Three Phase Contactor Testing

Introduction

This blog is an extension of the contactor testing I have carried out recently utilising the Keysight DAQ970A. It goes into the testing arrangements more and the results obtained, so that the Roadtest can concentrate more on the performance of the instrument.

DAQ970A Roadtest Review

I am particularly interested in testing the performance of used contactors, to see how they have deteriorated from a new contactor. Contactor assessment in the past, has been predominantly visual, looking at the level of damage to contact surfaces and arc chutes to determine their need for replacement. This was always a judgement call based upon experience. I aim to carry out these tests to ascertain if the decision to replace the contactor would be justified.

Contactors Tested

I have a number of contactors available to me for testing, predominantly from Schiele and Telemecanique. The contactors of this variety are just electrically operated three phase switches, usually utilised in motor control centres to provide power to pumps and fans. The rating of these contactors are 30kW (80A) and 50kW (130A), in the UK, new ones would retail for around £130 and £170 respectively.

The contactors I am using have come from either a main contactor from direct online starter (pictured above) or the star contactor from a star delta starter. Star contactors are quite interesting as some panel manufacturers, with deliberately install an undersized contactor, as the star contactor is only in circuit for 3 to 10 seconds whilst the motor runs up to speed. This allows them to reduce panel costs, but makes this contactor more sacrificial. The typical configuration of a Star-Delta starter is shown below, showing the undersized star contactor at the bottom of the panel.

The contactors will have seen service for 10 to 20 years. Direct online contactors will generally have a lower number of starts, but much longer running hours. Star contactors have few running hours, but significantly more starts. This changes the failure modes of the contactors, with more contact wear being seen on star contactors and more overall heat damage observed on direct online contactors.

Visual inspection

The contactors were partially dismantled to allow the thermocouples to be installed and this gives the opportunity to carry out a visual inspection. There were obviously no issues around the new contactors.

On the Telemcanique contactor, a pattern can be seen on the surface of the contact that is designed to reduce the arcing during the switching operation. The older style shield contactor does not have this, the contacts have a completely flat surface.

Both contactors also have a different style of arc chute, with the Schiele contactor possessing a more traditional multi-plate arc chute against the solid plate of the Telemecanique contactor. These differences are due to advances in contactor design over the years and the different approaches adopted by different manufacturers. Another notable feature is the full IP2x shrouding of the connections on later contactors, effectively making them 'finger proof, improving the safety when working inside energised panels.

The used contactors show varying different degrees and type of wear to the contacts and the contactors in general.

Contactors in direct online starters that are energised of long periods of time, have little damage to the contact surfaces but tend to suffer from heat damage. The screw connections can clearly seen to be discoloured in comparison to the new and star contactors. Flipping the contactor top over and inspecting the arc chutes reveals that they are practically new, with no signs of carbon contamination or contact spatter. Taking a look at the coil inside a DOL contactor, reveals that it too is suffering from heat, the covering has become brown and brittle. It is not uncommon to find this covering peeling off on very old contactors.

I have not really seen any issues with any of the DOL contactors that would necessitate the replacement of them, so I am not expecting the test results to show up any issues.

The same visual inspection was carried out on the Star contactor. Opening this contactor up reveals a different story, and there are obvious signs that there is significant deterioration to this contactor, to the extent that there are signs of this deterioration on the outside of the contactor as well.

The amount of carbon deposits within the contactor is significant. The deposits will make the surfaces of the moulding appear conductive, and enough build up could eventually lead to a flash over within between the phases. All the arc chutes are contaminated with the same carbon deposits, this reduces the effectiveness of the arc chute, that allows further deterioration of the contact themselves if the arc is not adequately quenched. The level of contamination, is starting to show through the top lid of the contactor. These carbon deposits can be cleaned off, to extend the life of the contactor.

Comparing a pair of the contacts from the DOL and star contactor side by side, shows the different types of deterioration. Very little over heating is seen on the star contact, but there are plenty of carbon deposits and spatter on the contact surface where globules of the contact have been melted and reformed at the edge, where the arc blast in towards the chutes. Conversely the DOL contactor has none of these deposits, but clearly shows heating around the screw connection.

One pair of contacts on the star contactor showed deep physical damage. One section of the contact pad has been melted away and it looks like it is starting to peel away from the copper connection plate.

I would condemn a contactor that is this kind of condition, and I am fully expecting this contactor to show higher operating temperatures and contact resistance on comparison to the DOL and new contactors.

Tests Conducted

I tried to collate as much test data as I could for the tests carried out. The Keysight DAQ970A is ideal for this, allowing all of the temperature, current and voltages needed to be collecting at the same time. Speed of data acquisition is not an issue for these tests.

I carried out load tests on each contactor for 10 minute durations at three different loads of 25A, 50A and 75A. Whilst my test set could deliver up to 100A, the cabling I had was rated at 70A, so I did not want to go up to that level of current.

I then carried out 30 cyclic operations of each contactor, switching a 10A load. This particular test was a response to observations made whilst carrying out the load tests, that the contact resistance was changing with every operation of the contactor. This test was conducted by manual operation of the coil voltage, so is subject to the odd human error or two.

Coil condition was assessed by ramping the voltage applied to the coil between 0.8 and 1.1 time its nominal rating and monitoring the current and contact resistance at every voltage adjustment. The Telemecanique contactor lent itself to having an adapter fitted to its auxiliary attachment that allowed the pressure on the contacts to be represented. As the spring loading was different between the pressure measuring rig and the actual contact mechanism, this isn't a direct measurement of the pressure on the contacts.

Above is how the test arrangement looks in practice. The coil voltage is controlled via an AC variable source, so that the adjustments can be made. A DC current injection test set provides the load current, in favour of an AC test, as I did not have an appropriate current transformer to connected into the DAQ970.

Al the data is recorded on the BenchVue DAQ application and then exported to Excel for further analysis. The real time display of the data on the computer and the data on the channel being monitored by the DAQ970 allow the operator to see if there are any immediate issues with any of the tests and act accordingly.

Test results

The following plots show the results of the load tests on some of the contactors.

The first plot shows the average temperature rise over the duration of each load test. Not surprisingly, the higher the current is, then the higher the temperature of the contacts gets to. The second plot shows how the three different phases of one contactor respond during a 75A load test. As can be seen, this isn't balanced across the three phases with Phase 2 clearly running at a higher temperature.

The plot below shows the temperature of each individual contact for the Star contactor. The contact found damaged in the visual inspection is represented by T3 on the plot. This is operating at the mid-range of temperatures and is not consistent with expectations, that would have placed it as having the highest temperature.

The left plot below provides a comparison of the contact temperature reached for the new Telemecanique and DOL Schiele contactors. There isn't really much between them. The DOL contactor shows slightly lower temperatures for two of the phases, which may be because the contacts are bedded in and match one another better. The right hand plot, is a comparison of the contact temperature rise of all the contactors tested.

The temperature rise is fairly even across all the contacts with a notable exception for B Phase on the DOL starter contactor and to some extent A Phase of the star contactor. This is also not following expectations, where a faster rise to higher temperatures should have been seen for the used contactors.

Another aspect of contacts during maintenance is not to mix up contacts, as they bed into one another and become 'paired'. I decided to test this by swapping around the contacts in the Star contactor. Each moving contact was moved on a phase and then rotated 180 degrees before being re-installed. The expectation was that increased temperature and resistance would be observed. The temperature plot for this test is below and all phases can be seen to stay below 50 DegC, which is lower than all the tests on the other contactors.

Contact resistance was measured and compared across the various tests.

Thew first plot shows the variation of the resistance over the duration of a 75A load test. After the initial temperature rise, two of the phases settle down and the resistance remains stable for the duration of the test. Phase A of the contactor shows a little more variation and also has a higher resistance value, this may be the underlying reason as to why it also saw more deviation.

Maximum contact resistance during the load tests follows a similar pattern to the temperature rise plot, which would be as expected.

The two remaining plots compare the contact resistance across the contactor types and the level of deviation across this values. The DOL contactor that had the higher resistance values, saw the most deviation in those values. This was against expectations from the visual inspection that suspected the highest resistance would be seen on the damaged contact of the Star contactor. This was Phase C, and actually had one of the lowest calculated resistance across all the contacts.

As the histogram plots show, this was not the case and the results show some of the lowest resistances, but also the most tightly grouped set of resistances were seen across all the tests.

The new contactor showed a sporadic set of resistance values, with predominantly low values and a number of excursions into higher values. This would be expected behaviour for contacts that had not bedded in to one another. In contrast, the DOL contactor showed the majority of readings were in a tight spread, towards its lowest set of values. There were a few excursion out to the opposite end of the chart for comparatively very high values of resistance. A bedded in contactor would be expected to have a more stable set of resistance values.

The star contactor showed a wider spread of values, moving more towards a typical bell curve seen in statistical analysis. More readings appeared to be at the lower end of the values, indicating that they were still somewhat bedded into one another. There were however, an increasing number of excursion towards higher resistance values, typical of contacts that are ageing. The results though, do still seem to be exceptionally good in comparison to the visual assessment of the contacts.

As expected, changes in the contact pressure were seen during the cyclic operation of the contactor, but these changes did not match the pattern in deviation of the contact resistances. Therefore, for this test, there seemed little reliance on good contact resistance in relation to contact pressure.

Variation of the coil voltage was carried out to ascertain if there were any changes in the pressure applied to the contacts, or the contact resistance.

The plots show very little change as the coil voltage is varied. The main concern would have been increases in resistance seen as the coil voltage was reduced. But the pressure observed remained constant across the voltage range and the actual contact resistance value slightly decreased, probably as the contact moved slightly due to thermal expansion. This would indicate that no detrimental behaviour would besides in a contactor due to voltage dips, generally seen as large electrical loads are switched on in industrial plants.

Part way through raising the coil voltage an upward spike in contact resistance was seen on all three phases. The reason for this unknown, a corresponding, much smaller spike was also seen on the contact pressure. The pattern from then on followed a slow decreasing resistance value as seen for all the other tests.

Conclusions

These contactor tests have been very interesting to me. The expectations based upon my training and previous experiences did not match these particular test results. Contacts that would have been condemned actually seemed to have good electrical characteristics, indicating that they had not actually reached the end of their life. Some caution must be applied here, as the test set is very small, and I would like to continue with these tests by obtaining more contactors, particularly star contactors, that always seem to have the most damage, due to their under-rating by the panel manufacturers.

However, the test results do justify the use of an unrated contactor for star contactors and is not necessarily a cheap panel building option.

Pressure contact variation was not really evident over the coil voltage range and as a consequence, contact resistance was reasonably stable. Larger variation in contact resistance was seen over cyclic operation of the contactor, but there was no correlation between resistance and pressure applied, probably due to the mechanical construction of the moving contact arrangement that applied pressure through quite strong compression springs.

As more contactors become available to me, I will carry out further testing. I would also like to obtain an actuator card for the DAQ970A, so that the cyclic operation of the contactors can be carried out as a control sequence instead manually.