Saturday, 31 October 2009

Analysis of the MFX Sonic Screwdriver and Inaccuracies to the Prop

The MFX sonic screwdriver has been two years in the making and was vaunted as a 'screen accurate' 1:1 replica of one of the screen used hero props. The initial shots of the protoypes were very, very promising and at least one of the prototypes was virtually 100% in terms of details (though there was still room for improvement). As can be seen here:




The production versions were released in early october and having obtained one for closer examination I can reveal that it is anything but the 1:1 replica promised. Here are my findings:


CONSTRUCTION

The filming prop emitter head consisted of two pieces: the ball jointed central section screwing into the head strutted 'cage'. This featured a circular join at the base of the head recess below the centre pipe. Look here at a shot of the prop from the Visual Dictionary:




The pencil is pointing to the join I refer to. There is also a join on the upper part of the inner recess too. This is where the ball join screws into the emitter head cage. Here is the head without the emitter cage screwed in:



Does the MFX have it? I'll let the following pic speak for itself.




Mine does though :-) 


The MFX head is not made in the correct manner the prop was. The head on the props were CNC machined from a single cylinder of solid aluminium bar stock.

It is my belief that the head blank was cast, the windows were drilled out and then machined to shape it. Several observations lead me to this conclusion.

  • The colour of the head does not match the rest of the screwdriver, in particular the collars at either end of the main barrel. The head is noticeably whiter and duller than the rest of it. This is because the metal was melted to cast the head and was oxidised and/or contained more impurities from the casting process. However, there is another theory for this which I will go into later.


  • The head shape is very indistinct and lacks the sharp defined edges characteristic of CNC milling. Contrast with my emitter head which is pictured below which has nice sharp, defined lines:





  • The finish on the head is not consistent with CNC milling. The prop it is a shiny satin finish whereas the MFX is dull matte. If the MFX head was CNC machined, why would they destroy the finish? My emitter head above was fresh off the machine, 10 minutes in fact, when the above pic was taken. That is the exact finish that the props have. I do not believe for one second that MFX destroyed a machined satin finish in order to make it more screen accurate as we know that is utter bullshit. For them to say otherwise is a face saving exercise. Let me quote you something Neill Gorton said about a year ago when called out on the matte finish of his original spring loaded prototype at SDCC. Speaking about his prototype:


    "The production sonic we used as reference has two yellow wires.

    The emiter was duller than it should be as the aluminium hadn't been fully polished. That will be rectified also."


    Basically, he admits that the prop sonic he had access to had a polished look. His engineering prototypes had the same finish as the prop as does mine. Every picture seen of the prop shows a satin finish. So why didn't they come through on their production model?  I believe that the answer is simple and compelling. The head was cast. The reason why the finish is the way it is is because the casting process fundamentally destroys the metal's chemical properties through oxidation (hence the colour differences) and impurities. The main consequence of this is that it makes the bare die cast aluminium very difficult if not impossible to polish to a high shine unless the alloy was  very, very pure. Sure the metal will brighten up, but you will not achieve the reflectivity needed unless you chrome or plate it.   The metal could also have been anodised so why didn't MFX do that? Well because die cast alloy does not take anodising well due to non-conductive impurities. It will end up mottled. Chrome plating will ruin the finish as it is too reflective. Realistically only expensive HG15 aluminium can be anodised consistently. Not even HG30 will suffice and the head is not even that. In fact, under a microscope I am seeing spots of what look like RUST on the head. To those that own an MFX, take a loupe and have a look, especially around the corners of the cutouts, on the inside where it has been drilled out. This is disturbing news. I have a strong suspicion that the metal used is a cheap aluminium alloy with added iron or tin to improve casting flow.  The finish does not match the screen used prop which features a shiny satin finish that is consistent with CNC milling. This is plainly evident when two pictures of the prop and the MFX replica are placed together:It is pretty obvious that the filming props are substantially shiner than the replica. .


  • Another giveaway is that the head is not 100% circular. If it has been turned, it would be the same diameter all the way around. The MFX isn't. It varies by several hundredths of a millimetre from the top to the bottom of the head and depending on rotational orientation. Sure, these differences are miniscule, but they are evidence of the head being cast in a flexible mold or the die was ever so slightly off. If the head was milled, it would have been spun on a fixed axis and the head should be perfectly circular. It isn't. Neill Gorton has stated categorically that the MFX Sonic has been CNC milled. I don't dispute that the body, end caps etc...basically everything EXCEPT the head has been CNC milled. However, I find it quite telling that he refuses to categorically confirm or deny whether the head has been. I do not believe it has been, though I am willing to accept that CNC milling was used for drilling out the head windows. The findings above have not been addressed adequately such as the discrepancy in diameters and the finish and colour differences.

Some will counter the argument by pointing out tiny circular machine marks on the head itself. There are two reasons for this. First, these marks could have been from the master they cast the die from or from the die itself as they are usually machined. Die casting has a high enough resolution to take in surface detail like that. Secondly, these marks could have resulted from secondary shaping on a machine after casting.


DIMENSIONAL INACCURACY

By definition, a screen accurate replica should be just that: accurate to the prop seen on screen, not just in features, but dimensions also. This is where the MFX is lacking in several respects.

Emitter Head Windows:
The window on the emitter head are in the wrong place. They are practically centralised on the MFX when they should be slightly below centre. Look at this picture from the visual dictionary and the area you should look at to give you an idea as to how 'off' the MFX one is:



It is visually apparent from the above that the windows are in the wrong place. However, to remove any doubt, I have used the very useful Ruler tool from Photoshop to measure the position of the emitter windows. For scale, the distance from the edge of the top top bevel to the edge of the lower bevel line at the bottom of the strut bevel should be 21mm. The ruler tool has been scaled to this length prior to any measurement. The pertinent measurement (in mm) is circled in the screenshots below. (Bear in mind that this tool is not highly accurate or as accurate as physical measurement using digital Verniers due to picture resolution, perspective and other small distortions, but will give an accurate measurement of relative lengths for comparison purposes)

First, the prop:




Here is the length of the top section above the window on the prop, result: 6mm:




Here is the length of the lower section below the window on the prop, result 3.95mm:



Result: The differential in length between the section above and below the emitter window of the prop is 2.05mm.

Let's examine the MFX the same way:

First, scaling:



Length of section above the window, result: 5mm:




And below the emitter window, result 4.55mm:



Result: The differential in length between the section above and below the emitter window of the MFX is 0.45mm! A staggering amount of error for something that purports to be screen accurate!

In case you think I am making this up, allow me to offer further analysis based on other pics of different props. I have yet to find a single pic of a prop that has the emitter windows as centralised as MFX's version does.

Scaling:




Upper length on prop 5.31mm:




Lower length of prop 3.65mm:




Differential between the two lengths on the prop: 1.66mm.

On the MFX:

Scale:



MFX Upper length 4.88mm:



MFX Lower length 4.57mm:



Differential between the two lengths on the MFX: 0.31mm.

Having established a clear and INDISPUTABLE difference, I decided to perform the analysis on another filming prop and also on my MFX from a different angle to see if the differences are caused by other factors.

Scale:



Upper length 6.01mm:



Lower Length 4.02mm:


Differential on this prop: 1.99mm

The same analysis on the MFX:




Upper length 5.15mm:



Lower length 4.57mm:


Differential on the MFX: 0.58mm

I thought I would run the analysis on other screwdrivers just to prove that this feature has been consistent throughout the history of the prop. The results are of the emitter head of one of the Eccleston props:
Scale:



Upper Length 6.37mm:



Lower Length 3.87mm:




Differential: 2.5mm

What this has proven without question is that the window positioning has remained consistent since the inception of the new prop. It is incredible that MFX could have failed to notice such an obvious and integral design feature such as this. Indeed, they DID get it right on their prototypes. Check out the following analysis which has been aided by the graph paper they laid down which allows for pinpoint scaling accuracy:

Scale:



Upper length 6.27mm:



Lower length 3.98mm:



Overall differential: 2.29mm - which is within a few 10ths of a mm of the original prop! Staggeringly accurate

But this is not all that is wrong.

Ball Joint:
The ball join on the MFX is incorrect. It sits too tightly into the crenellated recess of the emitter and is becuase it is too far in and also the walls of the recess are too thick. Please see the following pic:



Where the pencil is pointing is a distinct black area indicating a gap in between the ball and the crenellated wall where the ball is full screwed into the emitter head. This gap is very noticeable and you can fit your thumbnail into it easily on the prop. On the MFX the joint is flush.

My sonic, however, does have this correct fitting. Please see the following closeup of my emitter head and the join:


And in comparison to the MFX:




The following picture was posted to apparently disprove the above finding. I have performed analysis to show how erroneous this is. The picture has been scaled to 6.75mm, which is the length from lower edge of crenallation to the bottom narrow diameter of the ball on the original Eccleston prop. Bear in mind there is a slight perspective distortion in the pic below due to the sonic being angled towards up which will skew the prop measurement slightly towards the shorter end of the scale by about 0.5mm but is enough to demonstrate my point.

Here is the measurement on the prop: 5.33mm:




Measurement on MFX 4.2mm:




This is over 1mm difference! It certainly lends the lie to the assertion that the prop had the same fitting on the ball join that the prop did. In any case, the above pic is inconclusive as it does not show the join from below, looking up into the crenellated socket.

Strut width:
The struts on the MFX are also the wrong width. The struts measured exactly 5mm across on the prop. On my MFX they measure 4.39mm, 4.42mm, 4.37mm and 4.41mm. Over half a mm off!!!!! On a prop this small, these tiny amounts are huge when looked at as a percentage of overall dimensions. By this token, the struts on the MFX are over 10% wrong in width!


Strut Edge Thickness:
The edge thickness of the struts are also too shallow on the MFX and lack a seriously distinctive design feature that is obvious on the original prop and was an inherent part of the rather unique geometry of the head. The way in which the original was constructed meant that the windows were drilled in a certain way to bisect the support struts halfway into the thickness of the strut. Look at the following comparison of my Sonic and the MFX one and the difference will be obvious:





I have circled in yellow the area of attention. The MFX is on the left, mine is on the right. You can see that not only is the MFX struts too thin, the strut is not cut into by the window as it should. In fact, the window dows not cut into the strut at all but stops at the natural strut/body join.

To give you a better view of what is going on, here is a pic of my CAD model of my Season 1-2 emitter:




Notice how the emitter window cuts into the strut at exactly 0.5mm. The geometry of the design is such that on the original prop you cannot pick up the sonic by gripping the strut by the finger and thumb nail. It will slip out. You can do that on the MFX as due to the error of geometry, the lengthwise edge of the strut is too narrow to allow for this bisection since the original prop was milled in a specifically counter intuitive way due to equipment limitations.

The odd thing is about this whole feature is that MFX got it spot on on one of their prototypes. Compare and contrast the following:



The pic on the left clearly shows the emitter window cutting into the strut as it should. The picture on the right shows that the window does not cut into the strut but stops at the strut/body join. What is even stranger is that the strut on the right hand pic is clearly of the correct thickness to allow the window to cut into it, something that even the production version lacks. Allow the following analysis to demonstrate:

Scale:


Thickness:


The struct edge thickness of the MFX prototype is 0.99mm. 1 hundredth of a mm out! Staggering accuracy.

So how off is the production MFX? The strut edge thickness on the MFX is on average 0.35mm (it is as low as 0.30mm or as high as 0.37mm depending on the strut you measure and whether you measure above the window or below the window due to the MFX not being entirely circular). On the prop it is exactly 1mm. A noticeable and crucially missing feature. So where exactly has the missing 0.65mm gone? I have a good idea. You see, the next diameter in from the struts, ie. the section in between the struts is off by a +0.65mm from the prop. So obviously this middle section is too thick and perfectly fits the missing amount!
It could be argued that this feature was omitted in subsequent props from Seasons 3-4. However, the following pictures will show otherwise:




The conclusion is that the MFX strut is too thin by as much as 70% or 0.7mm. Again, a staggering amount!

Lip Shape:
The shape of the MFX emitter head lip is squared. On the actual prop it has a rounded 'D' profile as does mine. Please compare the following pics to illustrate the point (MFX on left, Mine on right):


Compare and contrast with the following shot of the prop where the lip is obviously curved in profile:



Ridge Shape:
The shape of the ridges of the MFX are of the wrong geometry. They are too sharp in their transitions and too deep which gives the tops of the ridges a flattened trapezoidal geometry that is incorrect. The extended head prototype on the MFX above has the correct shape. The second closed prototype has the incorrect ridges as per the production version.

Look at the following picture which demonstrates this subtle but crucial difference in shape as once you know what to look for, you will understand the problem is quite a serious one. The prop is above, the MFX is below.




Look at the areas circled below and the problem can be identified:




Essentially the ridge transitions on the MFX are too harsh and abrupt giving a defined straight edge to the ridges that is not present on the prop where the rounded edge transition disperses the light and shadow in a much gentler manner. Consider the following shots of the MFX under various lighting conditions and the straight edge can be seen in dramatic detail:




In addition the depth in between the ridges are 50% deeper than they should be, contributing to the pronounced banding not present in the prop.

MFX must surely be aware of this as they got it right on one of their prototypes. See the following:




The pic of the left show the prototype with the correct curved ridges. The right hand one shows the incorrect sharp ridges that have also shown up on the production version.

Diametric Differences:
Diameters are all over the place on the MFX.



At point A, the lip diameter is correct.

At point B, the diameter is off by +0.49mm

The inner diameter at point C is off by +0.65mm

The outer diameter at point D is incorrect by +0.85mm, or as little as +0.75mm, depending on the rotational orientation and height you measure at. This means it is off by as much as 4.2%-4.7% in overall diameter.

Point E should be EXACTLY the same diameter as point B but is +0.80mm off overall and +0.31mm off from point B.

Point F is +0.44mm off.

It can be argued that such seemlingly small tolerances make little difference, however, these small differences add up to quite serious differences especially in a prop this small. I suspect a mixture of inaccurate construction methods and sloppy measurement is to blames, especiall since one wrong measurement can compound an inaccuracy if one is to work using proportionate dimensions.


Paint Finish:
The paint finish of the MFX, though correct colourwise and despite using the correct paints, is a little too thinly applied. I appreciate whilst that certain compromises have to be made for the sake of durability and consistency, the thickness of the paint is quite evident on the prop. This was one of the things MFX got bang on with their engineering prototypes

Slider Plate:
The slider plate on the MFX is incorrect. It does not possess the correct curvature, clearance from the body, size and rounded corners that the prop features.

The size of the plate is correct lengthwise but is off by -1.52mm widthwise

The curvature is also wrong. Imagine the plate is a section cut from the wall of a tube of  Xmm diameter. The plate on the MFX is cut clearly from a much large diameter tube, hence the shallower curvature. This has the knock on effect of affecting the clearance from the body as the width wise ends of the plate do not curve into the body as much.

CONCLUSIONS
Having examined the MFX sonic screwdriver in detail I can conclude that whilst it is a fine effort and a worthy replica, it falls short in too many different areas to be considered truly screen accurate or 1:1.

As a product it is tremendous for what it is and I applaud them for bringing it to market however it would be foolhardy to think that such an undertaking would not necessitate some compromises. What puzzles me is how they can produce such amazing prototypes then fall woefully short of delivering on the promise that those prototypes showed. Various bits of visual information lead me to suspect that EFX made the MFX prototypes shortly before their relationship collapsed and the MFX version is essentially a copy of the EFX prototype, in which case the MFX is at least a generation removed from the filming prop - a copy of a copy.

Let's make no mistake, the MFX screwdriver is a wonderful piece but it is nowhere near 'screen accurate'. It is easy to make mistakes when working in the kind of ranges that this piece entails. All that is needed is one misread measurement or the Verniers not being zeroed and it is enough to throw off the entire piece - something that I suspect has happened here.

It has been said (usually by MFX owners to justify the buyers remorse they feel), that these differences are caused by a variance in the handmade nature of the props and each one differed.  Sure they did. In terms of paint finish and minor detailing. The overall shape does NOT change as the machinists have a defined rigid template to work to that does not allow room for error in terms of shape. In this world, 1mm is a very large margin. MFX claims that their sonic is machined is misleading and very suspicious. It is very easy to build a 3D model, load it into a CNC machine and basically knock out multiple replicas of the same parts to within micron accuracy time and time again.

If you want a screen accurate sonic screwdriver this is most definitely not it. If you want a quality collectible that has playability factor and are not adverse to paying over the odds, consider the MFX. However, I would order one of mine :-)