Hello,
We are going to use the TPS2492 for current protection in a security system for our company. The inputpower is always on and the loads are always on. (PIR detectors and other divices).
We have test the evaluation board (EVM-003) for 48V. Power supply is at that moment 47 Volt and there is now load. S1 switch is on Enable.
I provide a load so the current is 2A. Everything is OK.
When I trigger the S1 Switch to Disable the power falls down, still OK.
When I trigger the S1 switch to Enable again, the TPS2492 direct goes to Fault. Is this normal? The currentload is at that moment 2A.
I can only restart the TPS2492 when I take away the load of 2A and trigger the S1 switch (Disable and Enable) and reconnect the load.
This is not possible in our application, the load and the input power are always on.
How can we resolve this problem?
Desmedt Augustin
Product Development
augustin.desmedt@colruyt.be
A common mistake is to use a constant current load (an active load) on the output of a hotswap and expect normal behavior. This is usually a problem for several reasons: 1) The loads typicaly act like a short when voltage is first applied until their loops react, 2) Most real loads are not constant current, 3) Minimal hotswap solutions are achieved by keeping the load off while load capacitance is charged.
More than likely your load's power I/V characteristic has an intersection with the source's output I/V characteristic at a V/I point other than along the fully on point.
The hotswap output characteristic is bounded by the input voltage until current limit is active (50mV/0.01Ohm = 5A), then by the 5A line until the power limit is active (P = 10*Ilim*Vprog = 26W). Note that this is a type of foldback.
If your load was a resistor (r = 48V/2A = 24Ohms), its input characteristic is I = V/24.
Note that I = P/V = V/R or 26/V = V/24, or V = 25V. This is most likely the point at which the output charge locks up and the timer terminates, latching the TPS2492 off.
Assuming the hotswap was proper for the application (not just the EVM), the usual solution to the load-line lockup is to hold the converter off with PG to separate the capacitor charge operation from normal operation. The next option is to enlarge the output IV characteristic to avoid the lockup
Solve the startup by:
1) Creating a set of load lines that do not intersect at an intermediate voltage a) Higher P limit for shorter time b) a) might take a larger MOSFET, or better heatsinking, or lower starting ambient c) Minimize output capacitance d) Minimize the timer period (the EVM period looks like it can be smaller) as a higher MOSFET stress can be tolerated for a shorter time
The design tool (http://www.ti.com/tool/tps2480-81-90-91-92-93_calc) and simulation model (http://www.ti.com/litv/zip/slum211) can be of great help.
See also http://www.ti.com/litv/pdf/slva158