3 Phase Harmonics and Solid State Relay Issues

[Phynoxum]

I have a 3 phase problem that I think might be applicable to this forum. The setup is equipment which contains 5 resistive heaters (A through E). The unit is powered by 208V 3 phase and each of the heaters are hooked up line to line in an unbalanced configuration as follows:

L1-L2: Heater D (21A FLA)
L2-L3: Heater B (21A FLA)
L3-L1: Heater A (4.3A FLA), Heater C (4.3A FLA), and Heater E (7.2A FLA) for a total of ~15.8A FLA

The power to each of these heaters are controlled via zero-cross SSRs with each using their own dedicated 4-20mA control signal. I am measuring a heavy degree of all harmonics (including DC component) when all heaters are running in PID control to adjust current to maintain temperature. These harmonics disappear when the SSRs are forced to 20mA (full power) and when they are driven to 4mA (no power). The issue I am seeing is that randomly, after anywhere between 3-10 days, some or all of my SSRs will just stop working. They become completely unresponsive to command signal changes and can only be recovered by removing the 4-20mA signal completely for some duration of time and then re-applying.

During the past 3 months of being plagued by this issue, I have been monitoring the voltage and current of all 3 phases and do not see any voltage spikes, dips or other anomalies which could be the cause of failure. I was also continuously monitoring the 4-20mA signal to see if control power is timed out by chance which might have caused this issue. I am currently investigating to see if the harmonics present might be tripping up the zero-cross detector circuit internal to the SSRs. Can anyone give me some feedback about potential issues with this configuration or if they have ever heard of such an issue affecting SSRs?

The SSRs that I am using are Crydom PMP series.

 

[berkeman]

The power to each of these heaters are controlled via zero-cross SSRs with each using their own dedicated 4-20mA control signal.

Welcome to PF.

What does this statement mean? do the SSRs stay off for some half-cycles to lower the total current drawn (below the Full Load Amps value)?

 

[berkeman]

Also, how are you monitoring the currents in the 3 phases? With an AC clamp on each, or Hall Effect probes clamped on each?

 

[Phynoxum]

Welcome to PF.

What does this statement mean? do the SSRs stay off for some half-cycles to lower the total current drawn (below the Full Load Amps value)?

A zero-cross detector is an integral function to SSRs which drive them to switch their output when the incident voltage is at the 0V point. This is done to help minimize harmonics. SSRs can also be controlled via phase angle, which is similar to a shutter effect. Phase angle control tends to lead to greater harmonics.

SSRs regulate the amount of current to a load (and thus power). With these particular SSRs, the AC output current is controlled by 4-20mA. For example 4ma = 0% full load current, 12mA = 50% full load current, 20mA = 100% full load current.

The failure that I am seeing is that after some time of running a random sampling of these 5 SSRs will stop responding to the 4-20mA command signal and have their AC output latched OFF (0% full load) regardless of 20mA being applied to their control terminals.

To give further details on this issue, I have another piece of equipment which has only 3 of the 5 heaters equipped using the exact same SSRs and there has never been an issue. In this other equipment heaters A and B are not present.

I have a power data logger with a current clamp on each line and a voltage probe connected to each of the 3 phases. This power data logger also uses the voltage and current data to calculate each harmonic as a % of RMS current.

 

[anorlunda]

There is much unclear from your question.

1. How are the 3 heaters in phase L3-L1 wired up? Series, Parallel?
2. Any unbalanced 3 phase system has positive, negative and zero sequence components. You can calculate the magnitudes. Have you done that?
3. How are you seeing harmonics? What percentage harmonics in which phases?
4. What is the 4-20 ma control signal? What is it doing?
5. Do you have a spec sheet on the relays.
6. What instrumentation are you using, and how is it hooked up?
7. Do the resistive loads cycle, or turn on/off?
8. How many outages in 3 months? Are the outages correlated with any load events, such as peak powers, minimal power?

A schematic of your whole setup, including power, control and instrumentation might be needed to figure out the details.

We would like to help you, but the quality of answers on this forum depends on the quality of the questions. The devil is in the details.

 

[berkeman]

A zero-cross detector is an integral function to SSRs which drive them to switch their output when the incident voltage is at the 0V point. This is done to help minimize harmonics.

Sure, but how is total power throttled from 100% back to 0%? Presumably by only turning on for some half-cycles, right? That will generate some (sub) harmonics, and it will be worse if the 3 SSRs that share the same phase are not coordinated in their half-cycle skipping, I would think.

I am measuring a heavy degree of all harmonics (including DC component)

That would be from the way the half-cycles are being skipped, right?

Do you have access to a Hall Effect clamp-on current probe and an oscilloscope? It might be good to see what the current in that 3rd phase is doing with a high bandwidth current probe...

 

[Phynoxum]

I have 3 of the following current transformers clamped around each of the 3 phase feeder lines.

https://shop.dentinstruments.com/products/hinged-current-transformers

The SSRs have a time base control as you said, similar to duty cycle control for controlling average power over time.
4mA (0% power) = 0 cycles passed per 20 cycles
12mA (50% power) = 10 cycles passed per 20 cycles
20mA (100% power) = 20 cycles passed per 20 cycles

 

[berkeman]

12mA (50% power) = 10 cycles passed per 20 cycles

Are they skipping full cycles or half cycles? Do you have a link to their datasheet? Thanks.

 

[Phynoxum]

There is much unclear from your question.

1. How are the 3 heaters in phase L3-L1 wired up? Series, Parallel? Parallel
2. Any unbalanced 3 phase system has positive, negative and zero sequence components. You can calculate the magnitudes. Have you done that?
3. How are you seeing harmonics? What percentage harmonics in which phases?
4. What is the 4-20 ma control signal? What is it doing?
5. Do you have a spec sheet on the relays.
6. What instrumentation are you using, and how is it hooked up?
7. Do the resistive loads cycle, or turn on/off?
8. How many outages in 3 months? Are the outages correlated with any load events, such as peak powers, minimal power?

A schematic of your whole setup, including power, control and instrumentation might be needed to figure out the details.

We would like to help you, but the quality of answers on this forum depends on the quality of the questions. The devil is in the details.

1. How are the 3 heaters in phase L3-L1 wired up? Series, Parallel? Parallel
2. Any unbalanced 3 phase system has positive, negative and zero sequence components. You can calculate the magnitudes. Have you done that? No, I have not
3. How are you seeing harmonics? Using the power data logger, the harmonics are calculated within the program using the raw current and voltage data from each of the 3 phases.
4. What percentage harmonics in which phases? Voltage harmonics have never been greater than about 5%, however current harmonics have been in excess of 60% on each of the 3 zones at different times. The difficult part about this questions is that the harmonic profile is constantly changing due to each of the SSRs getting a 4-20mA signal that works to change the on/off duty cycle to regulate the temperature.
5. What is the 4-20 ma control signal? 4-20mA essentially controls the duty cycle for the SSR and therefore controls the current and power to the heaters. 4mA = 0% power, 20mA = 100% full power and is proportional in between.
6. Do you have a spec sheet on the relays. http://www.crydom.com/en/products/catalog/pmp-nova22-proportional-control.pdf
7. What instrumentation are you using, and how is it hooked up? The DENT power data logger monitors current and voltage on each of the 3 phases. Current is monitored using current transformers around the feeder lines.
8. Do the resistive loads cycle, or turn on/off? Yes, the SSRs continuously turn off and on to regulate the current to the heaters. The amount they regulate is dependent on the 4-20mA signal. The 4-20mA signal is generated by a controller which monitors temperature, so a PID loop is used to work to maintain temperature using the heaters.
9. How many outages in 3 months? Are the outages correlated with any load events, such as peak powers, minimal power? In 3 months I have seen about 35 outages. Of the 5 zones, sometimes it is all 5 zones, sometimes it is only 1 zone, it is very random which zones get affected. I was unable to find data where the outages occurred at relatively the same time as a peak or valley in either the 4-20mA signal or any AC voltage swings. I also looked into magnetic noise as a potential cause inside my enclosure and have that data if it would help.

I will work on trying to get a schematic of the setup.

 

[Phynoxum]

Are they skipping full cycles or half cycles? Do you have a link to their datasheet? Thanks.

Skipping full cycles

http://www.crydom.com/en/products/catalog/pmp-nova22-proportional-control.pdf

 

[berkeman]

Skipping full cycles

http://www.crydom.com/en/products/catalog/pmp-nova22-proportional-control.pdf

Interesting. I need to think more about how the 3 SSR units' current draw waveforms can align (or misalign) with them at different duty cycles...

 

[anorlunda]

What about ambient temperature? If your are using the 25W model, it approaches 21A limit at a temperature below 40C.

What about the heat sink? The spec sheet says,

 

[Phynoxum]

Here is the current setup/schematic for the AC power system.

Each of the SSRs are driven by their own dedicated 4-20mA control signal and each varies depending on the temperature at the heater. The goal is to control power to the heaters to maintain a set temperature. There are only 3 contactors which serve as a safety disconnect in the event of an over-temperature condition. These have been analyzed as being a possible cause, however were ruled out after multiple times only heating zone E and C stopped working. If the contactor was the cause, we would expect to see a failure with D and E simultaneously.

 

[Phynoxum]

View attachment 287039
What about ambient temperature? If your are using the 25W model, it approaches 21A limit at a temperature below 40C.

What about the heat sink? The spec sheet says,
View attachment 287040

SSR-B and SSR-D are both the PMP2450 versions with larger heat sinks. Temperature was measured using a thermal imager and the temperature on any of the heat sinks at full power or during steady state/PID control were never above 32C.

In reaching out to Crydom for technical support, they suggested the problem was with their control circuit and recommended a new lot which had updated firmware. We replaced the SSRs with the version with updated firmware and the fault still occurred. It is important to note that this failure was never seen on the machine which only uses 3 of the 5 heater zones (zones A and B don't exist on the model which does not have the failure).

 

[DaveE]

Your instrumentation is all pretty low bandwidth. It won't be adequate for diagnosing problems related to switch transients. I think it's worth looking at this with an oscilloscope and their appropriate probes. I am suspicious that your problem is some transient based problem because of your description that the SSRs shut down but can be reset by removing the control input. I suppose this could be an intentional protective response from the SSRs, but I doubt it. You should try contacting the application engineers at Crydom (if such a thing still exists there), and see what they suggest. OTOH, I don't have a good story about why removing a couple of heaters would eliminate the problem.

Is it a single system you are having problems with or multiple?
If it's just one, then I would also check your components and construction. Double check your assumptions. Things like a heater leaking current to ground, for example.

Do all of the SSRs shut down, or just individual ones?

Do the SSRs have snubbers to help with switching transients?

Have you tried removing the extra heaters to make the two systems the same to see if that helps?

 

[anorlunda]

Look at that green line. Even at 32C, the derate is close to 21 A. A thermal explanation is plausible. Might there be an installation error causing thermal resistance >3 C/w? See their installation requirements.

Their derate curves are pretty extreme. I wonder if it is not component protection, but rather an instability in the feedback control loop sensitive to temperature and that thermal resistance parameter R=C/w. A locked up loop sounds consistent to your description of the symptoms. It might not come out of saturation until you do what you said.

can only be recovered by removing the 4-20mA signal completely for some duration of time and then re-applying.

If the phases are all independently regulated, I'm unclear on how much the maximum phase unbalance is. As you know, unbalance is related to harmonics.

 

[Phynoxum]

Your instrumentation is all pretty low bandwidth. It won't be adequate for diagnosing problems related to switch transients. I think it's worth looking at this with an oscilloscope and their appropriate probes. I am suspicious that your problem is some transient based problem because of your description that the SSRs shut down but can be reset by removing the control input. I suppose this could be an intentional protective response from the SSRs, but I doubt it. You should try contacting the application engineers at Crydom (if such a thing still exists there), and see what they suggest. OTOH, I don't have a good story about why removing a couple of heaters would eliminate the problem.

Is it a single system you are having problems with or multiple? I have two systems, one only has heaters C, D and E. The other system has all 5 heaters and is the problem child.
If it's just one, then I would also check your components and construction. Double check your assumptions. Things like a heater leaking current to ground, for example.

Do all of the SSRs shut down, or just individual ones? Sometimes all 5 will fail in a cascading fashion within hours of each other. Sometimes only 1 zone will fail, sometimes 2 zones. Which zones fail seems very random and I haven't really be able to say for sure that heaters connected to L1-L3 fail more than the L2-L3 heater, etc.

Do the SSRs have snubbers to help with switching transients? There is a line RF line filter which supplies the incoming power. That filter was swapped out with a known good and the issues still occur. I do not have any chokes on the system to limit current rise.

Have you tried removing the extra heaters to make the two systems the same to see if that helps? On the unit with 5 heaters, there were no problems recorded when the unit ran with only 3 of 5 zones operational. When running 3/5 zones it was always the zones common to a single contactor which were de-energized. This was the original reason which made me start looking at harmonics. If either heater B or D is physically disconnected, the delta will be missing 1 of its 3 sides. This is my best guess as to why the faults are seen when all 5 zones are running vs. having only 3 of the 5 running. The full delta is not formed unless all 5 heater zones are operational.

Again that is just my SWAG, but I am hoping that some of you fine engineers and physicists might be able to give be better insight on this theory. Given that there is no neutral conductor I am wondering about circulating currents due to the delta connection that is made when all 5 zones are running. When only 3 zones are running, only 2 of the 3 legs have a load and without a neutral I feel like circulating currents can't occur.

 

[DaveE]

In reaching out to Crydom for technical support, they suggested the problem was with their control circuit and recommended a new lot which had updated firmware. We replaced the SSRs with the version with updated firmware and the fault still occurred.

Recognize that the very last thing those guys will say is "why don't you try our competitor's parts". If you are fairly sure it's not a problem with your design, you probably ought to at least start the process of evaluating alternatives. That whole 4-20mA reset behavior looks very suspicious to me.

 

[Phynoxum]

Recognize that the very last thing those guys will say is "why don't you try our competitor's parts". If you are fairly sure it's not a problem with your design, you probably ought to at least start the process of evaluating alternatives. That whole 4-20mA reset behavior looks very suspicious to me.

Yes, that is where I am currently at. I am going with Chromalox SSR1P-2511 type SSRs. They seem to have less "intelligence" packed into the control circuit design so I am hoping they are more reliable. Even if I don't end up seeing the fault after 2-3 weeks with these new SSRs, I still won't sleep well unless I am able to explain why the Crydom SSRs were failing randomly when all 5 zones were running when compared with only running 3 of the 5 zones.

 

[Baluncore]

The problem will be caused by something very simple, so simple that no-one has checked it.
Get someone else to check that the temperature sensing for area X is actually connected to the controller for area X, and to heater for area X. Ask them to “confirm” a subtle unspecified wiring fault. Do not give them a circuit diagram.

 

[Tom.G]

From your description of the problem and of the SSRs, and later from the datasheet, the SSRs are microprocessor controlled whose supply is the Control Input.

Correcting the lock-up by removal of the Control Input forces the processor to reboot from a cold start. That indicates a software crash in the processor.

The two most common reasons for a crash are bad programming and electrical interference. The first, bad programming, has been partially ruled out by the manufacturers admitted bug.

Remaining is electrical interference. I was going to suggest a series R-C Snubber across the switched terminals. I see @DaveE beat me to it. That makes it a 2-for-0 vote to give it a try. Some small inductance in series with the load could also help suppress transients and harmonics. The Snubber capacitors MUST be line-rated, Class Y. The Class Y capacitors are designed to fail Open, whereas Class X capacitors are designed to fail Shorted.

Also try some capacitance across the Control Inputs at the SSRs to bypass any noise pickup there. That is probably the cheapest and easiest first approach. Any paper, film or ceramic type capacitor can be used here.
My first wild guess would be 0.1uF; if that helps but doesn't completely cure the problem, parallel it with a 0.01uF disc ceramic.

Oh, the obvious, keep the Control Input wiring away from the power wiring, and it should be twisted-pair to reduce noise pickup, and maybe shielded too! If this is a typical industrial control panel with 16ga, TW wiring bundled all together, this may be all that's needed.

Please keep us updated on successes/failures.

Hope this helps.

Cheers,
Tom

 

[Phynoxum]

From your description of the problem and of the SSRs, and later from the datasheet, the SSRs are microprocessor controlled whose supply is the Control Input.

Correcting the lock-up by removal of the Control Input forces the processor to reboot from a cold start. That indicates a software crash in the processor.

The two most common reasons for a crash are bad programming and electrical interference. The first, bad programming, has been partially ruled out by the manufacturers admitted bug.

Remaining is electrical interference. I was going to suggest a series R-C Snubber across the switched terminals. I see @DaveE beat me to it. That makes it a 2-for-0 vote to give it a try. Some small inductance in series with the load could also help suppress transients and harmonics. The Snubber capacitors MUST be line-rated, Class Y. The Class Y capacitors are designed to fail Open, whereas Class X capacitors are designed to fail Shorted.

Also try some capacitance across the Control Inputs at the SSRs to bypass any noise pickup there. That is probably the cheapest and easiest first approach. Any paper, film or ceramic type capacitor can be used here.
My first wild guess would be 0.1uF; if that helps but doesn't completely cure the problem, parallel it with a 0.01uF disc ceramic.

Oh, the obvious, keep the Control Input wiring away from the power wiring, and it should be twisted-pair to reduce noise pickup, and maybe shielded too! If this is a typical industrial control panel with 16ga, TW wiring bundled all together, this may be all that's needed.

Please keep us updated on successes/failures.

Hope this helps.

Cheers,
Tom

Yes, thank you all for you feedback and input! I am hoping that this issue might provide me with a learning experience as well. I have never encountered an issue such as this and look forward to finding a root cause.

Right now I am doing testing to create additional magnetic/EMI noise in the vicinity of the SSRs inside the enclosure, along with connecting one of the heaters from L2-L3 to complete the delta. I hope to reproduce the problem on the unit which has never seen this issue.

On the other unit which uses all 5 zones, I changed out the SSRs for a "dumber" version which are not microprocessor based. So far after about 1 week of run time, there have been no issues.

I will definitely keep you all posted, and please let me know if anything else might come to mind.

 

[Phynoxum]

The problem will be caused by something very simple, so simple that no-one has checked it.
Get someone else to check that the temperature sensing for area X is actually connected to the controller for area X, and to heater for area X. Ask them to “confirm” a subtle unspecified wiring fault. Do not give them a circuit diagram.

An independent review of the system was performed by a fellow electrical engineer. There were no wiring or configuration issues found. Additionally, all wiring was checked for tightness (I know that loose terminals can affect operation drastically).

 

[Rive]

It's jut a gut feeling only, but the issue feels like some control circuit reference (ground?) drift. How are the control circuits organized? On what kind of power source they are hanging on? Are they independent or has some common point?

 

[Phynoxum]

It's jut a gut feeling only, but the issue feels like some control circuit reference (ground?) drift. How are the control circuits organized? On what kind of power source they are hanging on? Are they independent or has some common point?

The controller which supplies the 5 independent 4-20mA control circuits were originally connected to a common DC ground (designed to minimize the need to run 10 control wires, and instead only run 6 control wires). This was scrutinized as a potential source of the noise which could have been causing the SSRs to fail, if the DC ground were somehow floating or noisy. The control signal is a 4-20mA current loop, which in my experience are more resistant to EMI noise compared to 0-5V or 0-10V DC control signals which would be more susceptible to drift. The 4-20mA control signals were hooked up to a data logger to monitor their currents continuously and there was never a dropout or loss of control power measured.

To further troubleshoot, we ran 10 independent wires (5x 4-20mA control loops which were not connected to a common bus as they were in the 6 wire configuration). This change had no effect in mitigating the SSR control circuit failures seen. We also conducted a hi-pot test on the entire system to make sure that there were no broken wires or leakage to ground which might be causing ground fault issues.

 

[Dullard]

I'm with Tom G.

In the mean time:

If you're running PID from a PLC, you obviously have temperature feedback. It shouldn't be a big deal to detect when the condition occurs and briefly take the 4-20 output to '0' (some PLCs can't go below 4 mA). You could add a relay(s) to open the loop(s) if your PLC can't get to zero. That should change this from a disaster to an annoyance while you run it down. A 'Gordian knot' approach, but... Production must go on!

 

[Baluncore]

The controller which supplies the 5 independent 4-20mA control circuits were originally connected to a common DC ground (designed to minimize the need to run 10 control wires, and instead only run 6 control wires).

I assume the control circuits of the SSR modules are all powered by one shared DC supply.
How is the 0V (ground) of that supply connected to the 0V (ground) of the controller that generates the loop currents?