wmh829386 wrote: ↑Tue Jan 18, 2022 6:44 pm
2) According to Friedman, it's not a simple FtP. Rather a feedback process more similar to rate control with the unfortunate caveat that it is subjected to the limitations of HACS assumptions (and god help you if a fictitious input have been used for wind correction)
Ships completed in 1940 had a new Gyro Rate Unit (GRU) mounted on the director, to measure horizontal and vertical angular rates directly. Although this measurement might be described as tachymetric, it could not transform the HACS into something capable of dealing with climbing or diving targets. Nor did it solve the problem of measuring target inclination. That was still fed in by the control officer.
The Gyro Rate Unit Box (GRUB) in the calculating position below decks received the two rates and the angle of presentation from the GRU, ultimately to feed data into the HACT nearby. This was still a feedback process. The assumed target ground (plan) speed was set on the GRUB. Given an angle of presentation and a target height and range, this speed implied particular vertical and horizontal rates. It was adjusted until these rates matched the observed rates, the GRU acting in effect as a feedback mechanism.
Your quote of Mk37 manual just shows the time for FC solution to converge in Automatic mode for Mk37 is typically faster tham matching rates even when looking directly at the target.
3) This is hilarious.
The mode of operation without GRU:
HACS output director sight angles and (clock=radar) range to electric height finder. So the GO adjust the speed and angle of presentation until the electric height finder stays on target. In other words,
rate control without provision of climb rate.
4)According to Mk37 manual, automatic mode is more suitable for manuvering targets anyway.
5)With that in mind, I doubt he is calculating for targets that violates assumptions of HACS, a case which is amost impossible to quantify.
2) it is basically FtP because the operators are matching rates and the machine is doing the calculation just as Mk1 does the calculations in rate control the difference being that GRUB was optimized for this one process using relative rate linkages (as in HACS/FKC) rather than linear rate as in Mk1. In practise Mk33/37 could not deal with turning, climbing or diving targets either (ToF, target maneuver and computer lag times defeated accurate FC solutions) and we know this for a fact from the drone trial results and real world results
3) see 4.
4) If the target changes course/accelerates/changes altitude/the layer and/or trainer allow their sights to wander due to ship motion/etc. the Automatic Rate Control is then presented with an unsolvable problem.
Real world results show that Mk37 was ineffective when firing MT ammo and applying MT fuze and DT errors to drone trial results verify this. Friedman:
...In this connection it was argued that long range fire had inherent limitations; a pilot could
always steer away while the shell was in the air. New bombsights drastically reduced the time he had
to stay on course before dropping a bomb. On that basis DNO wrote in 1942 that there was little point
in adding a tachymetric element to the HACS and FKC (actually, to replacing them)...
Ivan Getting (helped develop SCR584 and Mk56):
Between World War I and I1 this range-keeper had been modified
for antiaircraft use by adding the third dimension airplane
altitude. But the Mark I computer or range-keeper in use with
the Mark 37 director and the FD radars had all the limitations
built into the system. For example, the equipment was so large
that only one could be installed in a destroyer which usually
had three twin gun turrets. So, while the guns could in principle
engage three different targets, the overall system was limited
to one attacking plane at a time. It was necessary to insert the
estimated heading and speed of the aircraft; and, even though
the radar could provide bearing, elevation and range, the computer
arrived at the solution exponentially. If the airplane
changed course, or if the original estimates of course and speed
were in gross error, the correct solution was not available until
the airplane had released its weapons. Many of these deficiencies
of the Mark 1 range-keeper turned up when a dynamic
tester developed by Section D-2 of NDRC was made available
to the Radiation Laboratory. These tests further demonstrated
that in the interval between World War I and 11, when the modifications
had been made to adapt the range-keeper to three-dimensional
use, simplifications in the computation of the roll
of the ship resulted in substantial errors. These limitations were
not fully understood within the Navy; and, in any case, for substantive
reasons, the Bureau of Ordnance was loath to make
any changes in the combination of the FD radar, the Mark 37
director, Mark 1 computer directing the five-inch guns. These
not fully understood within the Navy; and, in any case, for substantive
reasons, the Bureau of Ordnance was loath to make
any changes in the combination of the FD radar, the Mark 37
director, Mark 1 computer directing the five-inch guns. These
equipments were in large production, all the ships at sea had
been equipped, training schools had been established and spare
parts and maintenance were available. Ships return only every
two years or so for major overhaul and the prospect of making
any serious changes in such a complicated and integral system
was more than the Bureau of Ordnance was willing to accept.
5)Pout started his analysis by making a series of estimates for the various errors that effect the accuracy of HAA fire using an unstabilized HADT. These include range errors (no radar), stabilization errors, layer and trainer errors and MT fuze timing errors, etc. He arrived at a number which roughly matched actual HACS performance at long range before radar:
Case1
First, experience in the field. An operational analysis of the performance
of ships carrying the HACS, which were generally of light cruiser size
upwards, during the early years of the war suggested that about 10000
shells were fired per kill achieved, that is, Pk = 1 in 10000. The analysis
has been related to the light cruiser situation with a future range of 6kyds
in moderate sea states, and should therefore not be too far removed,
on average, from this result.
And by definition these are maneuvering targets that were trying not to get hit, and many of which were high altitude, high speed targets. Therefore Pout greatly exaggerates the various errors to match observed kill rate (Pk) results which he states as:
ToF
16 sec = 1/10k Pk (Rp=7.6K yds and Rf=6K yds)
10sec = 1/3.2K
_6sec = 1/2K
Pout then analyses various improvements in ranging, stabilization, etc to come up with his estimates in later Cases. Given his starting point the analysis includes all targets, even those not flying straight and level,
[which is also the case for Lundstrom's analysis cross referenced with 5in ammo expenditure per engagement.
But we already agree on this, right?
I can understand the appeal that Mk37 must have had to the RN, as it seemed to solve so many problems, and was so well engineered that it just had to better than HACS/FKC, and BuOrd claimed it to be so...yet it wasn't because the fundamental issue centred on MT fuze errors and ToF, which allowed a target to evade predicted fire. Even after VT was introduced ToF defeated Mk37, if the target didn't cooperate.