Can we learn more about making RTR models?

[andyman7]

Firstly, I definitely couldnt find any other topic heading that this matched.....

 

The purpose of this thread was to create a space where hopefully those with actual (as opposed to armchair) knowledge can help us all understand better the factors that affect whether an RTR model is made; for what market(s); what a realistic production run/lifespan of that market is; etc

 

It's been inspired not just by the demise of DJM but also by the endless speculation on other threads of 'why don't they make x'; 'why don't they reissue x'; 'the tooling is worth y/has been lost/has got damaged and would be too expensive to repair/remake' - as well as the endless comment on prices.

 

My only request is that this is used for facts that people who have actually been involved in various processes can reasonably attest to. So please no 'I think...' or 'I heard somewhere....'. 

 

I'll start with a few basics to provide a framework, all of which I'm keen are either refined or indeed challenged:

 

• In making RTR models, design still tends to be carried out in the country of origin, but the skills and capabilities required to produce scale models to any sort of standard mean that manufacture is likely to have to be in China if the final cost is going to be a viable proposition
• Model manufacture is a fairly niche part of Chinese manufacturing so sourcing and securing production slots and the support/dialogue to realise the final product is a specialist skill in its own right
• The market for fine scale RTR UK outline models is very specialist too, so production runs in batches of less than 1000 are the norm as opposed to the mass production of the 50s, 60s and 70s
• The main costs of fine scale RTR are design/CADs are; production of the steel moulding tols and ancilliary parts; assembly; shipping and distribution; - but it would be great to get a feel for how this breaks down!
• Tooling costs are in real terms lower now than 30 years ago, making production of sub-1000 piece items viable in the right circumstances (is this true?)
• If a mould has been lost, is it easier or viable to recreate it these days now that the item itself can be scanned? (I'm thinking for example of the Norev/Atlas Dinky Toy reissues which are all from new tooling for long defunct models).
• What sort of difference in cost is there for a tool needed to run 5000 items rather than 500,000 (if any)?

 

 I'm really interested to learn more about this!

'

Edited June 28, 2019 by andyman7
Spelling

[34theletterbetweenB&D]

2 hours ago, andyman7 said:

Tooling costs are in real terms lower now than 30 years ago, making production of sub-1000 piece items viable in the right circumstances (is this true?)

 

If a mould has been lost, is it easier or viable to recreate it these days now that the item itself can be scanned?

An aspect relating to the first question quoted above. Bachmann informed us - probably a decade past now - that the tooling of the mechanism was a much smaller proportion of the complete model cost than it had been in the past. The large majority of the expenditure went on the process of research through to manufacture of the externally visible parts of the model. Thus 'having a mechanism under the 'clanker' class that's also right for the 'basher' class for which there is currently no RTR model', did not make the 'basher' class significantly cheaper to introduce to the range.

 

(Opinion now, it is not far to look in mechanism layout comparisons, to see that there are uniform features in mechanism layout within a manufacturer's range. This is the sort of thing that CAD/CAM excels at, producing layouts from a template by input of a defined set of dimensions, potentially all the way through to cutting metal for the diecasting tools.)

 

Second of your questions above, specific to model railway product the likes of the 'Great British Locomotives' series clearly shows that it is possible for the body shell. Must have been relatively economical too, for a low volume magazine rack price production to be possible.

Edited June 28, 2019 by 34theletterbetweenB&D
attempt to correct weird formatting

[The Johnster]

I think we all benefit from a basic understanding of how RTR models are  designed, produced, assembled, and distributed in the hope that such understanding will limit to ‘reasonable’ our demands and requests to those who sell them to us. 

 

Assisted by CAD, tooling is much less a barrier to producing models than even 30 never mind 50 years ago; look at the range available.  Yet models constantly demanded for many years are still not available while others are duplicated; the market works well but not perfectly.  

 

The big costs are labour, particularly in the assembly facilities and rising inexorably, and R&D.  Anyone can draw a model to scale and some of us can build ‘em, but working out an efficient method of producing them in volume, to cost, and within time constraints because the production and assembly facilities are booked in advance for specific time slots, is a different sort of juggling trick!

 

In a way, I don’t really care so long as the models turn up at the time and price advertised, but a bit of understanding of what can and sometimes does go wrong is no bad thing!!!

Edited June 28, 2019 by The Johnster

[Miss Prism]

12 minutes ago, The Johnster said:

Anyone can draw a model to scale

 

The history of UK RTR models over the last 20 years or so indicates this presumption is not particularly well-founded.

 

[mdvle]

17 hours ago, andyman7 said:

My only request is that this is used for facts that people who have actually been involved in various processes can reasonably attest to. So please no 'I think...' or 'I heard somewhere....'. 

 

To be clear I have no direct involvement, just a model train person, but I will provide a valid source for some of what follows.

 

One of the problems is that most of the companies involved don't really provide much of a view of the "behind the scenes" aspect.  The big exception to this is Jason Shron, founder of Rapido Trains, who has from the beginning providing much information on how things work (at least for Rapido Trains).

 

Not everything posted by Rapido is behind the scenes, but if you follow their Facebook page they frequently post factory photos, as well as provided information in their newsletters and YouTube page(*)

 

17 hours ago, andyman7 said:

 

I'll start with a few basics to provide a framework, all of which I'm keen are either refined or indeed challenged:

 

• In making RTR models, design still tends to be carried out in the country of origin, but the skills and capabilities required to produce scale models to any sort of standard mean that manufacture is likely to have to be in China if the final cost is going to be a viable proposition

 

The research certainly happens in the country of origin, and some companies do a lot of the CAD / drawings in country of origin, but some companies also outsource the CAD to China.

 

Correct that the actual making needs to be in China - this is for several reasons including the (still relatively) low cost of labour but also includes things like the supply chain being next door - the proverbial company making the motors being done the street, the company doing etching being across the road, etc.

 

17 hours ago, andyman7 said:

• Model manufacture is a fairly niche part of Chinese manufacturing so sourcing and securing production slots and the support/dialogue to realise the final product is a specialist skill in its own right

 

Think this is a bit unclear.

 

Model trains is a very niche market, and likely considered a rounding error in terms of Chinese manufacturing.  But it is typically done in specialized factories that are dedicated to model production - there is a specific skill the workers need.  So when a company talks about production slots they are talking about slots in these small, specialized factories and not slots in some big factory that also makes other items.  If you look at the videos and photos from Rapido you see what is a (relatively) small space that is set up for making model trains and not much else.

 

What Rapido hasn't shown is the injection process itself which may well be done in places that also do non-train related stuff.

 

17 hours ago, andyman7 said:

• The main costs of fine scale RTR are design/CADs are; production of the steel moulding tols and ancilliary parts; assembly; shipping and distribution; - but it would be great to get a feel for how this breaks down!

 

Nobody I believe have publicly come out with exact figures, though tooling is often said to be in the $100k mark.  One of the few / only specific examples that I am aware of is from Rapido.  They announced and tooled a US model (Alco RS-11), and after the tooling was done and the samples were being shown publicly somebody queried the height of the cab...

 

Rapido confirmed that they had made a mistake and tooled the cab too high, and they fixed the error.  Whether is was new tooling or modifying existing tooling is unknown, but the cost to fix the tooling was $15k - https://myemail.constantcontact.com/Rapido-News-109---New-freight-cars--new-RS-11-order-date--and-we-re-hiring-.html?soid=1101318906379&aid=g-SzXvHaIjQ

 

As for a break down, my guess is that tooling is the most expensive part and the others start in a distant 2nd place.  If I had to guess assembly would come in second - quite a number of people involved and it is all repetitive manual labour - cutting out parts from sprues, painting, pad printing, assembling all those parts takes time.

 

17 hours ago, andyman7 said:

• Tooling costs are in real terms lower now than 30 years ago, making production of sub-1000 piece items viable in the right circumstances (is this true?)

 

Difficult to know, in part because the models themselves have changed.  Those models of 30+ years ago needed far less tooling, and perhaps simpler tooling, as they weren't necessarily as accurate in shape or had as many added parts.

 

As for production numbers, that is complicated.

 

The first thing is are you talking total number of models being made, or the minimum number of one particularly paint scheme?

 

The second aspect is the number of items needed to break even, and that depends on the finances of a given model.  Obviously fewer product made means a higher price is needed to cover the costs, whereas if you have a model that sells in large numbers you can get away with selling at a lower price, or making more profit.

 

On June 24th Rapido did a live Facebook video announcing some new Canadian products, and at the end Jason answered some questions that viewers asked.  One of the questions was about making some steam era passengers cars, and Jason's answer was that Rapido needed to sell about 5,000 cars to break even.  Now that 5,000 could be divided up covering 5 different paint schemes, etc but you would still in the end need 5,000 items sold.

 

On the other hand 5 years ago the Pendolino in N with Revolution/Rapido needed only 1,000 units (**).  But that was 5 years ago, with 5 year old Chinese labour costs, and perhaps other financial differences.

 

When it comes to additional production runs, then the number of items comes down to how few the factory is willing to make given the costs and time involved.

 

17 hours ago, andyman7 said:

• If a mould has been lost, is it easier or viable to recreate it these days now that the item itself can be scanned? (I'm thinking for example of the Norev/Atlas Dinky Toy reissues which are all from new tooling for long defunct models).

 

Common misconception.  Scanning, while it can be a help in getting a correct shape, doesn't output a set of CAD drawings.

 

Laser scanning creates what is called a point cloud(**), which is a 3D object made out of thousands/millions of little dots which is the result of the laser beam hitting the prototype and bouncing back, resulting in a "point" or a coordinate in X,Y,Z space repeated across the entire prototype.  This point cloud object must then be transformed into traditonal CAD drawings by a human being.  You can see an example of a point cloud image in the Cavalex Class 91 thread - https://www.rmweb.co.uk/community/index.php?/topic/142787-all-new-rtr-class-91-and-mark-4-carriages/&do=findComment&comment=3571582

 

So if a mould has been lost or damaged, a new mould would need to be created.  If the CAD already exists then it should be reasonably easy otherwise new CAD drawings would need to be created whether converting from a scan or by doing the more traditional way of using measurements and drawings.

 

 

* For those interested:

Rapido YouTube channel - includes some factory tours - https://www.youtube.com/user/rapidotrains/videos

 

Rapido Newsletter archive (North American) - various newsletters feature factory and other production information - https://www.rapidotrains.com/content/rapido-newsletters

 

** Pendolino thread, first message gives the 1,000 unit number - https://www.rmweb.co.uk/community/index.php?/topic/92432-n-gauge-class-390-pendolino-post-kickstarter-with-rapido/&tab=comments#comment-1657148

 

[melmoth]

19 hours ago, andyman7 said:

My only request is that this is used for facts that people who have actually been involved in various processes can reasonably attest to. So please no 'I think...' or 'I heard somewhere....'. 

 

There are a few other threads on here that could benefit from that sort of intellectual discipline...

 

19 hours ago, andyman7 said:

• What sort of difference in cost is there for a tool needed to run 5000 items rather than 500,000 (if any)?

 

I have no idea, but I would perhaps have expanded this a bit, e.g. "Does the rate at which the tooling is to be amortised have a bearing on the initial cost (and therefore, viability) of a project?".

[Dunsignalling]

21 hours ago, andyman7 said:

 

• If a mould has been lost, is it easier or viable to recreate it these days now that the item itself can be scanned? (I'm thinking for example of the Norev/Atlas Dinky Toy reissues which are all from new tooling for long defunct models).
• What sort of difference in cost is there for a tool needed to run 5000 items rather than 500,000 (if any)?

 

 I'm really interested to learn more about this!

'

 

3 hours ago, mdvle said:

 

Common misconception.  Scanning, while it can be a help in getting a correct shape, doesn't output a set of CAD drawings.

 

Laser scanning creates what is called a point cloud(**), which is a 3D object made out of thousands/millions of little dots which is the result of the laser beam hitting the prototype and bouncing back, resulting in a "point" or a coordinate in X,Y,Z space repeated across the entire prototype.  This point cloud object must then be transformed into traditonal CAD drawings by a human being.  You can see an example of a point cloud image in the Cavalex Class 91 thread - https://www.rmweb.co.uk/community/index.php?/topic/142787-all-new-rtr-class-91-and-mark-4-carriages/&do=findComment&comment=3571582 

 

So if a mould has been lost or damaged, a new mould would need to be created.  If the CAD already exists then it should be reasonably easy otherwise new CAD drawings would need to be created whether converting from a scan or by doing the more traditional way of using measurements and drawings.

 

 

This is more of a supplementary query than a contribution, and the suggestion (if practical at all) might only be applicable to relatively simple castings such as Dinky re-creations rather than the more complex mouldings used in model locomotives.

 

My question is, if an old model is scanned with a view to re-creating it, is a CAD always necessary? Might it be possible to use the scan output to cut a mould directly, using water cutting, spark erosion or some even newer process that I've not yet heard of?

 

If this were viable, removing an intermediate stage (the CAD) also removes an opening where errors/changes  might be unintentionally introduced. 

 

John

[griffgriff]

55 minutes ago, Dunsignalling said:

 

This is more of a supplementary query than a contribution, and the suggestion (if practical at all) might only be applicable to relatively simple castings such as Dinky re-creations rather than the more complex mouldings used in model locomotives.

 

My question is, if an old model is scanned with a view to re-creating it, is a CAD always necessary? Might it be possible to use the scan output to cut a mould directly, using water cutting, spark erosion or some even newer process that I've not yet heard of?

 

If this were viable, removing an intermediate stage (the CAD) also removes an opening where errors/changes  might be unintentionally introduced. 

 

John

I don’t think it can quite that. Data is taken from several scans (one side then the other, the ends and so on) and a  drawing (CAD) ties them all together. So I’m pretty sure it’s a logical and necessary part of the process.

 

Then again, technology may change all this in the future. It seems a little bit sci-fi for now. 

 

Griff

[justin1985]

2 hours ago, Dunsignalling said:

 

My question is, if an old model is scanned with a view to re-creating it, is a CAD always necessary? Might it be possible to use the scan output to cut a mould directly, using water cutting, spark erosion or some even newer process that I've not yet heard of?

 

 

Basically, CAD will always be necessary because CAM instructions (often "gcode" format) have to be valid, and configured for the resolution of the machine in question. (Regardless of whether that's a spark eroder, laser cutter, 3D printer, or whatever). As in, there can't be points that aren't supported by surfaces, sections that are too thin, too small for the resolution of the device, or whatever. A point cloud or similar is too random - things like reflection, shadow, and dust can create spurious points. Software can even these points out into surfaces suitable for CAD, but it will always be a compromise between spurious data creating lumps etc, or the smoothing being so smooth that detail is lost. A quality result needs human input. Perhaps AI type tech will be able to help in future.

 

Perhaps a bit more common use of this tech is the use of airborne laser scans in mapping. The process to get from a messy point cloud which includes leaves on trees etc, to the nice smooth contours on a map or 3D view, is pretty damn complex, and if automated, it gets all kinds of spurious artefacts.

[mdvle]

7 hours ago, melmoth said:

I have no idea, but I would perhaps have expanded this a bit, e.g. "Does the rate at which the tooling is to be amortised have a bearing on the initial cost (and therefore, viability) of a project?".

 

Given I am guessing nobody in the know will comments, some (somewhat educated) guesswork.

 

The first variable is the state of the company.  A long established company with in essence a "library" of existing tooling that can provide a steady stream of guaranteed (1) profit can take a different view than a young company with no easy source of revenue.

 

A second variable will be the popularity of the model, not so much in its viability (unless there is lots of competition), but in the amount of secondary runs that can be expected to be successful.  The example I would give here would be ScaleTrains.com, who in a podcast to paraphrase their comment called their SD40-2 (2) model a "decade product" because they expect to be able to continue making models with that tooling for at least 10 years.  So if you expect to be able to use the tooling extensively, then it might be worth spending a bit more on the tooling quality if that is possible.

 

The perceived rule of thumb is that tooling isn't amortized, that each project is expected to cover its costs and at least break even on the first production run.  I suspect that this is true, for there really is no guarantee is this hobby that (with perhaps a few exceptions) you can bank on subsequent runs.  I base this guess on several things:

 

a) the obvious example, the new and young companies can't afford to have that much debt hanging over their companies given they don't have the existing models to generate revenue to cover the borrowing costs.

 

b) not every project is going to be a success financially.  Whether the market has been misjudged and they don't sell as many as expected, or it turns out someone else is also making a new version and cutting into sales, some projects are going to make a loss at least on the first run.  If you aren't aiming to at least break even, then this would simply be even worse, causing more financial issues.

 

c) depending on the model, it may be several to many years before the market can support an additional run.  You aren't going to only cover half your costs on a model if you can't make another run for 5 years.

 

As for viability, I suspect the above covers it - someone needs to do some numbers and decide if the number of expected sales will cover the costs in developing the model.

 

But I will point out that an established company, that is financially healthy, can fudge the decision making process in determining the viability of a given project.

 

Again (just because he has been so open over the years), an example is Rapido Trains where the needs for the owner's personal layout or personal interest, combined with the revenue from an established base of models, means there are projects being done where the expectation is that the model won't break even, or that it is a gamble worth taking.

 

(1) this assumes that they are reasonably prudent in their secondary runs and don't produce an amount that leaves them with unsold inventory in the warehouse costing them money

 

(2) for those unfamiliar with US trains, the EMD SD40-2 was such a successful diesel that just under 4,000 were made between 1972 and 1989, and they were so common that you really can't model the mid 70s to mid 90s without having that engine, and more likely lots of them.

[andyman7]

18 hours ago, melmoth said:

I have no idea, but I would perhaps have expanded this a bit, e.g. "Does the rate at which the tooling is to be amortised have a bearing on the initial cost (and therefore, viability) of a project?".

To clarify, the original question was prompted by an assertion somewhere (which may or may not be true) that the moulds for some of the more obscure prototypes that Heljan had modelled were designed for fairly short production lives (the implication that they were capable of being amortised over a small run); by contrast, some of the 60 year old Kitmaster tools still be used by Dapol were running 100,000 shots a year in their Kitmaster days (as noted in the Kitmaster book 'Let's Stick Together', from first hand evidence).

[billbedford]

Moulds can be made of aluminium instead of steel, and they would have a shorter life, up to 100,000 shots, depending on the complexity.