Understanding PXI Switch Module Current Capacity
Pickering Interfaces manufactures a large range of PXI modules, some of which have high current handling capabilities. Individual switch paths will have a specified current capacity—based on whether the path is hot switched or cold switched (switched when no voltage is applied) and the relay that is being used. Under hot switch conditions capacity, may be limited when switching high voltages—sometimes limited by power but often with more aggressive reductions as DC voltage increases.
There is however, other limits imposed at the module level. Limits could be applied because:
- The signal power dissipated in the switches, plus the power required to provide the PXI interface and coil drives exceeds the recommended power dissipation in a PXI slot
- Components which are close together can all dissipate heat and create hot spots under certain conditions
- Components within the module reach their temperature limit because restricted heat sinking when other surrounding components are also dissipating power
- Restriction of airflow by large components reduces the ability of a module to dissipate the full power of a slot
The PXI standard recommends a slot power dissipation of 25W but indicates no basis for this. For example, a maximum temperature rise or how much heat might spill into an adjacent slot. Modern chassis can provide cooling for more than 25W on some slots, but if all slots were doing so, the chassis might struggle. If it is just one or two well-spaced out slots dissipating high power, the chassis will cope better than if they are adjacent to each other.
When designing modules with relatively high power loads, we always test for temperature rise and keep these consistent with the specified operating temperature limit of 55C. Different parts of the module can have a ,different temperature rise, the rating at a module level is whatever the weakest link in the design is. To find out what the module limits are we have to test.
How We TestTo test a module we first build a representative sample of the most stressed version that is being offered. We then conduct a series of tests, the most significant of which for understanding current ratings is based on a series of temperature sensors.
We measure temperature rise under a set of conditions with multiple paths carrying current. We measure the temperature rise and the time constant of the temperature rise. Usually, we know where to place the temperature probes, but we always experiment to find the worst locations.
We also measure the path resistance so we can see the spread of resistance values that a user might encounter; this is important because the power loss is directly proportional to the path resistance, and any increase has an impact on the module thermal load. We also have to take into account the aging of the relays, which varies with different types.
Design CompromisesIn principle, it is possible to design a switching module that can carry the maximum rated current on all paths at the same, for example, an array of 10A SPST switches. However, to do that might restrict the number of channels that are supported—and that might not be in the users best interest because it increases cost. In many test systems, the required current specification is driven by just a few paths at any one time, those paths might vary as to where they route through the switching system, but at any one time many paths operate at less than the maximum current.
For lower current switching systems (typically 2A and lower) many do not have restrictions simply because it is not possible to establish enough paths carrying the rated current to create any issues in the product. An x8 matrix, for example, can only concurrently have 8 paths carrying full current, a multiplexer can only carry full current on one selected path, so the total number of paths are limited by the number of multiplexer banks.
Wherever there are module-level restrictions, we identify them on the datasheet.
Module Level Restrictions.
On older datasheets we identify the total number of channels that can carry full current assuming all other paths are carrying none. This is not very informative for use cases with mixed current loads. On more recent datasheets we indicate the sum of the squares of the current.
The sum of the squares route of the currents is used because if the switch system has reasonably consistent path resistance between the available paths, then the sum of the squares of the current is related to the power loss in the matrix under many different load conditions.
To take a specific example, if an SPST switch system rated at 16A per channel has 16 SPST switches and is rated at 1536 Amps squared then some use case some examples:
- Example 1. 6 channels can operate at full current with the remainder carrying no current (each channel is 256 Amps squared, 6*256 is 1536 Amps squared)
- Example 2. 16 channels can operate at 9.79 Amps at the same time (9.79 squared is 95.84 Amps squared, multiplying by 16 channels results in 1533 Amps squared)
- Example 3. To estimate how many channels can carry 12A at the same time each channel is 144 Amps squared, which factored into a 1536 Amps squared for the module indicates 10 channels can support 12A at the same time (1440 Amps squared)
Not all modules can respond to this approach, it works best when the variance between channel path resistances is lowest. The greater the spread in path resistance between channels, the less likely that it is the power loss in the module due to path resistance corresponds to a simple estimate like this. The typical resistance is used for a channel rather than the highest resistance value since there is always going to be a spread of values involved.
Where the channel path resistance is more variable a more conservative approach has to be taken by assuming that the higher resistance channels are the ones the user elects to use.
For older style datasheets that indicate the number of channels that can carry full current, the sum of the squares current can be derived directly from that information to calculate the allowed current squared rating.
Short Term Overloads and Thermal Time ConstantThe thermal inertia of a switching system means that it can be more highly loaded for short time intervals provided the average remains within limits and the time that this occurs for is not too long. That is why knowledge of the thermal time constant in a typical Pickering Interfaces chassis is useful.
If the same 16A SPST module carried 16A on all channels for 30 seconds and then nothing for 90 seconds, then over the 120 seconds, the current squared average is 1024 Amps squared. If the thermal time constant is 300 seconds, then the switching system is unlikely to have a thermal overload issue.
For that reason, knowledge of the thermal time constant can be very useful and is included in some datasheets, we can provide additional information for most designs as we record the temperature change with time under load conditions in a typical chassis