How much does a medium format digital sensor actually cost?

[Brian Hirschfeld]

I've heard rumors that the Sony 50mp CMOS sensor costs around ~$2500 dollars a pop. I know that the pricing depends on the number of units ordered and other factors. But, for other digital backs like the 33-50mp sensors, how much do we think that the actual sensor itself, as a component costs? Can't seem to find any information about this online. Would love to hear anyones thoughts.

Best,
BH

[torger]

There was a time whenever a new chip was released it said "costs xxxx when ordered in volume", but this is rare these days.

When Kodak released their KAF-50000 (50 megapixel 49x37mm) the pricing was official, unfortunately I can't find the press release now but the cost was somewhere $2500 - $3500 to the manufacturers.

Sensors are also sold in various quality levels (depending on how many broken pixels they have), hopefully the manufacturers put the highest quality sensors in their cameras and I think they do. There's still usually a few hundreds of broken pixels mapped out in the meta data., at least for CCDs.

I think the new Sony CMOS 44x33 sensor is cheaper, heard $1000 but no source, anyway the Pentax pricing indicate that this is not too unlikely.

Anyway it's not the sensor cost that make digital backs expensive, it's the low volume business model that makes you pay a large part of R&D costs per unit. About $20-$25K extra compared to if it had been a product with reasonable volume, just compare IQ250 and 645Z. I really hope Pentax pulls off their stunt and show that MF systems can be sold in volume, CMOS (makes product more all-around and attractive to more users) and volume sales (makes it more affordable) could be a new beginning for MF.

[FredAz]

12-inch wafer starts are $5k to $8k PER WAFER depending on the process.
If it's a low volume process, these costs can be higher.
This is INDEPENDANT of yield.

Figure, at BEST, a wafer will yield 10 working sensors -> cost / sensor is already pushing $1k.

[eronald]

You can work things out for yourself.

Let us imagine the process of making a 48x36 sensor. We fish one out of the bin and test. [Wrongly] assuming non-correlation, your chance of getting a good 48x36MF sensor is that of having the "second 35mm half" of the sensor work when the first one does. so it is the 35mm yield y3624 [get first half] *y3624 [get working second half if you have the first].  

But our 48x36 sensors are also twice as large, so we carve at most 1/2 as many out of a [large, avoid idiotic edge effects] wafer anyway.

So in the end we scale down our number of good sensors per wafer from NgW3624 to 0.5*NgNW3624*y3624.

Unless I'm wrong these numbers aren't too bad if your yield for 35mm is ok, but they are horrid at the other end: at yield 0.7 you get 0.35 and your MF sensor is 3x more expensive than 35mm. At yield 0.5 you get 4x, at yield 0.2 you get 10x and at yield 0.1 you get 20x which is probably around where we were when the P25 first came out. Yields of course go well up with time, but Phase One's prices seem to defy progress.

Tweener 44x33 sensors don't have such a high scale up, I leave them as an exercise for the reader

If we want we can feed in some dollar costs derived from the retail price of 35mm fullframe; we can see that Pentax will probably be very decently profitable, reasoning strictly from the BOM point of view.

Edmund

[eronald]

And from the above reasoning you can derive an interesting marketing strategy: Have a large-sensor product in your range that is really expensive and which you don't sell, and actually sell a somewhat smaller sensor at per/megapixel pricing computed from the top. The expensive top price is justified by the strongly dwindling yield, the rebate is only proportional. And your pricing looks ok because the competitors cannot field a similar high end product more cheaply than you. Hmm, I wonder if that hasn't really been done


Edmund

[BernardLanguillier]

And from the above reasoning you can derive an interesting marketing strategy: Have a large-sensor product in your range that is really expensive and which you don't sell, and actually sell a somewhat smaller sensor at per/megapixel pricing computed from the top. The expensive top price is justified by the strongly dwindling yield, the rebate is only proportional. And your pricing looks ok because the competitors cannot field a similar high end product more cheaply than you. Hmm, I wonder if that hasn't really been done

That's a great idea, but does it still work if a competitor prices a full camera using the same sensor at less than 1/4 th the price?

Cheers,
Bernard

[eronald]

That's a great idea, but does it still work if a competitor prices a full camera using the same sensor at less than 1/4 th the price?

Cheers,
Bernard

May I correct you:
 Does it still work if a competitor prices a full BETTER camera using the same sensor at less than 1/4 th the price?



If a competitor breaks the Omerta' and prices fairly, you can sue, cash out, or tell people that your product is expensive because it is a luxury product, has never been cheap and never will - the Hermés, Vuitton or Bordeaux strategy which you are certainly familiar with, as people come to french business schools to learn branding. Margins and high prices don't count against you in luxury, and the dealers are fiercely on your side.  Of course Phase and Hassy know about fashion and luxury

Edmund

[Paul2660]

I think the actual cost per sensor would be in the 1K to 1.5K range, but the actual cost of the sensor is such a small part of the total equation I feel it gets a bit blown out of proportion. 

A few more things to consider for now:

1.  Phase One is the only company actually shipping this chip, so for now they can set the price point anywhere they want.  I don't think the blad version has shipped, please correct me if I have that wrong, and I have not even seen any early ship reviews from the 645z.  It was supposed to be shipping NOW, so time will tell.

2.  I believe that Phase One's entire market strategy is setup around smaller volumes, thus higher unit prices.  They have a smaller infrastructure than Pentax/Ricoh, or Nikon/Canon.  Their current go to market as pretty anyone who has dealt with them knows, is use our dealer channel, both for prices, and support.  You can't pick up the phone and call Phase One and get a quote very easy.  You also don't go into your local camera store for a price.   If I have learned anything since 2008 it's pick a dealer, and use them for your entry point for everything.  If Phase One is driven to a lower pricing model by huge Pentax 645z sales, then so be it, they will react with pricing.  Right now it's clear that Phase One is attempting to clear the channel of the 40MP chips with for them aggressive pricing.  If the 645z starts leaving a lot of IQ250's on the shelves, then Phase One will either make the decision to sell at a lower price or pull the product.  They have a pretty steep R&D curve to re-coop so I don't think they will pull the product. 

3.  Ask yourself, where are the IQ250's going?  I have not seen a lot of posts from new owners on this site or others, where as, the number of posts from credo 40 and IQ140 users has increased.  I am sure they are selling them, but the total volume only Phase One knows.  The tech camera market, which should be at least 1/4 or 1/5 of MF back users have not embraced the 250 yet, again at least no one is talking if they have. 

4.  Don't' forget Silverfleet's ownership of Phase One.  They are in this to make money so I am sure they are monitoring sales of backs. 

5.  Lastly, it's small market.  You can count the number of authorized Phase One (ACTIVE) dealers in the U.S. on one hand max 2. (CI, DT and Bear are the ones that come to mind, but I am sure there are others).  That's the marketing model that Phase One is working with.  I also feel that Phase may be selling a lot of product to the Asian markets, mainly China.  I know that China has recently been very active with Schneider glass, in the 100 and 120mm range, so they are more than likely making a lot of back purchases also.  These are sales that only Phase One knows and accounts for. 

Paul

[Doug Peterson]

3.  Ask yourself, where are the IQ250's going?  I have not seen a lot of posts from new owners on this site or others, where as, the number of posts from credo 40 and IQ140 users has increased.  I am sure they are selling them, but the total volume only Phase One knows.  The tech camera market, which should be at least 1/4 or 1/5 of MF back users have not embraced the 250 yet, again at least no one is talking if they have. 

We've been happy with our IQ250 sales, but very few have been to the market that most frequently come to this board. Significantly more in the wedding, on-location fashion, and still-life markets than the landscape market.

I've shot three weddings with IQ250 units. The increase in range of high-quality ISOs is welcome in all markets to be sure, but is felt most in uses like weddings where having a great full-res ISO3200 is a real game changer for using medium format. 

[JV]

1.  Phase One is the only company actually shipping this chip, so for now they can set the price point anywhere they want.  I don't think the blad version has shipped, please correct me if I have that wrong, and I have not even seen any early ship reviews from the 645z.  It was supposed to be shipping NOW, so time will tell.

The H5D-50c shows as In Stock at B&H so I assume it is already being shipped.

The Pentax 645Z as far as I know is not being shipped yet.

[Paul2660]

JV,

Thanks for the correction, I haven't kept up with it as much.  B&H has "free" shipping also.  It's definitely priced more aggressive than the IQ250 @ 27K, which included the camera body, the HVD-90X Eye-level Viewfinder.  I don't know blad at all as I have never used it, but it appears this is setup totally dependent on the leaf shutter in the lens?  It looks as if the back is removable for us on a tech solution but can't tell for sure.

I wonder if the Phase One "new" body will be like this, i.e no integrated focal plane shutter.  

As they say "let the games begin"

Paul

[gerald.d]

3.  Ask yourself, where are the IQ250's going?  I have not seen a lot of posts from new owners on this site or others, where as, the number of posts from credo 40 and IQ140 users has increased.  I am sure they are selling them, but the total volume only Phase One knows.  The tech camera market, which should be at least 1/4 or 1/5 of MF back users have not embraced the 250 yet, again at least no one is talking if they have. 

Paul

My understanding is that they are selling very well indeed. If the figure I've heard is correct, then the 250 is delivering significant incremental revenue to Phase One.

I don't think they make a huge lot of sense for the traditional tech camera market (in the main, due to the crop sensor, but as we have learned there are also considerable issues with regards utilising the tech wides on this sensor), but once you consider the ALPA FPS and the ability to use it with the Canon TS-E's? Well that changes the situation entirely.

I'm creating a library of 8K time-lapse content with mine, and I very much doubt I'm the only one. Even when rendered at 4K, the footage is making the Dragon shooters in the office drool. I just wish it supported a higher frame-rate, because it cannot get even close to the claimed 1.2FPS over the longer term (pretty sure it can only support that rate until the RAM is full, and then it simply can't get the data out of the back fast enough).

[Doug Peterson]

My understanding is that they are selling very well indeed. If the figure I've heard is correct, then the 250 is delivering significant incremental revenue to Phase One.

I don't think they make a huge lot of sense for the traditional tech camera market (in the main, due to the crop sensor, but as we have learned there are also considerable issues with regards utilising the tech wides on this sensor), but once you consider the ALPA FPS and the ability to use it with the Canon TS-E's? Well that changes the situation entirely.

I'm creating a library of 8K time-lapse content with mine, and I very much doubt I'm the only one. Even when rendered at 4K, the footage is making the Dragon shooters in the office drool. I just wish it supported a higher frame-rate, because it cannot get even close to the claimed 1.2FPS over the longer term (pretty sure it can only support that rate until the RAM is full, and then it simply can't get the data out of the back fast enough).

You're using USB3 with an SSD and IIQ-S file format?

[gerald.d]

You're using USB3 with an SSD and IIQ-S file format?

Nope - SanDisk Extreme Pro (UDMA7 - 160MB/s) 128GB CF card (and IIQ-S). I don't trust it under 0.5fps for extended work.

One problem is that it simply isn't practical to tether in all situations, but out of interest, have you successfully tested 4000 images at 1.2fps over USB3?

Phase One's own marketing material states "Your uninterrupted capture flow is ensured by a capture rate of up to 1.2 frames per second and a large 2 GB image buffer in the digital back", which kind of implies the buffer is required for that rate - i.e. once the buffer is full, it will start to stutter (which is my experience).

Kind regards,

Gerald.

[BernardLanguillier]

I think the actual cost per sensor would be in the 1K to 1.5K range, but the actual cost of the sensor is such a small part of the total equation I feel it gets a bit blown out of proportion. 

A few more things to consider for now:

1.  Phase One is the only company actually shipping this chip, so for now they can set the price point anywhere they want.  I don't think the blad version has shipped, please correct me if I have that wrong, and I have not even seen any early ship reviews from the 645z.  It was supposed to be shipping NOW, so time will tell.

The 645z will ship on June 27th in Japan. This is an official Pentax announcement.

So P1 has 24 days left of exclusivity.

I doubt they will lower the price of the IQ250 a cent on June 28th, so I would think that the discussion on the price difference can safely be had already.

I agree that China is probably buying in significant amounts, status is important for the new urban rich in Eastern Chinese cities. They learn fast and get bored quickly though.

Korean and Japan are obviously much more mature markets and I have never seen a single P1 back used for landscape during my many outings to key locations in Japan, there are of course some in fashion.

Cheers,
Bernard

[eronald]

The 645z will ship on June 27th in Japan. This is an official Pentax announcement.

So P1 has 24 days left of exclusivity.

I doubt they will lower the price of the IQ250 a cent on June 28th, so I would think that the discussion on the price difference can safely be had already.

I agree that China is probably buying in significant amounts, status is important for the new urban rich in Eastern Chinese cities. They learn fast and get bored quickly though.

Korean and Japan are obviously much more mature markets and I have never seen a single P1 back used for landscape during my many outings to key locations in Japan, there are of course some in fashion.

Cheers,
Bernard

Bernard,

 There is no sane discussion of MF prices to be had here. However, I guess we are going to see some interesting cameras thanks to the Sony chip, and that will provide material for talking. Starting with the Pentax, a Leica announcement at Photokina, and also larger size chips from Sony this year. If Rollei still had any financial life, they could commission an open source factory back for the Hy6, I would love to collaborate on such a project.

Edmund