The propagation delay calculator above divides a length by a signal speed. That is genuinely all it is, and the reason a page exists for it is that the signal speed is the part people get wrong, and the consequences of getting it wrong are larger than the arithmetic suggests.
Arb Digital does not publish a table of cable velocity factors here, and that is deliberate. Velocity factor is a property of a specific construction, and it varies between cable types, between manufacturers of nominally the same type, and with the dielectric a particular batch used. A generic figure attached to a family name is a guess dressed as data, so this tool asks for the number from your datasheet instead.
What This Propagation Delay Calculator Does
It converts your length to metres, works out the signal speed from whichever description of the medium you supplied, and divides. The signal speed can come from a velocity factor, from a published delay per unit length, from the relative permittivity of a surrounding dielectric, or from the free-space case where the speed is that of light in vacuum.
That last figure is fixed by definition rather than measured. The NIST reference on constants gives the speed of light in vacuum as exactly 299 792 458 metres per second, with no uncertainty, because the metre is defined in terms of it. Everything else on this page is that number multiplied by a fraction less than one.
The permittivity mode uses the relation that a wave in a uniform dielectric travels at the vacuum speed divided by the square root of the relative permittivity. That is exact for a fully embedded conductor such as a stripline. A microstrip on the surface of a board has part of its field in air, so the effective permittivity is lower than the laminate's and the real delay is shorter than this mode predicts; that effective value comes from a field solver or from your fabricator, not from a formula on a web page.
How to Use It
- Enter the length along the actual route. Cable follows trays, conduit and slack loops, and traces follow serpentine routing, so the physical run is always longer than the straight-line distance.
- Choose how you are describing the medium. Velocity factor and delay per unit length are two ways of writing the same thing, and datasheets use both.
- Replace the default velocity factor with your cable's figure. The value in the box is a placeholder to make the page show a result, not a claim about any particular cable.
- Read the round-trip figure for anything with an acknowledgement. A request and its reply both cross the link, so the delay that matters is twice the one-way number.
- Check the electrically-short frequency if you are designing rather than measuring. Above it, the line has to be treated as a transmission line rather than a wire.
The Formula: How the Delay Is Computed
The signal speed is v = VF × c, where c is 299,792,458 metres per second, or v = c / √εr when you are working from permittivity. The one-way delay is then t = L / v, and the round-trip delay is twice that.
Work the default. A run of 100 metres with a velocity factor of 0.66 gives a signal speed of 0.66 × 299,792,458, which is 197,863,022 metres per second. Dividing 100 metres by that speed gives 505.4 nanoseconds one way and 1,010.8 nanoseconds there and back. Per metre that is 5.05 nanoseconds, which is the figure a datasheet would quote directly.
The last grid figure comes from the electrically-short rule of thumb: a line can be treated as a lumped connection while its one-way delay is under about a tenth of the signal period, which puts the limit at f = 1 / (10t). For this run that is around 198 kilohertz. Above it the wire is a transmission line, with reflections, termination requirements and impedance to think about, which is exactly the material covered in the readings for MIT OpenCourseWare 6.013, Electromagnetics and Applications, whose chapters on TEM transmission lines and fast electronics deal with precisely this transition.
Why Bandwidth Never Fixes Latency
This is the most useful thing on the page. Propagation delay depends on distance and medium and on nothing else. Doubling a link's data rate does not halve its propagation delay, because the bits were never travelling slower than the medium allows; there are simply more of them per second.
The practical consequence is a hard floor under every round trip. A protocol that waits for an acknowledgement before sending more data is limited by that floor no matter how wide the pipe is, which is why long high-speed paths need large windows to stay busy. The bandwidth-delay product calculator works out how much data has to be in flight to keep such a link saturated, taking the delay this page computes as one of its inputs.
The distinction also explains a common frustration. Upgrading a connection makes large transfers finish sooner and does almost nothing for the responsiveness of a chatty application, because responsiveness is dominated by round trips and round trips are dominated by distance. If the two endpoints are far apart, the only real fix is moving one of them closer.
What This Number Leaves Out
Propagation is only one term in end-to-end latency, and on a real network it is often not the largest. Serialisation delay is the time to clock the bits onto the wire, which depends on packet size and data rate. Queueing delay is time spent waiting in buffers, which is highly variable and is the usual cause of latency that changes minute to minute. Processing delay covers switching, routing and any protocol work at each hop.
A long-distance path also rarely follows the map. Optical fibre routes follow rights of way, sea beds and existing ducts, so the fibre length between two cities is commonly well above the great-circle distance between them; the great circle distance calculator gives the geographic floor, and the real route is longer. Repeaters, amplifiers and regeneration add their own small delays on top.
Inside equipment the same layering applies. A memory access is dominated by device timings rather than by trace length, which is what the RAM latency calculator works with, and a serial link's per-character time comes from the framing and the rate rather than the cable, which is the baud rate calculator's territory. Trace delay matters at those scales for skew between parallel signals, not for throughput.
Where a Few Nanoseconds Genuinely Matter
Length matching on a circuit board is the clearest case. Parallel buses require the bits of a word to arrive within a tight window of each other, so traces are deliberately serpentined to equalise their lengths. At roughly five nanoseconds per metre, a centimetre of extra trace is about fifty picoseconds of skew, which is a meaningful fraction of the budget on a fast bus and nothing at all on a slow one.
Time-domain reflectometry inverts the calculation to find faults: send a pulse, time the reflection, and the round-trip delay divided by twice the signal speed gives the distance to a break or a short. The accuracy of that distance depends entirely on the velocity factor used, which is another reason to take the figure from the datasheet rather than from a rule of thumb.
Clock distribution, precision timing protocols and any system that compares timestamps across a link all care too, because an uncompensated cable delay shows up directly as a timing offset. In each of these cases the arithmetic is trivial and the input data is what determines whether the answer is useful.
Arb Digital builds free tools like this one because useful pages earn attention. If you want tools, calculators or content built for your own audience, we can help.
Browse All Free Tools Talk to Arb DigitalCommon Mistakes to Avoid
- Using a velocity factor from a family name — it varies by construction and manufacturer, so it has to come from the datasheet for the cable you actually have.
- Using the straight-line distance — cable follows routes and traces are serpentined, and both are longer than the map suggests.
- Quoting one-way delay where round trip is meant — anything with an acknowledgement pays the delay twice.
- Applying the permittivity formula to a microstrip — part of its field is in air, so the effective value is lower and the real delay is shorter.
- Expecting more bandwidth to reduce it — propagation delay is set by distance and medium, and data rate does not appear in the formula at all.
Related Free Tools From Arb Digital
For how much data must be in flight to keep a long link busy, use the bandwidth-delay product calculator. The great circle distance calculator gives the geographic floor under a route, the length converter handles runs that arrive in the wrong unit, and the speed converter rescales the velocity. For delays inside equipment rather than along a cable, see the RAM latency calculator and the baud rate calculator, and for uncertainty propagation, which shares a word and nothing else, the error propagation calculator. Browse the rest in the free online tools hub.
Frequently Asked Questions
It is the time a signal takes to travel from one end of a cable or trace to the other. It is the length divided by the speed the signal travels in that medium, and it depends on nothing else, least of all the data rate being carried.
Because velocity factor is a property of a specific construction and varies between manufacturers of nominally the same cable. A figure attached to a family name would be a guess presented as data, so the tool asks for the value from your datasheet instead.
No. Data rate does not appear in the formula. A faster link puts more bits per second onto the same medium, and each bit still travels at the same speed, so the transit time across a given distance is unchanged.
The fraction of the speed of light in vacuum at which a signal travels in a particular cable, always less than one because the surrounding dielectric slows the wave. It is published on the manufacturer's datasheet, sometimes instead as a delay in nanoseconds per metre or per foot.
Round trip for anything involving a reply: an acknowledgement, a request and response, or a reflection measurement. One way for a broadcast or a timing offset, where the signal only has to arrive once.
Because propagation is only one term. Serialisation, queueing and per-hop processing all add time, the physical route is usually longer than the map distance, and any active equipment in the path contributes its own delay.
It is where a connection stops behaving like a plain wire. Below it, the delay is a small enough fraction of the signal period that the line can be treated as a lumped connection. Above it, reflections and termination matter and it must be treated as a transmission line.
Only for a fully embedded trace such as a stripline. A microstrip on the board surface has part of its field in air, so its effective permittivity is lower than the laminate figure and the real delay is shorter. Use a field solver or your fabricator's data for that case.
This tool performs an idealised transit-time calculation from the figures you supply. It is not a substitute for measurement or for signal integrity analysis, and it does not account for dispersion, skin effect, dielectric loss, connectors or any active equipment in the path. Design decisions should rest on manufacturer data and on measurements of the actual installation.
