You just answered your own question.... 
Cost per chip = ((Development cost + fixed production costs (e.g. fab building, setup costs etc) + other fixed costs) / sales volume) + (production cost per chip + other variable costs)
Development cost and other fixed costs are very high (in the range of 10's of millions of USD) and need to be offset by high volumes.... All variable costs are relatively low (in the range of single dollars or less).
Hence the impact of sales on total costs is very high, therefore the impact on price.
EDIT: I think we are making the same point here...
Cost per chip = ((Development cost + fixed production costs (e.g. fab building, setup costs etc) + other fixed costs) / sales volume) + (production cost per chip + other variable costs)
Development cost and other fixed costs are very high (in the range of 10's of millions of USD) and need to be offset by high volumes.... All variable costs are relatively low (in the range of single dollars or less).
Hence the impact of sales on total costs is very high, therefore the impact on price.
EDIT: I think we are making the same point here...
Last edited:
No, I rather think my point is different
I'm saying small or medium volume (projected) guarantees no chip design is even begun.
Once we begin a chip design we're fairly sure we're looking at a million chips. As volumes go beyond this then the upfront costs (development) drop down into the noise and the costs of production take over. Testing time is very expensive coz its done on equipment of very high capital cost which is being depreciated.
Thus given two chip designs, both produced in the millions, the cost of development is not the major factor and hence volume is not the key determiner of price.
Actually the 10's of millions cost I cited before does not apply to the DAC chips we're talking about, it applies to state of the art designs at the current bleeding edge process node. DACs are not produced on such processes to my knowledge so perhaps the break-even point comes down below a million, I don't know for sure.
I'm saying small or medium volume (projected) guarantees no chip design is even begun.
Once we begin a chip design we're fairly sure we're looking at a million chips. As volumes go beyond this then the upfront costs (development) drop down into the noise and the costs of production take over. Testing time is very expensive coz its done on equipment of very high capital cost which is being depreciated.
Thus given two chip designs, both produced in the millions, the cost of development is not the major factor and hence volume is not the key determiner of price.
Actually the 10's of millions cost I cited before does not apply to the DAC chips we're talking about, it applies to state of the art designs at the current bleeding edge process node. DACs are not produced on such processes to my knowledge so perhaps the break-even point comes down below a million, I don't know for sure.
True, the resistive part of the multibit doesn't play well with the CMOS process. The laser timming part that I said about.
And what is not a difference that is semnificative? Even then... how would you know which is the "good" signal? And probably you still could blame the ADC (that is S-D) for lack of signifiance, forgetting that most of the digital music comes from the same exact ADC.
Last edited:
Its a good question - I reckon the nature of the difference signal would give us a clue. But its just a hypothesis Also I'd try subtracting the original digital signal from both too to see what the difference signal there looked like.
No, when I eventually get around to doing this, I'll be sure to use a multibit (SAR) ADC.
Also I'd try subtracting the original digital signal from both too to see what the difference signal there looked like.No, when I eventually get around to doing this, I'll be sure to use a multibit (SAR) ADC.
Since it looks like the chip this thread's about isn't going to happen, is there something wrong with taking it a bit meta about the whys and wherefores of S-D vs MB?
@SoNic - I do have my own musical instrument and microphones if I can't find anything suitable I believe some of the earliest Neve digitals used MB - not for sound as I understand, but for latency. Have no idea what they do now.
@SoNic - I do have my own musical instrument and microphones if I can't find anything suitable
I believe some of the earliest Neve digitals used MB - not for sound as I understand, but for latency. Have no idea what they do now.The problem is no one can say what the volume would be.
This thread is full on nonsense about a nothing datasheet. Its not going to generate enough interest to tool up for a MB chip, forget it.
Go with a Far Eastern TDA1541A, no BS and they sound good. Sometimes counterf**ting is a good thing and a 2012 MB DAC chip is the perfect example.
They do DXD 384kHz and DSD in the 88D. "Proprietary ADC". Like everybody
TI has PCM5100 family...
Last edited:
Oh, thank a lot for your reply.
I had misunderstood that it had something to with DSD-Wide.
Bunpei
Neve is not there. He is here:
Rupert Neve Designs
They only use his name, it's like Fender was after Leo left the company.
Hi, gusp!
Are you still retaining your interest on the ADC chip?
If so, please look at this post!
http://www.diyaudio.com/forums/digital-source/177336-usb-dsd-dac-adc-electrart-6.html#post3755634
Bunpei
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.