Acceptable SSR switching patterns for mains heater

[Arlet]

Hello,

I need to regulate a 220V resistive heater with a Solid State Relay.
The heater power is a few kW. The SSR is of a type that only switches
on or off during zero crossings (which is fine, because I want to
minimize interference).

I'm switching the SSR with a microcontroller. I have an opto coupler
attached to the mains, so the microcontroller can detect the phase and
frequency, and control the SSR in sync with the mains.

My goal is to control the heater output as smoothly as possible. I want
to do this by switching by half-period intervals, but I am wondering
which switching patterns are considered acceptable for (power company)
transformers.

For instance, a 50% duty cycle can be created by this waveform:

_ _ _
/ \ ______/ \ ______/ \
\_/ \_/ \_/

But also by this one:

_ _ _
___/ \___/ \___/ \___

However, the second pattern, which only uses positive half periods,
results in a DC current, and I understand this can cause losses in the
transformer.

Now, I'm wondering about this pattern, for a 75% duty cycle:

_ _ _
___ / \___/ \ / \
\_/ \_/ \_/

Would that be acceptable ? I assume so, since there's no DC component.
Or this one for a 25% duty-cycle, which has a DC component, but it's
relatively small:

_ _
___/ \_________/ \______


Basically, I'm looking a simple guideline as to which patterns are
considered acceptable, and still get the most switching flexibility. Or
do I need to worry at all ?

 

[John Popelish]

Hello,

I need to regulate a 220V resistive heater with a Solid State Relay.
The heater power is a few kW. The SSR is of a type that only switches
on or off during zero crossings (which is fine, because I want to
minimize interference).

I'm switching the SSR with a microcontroller. I have an opto coupler
attached to the mains, so the microcontroller can detect the phase and
frequency, and control the SSR in sync with the mains.

My goal is to control the heater output as smoothly as possible. I want
to do this by switching by half-period intervals, but I am wondering
which switching patterns are considered acceptable for (power company)
transformers.

For instance, a 50% duty cycle can be created by this waveform:

_ _ _
/ \ ______/ \ ______/ \
\_/ \_/ \_/

But also by this one:

_ _ _
___/ \___/ \___/ \___

However, the second pattern, which only uses positive half periods,
results in a DC current, and I understand this can cause losses in the
transformer.

Now, I'm wondering about this pattern, for a 75% duty cycle:

_ _ _
___ / \___/ \ / \
\_/ \_/ \_/

Would that be acceptable ? I assume so, since there's no DC component.
Or this one for a 25% duty-cycle, which has a DC component, but it's
relatively small:

_ _
___/ \_________/ \______


Basically, I'm looking a simple guideline as to which patterns are
considered acceptable, and still get the most switching flexibility. Or
do I need to worry at all ?

If your system has the intelligence to avoid DC, completely,
do it. The distribution transformers will also like you
better if you don't hit the same polarity of half cycles
twice in a row, even if you balance the DC out, eventually.

The more evenly distributed your power pulses are, the less
annoying light flicker other users of the line will
experience. You might use every third half cycle for 33%
power, every 5th for 1/5th power, etc., instead of every
third cycle or every 5th cycle. For 1/2 power, I think
every other cycle is as good as you can do, but for 1/4
power you can use every 4th cycle, or a positive half cycle,
a skipped full cycle, a negative half cycle, and two skipped
cycles.

 

[Arlet]

If your system has the intelligence to avoid DC, completely,
do it. The distribution transformers will also like you
better if you don't hit the same polarity of half cycles
twice in a row, even if you balance the DC out, eventually.

The more evenly distributed your power pulses are, the less
annoying light flicker other users of the line will
experience. You might use every third half cycle for 33%
power, every 5th for 1/5th power, etc., instead of every
third cycle or every 5th cycle. For 1/2 power, I think
every other cycle is as good as you can do, but for 1/4
power you can use every 4th cycle, or a positive half cycle,
a skipped full cycle, a negative half cycle, and two skipped
cycles.

Thanks. So would you suggest that for 3/4 power it's better to use 3
full cycles, followed by an idle cycle, rather than the pattern I had
in mind (which had two positive half cycles in a row) ?

It's not hard to implement. I'm basically using a delta-sigma
modulator, extended with a small state machine that keeps track of
previous actions. If a certain pattern is not allowed, it will skip a
pulse, letting the integrator add up the error, and compensate it
later.

It seems that a simple rule that enforces alternate polarity for each
pulse would produce the behaviour you've suggested.

 

[John Popelish]

Thanks. So would you suggest that for 3/4 power it's better to use 3
full cycles, followed by an idle cycle, rather than the pattern I had
in mind (which had two positive half cycles in a row) ?

I would.

It's not hard to implement. I'm basically using a delta-sigma
modulator, extended with a small state machine that keeps track of
previous actions. If a certain pattern is not allowed, it will skip a
pulse, letting the integrator add up the error, and compensate it
later.

It seems that a simple rule that enforces alternate polarity for each
pulse would produce the behaviour you've suggested.

I think so.

 

[Rich Grise]

Hello,

I need to regulate a 220V resistive heater with a Solid State Relay.
The heater power is a few kW. The SSR is of a type that only switches
on or off during zero crossings (which is fine, because I want to
minimize interference). ....
Basically, I'm looking a simple guideline as to which patterns are
considered acceptable, and still get the most switching flexibility. Or
do I need to worry at all ?

You might try some of these links:
http://www.google.com/search?q=integral+cycle+proportional+control

Have Fun!
Rich