PLA next/6th generation fighter thread

Kejora

Junior Member
Registered Member
I'm seeing in multiple instances of weibo discussion under 6th gen fighter it's mentioned both CAC and SAC are doing competing designs. CAC is faster and their prototype is about to fly hence all the excitement. SAC isn't too far behind and their prototype is due to fly sometime next year.

Is this where all these J-XD1 and J-XD2 discussions are coming from?
Is this the first time China has two competing designs with flying prototypes? During J-XX competition Snowy Owl never left drawing board.
 

ACuriousPLAFan

Brigadier
Registered Member
I'm seeing in multiple instances of weibo discussion under 6th gen fighter it's mentioned both CAC and SAC are doing competing designs. CAC is faster and their prototype is about to fly hence all the excitement. SAC isn't too far behind and their prototype is due to fly sometime next year.

Is this where all these J-XD1 and J-XD2 discussions are coming from?

Yes, frankly speaking.
 

Temstar

Brigadier
Registered Member
Is this the first time China has two competing designs with flying prototypes? During J-XX competition Snowy Owl never left drawing board.
I think so, in previous cases the loser gets dropped while still in drawing stages, eg Snowy Owl, J-13 etc.

But I think that's a cost thing. Back then China couldn't financially support multiple prototypes proceeding to physical flying article stage, but today things may be different. It's like how US is rich enough to have many X planes - one off prototypes that lets you learn a lot in the process of building them without them being intended for production. Yankee has argued that China has reached this stage and he points to CAC's J-20S as an example of Chinese X plane (not intended for production at onset of project, laid the foundation for later WS-15 powered J-20). So if you take that view then competing 6th gen prototype seem plausible.

J-10 TVC I think would be another good example of a Chinese X plane.
 

AndrewS

Brigadier
Registered Member
keep in mind of the power requirements of 6th gen.

F-35 right now is having a hard time because they are trying to future proof the demand by increasing thermal management to 80kw.

The major consumption of power on an aircraft is the radar.

Here is APG-79 for super hornet

View attachment 140871

This uses GaA T/R modules and it already requires 15kW of power requirement so liquid cooling needs to be 15.6kW. Since APG-81 is from the same generation and uses GaA T/R modules also, it's probably at this mark or a little more just for the radar. I would assume there is some more consumption from computers and processors, communication gears, powering of EW pods, EODAS and such, but the maximum usage is the radar. One could see how they got to the current requirement of 30kW max of cooling and want to increase it to 80kW

Please, Log in or Register to view URLs content!

1600 T/R module at 16kW would be 10 watt per T/R module, that I would imagine is the maximum possible for GaA modules from back in 2000s.

But nowadays, we are using GaN-on-Si, which should be at least several times that of GaA. In the future, we need to get ready for GaN on Diamond (or GaN-on-Sic if you are less ambitious)

Just a cursory glance on google search, you will find a GaN T/R module can have peak 100W power.

If you plaster 3000 T/R module. That would be capable of using 300kW of power.

View attachment 140872
Another look here, you can see the thermal conductivity of GaA is just 0.46 vs 1.3 on GaN vs 22 for diamond.

melting point is also much lower in GaA vs GaN vs Diamond.

There is a reason why GaA is only used for mobile phone power amplifiers now. The 5G cell towers all use GaN.

Remember from this
Please, Log in or Register to view URLs content!


Guobo has capacity for at least 20 million GaN chips and packaging for cell tower. You need more power on your cell tower, so you have to use a material that's capable of handling more power. That would be GaN, not GaA

Notice SiC's thermal conductivity is 3.5 and Si is 1.5? That's why GaN-on-SiC has better performance than GaN-On-Si

Macom was awarded money by DoD from CHIPS act for GaN-on-Sic.



So, if GaN-on-Si is at 8x GaA and GaN-on-Sic is 2x of Si and GaN-on-Diamond is 3x of Si.

Then, it would reason that 3000 T/R module GaN-on-Diamond may emit something close to 1MW in power.

I think all the computation power is probably going to be at most 10% of that. Even 8 Ascend-910B GPUs + 16 CPUs @ 300W each would be at most 7.5kW. Not in the same ball park.

Thermal management system needs to be able to handle 1MW in cooling requirement if you want to use the most powerful ever radar. Which is GaN-on-Diamond.

So peak power consumption requirement might be higher than 1MW!

so how much thrust is needed to support that much power generation?

Now, let's say you have nice battery and electric generator technology that is highly efficient and can shave peaks of your demand and that you only need to sustain 800kW power

LM-2500+ at 5.25t can generate 35MW of power
Please, Log in or Register to view URLs content!

A modern large turbofan of 1.7t (around WS-10C/WS-15 weight?) should be able to generate 10-15MW of power if that's all it did. As such, it doesn't really need to divert that much of its thrust to power the electrical stuff.

So, the key is still to have enough space for the plumbing needed to cool all that power generation from RF coming out of the aircraft.

Just bear in mind that GaN is a lot more efficient at converting electricity into RF energy, so cooling requirements are a lot less than for GaA.

If they did go with 3000 T/R GaN modules, it could track many more targets simultaneously which would be useful for directing CCAs.

Or alternatively, let's say they divert all that extra energy to detect a small number of targets. I'm thinking this gives it significantly more detection range against opposing stealth 2 engine stealth aircraft.
 

tphuang

Lieutenant General
Staff member
Super Moderator
VIP Professional
Registered Member
This is even more interesting given that USAF also sees 1MW power generation and cooling as a benchmark for its own future fighter as well.

View attachment 140874
okay, I have not seen this before, but now I'm thinking that maybe 1 MW power requirement is too conservative. Maybe the DEW and additional EW system placed in other part of the aircraft will take up a lot of power too.

but one thing clear to me is that F-35 is really going to suck in a few years. 80kW is nothing. That's at most 4x the power supply of the radar.

but if you compare a 1MW radar unit vs 17kW radar unit for peak power. That's 58x more
range will be 7.6x longer vs same target

Your processor will be processing 58x more information. Which for modern processor actually isn't so difficult actually.

google AI says
To effectively run inference on GPT-3.5, most experts would recommend a cluster of 8 NVIDIA A100 GPUs; this allows for a good balance between performance and cost for typical usage scenarios, with the possibility of scaling up or down depending on the required throughput and latency demands.
and it already had 175m parameters. Many of the modern large open source models, you can run on 1 or 2 4090 as long as you have enough VRAM.

even if we 2x that, you are looking at most 15kW for computation. Still just a fraction of the radar power.
 

Warwolf27

Just Hatched
Registered Member
Hello. I'm new here and I'm mostly a reader. But the discussion here is interesting. Aren't the radar power electronics and its processing system one thing and the mission electronics another? Also, they're talking about integrating AI to manage everything and graphics processors as the engine for that. Is this (broadly speaking) correct or is there some conceptual error? I understand that the normal thing is to have hardened but modular electronics for the systems and integrate them with a system that manages them. Everything around a data bus. Flight electronics are one thing, mission electronics are another, etc. Are they talking about integrating everything with a dedicated AI that can manage everything? I thought the current focus of AI was something around process automation and support in pilot decision making. Greetings to all and apologies for the bad syntax, Google translate has a tendency to talk like Tarzan and my English (zero) is insufficient to correct it.
 

AndrewS

Brigadier
Registered Member
okay, I have not seen this before, but now I'm thinking that maybe 1 MW power requirement is too conservative. Maybe the DEW and additional EW system placed in other part of the aircraft will take up a lot of power too.

but one thing clear to me is that F-35 is really going to suck in a few years. 80kW is nothing. That's at most 4x the power supply of the radar.

but if you compare a 1MW radar unit vs 17kW radar unit for peak power. That's 58x more
range will be 7.6x longer vs same target

Your processor will be processing 58x more information. Which for modern processor actually isn't so difficult actually.

google AI says

and it already had 175m parameters. Many of the modern large open source models, you can run on 1 or 2 4090 as long as you have enough VRAM.

even if we 2x that, you are looking at most 15kW for computation. Still just a fraction of the radar power.

Yes, a target would be painted with the same amount of RF energy (and at a wider range of frequencies) at a distance 7.6x further away.

But the return signal would also be 7.6x weaker and presumably there is no difference in receive module sensitivity.

So the same amount of information is being processed, but you need better algorithms to pick out such weak signals from the noise.

---

In any case, my guess is a bare minimum of a 2x improvement in X-Band detection range against something like an F-35.
 
Last edited:

AndrewS

Brigadier
Registered Member
I think you might be on to something here, this could very well be it. I’ve actually not come across this in detail before. It’s a very simple and elegant solution.

If it indeed turns out to have 3 “engines” - I’m throwing my hat in the ring that this is the reason.

The turbo generator would be the middle “engine”, providing electric power to spin engines 1 and 3 in turbofan mode (and maybe producing some tiny thrust itself?). For the jet cycle it would continue to provide power but with fuel now pumped into combustors of engines 1 and 3.

Engines 1 and 3 only need to be uprated and then heavily modified WS-15s with the electric motors added. But none of this is actually bleeding edge technology and materials science.


In a combat scenario, you want as much thrust as possible.

Given a heavy airframe and 3 engines, then you want all those engines to be capable of low-bypass afterburn to maximise available thrust.
Plus presumably the radar (or weapons) are consuming a lot of electricity as well.

But in cruise mode, perhaps only the centreline engine is combusting fuel and the other engines are shut down. But the fans on the other engines are being powered by electricity. So they are also pushing air through the engine and generating thrust.

So you end up with a much higher bypass ratio and therefore fuel efficiency?
 

ZeEa5KPul

Colonel
Registered Member
Yes, a target would be painted with the same amount of RF energy (and at a wider range of frequencies) at a distance 7.6x further away.

But the return signal would also be 7.6x weaker and presumably there is no difference in receive module sensitivity.

So the same amount of information is being processed, but you need better algorithms to pick out such weak signals from the noise.

---

In any case, my guess is a bare minimum of a 2x improvement in X-Band detection range against something like an F-35.
You take the fourth root to find the detection range, so the detection range with 58x power is 2.76x.
 

Kich

Junior Member
Registered Member
I'm seeing in multiple instances of weibo discussion under 6th gen fighter it's mentioned both CAC and SAC are doing competing designs. CAC is faster and their prototype is about to fly hence all the excitement. SAC isn't too far behind and their prototype is due to fly sometime next year.

Is this where all these J-XD1 and J-XD2 discussions are coming from?
If PLA is able to introduce and fly two prototypes for a 6th gen fighter, they should keep both; to complement each other.

Like currently many in the US are regretting ending the YF-23 development after it lost to YF-22. The YF-23 could have solved the range issue that US faces in the Pacific with its nearly 3K miles and 65K ft ceiling versus 2K miles and 50K ft ceilings for the F-22.

I don't know but having two competing air frames would keep both industry busy as well. It would be short-sighted if they just picked one and left the other into the dust.
 
Top