MiG-29 Avionics

[Aeria Gloria]

Idk how that’s related to look up/notch/4/4 but okay

 

[0minutes]

Idk how that’s related to look up/notch/4/4 but okay

I was talking about the radial speed limit you mentioned in RNP

 

[Aeria Gloria]

I see

 

[Aeria Gloria]

What would be a computer too weak to do this? How do you quantify or know the needed computations for N-019?

 

[0minutes]

BZPP-4D allows for radial closures of 1200kmh
BZPP-4 allows CC Radar locks of beyond 10km

What would be a computer too weak to do this? How do you quantify or know the needed computations for N-019?

Again, this is still WIP research, but a contrary theory to a no radial limit is:

The target speed range of −2200 to +1100 km/h in “Tail-on” (Догон) mode was may be limited by the radar’s digital processing system, rather than by what the radar could physically measure. The radar’s analogue tracking system is capable of detecting higher speeds, but the digital system used a 15-bit fixed-point format that imposed limits.

There are three possible reasons why the software may have these imposed limits:

- Resolution vs. speed range: The digital system represented values between −1.0 and +1.0. Increasing the maximum speed would require a larger scaling factor, meaning each digital step would represent a larger change in speed and reduce accuracy. Limiting the target speed to +1100 km/h allowed the system to use a smaller scale, giving a resolution of roughly 0.067 km/h per step. This would help with accurately tracking and filtering small changes caused by noise.
- Preventing overflow: If a value exceeded the limits of the 15-bit format, the binary value could overflow and wrap around, potentially turning a large positive number into a negative one. The software likely kept the speed within a safe range to prevent this kind of sign error, which could cause serious problems in later calculations.
- Closure-rate calculations: The radar needs to calculate the relative or closure speed between the fighter and the target. By restricting the target's allowable speed range, the software could ensure that subtracting the two velocities stayed within the numerical limits of the system. This avoided overflow and reduced the need for additional error-handling in the guidance and tracking calculations.

Again, these are educated guesses since documentation is not available, however signed 15bit should be enough for +2500/-2500kmh and no manual mentions limitations in Lock mode, only in search, so most likely it doesnt have limits, but there are always other possibilties.

 

[Aeria Gloria]

Idk how we can say x number of bits is enough to have no radial speed limits but I’m not a radar math person either. Su-27 uses C100, it can track through notch in over 3° lookup under 30 km range.And its manuals don’t even mention speed limitations in search mode except that target must be over 180-200 kmh.

The C100 is also not the only computer for radar. The radar is its own analog-digital computer that does things on its own. The C100 is merely an “executive” device sending out “flags” of different kinds to different devices depending on read conditions changing. I would bet the radar is basically doing all the target finding and processing, the C100 is telling radar to turn on, switch to this mode and this position, oh you found targets here I’ll display it for the pilot on HUD. Send flag to go into lock mode. Pilot presses break lock send flag to return to search and reset.
That kinda stuff. I’ve heard this from others who looked much deeper into MiG-29 then I such as “Kosh.” So It’s not my own assumption, but reading MiG-29B it makes perfect sense

 

[0minutes]

reading MiG-29B it makes perfect sense

The information I talked about is from the MiG-29B, and also I prefixed this entire conversation with the fact that these are all educational guesses. There might as well be no constraints in lock mode. Also I love how we were able to change your mind about the radial speed limit, you used to be so hard set on the idea that it did now you defend it so much , nice to see people can change their minds...

 

[Aeria Gloria]

I don’t ever remember being set against it. I gladly conversed with the first people who brought it up to see why they thought this and posted a bug report on ED forums within 24 hrs. Please, you consistently mistake my caution and waiting to get all the facts I reasonably can for things that are far more sinister, when they aren’t! But whatever. You learn patience and caution with time.

I may have seemed to be “against it” when someone randomly says “I’m pretty sure it’s doesn’t have these limits” out of the blue, which is why I listened to his explanation.
No one should be changing their mind on a moments notice from stranger input, even if you want it to be true or it “makes sense”.

R-27R antenna is 4 mini cassegrains in one.,

 

[LukaszK]

If anyone here has documentation on the Ts100 in detail, and the firmware, I'd really appreciate it! Thank you!

Look at the topic related to soviet vs western computer comparison. There was some block diagrams from hardware point of view. I doubt there are some documents showing programming architecture.

BZPP-4D allows for radial closures of 1200kmh
BZPP-4 allows CC Radar locks of beyond 10km



Again, this is still WIP research, but a contrary theory to a no radial limit is:

The target speed range of −2200 to +1100 km/h in “Tail-on” (Догон) mode was may be limited by the radar’s digital processing system, rather than by what the radar could physically measure. The radar’s analogue tracking system is capable of detecting higher speeds, but the digital system used a 15-bit fixed-point format that imposed limits.

There are three possible reasons why the software may have these imposed limits:

- Resolution vs. speed range: The digital system represented values between −1.0 and +1.0. Increasing the maximum speed would require a larger scaling factor, meaning each digital step would represent a larger change in speed and reduce accuracy. Limiting the target speed to +1100 km/h allowed the system to use a smaller scale, giving a resolution of roughly 0.067 km/h per step. This would help with accurately tracking and filtering small changes caused by noise.
- Preventing overflow: If a value exceeded the limits of the 15-bit format, the binary value could overflow and wrap around, potentially turning a large positive number into a negative one. The software likely kept the speed within a safe range to prevent this kind of sign error, which could cause serious problems in later calculations.
.

I doubt if speed limits have source with limited resolution of digital words in C100.
What you said 0.067km/h - this resolution has nothing in common with speed resolution of radar.

Doppler band is analyzed by analog , or digital-analog filter bank.
I am writing from memory but there was like 20 filters for 2.5kHz bandwith (total 4x2.5 -> 10kHz, and 6 such filter banks: so 60kHz x2 for HPRF or 6x, or rather 12. So total 20 x4 x12 so total 960 filters.
But returning to point: 2.5kHz /20 filters - means each filter has central frequency separated by 125Hz.
This results in speed resoultion : (let assume for a while carrier freq fc= 10GHz - so 10^10)

f_doppler = 2* v/c * fc -> v = (f_dopp *c) /(2*fc)

v_resolution = (125Hz * 3 *10^8) / (2 * 10^10) = 125 *3 /2 *10^(-2) = 1.875 m/s -> 6.75 km /h

Possible limits +1100 / -2200 may be from spectrum wrapping (aliasing) in MPRF or some way of processing ( as you know there are needed multiple PRF to find real value of distance and closure speed)

 

[0minutes]

Su-27 uses C100

Yes this likely means that there is no radial limit in lock,
thanks!

 

[0minutes]

Possible limits +1100 / -2200 may be from spectrum wrapping (aliasing) in MPRF or some way of processing ( as you know there are needed multiple PRF to find real value of distance and closure speed)

When you use STT in the MiG-29 the radar uses only one single velocity gate tracking filter that continuously tunes and centers its frequency on the target's predicted Doppler speed using Kelmans FIlter.

The radar cycles between PRFs at different ranges and this is done only to eliminate range blind zones, most importantly the official N019 radar technical manual does not contain any method or precaution for shifting PRFs to avoid speed/frequency blind zones.

Because the PRF width is narrower than the actual spread of true velocities, the radar's internal spectrum folds over on itself. This creates a Composite Doppler Profile where the high-power Main-Lobe Clutter (MLC) peak repeats at exact intervals.

This causes MLC peaks to fixed and repeated in the folded spectrum "every 1100kmh"

In lookdown there are 3 scenarios:
When the target has no radial closure (0 closure), their doppler drops down to 0 and the target is burried within the MLC peak, this causes the radar to start tracking the ground instead, (which can be noticed by the rhombus and targetting circle being stationary/moving only in relation to the ground)
When the target has +1100kmh radial velocity the target is closing, which causes its doppler frequency to fold exactly into the next MLC peak, which again causes the radar to start tracking the ground OR the target IF and only IF their signiture is strong enough at very low ranges (under 10km like the manual cites) because bare metal reflects EM waves a lot better than dirt, so the relative target signature is stronger than the MLC
When the target has -1100kmh radial velocity the same occurs as with +1100, their true doppler frequency folds again onto the next negative MLC peak, causing the radar to start tracking the ground or target at low low ranges (under 10km)

However, in lookup situations everything changes.
Since the 1100kmh radial velocity limit only applies when MLC is present, when the radar antenna is pointed upwards, the MLC disappears, this means the target doesn't fall into any doppler frequency bright spots caused by MLC and only has to combat the SLC. This can also occur in shallow lookdown cases, or at high altitude lookdown cases as the MLC is going to be weaker than the target signature or the lesser your radarn is pointed downwards.

All this information was obtained from the MiG-29B on filtering algorithms for ОБЗОР and РНП 3.3 Измерение координат в режиме непрерывной пеленгации page 138.

 

[0minutes]

in ОБЗОР (search) mig-29 radar scan frequencies just between 2 nearest MLC peaks

The N019 uses a 3-PRF schedule to cross-reference and unwrap true velocities. Because it can only resolve 3 adjacent zones, it has a strict physical speed limit spanning -2,200 km/h (receding/tail-chase) to +1,100 km/h (closing/head-on)

 

[Aeria Gloria]

which causes its doppler frequency to fold exactly into the next MLC peak, which again causes the radar to start tracking the ground

Idk what passage you are referencing for it to transition from 4/4 to ground track, as I answered you on ED forums this is only for look down situation in modes D/AUTO in ZPS, which would be angles of 3/4-0/4.

And I don’t think MLC ever truly disappears. Only becomes small enough for high enough SNR to track in it.

 

[0minutes]

Idk what passage you are referencing for it to transition from 4/4 to ground track, as I answered you on ED forums this is only for look down situation in modes D/AUTO in ZPS, which would be angles of 3/4-0/4.

And I don’t think MLC ever truly disappears. Only becomes small enough for high enough SNR to track in it.

This isn't the ED forums, I'm talking about the Persuit radial speed limit in general right now, not specifically the lookdown 4/4 track.

 

[0minutes]

Regarding tracking in 4/4 aspect (notch/0 radial velocity):

The RNP/STT Doppler processing is not simply a fixed "notch filter" that rejects everything below a certain radial velocity. The received target signal is first selected by range and passed through the Doppler-processing channel. The system then uses an extrapolated Doppler frequency (F_de), calculated from Doppler measurements from the previous tracking cycle, to control a local oscillator. After mixing, the resulting frequency is essentially determined by the difference between the actual target Doppler (F_d) and the predicted Doppler (F_de).

The signal is then passed through a narrow 10 kHz filter centred on the reference frequency. When the actual Doppler matches the predicted Doppler, F_d = F_de, the signal falls directly in the centre of this filter. Therefore, the filter effectively follows the predicted Doppler of the tracked target rather than remaining at one fixed velocity. If the target's Doppler changes, the signal moves away from the filter centre until the system updates its prediction during the next tracking cycle.

This is particularly important for look-down/notch situations. If the target performs a notch manoeuvre and its radial velocity decreases, its Doppler frequency also decreases. Since the tracking channel follows the predicted target Doppler, the filter will follow this decreasing Doppler as well. If the predicted target Doppler enters the same Doppler region occupied by MLC or SLC, the 10 kHz tracking filter can overlap the ground clutter. The issue is therefore not necessarily that the radar has a fixed "same-speed filter" that rejects the target; rather, the target's predicted Doppler can move into the clutter region, allowing the much stronger ground return to compete with or overwhelm the target return.

 

[Aeria Gloria]

Or don’t forget how SLC altitude return line stops reception of targets with similar radial velocity or small closure in rear aspect, which any pulse Doppler radar is going to have issues with as the sidelobe is hitting ground directly below you.

 

[0minutes]

Su-27 uses C100

Su-27 has 2 C100 compute blocks, that's double the compute power.

The C100 is merely an “executive” device sending out “flags”

Ts100.02 determines the required operating modes and parameters of the radar blocks every radar cycle and sends them over a 16-bit digital bus such as selecting the PRF and in РНП its responsible for the Filter. The Radar doesn't have an internal computer per say except for the primary signal processing is done by the N019-03 receiver by taking radar input and turning it into information that the Ts100 computer can use,

RLPK-29 = N019 radar + Ts100 + associated interface/communication equipment

N019-01 - antenna/waveguide system
N019-11 - antenna control and synchronisation
N019-02 - transmitter
N001-22M - master/setting generator
N019-09 -high-frequency receiver
N019-03 -receiver / primary signal processing
N001-25 -information conversion unit
N001-35 -I/O unit
N001-45 -communication with SEI
N019-05 -Ts100.02.01 computer

 

[Aeria Gloria]

Su-27 has 2 C100 compute blocks, that's double the compute power.

What makes you think MiG-29 only has one? Or that both are for radar?

MiG-29B book BCVM chapter near beginning gives all the info and more you probably want on C-100 starting at page 29

There is C-100.02.06 which handles Radar and KOLS, and C-100.02.01 that handles the S-31 computer (OEPrnk), SEI, and other analog devices such as missiles.

C-100.02.01 also does radar work for the COOP mode and other cases

 

[0minutes]

Regarding tracking in 4/4 aspect (notch/0 radial velocity):

The RNP/STT Doppler processing is not simply a fixed "notch filter" that rejects everything below a certain radial velocity. The received target signal is first selected by range and passed through the Doppler-processing channel. The system then uses an extrapolated Doppler frequency (F_de), calculated from Doppler measurements from the previous tracking cycle, to control a local oscillator. After mixing, the resulting frequency is essentially determined by the difference between the actual target Doppler (F_d) and the predicted Doppler (F_de).

The signal is then passed through a narrow 10 kHz filter centred on the reference frequency. When the actual Doppler matches the predicted Doppler, F_d = F_de, the signal falls directly in the centre of this filter. Therefore, the filter effectively follows the predicted Doppler of the tracked target rather than remaining at one fixed velocity. If the target's Doppler changes, the signal moves away from the filter centre until the system updates its prediction during the next tracking cycle.

This is particularly important for look-down/notch situations. If the target performs a notch manoeuvre and its radial velocity decreases, its Doppler frequency also decreases. Since the tracking channel follows the predicted target Doppler, the filter will follow this decreasing Doppler as well. If the predicted target Doppler enters the same Doppler region occupied by MLC or SLC, the 10 kHz tracking filter can overlap the ground clutter. The issue is therefore not necessarily that the radar has a fixed "same-speed filter" that rejects the target; rather, the target's predicted Doppler can move into the clutter region, allowing the much stronger ground return to compete with or overwhelm the target return.

View attachment 823885

A few more additions, and more raw look at what everything means:

1. When signal is recieved it goes through a 130MHz filter passband from a f_pr2 = 28.055MHz centre (so +/- 65Mhz on either side)
F_c = (f_pr2 + F_doppler)

if its outside the range, it is rejected

2. 10 kHz filter used for speed measurements where

f_pr3 = 5.0475MHz (centre)
ΔF = (F_doppler - F_extrapolated)

ΔF_c = f_pr3 + ΔF

then the signal then goes through the 5.0475 MHz quartz filter with a 10 kHz passband

if ΔF = 0 then the target is placed at the centre of the f_pr3 and therefore prediction was good, or if it is inside the 10kHz filter the signal is passed onto the next stage

if ΔF_c is too far from f_pr3 then the signal is rejected, though the manual is unclear and just says that it filters inside a 10kHz passband gate filter
"фильтра с полосой пропускания …(Δf = 10) кГц"

3. then this ΔF_c is passed onto the 1.6kHz passband gate filter which is used for the angular/ direction-finding receiver
where only now all the signals outside the +/-800Hz range are "supressed"

received Doppler -> 130 kHz filter -> 10 kHz filter -> 1.6 kHz narrow-band selector

The 1.6kHz filter is the main selector for selection based on speed

 

[Aeria Gloria]

Yes, just Those search mode filters only apply to High PRF mode and it’s 130 kHz, your not gonna subtract -65 MHz from 28 MHz and look at negative hz. Medium PRF does it a bit differently.

 

[0minutes]

Yes, just Those search mode filters only apply to High PRF mode and it’s 130 kHz, your not gonna subtract -65 MHz from 28 MHz and look at negative hz. Medium PRF does it a bit differently.

My mistake, the 130kHz one is a pre-selection filter, and, its +/- 65kHz not MHz. Honest mistake

Target velocity tracking and closing-speed measurement channel

For measuring closure speed a signal from the first channel of the antenna's system (I__k) is used. Signal is created and amplified in the high frequency unit (в.ч.ПРМ), amplified in the УЧП28 and fed to distance gate (СД5). This gate is opened with a strobe СТР.Дпел which is created the moment the pulse arrives back from the tracked target. This way a filter based on distance is done. Following the gate is a pre-selection filter with a passband of 130 kHz and a centre frequency of 28.055 MHz. Signals across the entire possible range of Doppler frequencies pass through this filter.

From the output of the pre-selection filter, a signal with a frequency fс equal to

fс = fпр2 + fд

where fпр2 = 28.055 MHz; fд = (2Vсб)/λ is the doppler frequency shift Vсб is the closure speed and λ is the wavelength, is fed to the СМ5 mixer

 

[0minutes]

Medium PRF does it a bit differently

130 kHz filter: very wide pre-selection filter centred at 28.055 MHz. It passes the entire possible Doppler-frequency range.
10 kHz quartz filter: after mixing, it provides the intermediate narrow filtering around f_пр3=5.0475 MHz. This is the stage that converts the pulsed Doppler signal into a continuous sinusoidal signal.
1.6 kHz narrowband filter: later in the angular-bearing receiver, it is explicitly described as the main speed-selection element. Signals outside its ±0.8 kHz linear/discrimination region are suppressed.
The 1.6 kHz filter is therefore the much narrower final velocity-selection filter, while the 10 kHz filter is an earlier intermediate-frequency filter used in the Doppler processing chain.

those pages do not explicitly say that the described Doppler-processing chain is HPRF-only, nor do they mention MPRF. If you can tell me where it says MPRF does it differently I'll look into it! Right now im doing this one by one.