Estimate a BOM delta for PS5 and XBSX

There's the gyroscope, touchpad, microphone array, and speaker in the DualSense. So there's definitely some amount of additional cost there. What amount though? $1? $10?

Former aerospace engineer and gyroscope connoisseur here! The gyroscopes that Sony would use are not expensive, probably under 25 cents, maybe even less. They don't need anything more sophisticated. You don't need 1,440 degree precision at 5,000hz!
 
Former aerospace engineer and gyroscope connoisseur here! The gyroscopes that Sony would use are not expensive, probably under 25 cents, maybe even less. They don't need anything more sophisticated. You don't need 1,440 degree precision at 5,000hz!
Yeah today they make all-in-one chips with 3 axis gyro, accel, mag, for less than $1. It wasn't the case when the DS3 came out though, that single axis gyro must have been a few dollars.

I remember when piezo-ring gyros were coming out in the consumer space (late 90's?), it was over $100 for a single axis to use on my RC Heli. The rate wasn't that impressive either compared to what we get today. I quickly learned about destructive oscillations and damping factors. Destructive. :LOL:
 
I remember when piezo-ring gyros were coming out in the consumer space (late 90's?), it was over $100 for a single axis to use on my RC Heli. The rate wasn't that impressive either compared to what we get today. I quickly learned about destructive oscillations and damping factors. Destructive. :LOL:
Yeah, it's all laser ring gyroscopes now. You can have miles of fibre optic coiled in a single gyroscope. It's incredible engineering. Expensive but incredible.
 
Both companies using TSMC doesn't mean it's the same cost. If we assume that the controller chip is roughly the same size as the chip used in the XBSX then the 2 major influencers of cost per wafer will be.
SB
I was not making a comparison to Phison and did not imply the cost of manufacturing their controller will be cheaper than or equal to Phison. My point was it will likely be cheap to make than people think and likely cheaper than buying a similarly spec controller from someone else. If we want to discuss bargaining and buying power, the logic layer for all Sony cmos sensors are manufactured by TSMC and they are in the process of moving some of their sensor manufacturing to TSMC as a backup fab. How many billion sensors do you think Sony sells each year between every smartphone manufacturer that uses Sony sensors which is just about all of them, from Apple, Samsung, Huawei to Xiaomi, Oppo? And that's not including other sectors of the cmos market that Sony also supplies. Or do you figure Sony cannot leverage their partnership in other parts of their business to get a good deal?
 
I was not making a comparison to Phison and did not imply the cost of manufacturing their controller will be cheaper than or equal to Phison. My point was it will likely be cheap to make than people think and likely cheaper than buying a similarly spec controller from someone else. If we want to discuss bargaining and buying power, the logic layer for all Sony cmos sensors are manufactured by TSMC and they are in the process of moving some of their sensor manufacturing to TSMC as a backup fab. How many billion sensors do you think Sony sells each year between every smartphone manufacturer that uses Sony sensors which is just about all of them, from Apple, Samsung, Huawei to Xiaomi, Oppo? And that's not including other sectors of the cmos market that Sony also supplies. Or do you figure Sony cannot leverage their partnership in other parts of their business to get a good deal?

That's assuming it's at TSMC and using the same node.

https://www.digitimes.com/news/a20191209VL202.html

Indicates that Sony are using TSMC for at least some of their CMOS image sensors, but that they are using the 40 nm process tech. So they could be making it there, but that would put it at a size and likely cost disadvantage to Phison who utilizes 28 nm for the rumored XBSX controller.

Also, this assumes they would be the same size on the same process node. But it's entirely possible that Sony's custom chip would be larger on the same node. The rumored chip for the XBSX is pretty bare bones.

Regards,
SB
 
Last edited:
That's assuming it's at TSMC and using the same node.

https://www.digitimes.com/news/a20191209VL202.html

Indicates that Sony are using TSMC for at least some of their CMOS image sensors, but that they are using the 40 nm process tech. So they could be making it there, but that would put it at a size and likely cost disadvantage to Phison who utilizes 28 nm for the rumored XBSX controller.

Also, this assumes they would be the same size on the same process node. But it's entirely possible that Sony's custom chip would be larger on the same node. The rumored chip for the XBSX is pretty bare bones.

Regards,
SB
They are moving some of their sensor layer manufacturing to TSMC but apart from that, TSMC also manufactures the logic layer in their stacked sensors. They don't use just one node, they use TSMC 40nm, 28nm for their ISP. It wouldn't even matter if they are on the same node, having business partnerships that brings in hundreds of millions every year should afford them some bargaining power in my opinion, no?

I'm not comparing it to what's presumed to be in XBSX. The only assumptions I'm making is that designing their own controller and having a fab make them offers them a lot of flexibility on cost than simply buying off shelf parts does. I believe that in the end, it will be a lot cheaper than people think and won't add a significant cost to their total BOM as the specifications imply.
 
I never thought the cost of manufacturing the SSD controller would be anything significant compare to COTS, but the real difference would be in the sunk costs of R&D Sony had upfront.
 
I never thought the cost of manufacturing the SSD controller would be anything significant compare to COTS, but the real difference would be in the sunk costs of R&D Sony had upfront.

They might share that out similar to Microsoft and its cloud teams.

A stupid fast ssd would be great in Sony Pro cinema camera business. Raw 4k
And 8k streams are bandwidth hogs, knowing you support that under all loads with qos or similar is probably a must.
 
A stupid fast ssd would be great in Sony Pro cinema camera business. Raw 4k
And 8k streams are bandwidth hogs, knowing you support that under all loads with qos or similar is probably a must.
Good point, but Sony's divisions haven't worked that way before. Wouldn't there be some mention of this in their imaging division or some other shareholder-focussed PR statement?
 
They might share that out similar to Microsoft and its cloud teams.

A stupid fast ssd would be great in Sony Pro cinema camera business. Raw 4k
And 8k streams are bandwidth hogs, knowing you support that under all loads with qos or similar is probably a must.

That doesn't wipe away the front-loaded cost of Research and Development to create a custom part. It makes the cost per unit lower if quantities can be ramped up to the next level in the pricing contracts. You still need to recoup the R&D costs somehow.
 
It will definitely be more expensive because there's 3 times more silicon area and IOs than a 4ch controller.

But I guess at this sort of volume the NRE shouldn't be a big expense for such a small chip. There's basically 3 IPs to use, the NAND PHY, the ARM cores, and the pcie 4.0 PHY. So $1 per console is $100 million to develop the controller. Sony have been co-authors of the ONFI specifications since it's introduction meaning they certainly have the expertise and silicon IPs necessary.

There isn't a big distinction between this controller and a custom south bridge. They could literally make the south bridge the main flash controller instead of a separate chip. They are developing it anyway, add nand PHYs and it's a pretty well integrated IO controller.
 
Back
Top