I am trying to understand what happened in the famous incident where the Patriot failed to intercept a Scud missle because of a software bug.

The reprot can be found here.

Why does the radar need to use range gates during the tracking mode (when it predicts where the target will be). Why can't the whole beam be processed? I doubt it is processing power limitation since during the search mode the whole beam is processed.

  • \$\begingroup\$ Tracking time filters increase the SNR and that reduces the error rate \$\endgroup\$ May 17, 2021 at 20:34
  • \$\begingroup\$ They used integer absolute time history with an OCXO clock better than 1e-11 error rather than relative time for each and every target. \$\endgroup\$ May 17, 2021 at 20:55
  • \$\begingroup\$ *purpose. Propose is to present an idea or offer of marriage \$\endgroup\$
    – DKNguyen
    May 24, 2021 at 18:36
  • \$\begingroup\$ A range gate is merely the along-the-beam prediction of where the target might be now. In a certain time window it can't move very far from where it was last detected, so there's no value in "processing the whole beam". But it can move a small distance because it's physically moving along the beam, and because each range reading is noisy, so they jump around a bit. \$\endgroup\$
    – tomnexus
    May 24, 2021 at 20:35

4 Answers 4


The term "range gate/gating" can mean different things depending on the context and the radar system being talked about.

In general, a "range gate" is a window in range observed for some purpose, mainly for detecting and tracking targets in a radar system.

In legacy radar systems, analog circuits were used to define a time window (which corresponds to range) where target energy was allowed to pass and fed to the detection and tracking circuits. Other energy existing outside of this window was not used. When it came to tracking, there are techniques where you can compare portions of the time window to detect if a target is in front or behind the center of the window. Usually an error signal is generated that would then drive some kind of servomotor which would move the window to center the target again. This action happened continuously when the radar had a good track. Many systems allowed the human operators to manually adjust this window as they saw fit, usually during the detection phase of an engagement.

With proper execution, jamming techniques like range-gate pull-off (RGPO) could be especially brutal.

In modern radar systems, tracking is accomplished digitally. Here, a "range gate" can refer to a few things including but not limited to:

  1. One sample from the sampled data after pulse compression/DFT. This definition is one of the most common.
  2. A collection of samples as defined in (1), who each are their own range gate, that defines a range span. This one can be confusing but with proper context it's usually not too bad.
  3. A particular value that limits the search space for detection/tracking algorithms.

Your question is related to (3). First off, you can totally search the entire range span if you wanted to. As a matter of fact, this is done in systems that are responsible for detecting and acquiring a target. You can see that being done in the "search action" phase in Figure 3 of the document you linked.

This is avoided in tracking because in many situations you simply do not have enough time to search the entire range span. The tracking algorithms/filters need to be fed at some required measurement update rate in order to keep the track alive. These rates are especially fast in missile-on-missile engagements were closing velocities are easily in the multiple of Machs.

  • \$\begingroup\$ I am not sure I understand the last part. What does "search the entire range span" practically mean? Why does the span have to be decided a-priory and not after all the energy returns received. Why can't the closest energy peak to the predicted target range be chosen as the center of the range gate (instead of the predicted target range). Sorry if I use wrong terminology, I know almost nothing about radars. \$\endgroup\$
    – Artium
    May 29, 2021 at 19:46
  • \$\begingroup\$ @Artium The reason is because you cannot observe an "infinite" range span. The receiver needs to be on to receive the return signal for some amount of time. You run out of time by either bumping into the ambiguous range of the radar and need to transmit your next pulse and/or you run into your limitation on the number of samples you collect. This is the problem in general regardless of the stage of the engagement. In tracking specifically, you self-impose a limit for the reasons I stated in the answer which are mainly for performance. \$\endgroup\$
    – Envidia
    Jun 5, 2021 at 18:13

Range gate gives to the direction tracking process the timing "compare the monopulse beams now".

Servo tracking range gate can drift away if there's a strong artificial echo which is timed to give at first a plausible "it's on the peak" indication and slowly moved away.

Clever anti-jam systems and also an experienced radar operator can manually force the range gate follow the right target, no matter there's a strong fake echo.

  • 1
    \$\begingroup\$ This is such a nicely accurate statement at the "tip of the iceberg," which tells me much is underneath what you wrote. I worked with experienced Naval radar operators (20+ yrs) and this ties well with what I understand, too. I almost wish you were able to consider writing more. \$\endgroup\$
    – jonk
    May 17, 2021 at 22:16

I doubt it is processing power limitation since during the search mode the whole beam is processed.

Hm the figures suggests you have a scanning beam: the radar can't observe the whole sky at once. So, until anything happens, you scan (as in: focus the beam at a spot, wait for the farthest sensible reflect, then you look at the next spot) the part of the sky from which attacks might be coming.

Typically that part of the sky is "empty", there's not going to be civilian aircraft in there, for example

Of course, once you detect something, you want to make sure where exactly it's going, and that it's really what you think it is. Hence, you shift your beam to a predicted position of the target. But that might include other radar targets – steel structures, your own aircraft, civilian aircraft, your own artillery fire...

So, it doesn't suffice there's something in your beam, it needs to be in the right spot.

Now, if you can't choose where you're looking, your SNR suffers – as said, there's clutter, there might be your own reflectors etc. So, this operation basically requires you to increase your radar's sensitivity by gating out anything that's not what you're expecting.


In order to track an object in 3D space, azimuth, elevation and range must be translated from 3 antenna with time and mobile spacial fixed reference.

It is synchronous to rev/sec stored in integer relative time with 12bit counters and position rotated by the phase shift of the reflected pulse relative to each rotor angle to triangulate position and compute velocity and acceleration to next expected position. The clock was counting relative time with an error less than 1e-11 to track all event history and current events with an expected use of 2h rather than 100h used. As a consequence, the triangulation history of the 3 phase array had a cumulative error from OCXO drift. Resetting the system count to calibrate a reference location for absolute accuracy was being used. The fix may have been to provide a selective resync/rezero count to null the history and start a new accumulated real-time travel history in order to accurately predict the future location travelling faster than the speed of sound in short range.

This selective gating blocks fake delayed echoes.

  • 1
    \$\begingroup\$ Not sure if you were implying this, but you don't need 3 antennas to track an object in 3D space. \$\endgroup\$
    – SteveSh
    May 17, 2021 at 21:29

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.