TPS2068DGNR
Product Specifications
TPS2068DGNR
Texas Instruments — Power Management & Signal Chain
Product Overview
The TPS2068DGNR from Texas Instruments is a power management / signal chain IC delivering high efficiency, precision regulation, and robust protection. Texas Instruments (TI) is a global leader in analog ICs, power management solutions, and embedded processors.
Superb Automation partners with leading brands including Qorvo, TI, ADI, XILINX, NVIDIA, Intel, ST, MACOM, Renesas, and Coilcraft to provide complete BOM and supply-chain solutions.
Key Features
- High-efficiency power conversion with low quiescent current
- Wide input voltage range with precision output regulation
- Integrated over-current, over-temperature, and short-circuit protection
- Small footprint packaging for space-constrained designs
- AEC-Q100 qualified options available for automotive applications
- Flexible sequencing and soft-start control
Key Specifications
| Input Voltage | V |
| Output Voltage | V |
| Output Current | A |
| Switching Frequency | kHz |
| Efficiency | % |
| Quiescent Current | µA |
| Topology | — |
| Package | — |
| Operating Temp | °C |
| Protection | — |
Applications
Request Quote — In Stock, 100% New & Original, Full Warranty
TPS2068DGNR single channel 1.5 A current limited power distribution switch in the 8 pin HVSSOP body
The TPS2068DGNR is the single channel 1.5 A member of the TPS206x power distribution family, in the eight pin HVSSOP body with an exposed thermal pad. It is the part a board uses between a 5 V or 3.3 V rail and a port that a user can plug into, short, or unplug while the product is running. One switch, one enable, one fault flag, and a fixed current limit that the vendor publishes as a window rather than a single number.
The power element is a 70 mOhm N channel MOSFET driven by an internal charge pump, so the switch works from 2.7 V and needs no external gate supply. When the load pulls more than the limit the part moves into constant current instead of opening, the overcurrent flag goes low after a short blanking delay, and a thermal circuit turns the switch off if the dissipation keeps climbing. Nothing in that chain needs a processor to be alive, which is the reason this class of part survives a field short that the firmware never sees.
Key parameters taken from the TPS2068DGNR document
| Rated continuous current | 1.5 A through one switch, with the thermal pad soldered to copper |
| Input range | 2.7 V to 5.5 V, which covers a 3.3 V rail with margin and a 5 V rail at minus ten per cent |
| On resistance at 5 V | 70 mOhm typical and 115 mOhm maximum over the full temperature range |
| On resistance at 3.3 V | 70 mOhm typical and 125 mOhm maximum |
| On resistance at 2.7 V | 75 mOhm typical, the price of a weaker gate drive at the bottom of the range |
| Current limit | 1.6 A minimum, 2.1 A typical and 2.6 A maximum, measured with the output shorted to ground and the device enabled into the short |
| Rise time | 0.6 ms typical into a heavy capacitive load, set internally by the charge pump |
| Fall time | About 10 ms into 100 microfarad with a 5 Ohm load |
| Overcurrent flag | Open drain, active low, with a 4 to 15 ms deglitch window so a hot plug does not report a fault |
| Undervoltage lockout | The switch stays off below roughly 2 V and the threshold row is printed as 2 to 2.5 V, so hot insertion cannot drive the output before the rail is valid |
| Thermal shutdown | 135 C trip, 125 C recovery, 10 C of hysteresis, and the vendor notes the circuit only acts under overcurrent |
| Standby current | One microamp maximum with the switch off |
| Reverse current | Blocked while the switch is disabled, so a charged output cannot feed back into the rail |
| Body and packing | HVSSOP DGN 8 pins with exposed pad, 2500 pieces on a large reel, NIPDAU finish, moisture level 1 with unlimited floor life |
| Approvals | UL listed under file E169910 and CB certified for the DGN members |
The table above is read from the vendor document, not from a distributor summary. Where a window is printed, the window is what a production test should use, and the typical value is only a starting point for the first article.
Where the numbers matter in a real build
The current limit window is wider than most people expect. The published rows are 1.6 A minimum, 2.1 A typical and 2.6 A maximum, on a part rated 1.5 A continuous. A load that draws 2.4 A will therefore pass on one unit and limit on another, and a load that draws 1.7 A sits inside the minimum of the window and is guaranteed to pass. The practical rule is that a port should be fused by the part for a load above 2.6 A, and anything between 1.6 A and 2.6 A is a margin question that has to be answered with measurement on the actual board rather than with the typical value.
The thermal path is the real rating. One and a half amps through 115 mOhm is 259 mW, and the DGN body with a soldered pad has a junction to ambient figure in the order of 50 C per watt on a modest copper area, which is a 13 C rise. Without the pad soldered the same body behaves like a plastic package with a thermal resistance several times higher, and the part reaches thermal shutdown at a load it should carry. Every board we see that reports a switch running hot has a pad with vias to an inner plane missing, or a pad connected to a plane that is itself isolated.
The blanking window explains most of the fault reports that look like false positives. A hot plug into a discharged 100 microfarad load draws a current well above the limit for a short time, and the part limits that current rather than reporting it. The flag only asserts after the deglitch window, which the vendor prints as 4 ms minimum, 8 ms typical and 15 ms maximum. A firmware that polls the flag every 10 ms can therefore miss the first assertion on one unit and catch it on another, which is not a part defect but a sampling artefact. Latching the flag in hardware, or polling faster than 4 ms, removes the ambiguity.
The temperature range in the document and the temperature range in the addendum do not match, and the difference is worth knowing. The feature list on page one states minus 40 to plus 85 C for the DGN members and 0 to 70 C for the D package, while the package addendum prints minus 40 to plus 105 C against the same order code. The addendum is the newer statement and it is the one a customer qualification document should quote, but a board that has to run at 100 C ambient should confirm the figure with the vendor before it signs, because the electrical tables are characterised over the narrower window.
- Inrush into the port capacitance, measured with the actual cable and the actual load capacitor, not with a bench resistor.
- The current limit at the hot and cold corners, since the window is 1.6 to 2.6 A and the load tolerance has to sit inside it.
- Fault flag timing against the firmware polling interval, because the deglitch window is 4 to 15 ms.
- Junction temperature with the pad soldered and with the copper area the production board actually has.
- Reverse blocking with a charged output and the rail removed, which is the case a battery powered product sees at every power down.
- Standby current of the whole port circuit, since the switch itself is one microamp and the divider or the flag pull up can be more.
Which projects this part fits
This part fits USB ports on industrial gateways and instruments, protected auxiliary rails on telecom line cards, sensor supplies that a host has to be able to reset, and any single port where 1.5 A is the ceiling and the board cannot afford a controller plus an external FET. It is also the part to reach for when a product already uses the dual member of the family and one of the two ports has been dropped in a later revision, because the pin out and the behaviour are the same and only the packaging changes.
Questions that come up before a purchase order is released
Does the disable path need a pull up or pull down on the enable pin?
The enable is active high on this member and the input is high impedance, so it must not float. Tie it to the rail through a resistor if a processor pin can be high impedance during reset, otherwise the port can come up on its own while the firmware believes it is off.
Can two outputs be paralleled to reach 3 A?
No. Two parts in parallel do not share the current evenly, because the one with the lower on resistance carries more until it limits, and the pair then delivers less than the sum while the hotter one reports a fault. Use a single part rated for the current, or a controller with an external FET.
What happens when the output is shorted while the switch is on?
The part moves into constant current at 1.6 to 2.6 A, the flag asserts after the deglitch window, and the dissipation raises the junction until thermal shutdown trips at 135 C. Recovery is automatic at 125 C, so a permanent short produces a hiccup pattern rather than a latched off port.
Is the exposed pad electrical or only thermal?
It is the thermal pad and it is not a signal node. It must be soldered to a copper area, ideally with vias to an inner ground plane, and it must not be connected to the output or to any node other than the one the vendor shows in its land pattern.
How much input capacitance does the port need?
The vendor application shows a ceramic capacitor at the input and a larger bulk capacitor at the output. The input capacitor has to supply the inrush pulse before the switch limits, and the output capacitor sets the rise time the load sees. A port with no output capacitor and a long cable sees the cable inductance ring against the switch, which is the case where a small series ferrite is worth the board area.
How SUPERB handles this line
We supply the TPS2068DGNR on original 2500 piece reels with the date code and the marking recorded on the packing list, and we can supply the D package member of the same family when a board has the room and does not need the minus 40 C floor. The power management cross reference page lists the single and dual members of this family by rating and body, BOM consolidation covers the case where several boards in one product family use different members of the same switch line, and kitting service is how we deliver a mixed set of these switches against one build.