The fundamental problem of managing hot-swap products is the in-rush current when a board is inserted into a backplane. The concept for the TPS2392 and TPS2393 is based on managing current. Managing that current during insertion and providing features that manage the current during operation are the challenges. Offering solutions for those challenges, along with the high voltage process that is capable of handling 100 V transients, provides a general overview of the capabilities of these products.
To manage an external power FET and the current coming into a board during insertion, TI uses a current ramp controller. That technology can manage a wide range of loads as opposed to a voltage ramp design that focuses on a specific load. These products manage the in-rush current directly so the system doesnęt see any spikes in the current.
The TPS2392/2393 modulate the external power FET with a linear current amp that is integrated on the IC and used to ramp the current. This provides precise control of the current and allows you to select the ramp through the Iramp pin. When you select the ramp rate you charge the reservoir capacitor of the DC-DC controller on the board to eliminate the in-rush current, and that provides a very high slew rate. Additionally, very few external components are required to achieve control of the current.
The basic need in an application that has high voltage and low current is to change the high voltage to low voltage and high current. To do that you need a very high frequency and very low impedance for the demands that occur after the DC to DC controller. The TPS2392 and TPS2393 provide the capability to charge and manage current very quickly. They donęt need the typical resistors and capacitors on the FET gates required by other designs to tailor the signal to a specific capacitive load. These extra components can also bog down the circuit.
The management of current enables these products to tolerate nuisance events or spikes. If you get a spike during operation, this product will set a pin called the fault timer. When a spike is detected the circuitry uses a fault timer that has a capacitor, with a pin to ground, allowing you to program the time duration of a fault. For example, you might decide that you want to set the timer so it goes into a fault condition with any signal more than 10 microseconds long. The fault timer lets you eliminate these nuisance events. It uses the internal linear current amp to clamp the signal and modulate the external power FET to ride through the 10 microsecond duration without going into a fault condition. If the fault stays longer than that time duration then the circuit goes into a latch mode or a retry mode. This helps manage the event so there isnęt a jump in energy during the in-rush occurrence. Additionally, if a crowbar event occurs (a dead short) or twice the current fault happens then a circuit break is performed. Therfore, these products manage events that donęt have a lot of energy and those that do and can cause a dead short.
Other designs donęt use the fault timer that TI uses. Typically, designs by other companies set up external circuitry for a fault timer. According to TI, you need to use more than just an RC network to make the fault timer. They say you need to clamp the current by using an internal linear current amp to modulate the external power FET. Specifically, there are two comparators that are constantly looking at the current going across the sense resistor. One of them is a linear current amp and the other is a very fast comparator. The linear current amp handles the management of the ramping during insertion, and operation during the nuisance event. The other comparator checks the sense resistor and trips when it detects a high current across the circuit. TI says this design is able to manage events directly and that gives the product an advantage of gentle power operation during insertion and helps the circuitry decide what to do when a nuisance event occurs.
You may have wondered why this technique hasnęt been used by others since it is so stable. It may be because the linear current amp takes experience to develop, and very few companies have the know-how. Call it fortuitous or call it good planning, but TI bought Unitrode, a company that developed the first IC for hot swap, about three years ago. Unitrode also had knowledge about power management using a linear current amp design. According to TI, other companies donęt have this design because using a linear current amp is more difficult than a design that uses a constant current source with a power FET. TI says the constant current design is easy to make because all you do is turn on an internal FET that trickles 10 micro amps and allows the FET to charge. Unfortunately, this design provides no control of the FET. TI says that the linear current amp approach is tricky and requires experience. The trick, they say, is to get the circuit to constantly check the current across the sense resistor and modulate the FET appropriately. Once you figure it out you get products that are very stabile and manage the current very precisely.
However, customers told TI they wanted more - they also wanted a part that was easy to use. Unitrode (before the acquisition) probably had similar requests but, for whatever reason, didnęt take the time to make it. Thatęs where the bankroll and experience of TI came in to make the acquisition of Unitrode pay off. They realized it made good business sense to invest in a technically superior product line that more people would use if they didnęt need an analog guru to implement the design.
The TPS2392 and TPS2393 will be very attractive to the marketplace with their capability to manage the current during board insertion and board operation, and the simplicity of setting it up for the application. It takes a few simple equations to calculate the value of the capacitor on the Iramp, calculate the trip current, and calculate the value for the sense resistors.
Finally, the -48 V function of these products will obviously be used in the telecom market but it will also find a home in optical networking and base station designs. The telephone companies will be attracted to the full complement of features offered but the other industries will like its simple-to-use approach and the smaller footprint it provides for their high density boards.
TPS2392 and TPS2393 are priced at $1.95 per 1,000 units, and are available today from TI and its authorized distributors in North America, Europe, Asia and Japan.
TPS2392
TPS2393