LMK62E0-156M25SIAR
Product Specifications
LMK62E0-156M25SIAR
Texas Instruments — Power Management & Signal Chain
Product Overview
The LMK62E0-156M25SIAR 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
LMK62E0-156M25SIAR: a 156.25 MHz LVPECL oscillator at 150 fs and plus or minus 25 ppm
LMK62E0-156M25SIAR is a factory programmed oscillator from Texas Instruments that puts out 156.25 MHz in LVPECL from a single 3.3 V rail, in a 6 pin QFM package measuring 5.00 mm by 3.20 mm. The part carries two numbers that decide most of its use: 150 fs RMS of phase jitter typical over the 12 kHz to 20 MHz band, and a total frequency tolerance of plus or minus 25 ppm that already includes the initial offset, the temperature swing, the supply variation, the solder reflow shift and five years of aging at 40 degrees. The E0 suffix is the 25 ppm grade of the family. The E2 suffix of the same frequency is the 50 ppm grade, and the price difference between them is smaller than the cost of a link that drops because the reference drifted.
156.25 MHz is the reference rate that sits behind 10 gigabit and 100 gigabit line cards, radio units and switch fabrics, so the part is usually found on a board next to a SerDes, an FPGA transceiver bank or a clock buffer rather than in a standalone clock tree. What a buyer of this number is really comparing is not the frequency, which is fixed, but the jitter budget and the tolerance grade, because those two decide whether the same footprint can serve a 10G port and a 25G port with the same bill of materials.
The values on this page were read from the vendor document SNAS691E, revised December 2023, which covers the whole LMK62XX family. The document is a family document, so every table in it carries several part numbers at once. The rows quoted here are the ones that apply to the E0 grade at 156.25 MHz, and the LVDS and HCSL output tables in the same document do not apply to this part at all.
What 150 fs and 25 ppm mean at 156.25 MHz
One period at 156.25 MHz is 6.4 ns. A jitter of 150 fs RMS is 2.3 parts in 100,000 of that period, or 0.0023 percent. That is the figure that matters when a SerDes eye is being closed by the reference, and it is the reason a part like this replaces a crystal plus a buffer: the buffer inside the package is already characterised with the resonator rather than added afterwards.
The tolerance figure works the other way. Plus or minus 25 ppm at 156.25 MHz is a window of 3.906 kHz around the nominal frequency, and the guarantee covers the whole window at any point in the life of the part, not just at 25 degrees. A 50 ppm grade doubles that window to 7.8 kHz. Both grades will lock in most systems. The 25 ppm grade is the one that keeps a margin when the same board has to pass a temperature sweep and a five year aging calculation at the same time.
Supply noise is handled inside the package. The vendor prints minus 60 dBc of power supply rejection from 50 kHz to 1 MHz with 50 mV peak to peak of ripple injected on the supply, and the same document gives the conversion formula for turning that rejection into a jitter contribution. Applied to this part at 156.25 MHz, minus 60 dBc works out to about 4.07 ps peak to peak of spur induced jitter. That is the arithmetic a power integrity engineer runs when deciding whether a switching regulator can feed the oscillator directly or needs a linear post regulator in front of it.
The supply, the start up and the enable pin
Device current is 95 mA typical and 110 mA maximum in LVPECL, which at 3.3 V is 313.5 mW typical and 363 mW maximum inside the package, before any termination loss is counted. The vendor states that roughly 40 mW of the total is dissipated in the 150 ohm termination resistors and should be excluded from the on chip figure, so the die sees about 273 mW typical. With a junction to ambient resistance of 94.5 degrees per watt on a four layer JEDEC board, that is a junction rise of about 26 degrees above ambient, which puts the junction at roughly 111 degrees in an 85 degree cabinet. The margin to the 125 degree junction limit is thin enough that the two thermal vias the vendor asks for under the ground pad are not optional.
Two timing numbers matter on the production line. Start up takes 10 ms from the moment the supply passes 3.135 V to the moment the output is valid, so a board that gates a reset on the clock has to wait longer than that. Output enable and disable each take 50 microseconds through the OE pin, with an input high threshold of 1.4 V and an input low threshold of 0.6 V, so the pin can be driven from a 1.8 V logic device without a level shifter.
The disable current is worth a second look. With OE held at ground the device still draws 70 mA, which is about three quarters of its running current. This part saves very little power when it is disabled, so a board that tries to use OE as a power saving measure will be disappointed, and a board that needs the clock off during a test should plan for the current to stay in the budget.
Where this part sits in the LMK62XX table
The family is organised by output format and by tolerance grade, and the frequency is fixed at the factory. Reading the table is how a board picks the grade it needs instead of paying for jitter it cannot use.
| Part number | Output frequency | Format | Tolerance |
| LMK62E2-100M | 100 MHz | LVPECL | plus or minus 50 ppm |
| LMK62E2-156M | 156.25 MHz | LVPECL | plus or minus 50 ppm |
| LMK62E0-156M | 156.25 MHz | LVPECL | plus or minus 25 ppm |
| LMK62A2-100M | 100 MHz | LVDS | plus or minus 50 ppm |
| LMK62A2-150M | 150 MHz | LVDS | plus or minus 50 ppm |
| LMK62A2-156M | 156.25 MHz | LVDS | plus or minus 50 ppm |
| LMK62A2-200M | 200 MHz | LVDS | plus or minus 50 ppm |
| LMK62A2-266M | 266.66 MHz | LVDS | plus or minus 50 ppm |
| LMK62I0-100M | 100 MHz | HCSL | plus or minus 25 ppm |
| LMK62I0-156M | 156.25 MHz | HCSL | plus or minus 25 ppm |
Three points come out of that table. The tolerance grade is carried by the letter and digit pair in the middle of the number, not by the suffix at the end, so E0 and E2 at the same frequency are not interchangeable in a timing budget even though they share a footprint. The LVDS and HCSL versions stop at 50 ppm except for the HCSL 25 ppm rows, which is why a 25 ppm LVDS requirement usually ends in a different family. And the frequencies in the table are the ones the vendor keeps as standard; anything else is a factory program with its own lead time, which is the question to ask before a schedule is fixed.
The figures a board engineer checks first
| Function | Factory programmed low jitter oscillator, single ended supply |
| Output frequency | 156.25 MHz, fixed at the factory |
| Output format | LVPECL differential |
| Total frequency tolerance | plus or minus 25 ppm, includes initial, temperature, supply, reflow and five year aging |
| RMS phase jitter | 150 fs typical, 250 fs maximum, 12 kHz to 20 MHz, output at or above 100 MHz |
| Supply voltage | 3.3 V plus or minus 5 percent, single rail |
| Device current, LVPECL | 95 mA typical, 110 mA maximum |
| Device current, output disabled | 70 mA with OE at ground |
| Output swing VOD | 700 mV minimum, 950 mV typical, 1200 mV maximum |
| Output common mode | VDD minus 1.45 V typical |
| Rise and fall time | 260 ps typical, 350 ps maximum, 20 to 80 percent |
| Output duty cycle | 45 to 55 percent |
| Power supply rejection | minus 60 dBc with 50 mV peak to peak ripple, 50 kHz to 1 MHz |
| Start up time | 10 ms from supply at 3.135 V to valid output |
| Output enable and disable | 50 microseconds each |
| Operating temperature | minus 40 to plus 85 degrees, industrial |
| Package | 6 pin QFM, 5.00 mm by 3.20 mm, SIA |
| Junction to ambient | 94.5 degrees per watt on a four layer JEDEC board |
Package, footprint and the 5032 industry standard
The six pin QFM is 5.00 mm by 3.20 mm and the vendor states the pin out is compatible with the industry standard 5032 crystal oscillator footprint. That single sentence is the commercial argument for the part: a board that was laid out around a 5032 can take this device without a new revision, which is how a product that is already in production improves jitter without requalifying a board.
The thermal path is through the ground pad and the vendor asks for two thermal vias of 0.3 mm diameter connected to ground. The 94.5 degrees per watt figure assumes a four layer JEDEC board with those vias in place. A two layer board with no vias under the pad will run hotter than the number suggests, and at 313 mW of dissipation the difference between 94.5 and, say, 130 degrees per watt is 11 degrees of junction temperature. That is worth a note on the fabrication drawing rather than a surprise at qualification.
Buying position: this grade is not recommended for new designs
The vendor lists LMK62E0-156M as not recommended for new designs while keeping it in production for existing customers. That status is a purchasing fact, not a quality fact: parts keep shipping, but the vendor directs new work elsewhere. For a product already in volume the practical response is a last time buy calculation that covers the remaining life of the product plus service spares, held against a date code plan so the stock does not age out on the shelf. For a product still in development the same status is a reason to look at the alternative the vendor names before the board is frozen.
Our position on a part in that state is straightforward. We quote the exact orderable number, we state the date code and the packaging we can supply, and we tell the buyer which of the two paths they are on: keep the current part with a controlled stock, or move to a replacement with a verification plan that includes the jitter measurement, not just the pin out. Buying an oscillator on pin compatibility alone is how a link budget quietly loses 100 fs.
What to send for a quote on this part
A quote on LMK62E0-156M25SIAR needs four things: the annual quantity and the first call off quantity, whether the reels must be full 2500 piece reels or cut tape is acceptable, the date code window the line will accept, and whether the incoming inspection includes a jitter or frequency check or only a visual and a solderability test. With those four we can confirm stock, packaging and the test paperwork in one reply.
Related pages that answer the questions this part usually raises: component obsolescence management for the plan once a status changes, last time buy and LTB planning for the stock calculation, pin to pin cross reference for the replacement route, and how to verify a drop in replacement for the test plan that keeps the jitter number honest.
Questions buyers ask about the E0 grade
The first question is whether the 25 ppm grade is worth the price over the 50 ppm grade. It is worth it when the same board has to hold a link at the top of the temperature range after five years, and it is not worth it when the reference is divided down and the loop bandwidth of the downstream PLL is wide. Look at the total window the standard allows, not at the initial accuracy.
The second question is whether the part can be driven from the same 3.3 V rail as the FPGA. It can, provided the ripple on that rail stays inside the 50 mV peak to peak figure the rejection number is quoted at. Above that, the rejection figure no longer describes what happens, and a small linear regulator in front of the oscillator is cheaper than a failed eye mask.
The third question is about the OE pin. Because the disabled current is 70 mA, disabling the output is not a power saving move. If the board needs the clock quiet during a test, OE does that in 50 microseconds; if the board needs the rail to drop, the rail has to be switched, not the enable pin.