Chinese Radar Developments - KLJ series and others

Stealthflanker

Senior Member
Registered Member
Okay, I see where the problem is now. I think you’re using the below equation incorrectly. You are missing that N^3 includes whole array mean power. It’s the multiple of whole array mean power with transmitter count and receiver count. Individual module power is a separate variable outdid whole array mean power.

Are you sure about that ? Individual power or small p there is a separate variable because you dont really emit power when receiving.
Also mean power here is basically peak power of the module multiplied with the duty cycle of the module.

Also the other source you mentioned is my writings. I am the developer of the AESA Radar Calculator.


I didn’t say “you not discuss this is too complex or you should just follow what I say”. Pointing out that an argument is too simple is not trying to shut down conversation or appealing to false authority. It’s simply taking note of potential sources of error.

Adding unnecessary technobabbler without proper usage is also a source of error.

I never said more modules aren’t better than less, just that you can’t get a good sense of how two AESA radars perform relatively to one another omly by looking at module counts. Anyways refer to the correction I’m pointing out to how you’re using the equation.

The problem is that you never really provide your own version on "how to get a good sense", particularly something which can actually be modeled or shown to general people.
 

latenlazy

Brigadier
Are you sure about that ? Individual power or small p there is a separate variable because you dont really emit power when receiving.
Also mean power here is basically peak power of the module multiplied with the duty cycle of the module.

Also the other source you mentioned is my writings. I am the developer of the AESA Radar Calculator.
Okay, so if this is your equation why isn’t mean power included in the calculations you’re presenting here…all I see is T/R count. There’s a missing term.

Adding unnecessary technobabbler without proper usage is also a source of error.
It’s not technobabble. You just don’t have an answer for someone pointing out complex operation factors in radar performance because it can’t be reduced to a simple equation. That doesn’t make it not real. Don’t need to attack me for pointing out that you can’t account for all the variables needed to justify drawing strong conclusions about performance comparisons.

The problem is that you never really provide your own version on "how to get a good sense", particularly something which can actually be modeled or shown to general people.
I can show people a model of how a high aspect vs low aspect wing works but it won’t tell anyone anything about how a Flanker performs against a Eurocanard. Having a model is never enough. Showing the model is representative is necessary if you want to draw hard conclusions about real performance.
 

Stealthflanker

Senior Member
Registered Member
Okay, so if this is your equation why isn’t mean power included in the calculations you’re presenting here…all I see is T/R count. There’s a missing term.

Well dont you see i included it ? Like the simplified version i used i mentioned Px and Pref which P for Power. This can be any power be it Mean or Peak. As i said the most important factor when you are using inverse square law factor is the factor and i forgot to mention consistency (e.g unit should be same Watt).

and i'm curious the missing term you mean ? Like do you actually want to see "N^3*P" ? It's already done.

It’s not technobabble. You just don’t have an answer for someone pointing out complex operation factors in radar performance because it can’t be reduced to a simple equation. That doesn’t make it not real. Don’t need to attack me for pointing out that you can’t account for all the variables needed to justify drawing strong conclusions about performance comparisons.

I apologize if i sound attacking you However. I'm criticizing you because you dont seem to have anything to bring to the table. My modeling so far is showing performance of AESA in terms of radar range equation. and you suddenly barge in pointing some superflous technological term which have no real relationship nor easily explainable. Doesn't that sound bit rude ?

I'll be honest this is not the first time i seen this overdemanding behavior. and i'm already so tired of it.

I can show people a model of how a high aspect vs low aspect wing works but it won’t tell anyone anything about how a Flanker performs against a Eurocanard. Having a model is never enough. Showing the model is representative is necessary if you want to draw hard conclusions about real performance.

And did you do what you are saying here ?

Naturally i when discussing stuff i expect people to also bring something in the table too.
 

latenlazy

Brigadier
Well dont you see i included it ? Like the simplified version i used i mentioned Px and Pref which P for Power. This can be any power be it Mean or Peak. As i said the most important factor when you are using inverse square law factor is the factor and i forgot to mention consistency (e.g unit should be same Watt).

and i'm curious the missing term you mean ? Like do you actually want to see "N^3*P" ? It's already done.
Shouldn’t two radars with different transmit powers have two different P-x and P-ref? If that’s the case shouldn’t those differences also be included in the N^3 term, not just module count?

I apologize if i sound attacking you However. I'm criticizing you because you dont seem to have anything to bring to the table. My modeling so far is showing performance of AESA in terms of radar range equation. and you suddenly barge in pointing some superflous technological term which have no real relationship nor easily explainable. Doesn't that sound bit rude ?
Saying that signal processing power and complex waveform management matter is “bringing something to the table.”

I'll be honest this is not the first time i seen this overdemanding behavior. and i'm already so tired of it.
I don’t think I’m being over demanding. I think you’re being a bit defensive about factors that you don’t have answers for.
And did you do what you are saying here ?

Naturally i when discussing stuff i expect people to also bring something in the table too.
Whether you want to acknowledge my point or not things like signal processing power and complex waveform management do matter for radar performance comparisons and it’s not sound to assume all radars are equivalent on these parameters. If you don’t have any specific answer for how to consider those factors that is fine, I don’t either, but these factors do exist as part of radar performance assessment, and lay people who don’t have technical context should be aware that they exist. Simply calling it technobabble when it’s not though is not encouraging informed discussion. Informed discussion should include the caveat of what factors matter even if you don’t have data to draw conclusions for them, not treating something that does matter as nonsense because you can’t stick it into a simplified model you like using.
 
Last edited:

BoraTas

Captain
Registered Member
Yeah this is good. Liquid cooling does indeed provide more cooling capacity than air cooled which limited to about 2-3 KW/sqm of antenna aperture. Rafale AESA however is also liquid cooled. The liquid cooling performance in other hand would be limited typically by the amount of fuel the aircraft can carry as that's the medium where the coolant solution will dump the heat.

Thus more fuel = more cooling potential.

In the other hand, air cooling is easier to implement, light and may have little impact to the aircraft's structure, thus why those AESA upgrades for legacy fighters are typically air cooled.

However the main driver of AESA performance would be number of TRM's as the performance gain, scales by cube of TRM. Radar that can pack more TRM's than the other will usually "win" in terms of performance. Radar with less number of TRM's when trying to match the performance of the radar with more TRM's will be "penalized" in cooling and power requirement and thus cost.

---------
Trying to compare J-10 AESA's with "other kids in the block" could start with comparing number of TRM's. Let's say, from a twitter poster i got 1200 TRM's for J-10B AESA, how does that compare to say an APG-79 with 1368. One can then use the 4th root law with cube of the TRM's. Using APG-79 as base.

Rfac=((1200/1368)^3)^(1/4)
Rfac=0.906

Basically the range of the J-10 AESA is 10% shorter than the APG-79. How about comparing with AN/APG-80 with 1020 TRM ? same method.

Rfac=((1200/1020)^3)^(1/4)
Rfac=1.13

Thus the J-10 AESA would have 13% range advantage over APG-80.

and how about Rafale ? It has TRM count of 838 according to photo although you will see people claiming 1000 TRM. If we stick to the google image it would be :

Rfac=((1200/838)^3)^(1/4)
Rfac=1.3

The 1200 TRM AESA will have 30% advantage in range. But hey how about element power ? Same methods. Let's say the French put TRM twice the power of the J-10's or a factor of 0.5 against the J-10.

Rfac=((1200/838)^3*(1/2))^(1/4)
Rfac=1.1

The French radar gained 20% (should be closer to 19% however) range but radar with larger amount of TRM still wins. In order to match or maybe reverse the gain from having more TRM's. The TRM power would need to be increased by factor of at least 3. This however will have direct impact on cooling, power requirement and cost of the TRM's.
These calculations are good but they assume TRM increasing without any compromises and changes. So your 2x TRM scenario is basically 2x power, 2x collection area and 2x gain scenario. Without platform-level changes you won't be able to grow your radar and its total power output much. TRM increase would still result in a gain increase but that wouldn't be fully realized since a higher gain also means a slower scan at the same dwell time. A dwell time decrease would be needed. Range gain would likely be around 12%.
 

Stealthflanker

Senior Member
Registered Member
These calculations are good but they assume TRM increasing without any compromises and changes. So your 2x TRM scenario is basically 2x power, 2x collection area and 2x gain scenario. Without platform-level changes you won't be able to grow your radar and its total power output much. TRM increase would still result in a gain increase but that wouldn't be fully realized since a higher gain also means a slower scan at the same dwell time. A dwell time decrease would be needed. Range gain would likely be around 12%.

Thank you and you know what, You can also add dwell time too as it is also a variable in radar range equation. Also in AESA you can at least in theory changes the antenna tapering e.g from Taylor -40 dB to something like Cos or if you want a full gain an untapered one. This will increase or decrease your beamwidth and sacrifice or put some more gain (within limits of course) and you can optimize your dwell time for desired coverage.

Shall we revisit our 1200 TRM AESA and 2400 one with 10 Watt emitted power ? to demonstrate the effect of dwell time. So you dont have to guesstimates or even use word "Likely" at all. Let's say we want to introduce the beam dwell time into the problem. This will introduce us into two terminology Dwell time and Frame time. The dwell time is how long your beam sticks in one area or one beam, this can usually be 0.025, 0.01 or 0.1 or even longer in seconds depending on requirement. the Frame time is your total time required to scan your area of interest.

As for the addition it would look like this :

Rfac = ((2400/1200)^3*(10/10)*(Dwellx/Dwellr))^(1/4)

Where dwell time is added

Now let's assume both AESA have pencil beamwidth which is the beamwidth where vertical and horizontal one is the same or close enough for not to be a Fan beam. We can find the beamwidth of each antenna by simple equation :

Beam in Deg = 100/SQRT(N)

for AESA with 2400 TRM

Beam=100/SQRT(2400)
Beam = 2 Degrees

AESA with 1200 TRM

Beam=100/SQRT(1200)
Beam = 2.9 Degrees

After that we can assume a scan sector.. let's say a 7 bar scans in an arc of 120 degrees and dwell time of say 0.025 seconds which typical for fighter aircraft with an X-band operating frequency. if we divided those area with beamwidth it will result with following :

Radar1200 TRM will complete the scan with 158 Beam Thus the total frame time required would be 158 * 0.025 = 3.96 Seconds
Radar 2400 TRM will complete the scan with 316 Beam Thus the total frame time required would be 316 * 0.025 = 7.92 Seconds

or twice the amount of the 1200 TRM Radar. So yeah the scan is slowed. However Since the dwell time is the same the factor would be 1, thus the growth is the same.

Rfac = ((2400/1200)^3*(10/10)*(.025/.025))^(1/4)
Rfac =1.68

But now, what if we want the frame time or your total scan time the same ? let's say we want to complete the scan in 3.96 seconds same as the 1200 TRM AESA ? We have to reduce the beam dwell time of our 2400 TRM Radar by half which is 0.0125 seconds to get the same frame time. Does our 2400 TRM AESA still have advantage ?

Rfac = ((2400/1200)^3*(10/10)*(.0125/.025))^(1/4)
Rfac = 1.41

So as you see, our 2400 TRM AESA Still have 41% advantage over the other with smaller TRM counts. although yes not as extreme as one with slower scan. and now we increased the power of the 1200TRM AESA by twice.. see how much it can offset the advantage of having more TRM's

Rfac =((2400/1200)^3*(10/20)*(.0125/.025))^(1/4)
Rfac = 1.18

The AESA with larger TRM counts still trump over the other although with markedly 23% "loss" compared to the one where the power of the modules are equal. If the 1200 TRM AESA also increase the dwell time then there would be point where it can compensate the smaller number of TRM's

Shouldn’t two radars with different transmit powers have two different P-x and P-ref? If that’s the case shouldn’t those differences also be included in the N^3 term, not just module count?

It's included already. if you properly read my writings you tried to quote you will see how the N^3 factor arise. I honestly confused on what you are trying to criticize here. Considering that people has been using 4th root laws the way i did for long time to compare radar.

Or maybe you mean i should do the calculations twice. So each radar will have its own factor then compared the result. Let's say.. now we calculate 2 AESA separately. One with 1200 TRM vs another with 1200 TRM but twice the power in this manner ?

So i have to write them separately like that ?


Saying that signal processing power and complex waveform management matter is “bringing something to the table.”
I am expecting something that is in Radar range equation. Let's say what it can do to reduce the required SNR. Something like pulse integration.. like when you integrate the pulse in coherent manner vs one with noncoherent. There are calculation methods for it and there is simplification too to allow introduction to Radar Range Equation.

and if you are into processing. Let's say you wish to introduce a complex algorithm like say a STAP to cancel clutter, you can show how much load it could have in computing maybe in terms of MIPS or maybe how a faster clock processor can handle that. Then we can actually have something to compare with.



And i will close this post with remarks from the book "Radar Techniques using Array Antenna" by Wulf-Dieter Wirth on AESA. You can find it in page 76 chapter 4 of the book.

Active Array.png

It is very revealing. The most effective means of increasing performance for AESA is having more TRM's.
 

latenlazy

Brigadier
It's included already. if you properly read my writings you tried to quote you will see how the N^3 factor arise. I honestly confused on what you are trying to criticize here. Considering that people has been using 4th root laws the way i did for long time to compare radar.

Or maybe you mean i should do the calculations twice. So each radar will have its own factor then compared the result. Let's say.. now we calculate 2 AESA separately. One with 1200 TRM vs another with 1200 TRM but twice the power in this manner ?

So i have to write them separately like that ?
Well…yes? If you’re talking about two different radars with two different transmit module powers then it should follow that their mean powers are also different?
 

Stealthflanker

Senior Member
Registered Member
Well…yes? If you’re talking about two different radars with two different transmit module powers then it should follow that their mean powers are also different?

So you want to do this ?

a 2400 TRM AESA with 10 Watt of power compared to other 1200 TRM AESA with 20 Watt of power. SO you think the application of the 4th root laws would be :


RfacAESA2400 = ((2400/2400)^3*(10/10))^(1/4)
RfacAESA2400 = 1

and then the 1200 TRM one

RfacAESA2400 = ((1200/1200)^3*(20/10))^(1/4)
RfacAESA2400 = 1.19

It only shows the advantage of the 1200 TRM AESA have its module power increased by factor of 2. But it doesnt say anything to compare with the one with 2400 TRM.

The purpose of having that 4th root equation is to see what happen if we change 1 parameter of a radar to the design. You can also compare 2 different radars with it assuming at least it works in same wavelength and the differences are limited to the factor being considered. That's what i am demonstrating so far, which you seem having difficulty to understand with.

You can add the factors on the go as long as it's part of the RRE.
 

latenlazy

Brigadier
So you want to do this ?

a 2400 TRM AESA with 10 Watt of power compared to other 1200 TRM AESA with 20 Watt of power. SO you think the application of the 4th root laws would be :


RfacAESA2400 = ((2400/2400)^3*(10/10))^(1/4)
RfacAESA2400 = 1

and then the 1200 TRM one

RfacAESA2400 = ((1200/1200)^3*(20/10))^(1/4)
RfacAESA2400 = 1.19

It only shows the advantage of the 1200 TRM AESA have its module power increased by factor of 2. But it doesnt say anything to compare with the one with 2400 TRM.

The purpose of having that 4th root equation is to see what happen if we change 1 parameter of a radar to the design. You can also compare 2 different radars with it assuming at least it works in same wavelength and the differences are limited to the factor being considered. That's what i am demonstrating so far, which you seem having difficulty to understand with.

You can add the factors on the go as long as it's part of the RRE.
I understand what you’re trying to do just fine, but two different radars are often different in more than one parameter. Assuming all else held equal but one parameter is not a realistic comparison except between two radars from the same product series with different module counts.
 
Top