SN65EPT21DGKR
Product Specifications
SN65EPT21DGKR
Texas Instruments — Power Management & Signal Chain
Product Overview
The SN65EPT21DGKR 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
SN65EPT21DGKR: PECL or LVDS to TTL translator with a one nanosecond delay and a mid rail reference on pin
SN65EPT21DGKR is a differential receiver that turns an LVPECL or LVDS pair into a single ended TTL output, with an internal reference that holds an unused input at the middle of the supply in a VSSOP body of 8 pins. It runs from 3.0 V to 3.6 V, translates at rates beyond 300 MHz with a propagation delay of one nanosecond, drives 24 mA of TTL output current, and carries an on chip reference at about 2.0 V for terminating an unused or an AC coupled input. The part is drop in compatible with the MC10EPT21 and the MC100EPT21, so a board built around the older emitter coupled logic part takes this one without a change to the copper.
Order code and package
| SN65EPT21DGKR | VSSOP DGK body, 8 pins, 2500 pieces on a large reel, Active in the addendum dated 10 August 2026, NIPDAU finish, moisture level 1 with no floor life limit, marking SSSI, minus 40 to plus 85 degrees |
| SN65EPT21DGK | the same body in an 80 piece tube, which is the row a prototype or a repair order comes from and the row to quote when a customer needs twenty pieces |
| SN65EPT21D | the SOIC D body, 8 pins, in a 75 piece tube. This document prints the SOIC in tube only, so a customer who needs this part on a reel gets the VSSOP body whether or not the footprint suits |
| SN65EPT21DGKG4 | the lead free declaration row of the tube, printed with the same NIPDAU finish and the same quantity, which is the string a material declaration has to name |
| MC10EPT21 and MC100EPT21 | the older emitter coupled logic parts this device replaces pin for pin. Both are still quoted in the market, and a substitution in either direction is a footprint match with an electrical comparison that has to be done |
Electrical parameters that decide the application
| Supply | 3.0 V to 3.6 V, with ground at 0 V, so the part sits on the same 3.3 V rail as the logic it talks to and needs no second supply |
| Speed | operation beyond 300 MHz, which is the figure the features list gives for the maximum frequency and the one to quote against a clock distribution requirement |
| Propagation delay | 1 ns typical, and the electrical table prints 1000 ps minimum, 1394 to 1481 ps typical and 1800 ps maximum depending on the temperature column, so the honest planning number is 1.8 ns worst case |
| Output drive | 24 mA of sink current at a 0.5 V low level maximum, and 3 mA of source current at a 2.4 V high level minimum, which is enough for two or three TTL loads or one terminated line |
| Output levels | TTL compatible, with the high level at 2.4 V minimum under a 3 mA source and the low level at 0.5 V maximum under a 24 mA sink, both over the full temperature range |
| Reference output | the VBB pin holds about 2.0 V, printed as 1910 mV minimum, 2009 mV typical and 2160 mV maximum, and it can source or sink no more than 0.5 mA, so it biases an input and does not power anything |
| Input standard | differential LVPECL or LVDS levels, with the internal circuitry holding an open input pair at the middle of the supply so an unconnected input does not float into an undefined state |
| Edge rates | 94 to 250 ps output rise and fall times depending on the temperature column, which is fast enough that a series resistor at the output is normally needed to keep the trace from ringing |
| Supply current | 8 mA typical and 26 mA maximum with the output high, and 7.5 mA with the output low, so the part costs about 85 mW at 3.3 V |
| Short circuit current | minus 100 mA to minus 180 mA into a short, which is a destructive current if it is held, so a series resistor or a controlled load is part of a robust arrangement |
| Temperature | minus 40 to plus 85 degrees on every printed row, with an internal temperature compensation network that holds the output levels across that range |
| Body size | 3.00 mm by 3.00 mm in the VSSOP against 4.90 mm by 3.91 mm in the SOIC, which on a dense backplane card is the difference between fitting and not fitting |
Family variants and how they differ
| EPT21 against the MC10EPT21 | pin compatible and functionally the same translator, with the TI part specified over minus 40 to plus 85 and the older part over a commercial range on most grades. A board that has been in production on the older number has a documented second source here |
| Against a resistor divider | a divider plus a comparator translates levels for a fraction of the cost at rates up to a few tens of megahertz. Above that the comparator delay and its jitter dominate, which is where this part starts to be the only answer |
| Against an LVDS receiver | an LVDS receiver produces an LVDS output. This part produces a TTL output, so it is the interface between a differential pair and a single ended logic input rather than a repeater |
| Against a PECL to LVDS translator | the SN65LVDS101 goes the other way, from PECL to an LVDS output, so a link that has to stay differential uses that part and a link that has to land on a TTL pin uses this one |
| Against a clock buffer with a translator inside | an integrated buffer and translator distributes one clock to several loads with matched delays. This part is one channel, so a fanout requirement is a different part number |
Points to check before this part goes on a BOM
The input pair needs a defined bias and the datasheet is explicit about it. Both inputs want a pull up, and the on chip reference at about 2.0 V is there to hold an unused or an AC coupled input at the middle of the supply. An input pair left floating without that bias sits at an undefined level and the output follows noise. This is the single most common bring up failure on this class of translator and it costs one resistor pair to avoid.
The reference pin is a bias and not a supply. It sources or sinks at most 0.5 mA, and a board that uses it to terminate a long pair, or that ties two devices' reference pins together, pulls the reference off its nominal 2.0 V and moves the input threshold with it. The current limit is printed in the electrical table and it belongs in the note that goes with the drawing set to the customer.
The propagation delay spreads from 1000 ps to 1800 ps across temperature and process, and that spread is what has to be planned with. A link that budgets 1 ns and gets 1.8 ns at the cold end is a link with a setup violation. Where the delay matters, the part to use is one with a tighter window, and where it does not, this one is fast and cheap.
The 24 mA sink capability is what makes the part usable as a clock driver into a terminated line. A 50 ohm line terminated at 50 ohm needs 66 mA at 3.3 V, which is beyond this device, so a 24 mA output drives a 100 ohm or a 120 ohm load comfortably and a 50 ohm load only for short stubs. A customer asking about driving a 50 ohm backplane from this part is asking for a buffer rather than a translator.
Temperature compensation is the reason the output levels hold across the range. An uncompensated emitter coupled logic receiver drifts with temperature, and the drift shows as a duty cycle change on a clock rather than as a level change, which is harder to see on a scope. The compensation network inside this part is what keeps the crossing point stable, and it is also why the part needs a clean supply: the network references the supply rail.
The drop in compatibility with the MC10EPT21 is a real advantage in a market where the older number is being discontinued. What it does not guarantee is identical timing at the corners. The delays are close enough for a link with margin and different enough that a link sitting exactly on its timing budget has to be re checked, which is the honest sentence for a second source conversation.
Where a translator still has to sit
Legacy backplanes, telecom line cards and instrumentation built around emitter coupled logic still exist in service, and the logic behind them is often 3.3 V CMOS. A translator is the only part that bridges the two without changing either side, and it is why a 2009 datasheet still has an active order code in 2026.
The other place is an AC coupled clock input. A pair that arrives through a capacitor has no DC bias of its own, and the reference pin on this part restores it. That arrangement is common on a card that receives a clock from a backplane connector where the two ends sit at different DC levels.
Reading the order code
SN65EPT21 reads as the family, EPT for the emitter coupled logic to TTL function and 21 for the member. The D at the end is the SOIC body, the DGK is the VSSOP body, and the R is a reel. The string does not carry a speed or a temperature grade, because the device has one of each. A buyer comparing this row with an MC100EPT21 is comparing two vendors rather than two options.
Where this order code fits
Data and clock transmission over a backplane and signalling level conversion for clock or data lines are the stated applications, and both of them are about getting a differential pair onto a single ended logic pin without adding a nanosecond of uncertainty. Teams comparing it with a neighbouring part start from our Pin to Pin Cross Reference page, and teams replacing a part that has gone end of life start from the How to Verify a Drop in Replacement page.
Questions buyers ask
Can the inputs be left floating? No. Both inputs want a pull up, and the on chip reference at about 2.0 V is there to bias an unused or an AC coupled input. A floating pair produces an undefined output.
Can it drive a 50 ohm line? No. The output sinks 24 mA, which suits a 100 or 120 ohm load. A 50 ohm terminated line needs more current than this device provides.
Is it a drop in replacement for the MC100EPT21? Pin for pin yes, and the timing has to be re checked against the budget, because the delays are close rather than identical at the corners.
SUPERB holds and sources this line for production and repair demand, quoted against the exact ordering string with date codes, packing and country of origin stated on the offer. When a part is moving, the Obsolete Part Cross Reference Hub page sets out the checks that keep a substitute from failing on the bench.