the line voltage, which is the 
time when the flux density should be zero, and given that it _is_ zero 
in the transformer that was off, the transformer simply starts 
operating, without any sort of strange behaviour, overcurrent, spikes, 
or anything. This would be the optimal situation, from the transformer's 
point of view. But if you power up the transformer at any other time, 
things are not as good. The worst situation is switching on the 
transformer at the voltage zero crossing. In this case, the core is at 
zero flux density and now gets a full semicycle of voltage in one and 
the same polarity. So, instead of starting from -1 tesla to go to +1 
tesla, it will start from 0 tesla and thus will try to go up to 2 tesla 
in that first semicycle! And 2 tesla causes total saturation of the 
core. When a core saturates, several things happen. One is that the 
inductance of the windings drops drastically, producing a strong current 
pulse in the primary. And another is that the core no longer 
concentrates most of the flux, so that a significant amount of the total 
flux leaves the core, and flows through the air around it, and specially 
through anything magnetic in the vicinity. If the transformer is housed 
in a steel cabinet, the panels of that cabinet will attract a huge part 
of this leaked flux.

During the next several cycles, the resistance of the windings, losses 
of the core, etc, play together to gradually center the magnetic flux 
around zero. That is, while in the first cycle the flux moved between 
zero and 2 tesla, during the second cycle it might fall between -0.2 and 
+1.8 tesla, in the third cycle it might go from -0.35 and +1.65 tesla, 
and so on. It can take ten or more cycles to center the flux well enough 
to make the transformer work normally, without spreading flux around any 

The cabinet panels will thus pick up lots of flux initially, and then 
gradually less, over several cycles. And they will act like 
electromagnets, while they are picking up flux. This makes them move, 
proportionally to the flux they get. And since the flux is alternating, 
the panels will vibrate.

And this, my dear friends, is why equipment with steel boxes and big 
transformers inside can make that "WHummpppp" at the moment it's 
switched on.

Of course, in equipment controlled by a simple mechanical power switch, 
it's impossible to decide when exactly in the cycle the switch will 
close. It's this a lottery how close we will get to the peak voltage 
point, or the zero crossing. That's why sometimes the "WHummpppp" is 
strong, sometimes weaker, and sometimes even nonexistant.

OK. Now somebody out there might say that it would be a good idea to 
make a circuit that always powers up an amplifier or power supply at the 
peak voltage, to avoid that "WHummpppp". But it ain't that simple... We 
aren't powering up just a transformer, but instead a combination of a 
transformer and filter capacitors. The filter caps would prefer to be 
powered up at the zero crossing... they don't agree with the transformer 
at all!

That's why large power supplies should have at least a step start 
circuit, or even better, a full soft start circuit. Full soft start can 
be combined with power factor correction in the same circuit.

About solid state relays: They exist with TRIAC output, SCR output, 
MOSFET output or IGBT output. Each of them can be made in versions that 
switch on immediately when they get the control signal, or at the next 
zero crossing of the applied voltage. All SCR and TRIAC SSR's switch off 
as soon as the current through them falls to a certain level, after the 
control signal has ceased. Usually that's close to the first current 
zero crossing after the control signal ceases. But note that this is the 
_current_ zero crossing, not voltage! MOSFET and IGBT SSR's can be made 
to switch off immediately when the control signal ceases, or at the next 
current zero crossing.

Handling inductive loads is more difficult than resistive ones, because 
of switch-off, not switch-on. When a SSR switches off, there is still a 
small current flowing in the load inductance. Even if small, it can be 
enough to induce a high voltage spike at the moment the SSR switches 
off, and this spike is easily strong enough to make the SSR switch on 
again, which collapses the voltage, leaves a tiny current flowing, so it 
switches off again, causing the next spike... leading to an oscillation, 
which places great stress on the small control areas of the 
semiconductor structure that forms a TRIAC or SCR.  This is what can 
kill SSR's or TRIACs when switching inductive loads.
The trick to safely switch inductive loads is to install a snubber in 
parallel with the power device. It's typically as simple as a 0.1 µF 
capacitor in series with a small 100 ohm resistor. This will absorb the 
small residual current of the inductor, avoiding the creation of a 
voltage spike, and will give the power device time to switch off fully. 
But this network will also pass a tiny current even while the SSR is 
supposedly off. That's why not all SSRs are fitted with that snubber.

Enough for now...


Visit my hobby homepage!

More information about the Amps mailing list