Cost of DSLR components (info for us all)

[subroc]

I have held the position of planner/estimator. It was at the job shop level and not at the corporate level with huge production runs.

I doubt you could accurately estimate their costs for screws never mind production of their alloy bodies.

You are starting this effort with a guess on chip cost. Great place to start. Your entire estimate process will be guesses.

This is an effort in futility that will tell you nothing.

Good Luck in your endeavor.

 

[Steve Balcombe]

I have held the position of planner/estimator. It was at the job
shop level and not at the corporate level with huge production runs.

I doubt you could accurately estimate their costs for screws never
mind production of their alloy bodies.

You are starting this effort with a guess on chip cost. Great place
to start. Your entire estimate process will be guesses.

This is an effort in futility that will tell you nothing.

Good Luck in your endeavor.

Many years ago I used to run a motorcycle dealership. I remember two interesting facts that people with more time to waste than me had established. One was that if you could strip down a bike and sell all the parts as spares, it would fetch around five times its retail price. The reason nobody ever did this is that most of the parts would never actually sell. It works in theory but not in practice.

The other was that a certain small bike which retailed at the time for over £500 (yes it was a long time ago) cost less than £100 to manufacture.

 

[subroc]

I have held the position of planner/estimator. It was at the job
shop level and not at the corporate level with huge production runs.

I doubt you could accurately estimate their costs for screws never
mind production of their alloy bodies.

You are starting this effort with a guess on chip cost. Great place
to start. Your entire estimate process will be guesses.

This is an effort in futility that will tell you nothing.

Good Luck in your endeavor.

Many years ago I used to run a motorcycle dealership. I remember
two interesting facts that people with more time to waste than me
had established. One was that if you could strip down a bike and
sell all the parts as spares, it would fetch around five times its
retail price. The reason nobody ever did this is that most of the
parts would never actually sell. It works in theory but not in
practice.

The other was that a certain small bike which retailed at the time
for over £500 (yes it was a long time ago) cost less than £100 to
manufacture.

I just chuckled

I rode motorcycles for 20 years. Although, I sold my last one 15 years ago. I had a BSA 500, a Suzuki GT 550, A Honda 750 (the single overhead cam) and 2 Harley Davidson’s, a Sportster and a Superglide.

Your point on what would actually sell as spare parts on a motorcycle has merit. I remember in the late 80s being in a used parts for motorcycle place, we were looking for some Triumph parts. I was shooting the breeze with the sales guy and asked what were people really looking for, or what was rare? His answer, big bore Japanese front ends. He could sell all he could get his hands on. I expect young guys were wrecking them and they became the commodity. The rest, of the parts would/could sit on the shelf for a long time

I expect motorcycle front ends always sold well as spare parts.

)

 

[David Hull]

I couldn't follow the link last evening for whatever reason.

You obviously have trouble with people who disagree with you. I don't find this surprising given that most of the stuff you post shows show a total lack of understanding of fundamental marketing principles (or you choose to ignore them). So you probably get that a lot. However, as bad as your understanding of marketing is, it is eclipsed by your lack of understanding of engineering.

Your chip price is an example. You say "it won't be far off" but you quote a sample price, which is almost always "far off". Canon makes their own custom parts so they have no mark up over the cost to make the part. Most semiconductor companies like to carry 50 points of margin (they don't always get to -- but that is what they shoot for). Canon doesn't have to worry about this for DIGIC II. Their BOM cost will be the packaged and tested part cost -- they won't need to carry margin. So you can cut the Fujitsu price to $22 – still WAY too high IMO.

The large chips that you find in set top boxes which incorporate several ARM processors, and a couple of MPEG decoders COST around $10 to produce and sell for around $15 in large quantity (we are talking 100k -- what was the qty price for the Fujitsu part?) the DIGIC II is smaller and doesn't need the exotic package that the STO parts have with heat spreaders etc. I would guess that DIGIC II is $10 at best (in the quantities that Canon is using).

Your wafer cost is high as well. Last, I checked, 12 inch wafers running in the 90 nm process were around $3500, that is down from $5000 a year ago and is still declining. 8 inchers are more like $1500 or lower. The question is, does Canon have their own FAB or are they using TSMC or UMC to do it for them like everyone else. If they have their own FAB it will probably be lower still.

 

[Badbatz]

Profit - 25% on the retail level, 10% on the wholesale level and 25% on the manufacturing level. Thus the camera that retails for $1100 (as the 30D does in the US) costs $825 wholesale, $740 sourced from the manufacturer, and $575 to manufacture (and that last figure includes absolutely everything including corporate overhead, R&D, raw materials, components, labor, packaging, marketing, depreciation, hookers for the CEO of Canon, etc.)

 

[Elwood_98034]

I used to work for a large multinational photographic company, and the landed (import) cost of various machines I used to sell were as follows:
sale price $38,000 landed $7,000,
$20,000 was about $4,500,
$5,000 was $1,650,
$145,000 was $75,000
$250,000 was $50,000

The 'landed' costs were the 'transfer' prices, which is the price the company head office sold them to us as a branch distributor, and after H/O had made their profit margin to shareholders, and recouped all the R&D and other costs involoved in getting it put on a ship for export.

The sale price that I could sell it for was the price I needed to get to pay for local infrastructure costs, warranty, et al, and to ensure that the local national office made no net profit on the sale as to avoid the need to pay tax to the local national government, and hence keep their prices as low as they could locally while giving the best return they could to H/O. These transfer prices were reviewed constantly against the exchange rate and market conditions.

Some items were priced with lower sales prices/ratios than others due to marketing reasons.

I won't say which ones, but two of the above were types of (digital) cameras, but not ones you would buy at B&H.

The 5x ratio quoted by an earlier poster was not far off the mark for the industry, but it will vary by product. Ironically, the cheaper the product, the higher the sales price ratio usually is, but its 'net' is eaten into by shipping and handling costs. It costs just as much for someone to go pick a $40 handswitch off a warehouse shelf, as it does a $1,500 body, so the $40 needs to be priced effectively higher to recoup those fixed costs.

Also, the 'transfer' price is as I said, is what the factory needs to make to keep the shareholders happy, so the actual manufacture cost is substantially lower again. What the hardware actually costs is irrelevant, as there are so many other things that go towards getting the item produced and into a container sitting at the home port for export.

If you want to nuts and bolts it, a 30D probably costs maybe $125 in hardware, but you need to pay for the person who puts it together, the factory they sit in, the guy who drives the truck to the port, fuel for the ship, the guy in the big office with the gold Rolex, etc. The transfer cost could be as low as about $200 - $300 by my guess. The sale price to the dealer will be much higher than that, as the local company infrastructure etc has to be paid for.

What the local national company distributor gets for the product is irrelevant as far as the 'factory' is concerned, as long as they keep shifting units at the agreed price, so the 'factory' can make its profit.

Also, if it does cost $125 in parts to make the camera, that is great, but then we would need to be billed for all of the other expenses that go to getting it into our hands, and keeping the company in operation as an economic entity. That is where the additional $1,000 comes in.
--
Elwood.

Light! Give me light!

 

[thecameraguy]

--
~ Being over-exposed can get you arrested ~

 

[dboreham]

As someone who once had a job that involved predicting the
manufacturing cost of various electronics products, I can say that
you are wasting your time with this. It would take someone
with extensive experience weeks to complete the exercise
and come up with an accurate estimate. You are much better
off simply working 'backwards' from the retail price, assuming
a certain markup from the ex-works cost (typically around 3x).

 

[ceoyoyo]

Wow. You're one of the most defensive people I've ever met.

So you think a chip is the price of a wafer of silicon times the ratio of wafer area to chip area? Why don't we take that a step further? Silicon wafers are made of, well, silicon, which is just beach sand with the oxygen cracked out. Beach sand is basically free. So chips are free!

Except that you have to have someone design them, then you have to have a multi-billion dollar fab to produce them. Even then, not all the chips you print are going to work. Besides the loss from throwing away that valuable silicon you have to pay someone (or buy a machine, or both) to test every single one of them.

Don't think my beach sand starting point is fair? Then why would your silicon wafer starting point be? As others have said, the only way you're going to get a remotely realistic idea of how much a camera costs to manufacture is by starting with the retail price and making three or four assumptions about markup at each level. That is opposed to starting with beach sand, oil and rocks (or anywhere else in the production process) and making zillions of assumptions.

If you want a good example of how badly wrong this process can go, take a look at pre-PlayStation 3 release speculation. The professional speculators, while trying to guess its release price, put its manufacturing costs all over the map. They had a much easier job too, because they were dealing with something that is effectively a PC with a couple of special parts. Anybody can build a PC from parts, so the component prices are well known.

--
Robb

 

[David Hull]

Current Q2, FY-07 wafer cost from TSMC for 8 inch wafers in the 150 nm process is about $975. Similarly for 12 inch wafers in the 90 nm process TSMC is at (currently) $4950. This is all dependent on FAB load, etc. and changes often. Another FAB, CSM, is quoting $3800 for 12 inch in the 90 nm process.

I would use $1k per for your wafer but once again, you don't know what Canon pays (and never will). These are quiotes from the largest FAB in the world, and are just quotes, Intel gets the best prices on 90 nm and 65 nm 12 inchers from TSMC, for obvious reasons -- their volume is huge.

 

[Alvin Nunley]

Maybe a little low. Looking at a parts book is going to give us little to go by. And even if you knew the price of part in the catalog, (they have their own markup path) that price would have nothing to do with what the parts in the camera cost.

Take the battery for instance. Canon package, $60.00 SterlingTeck $12.00. Some difference between them, but hardly enough to justify a 500% price difference. I'll bet SterlingTeck is making a 100% markup.

 

[Alvin Nunley]

This would tell us what the sensor in our XXD camera cost. And is that cost the same as 4 years ago.

 

[Elwood_98034]

Distribution was mentioned - but to where? West coast USA will be cheaper than Europe, as it is closer. Does Canada get theirs via the USA, or through Vancouver direct. So Vancouver prices should be cheaper than Toronto, as it is closer to the factory, but Canon Canada has a central distribution point, so it has to go there first, and THEN back to Vancouver, so the prices will be higher... Then there is import duty, and the price of shipping which will probably be contracted a year or two in advance, so are the shipping contracts worked out at high oil price levels, or low? And that goes for every other item in the chain of production. We can't possible guess the cost of shipping, as we don't know the volumes to each market.

Another thing is is Marketing paid for from a seperate budget direct from the Factory, or are the local Sales organisations liable for their own marketing costs? Or is there a mix, with the Marketing staff bedget paid for by the Factory, but the rest of the costs being local? And are different countries supported differently? I bet they are, so their unit cost will be lower.

It is a pointless exercise, no offence meant to the OP himself.
--
Elwood.

Light! Give me light!

 

[eelnoraa]

I must think the OP does NOT know much about the whole semi-conductor instrustry or the while hi-tech industry.

First, the cost of making a piece of equipment is way, way more the the cost of its individual component. There are assembly, yield, testing, packaging ..... This still hasn't include the cost of R&D.

In semi-conductor industry, testing alone can be as high as 20% of the total cost depending on the complexity of the chip.

2nd, assuming the yield of 85% on APC sensor may not be so reasonable. Because the area of APC sensor is considered huge in semi-conductor industry. As area goes up, yield decrease significantly. I don't have a good number for yield change vs area, but this is the general case.

Anyway, I don't know why I am even responding to this thread.

eel

 

[MichaelOuellet]

Just for fun.

You can make a rough calculation of the number of hour you have used your equipment.

Divide the cost of your equipment by this number and, I'm pretty sure that using your camera cost less than going out to see a movie or having a lunch with your friend.

If not! Stop reading this forum, take your camera and go shooting.

For me, value is more interesting than cost.

--
Michael Ouellet
Quebec city, Canada
http://michael.volcan.ca/

 

[RichW]

I got a good laugh at that one.

I'm not sure of the point of this exercise. Selling cost is based on market demand and has little to do with manufacturing cost as long you can sell at a profit.

Rich

Profit - 25% on the retail level, 10% on the wholesale level and
25% on the manufacturing level. Thus the camera that retails for
$1100 (as the 30D does in the US) costs $825 wholesale, $740
sourced from the manufacturer, and $575 to manufacture (and that
last figure includes absolutely everything including corporate
overhead, R&D, raw materials, components, labor, packaging,
marketing, depreciation, hookers for the CEO of Canon, etc.)

 

[Progress Lover]

Especially when there are a lot of guesses comprising the main total.

Some will be overestimates, some will be under some will be GREATLY over and some will be GREATLY under. On average they will ALL be about right.

This is how come it is so easy to estimate complex stuff so darned accurately. Bizarre quirk of the system.

You would know that as a planner/estimator.

I have held the position of planner/estimator. It was at the job
shop level and not at the corporate level with huge production runs.

I doubt you could accurately estimate their costs for screws never
mind production of their alloy bodies.

You are starting this effort with a guess on chip cost. Great place
to start. Your entire estimate process will be guesses.

This is an effort in futility that will tell you nothing.

Good Luck in your endeavor.

--
Keep photography wild.

 

[jman32]

while cost matters when a company determines whether or not it will be in a given business, pricing always comes down to supply and demand.

if nikon comes along and offers a vastly superior product or vastly cheaper product to canon, canon will be forced to lower its price.

for manufacturers that can make enough units to meet the demand, a lack of supply will sometimes raise the price in the beginning of a product cycle before the supply meets the demand.

and if you truly want to play the "cost game", there are costs that we can not begin to appreciate: R&D, customer service, overhead, advertising, firmware updates, software development, disposing of waste, currency changes, testing, and many more

--
jeremy
there is no such thing as a stupid question
only stupid people asking questions ;-)

 

[Progress Lover]

Yes pretty much my thoughts.

However, it would be nice to get an idea of the component prices themselves and to break things down if at all possible so we can see exactly where most of the money goes and why.

There is a lot of talk on these forums from people who belive that parts cost a fortune and they cause the price of DSLRs. This exercise can debunk that myth and show clearly that a DSLR with double the performance need only cost a small amount more to produce because the costs to the end user of assembly, distribution, margin, dealer's cut, packaging, marketing, etc remain almost entirely unchanged...

Your post goes most of the way to pointing that out anyway, but not as explicitly as a breakdown of costs could with actual figures for people to quote.

This information in the public domain has the potential to significantly enhance public understanding on the subject imho. Even if not, it will help mine and anyone else that's interested!

--
Keep photography wild.

 

[David Hull]

I doubt that they are using 12 inch, but that is pretty much my point -- nobody knows, so this exercise has limited value. There was a white paper from Canon a while back that had some of that information. I don't have the link handy.

You can get a rough idea of how many sensors come off a wafer from as follows: A=PI*(D/2)^2, divide that by the area of the APS-C and deduct a bit for the loss on the corners of the reticule. Then there is the yield -- I don't know what that might be for sensors of this type. For normal IC's we try to get in the 90's. If we end up much lower than that we have what we call product engineers who go to work looking for the root cause of the yield hit and a solution.

Incidentally, the mask cost for a 12 inch 90 nm part is $1 million -- yep, that is Million with a big M. So each time you FU%K up, cha-ching, TSMC gets another cool million. For the eight inchers in 150 nm, that is more like $600k. Obviously a rev B or Rev C is to be avoided. Engineers really get slapped hard if there is a Rev C. This is why Canon doesn’t come out with a new processor every year or so, it is really quite expensive although that cost is spread across all the cameras they make. My experience has been, that no company likes to spend a Million if they don’t have to – maybe oil companies would be an exception .