Mk3 and Type 284 radar

Guns, torpedoes, mines, bombs, missiles, ammunition, fire control, radars, and electronic warfare.
User avatar
Dave Saxton
Supporter
Posts: 3148
Joined: Sat Nov 27, 2004 9:02 pm
Location: Rocky Mountains USA

Mk3 and Type 284 radar

Post by Dave Saxton »

I'm not going to get involved in those other controversial threads, but there seems to be some confusion about the effectiveness of the decimetric fire control radars to determine usefull bearing data. These radars were not always "useless" in this regard. I think I can focus on the USN Mk3 and the British Type 284, without giving away prematurely my findings on German radars.

Bearing resolution is the ability to distiguish two or more closely grouped targets from each other -side by side. A wide (horizontal) beam will make this more difficult, because two or more targets might be within the beam. Actually, even if two or more targets are within the beam, they could be distiguished if they are not too close, because the beam is much more strong toward the center and weak toward the outboard. How focused and tight the beam is, is determined by the antenna gain. At the decimetric wave lengths, antenna gain is determined mainly by how many dipoles, as measured in wave lengths, are arrayed in front of the reflector. Determining antenna size, or gain, by just the area of the reflector will not result in correct specs (unless it a wave guide window), but a larger reflector does allow more dipoles to be arrayed. Doubling the dipole sum length; increases gain by a factor of four. Likewise, decreasing the sum dipole length in half; decreases gain by a factor of four.

Mk3 had a beam width of 9 degrees, using an antenna with a sum dipoles length of 4 wave lengths. The first Type 284 that the British battleships had at the time of the Bismarck chase, used send and recieve antennas with 12 wave lengths worth of dipoles each. This gave a beam width of 3*, (close to that of MK8 BTW). By aiming the antenna so that the max signal strength was recieved, the bearing accuracy could be attained within about 1* This however, isn't accurate enough for fire control purposes.

Bearing accuracy is quite a different thing from bearing resolution. Bearing accuracy required of gun laying can only be obtained by lobe switching. The most common lobing technique was first invented by Rudolf Kuhnholdt and Hans von Willisen as they devolped the prototype Seetakt, and they called it Peil Verfaren A/N. This type of lobe switching involves aiming two beams slighty to the right and left of center. Then if the signal strength of the two beams are matched on the bearing scope, the antenna is aimed right at the target. Bearing accuracies to within 0.1* are typically attained. Both Type 284 and Mk3 used this method to fix the exact bearings of the target.

There were draw backs to this method, because it essentially reduced the number dipoles being used in 1/2. This of course reduced the gain by a factor of 4, so it reduced the effective range, and it degraded the bearing resolution. For example, Mk3's beam width was increased from 9* to 15* while lobing. This created an nasty trade off for Mk3, because when it's bearing accuracy was good enough for targeting, the bearing resolution may have been such, that it could no longer distiguish the target from other targets or only to range that was unsuffciant for the task. This was why the German Navy decided to not use this type of lobe switching until a better method of lobing could be employed.

Type 284I used six yagi antennas so that Lobe switching could be employed, but these antenna arrays had less gain. Nonetheless, the Type 284I had limited blind fire capability, and it was generally more capable than the USN Mk3. DoY 's Type 284 still required the use of star shell, and it required other ships to help spot the fall of shot at North Cape, but it's bearing data was quite suffciant for the task.
Tiornu
Supporter
Posts: 1222
Joined: Mon Oct 25, 2004 6:13 am
Location: Ex Utero

Re: Mk3 and Type 284 radar

Post by Tiornu »

The British felt that numerous DoY shells at North Cape were wasted because, though they were correct in range, the bearing data was inadequate.
Brad Fischer
Member
Posts: 71
Joined: Thu Apr 21, 2005 4:48 pm
Location: USA

Post by Brad Fischer »

Hi Dave,

Are you working off of theoretical calculations or source documentation? I have two Type 284 manuals covering the basic Type 284 and the Type 284M/P (HANDBOOK on the USE OF RADAR FOR GUNNERY PURPOSES G.S. SET – TYPE 284. 1944). Both documents state clearly that the bearing resolution is +/- 4 degrees.

USN’s definition of beam width is different than the RN’s. Beam width is defined as “…the angle between the two directions for half power on the round trip.” The Mark 3 Mod 0 and Mod 2 – which was the battleships mostly used – is listed as having a bearing resolution of 5 degrees regardless of lobbing. The Mark 3 Mod 1 and 3 – the narrow 6’ x 6’ antenna – is listed has having a resolution of +/- 10 degrees.

Incidentally Duke of York’s gunnery narrative states that bearing track was considered good but there was no ability to spot in deflection and there was significant doubt that her salvos were “on for line” during the blind fire conditions (which was most of the time). The USN had similar problems with the Mk 3 sets out in the Pacific even without the weather conditions and the cross-roll problems that it may have presented.

Brad Fischer
User avatar
Dave Saxton
Supporter
Posts: 3148
Joined: Sat Nov 27, 2004 9:02 pm
Location: Rocky Mountains USA

Post by Dave Saxton »

Yes, I'm working from primary sources, and using data listed in those sources. A USN manual that gives the bearing width specs (RN literature also gives these specs) also states that Mk3 was good at spotting the fall of shot for range, but determining the bearing of the splashes was problematic.

There's another angle to the overall resolution question and particularly when it comes to spotting. This is the signal to noise ratio of the reciever and the overall noise levels. There will come a point when the strength of the return signal will fall below the noise level, and the true signal can't be distiguished (resolved) from the noise. Indeed the sensitivity of the reciever and it's nominal S/n ratio has as much or more to do with effective range than transmitted power. This is why the effective range of any radar will be dynamic. The sensitivity of the reciever is not a constant, but there comes a limiting point of reciever gain, when the noise level becomes too great.

The British literature quantifies a radar's ability to spot the fall of shot by signal to noise ratio. The ability to resolve shell splahes from sea clutter is profoundly effected by the quality of the receiver and it's signal to noise ratio. In adverse weather, with greater levels of sea clutter, with overall greater noise levels will reduce spotting capabilities, particularly for bearing. The noise levels and the ability to see through noise has much to do with resolution of a target on the displays, and the ability to deal with adverse weather, sea clutter, and noise jamming... ect...

Decimetric radars were only made possible by the advent of the RCA Acorn receiving pentodes. Previous to this, the max frequancy that could be recieved was about 200mhz (150 cm). These vacuum tubes were known to the worldwide electronics community and most of the top flight tube manfactures built copies. Philips in Holland built a very good copy, and both GEC and Mullard built copies in England. However, one problem with Acorns was that they developed a serious noise problem starting at 400mhz, increasing steadly through 1ghz. With pure vacuum tube technology it was found that the higher the frequancy the worse the signal to noise ratios of the receivers. The point were the S/n ratio became unmanagable was 27cm.

The British eventually switched to a different design with their 50 cm recievers. This was grounded grid triodes in place of pentodes. This lowered the noise floor allowing more aceptable performance up to 600mhz, but it also restricted the signal strength of the IF signals, and the receiver sensitivity, so it was only a marginal improvement. This did however, allow for somewhat better reception, and better resolution at 50cm. Many of these resolution problems with the 40cm and 50cm radars had to do with the S/n ratio problems.
Brad Fischer
Member
Posts: 71
Joined: Thu Apr 21, 2005 4:48 pm
Location: USA

Post by Brad Fischer »

The USN also rated “reliable ranges” as an S:N ratio of 4:1 or 2/3rds of maximum, quite similar to the RN. BuOrd’s assessment of the Mk 3’s spotting ability is mixed. The initial assessment was quite glowing particularly from the ships themselves with its introduction. However as the war progressed, the overall opinion declined, even from BuOrd itself as seen in the latter BOIs. There are several complaints discussing how spotting with the Mk 3 would tend to carry the MPI off the target even though it appears to be correct or even when the MPI appeared on and no change was called when in fact that pattern wasn’t even straddling.

Data that I’ve collected from the battleships and cruisers gunnery exercises show that the Mk 3 had an average spotting error of 2-3 times that of the Mk 8 or about 155yds. Interestingly, this corresponds to a good spotter’s direct spotting average on the Mk 48 26.5’ rangefinder at 20,000yds. Although the data isn’t as robust as the Mk 8 data there is that tendency for the MPI to be spotted off the target but the overall MPI control is better compared to optical shooting prewar. It should be noted, that the Japanese were impressed with the American control for range in the night battles off Guadalcanal where most of the ships had the Mk 3.

Brad Fischer
User avatar
Dave Saxton
Supporter
Posts: 3148
Joined: Sat Nov 27, 2004 9:02 pm
Location: Rocky Mountains USA

Post by Dave Saxton »

Hi Brad,

By all indications Mk3's range accuracy was very good. There are several accounts alluding to this, and this was one of the first things noticed with the first prototype. My sources indicate a typical range accuracy of +/- 40 yards. Actually this was about the average range accuracy attained by the centimetric Type 274 in trials.
User avatar
wadinga
Senior Member
Posts: 2471
Joined: Sat Mar 12, 2005 3:49 pm
Location: Tonbridge England

Excellent posts from very knowledgeable posters

Post by wadinga »

Gentlemen,

Can I just clarify my understanding on one or two items?

For radar ranging on a hard, good return target like a ship, +/- 40 yards sounds about right, because we are talking about transmission and re ception timing on a very accurate basis, related of course to frequency. Once we start bringing azimuth into consideration, I can see beam width presenting resolution problems, plus actual usage of the bearing data means referencing to ship's head or true north, and the assumption that a gyro conforms very closely with ship's heading over time. Which may not be true.

However, once we start talking about spotting error, ie spotting the fall of shot, we are attempting to differentiate between a number of relatively closely spaced water splashes which are not particularly good radar targets, and don't linger for very long. Also the spotter has to derive a complex weighted average, to be the MPI, on the fly.

Also I need to understand "spotting" in the context of the 26.5 ft rangefinder. My understanding was that the rangefinder was primarily concerned with ranging on the target vessel, and sampling that range as often as possible, and sending it to the fire control computer. This might occupy all the operator's time, and that given field of view constraints etc he would not be seeing fall of shot and estimating position relative to the target. My understanding is that the optical spotting is done by another individual with different optics who estimated the MPI of a salvo relative to the target. Maybe the crucial difference is that a stereo rangefinder allows the operator to do both, although if you are sharpening the image of the target, can you be sure to see and sharpen all the splashes at the same time?

I would like to understand better
Data that I’ve collected from the battleships and cruisers gunnery exercises show that the Mk 3 had an average spotting error of 2-3 times that of the Mk 8 or about 155yds. Interestingly, this corresponds to a good spotter’s direct spotting average on the Mk 48 26.5’ rangefinder at 20,000yds.
Is the "direct spotting average" a rangefinder range against a surveyed distance for calibration purposes? Is this +/- 155 yds at 20,000yds?

Hope you can help me grope towards a better understanding of the subject.

All the Best
wadinga
"There seems to be something wrong with our bloody ships today!"
lwd
Senior Member
Posts: 3822
Joined: Sat Jun 17, 2006 2:15 am
Location: Southfield, USA

Post by lwd »

There are a number of good articles on fire control here:
http://www.navweaps.com/index_tech/index_tech.htm
and here
http://www.eugeneleeslover.com/GS-USN-PAGE.html
User avatar
Dave Saxton
Supporter
Posts: 3148
Joined: Sat Nov 27, 2004 9:02 pm
Location: Rocky Mountains USA

Post by Dave Saxton »

However, we must not assume that USN procedures and equipment apply generally to the other Navy's as well. There are some parallels, but also some distinct differences.

The USN and the Royal Navy both mounted their firecontrol radar antennas to the existing optical firecontrol directors. This way the bearing indication for the radar was obtained by the relative position of the director, and the indication was communicated by an established method.

The radar displays and the operators were located inside the directors in USN practice. With Mk3 for example, the radar operator for bearing controled the traverse of the director. This way he could control the lobe switching procedures.

In the Royal Navy, the radar office and the operators for the fire control radars, were ofton romotely located from the optical directors themselves. With British firecontrol radars, we eventually see remote slave displays being located in other areas in the ship, and especially below decks. By 1944, the British usually had slave displays in the fire control computor rooms, so that the gun laying crews there could directly observe the radar data. Additionally in 1944, Type 284 was equiped with an additional scope specifically for spotting the fall of shot. This indication could be recorded by photograph at the time of fall in some cases.

The British continued to prefer A-scope or J-scope displays even for their late war centimetric firecontrol radars. They disliked Type B display for firecontrol radars unless it was just for coarse observation or general spotting. They beleived that the accuracy of data obtained from a Type B display was largely dependant upon the interputive skills of the operator. More finely precise data can be obtained from an A-scope or J-scope display it is true, but it also more abstract by nature. Spotting for bearing would probably have been easier with a less abstract type of display, and that might be why they went to seperate spotting displays. Post war, the British added a completely seperate radar system, exculsively to spot the fall of shot, to their centimetric firecontrol radar systems. Many major British warships continued to use the proven 50cm radar systems until well after WWII.
Brad Fischer
Member
Posts: 71
Joined: Thu Apr 21, 2005 4:48 pm
Location: USA

Re: Excellent posts from very knowledgeable posters

Post by Brad Fischer »

wadinga wrote:Gentlemen,

Can I just clarify my understanding on one or two items?

For radar ranging on a hard, good return target like a ship, +/- 40 yards sounds about right, because we are talking about transmission and re ception timing on a very accurate basis, related of course to frequency. Once we start bringing azimuth into consideration, I can see beam width presenting resolution problems, plus actual usage of the bearing data means referencing to ship's head or true north, and the assumption that a gyro conforms very closely with ship's heading over time. Which may not be true.

However, once we start talking about spotting error, ie spotting the fall of shot, we are attempting to differentiate between a number of relatively closely spaced water splashes which are not particularly good radar targets, and don't linger for very long. Also the spotter has to derive a complex weighted average, to be the MPI, on the fly.
The Mk 3 had excellent tracking accuracy but a poor resolutions cell (bearing x range resolution). This made spotting the fall of shot via the Mk 3 problematic as it was difficult to distinguish splashes from the target or even from individual splashes. With a range resolution of +/- 400yds even a 9-gun pattern will appear as one irregular pip and it is difficult to determine the actual MPI. Regarding spotting difficulties, keep in mind that optical spotters faced time limitations as well but perhaps a bit more than with radar.

Also I need to understand "spotting" in the context of the 26.5 ft rangefinder. My understanding was that the rangefinder was primarily concerned with ranging on the target vessel, and sampling that range as often as possible, and sending it to the fire control computer. This might occupy all the operator's time, and that given field of view constraints etc he would not be seeing fall of shot and estimating position relative to the target. My understanding is that the optical spotting is done by another individual with different optics who estimated the MPI of a salvo relative to the target. Maybe the crucial difference is that a stereo rangefinder allows the operator to do both, although if you are sharpening the image of the target, can you be sure to see and sharpen all the splashes at the same time?
Generally all stations with high powered optics were tasked with spotting. Usually one position was tasked as being the controlling spotter and would concentrate more on spotting than range taking but would still provide ranges. For battleships, all stations should be able to spot and still provide a good ranging density because the salvo interval is quite long. For light cruisers one or two stations will spot while the others would provide ranges and stations would even rotate duties between spotting and ranging.

Stereospotting is possible because of the ability to assess depth in a stereo field with set floating graduations in the stereofield. The operator can discern the distance between the splashes and the target by judging them against the floating unit graduations. Of course the instrument enhances the stereo acuity of the human eye – and hence judge distances – by a significant amount. The 46’ Mk 52 Mod 1 on the Iowas gave its operator 5,125 times the stereo acuity over the naked human eye for instance. This ability literally takes years of practice and skill to achieve good results.


Is the "direct spotting average" a rangefinder range against a surveyed distance for calibration purposes? Is this +/- 155 yds at 20,000yds?
“Direct spotting” is a method where the operator estimates the error of the MPI directly to the target. This technique differs from say a ladder or “bracket and halve” techniques where salvos are fired at different ranges and the operator estimates which is closest to the target and calls for that adjustment (ladder). What I meant specifically by average is the average expected by an experienced operator. A good operator might be expected to achieve results on the order of about 1.5 times the unit error of the instrument. At 20Kyds, the 15m stereoscopic RF in the Fwd RF on the Yamato should produce an average spotting error of +/- 85yds at 20Kyds and +/- 192yds at 30Kyds.

Brad Fischer
Tiornu
Supporter
Posts: 1222
Joined: Mon Oct 25, 2004 6:13 am
Location: Ex Utero

Re: Excellent posts from very knowledgeable posters

Post by Tiornu »

“Direct spotting” is a method where the operator estimates the error of the MPI directly to the target.
How does a scartometer differ from a rangefinder?
User avatar
Dave Saxton
Supporter
Posts: 3148
Joined: Sat Nov 27, 2004 9:02 pm
Location: Rocky Mountains USA

Band width

Post by Dave Saxton »

I want to go back a bit and address another issue involving the physics of vacuum tube radar reciever technology. This issue is band width.

The band width of the componants of the radar reciever, and that of any signal proccessing between the reciever input and the indicators, will greatly effect the potential range accuracy. Moreover, the potential range resolution of those radar designs that proccess the return pulse, and essentially only use the leading edge of the pulse, will be greatly effected by band width as well.

Pulse radars systems send square wave pulses, or as square as possible. How sharp the leading and trailing edges of a pulse is, is directly effected by the band width. Only infinite band width can process a perfectly square leading edge with a perfectly flat top. Additionally, the shorter the effective pulse duration, the greater the band width needs to be. The sharper the leading edge of the pulse, the better the potential range accuracy. If a particular radar exhibits more exceptional range accuracy, that would usually indicate a relatively high band width.

However, there are trade offs to be made when utilizing relatively high band width. The British found that the higher the vacuum tube componants's bandwidth, the greater it's noise. This was particulary true of the 400mhz to 1ghz noise problem of recieving pentodes. Even the grid grounded triode solution for the 50cm radar resulted in limiting band width to about 4mhz.

A radar design working in the decimetric wave length ranges, must balance the trade off between necessary bandwidth, and tolorable noise levels at the selected operating wave length, with the selected necessary componantry.
Bgile
Senior Member
Posts: 3658
Joined: Wed Mar 09, 2005 7:33 pm
Location: Portland, OR, USA

Post by Bgile »

Dave,

Aren't you confusing the bandwith of the video processor which forms the wave envelope into a pulse for use with the timing circuitry with the bandwidth of the transmitter, receiver, and antenna, which can be quite a bit less in order to produce the needed s/n ratio?

In any case, I'm not sure where you are going with this. It isn't terribly relevant unless one side has better stuff than the other in a particular area. Is that what you are trying to demonstrate?
User avatar
Dave Saxton
Supporter
Posts: 3148
Joined: Sat Nov 27, 2004 9:02 pm
Location: Rocky Mountains USA

Post by Dave Saxton »

No, I'm not confusing anything. I'm talking mainly about reciever band width, and the problems presented by British engineers in their technical papers on reciever design. Of couse bandwidth in any signal processing will also be important. This illustrates the physics involved and the trade offs that must be made.
Brad Fischer
Member
Posts: 71
Joined: Thu Apr 21, 2005 4:48 pm
Location: USA

Post by Brad Fischer »

Dave Saxton wrote: The radar displays and the operators were located inside the directors in USN practice. With Mk3 for example, the radar operator for bearing controled the traverse of the director. This way he could control the lobe switching procedures.

In the Royal Navy, the radar office and the operators for the fire control radars, were ofton romotely located from the optical directors themselves. With British firecontrol radars, we eventually see remote slave displays being located in other areas in the ship, and especially below decks. By 1944, the British usually had slave displays in the fire control computor rooms, so that the gun laying crews there could directly observe the radar data. Additionally in 1944, Type 284 was equiped with an additional scope specifically for spotting the fall of shot. This indication could be recorded by photograph at the time of fall in some cases.
The Mark 8 Mod 2/3 and Mark 13 all had auxiliary indicators with the latter having numerous independently controlled auxiliary scopes where the operator at that station can change the mode to his liking. All three types had the primary controls located down in plot with the director crews only using their scopes for spotting and by the trainer for adjustments generated director train
The British continued to prefer A-scope or J-scope displays even for their late war centimetric firecontrol radars. They disliked Type B display for firecontrol radars unless it was just for coarse observation or general spotting. They beleived that the accuracy of data obtained from a Type B display was largely dependant upon the interputive skills of the operator. More finely precise data can be obtained from an A-scope or J-scope display it is true, but it also more abstract by nature. Spotting for bearing would probably have been easier with a less abstract type of display, and that might be why they went to seperate spotting displays. Post war, the British added a completely seperate radar system, exculsively to spot the fall of shot, to their centimetric firecontrol radar systems. Many major British warships continued to use the proven 50cm radar systems until well after WWII.
The British perspective is interesting regarding plan-view “B-scope”. The Mark 8 Mod 0-2 is credited with a tracking accuracy in bearing ± 2 mils (± 7’) while the Mark 13 is credited with ± 1 mils (± 3’). These values are almost identical to the Type 284M (± 5’) and the Type 274 (± 3’). It does makes sense, however, that the A-scope might provide more accurate indication of train error. I do note that one limit is the director’s ability to track smoothly in train; this is a concern in certain situations such as when the director is in automatic and is being trained from increments of generated train from the rangekeeper solution in plot.

Brad Fischer
Post Reply