Advertisement

Blog

Tradeoffs: Should I Use an Integrated Analog IC or Individual ICs?

I've had some additional thoughts on creating integrated analog ICs. These thoughts came to mind following the publication of two of my blogs on debugging custom ICs and putting a DAS on your PCB to make it easier to trouble-shoot remotely; and Bill Schweber's blog on debugging tactics.

Sometimes, it may be more appropriate to use individual ICs that contain small- or medium-scale integration IP (intellectual property). Consider an example: Let's say you're designing a product that is a small stereo (2-channel) audio processing device for a consumer application. Let's assume it will accept several audio inputs, process the audio, amplify the audio, and drive two speakers.

Let's further assume that the “processing” means a multi-band EQ, balance control, and volume control. And you'd like a very simple user interface — perhaps a few push-buttons to increase and decrease the settings. That's all you need , but you wouldn't object if you could get a few more features thrown in for little or no additional cost — like reverberation, bass enhancement, and spatial separation tweaking.

After a look at various manufacturers' sites, you see that a few companies have the IP you need and the extra features that you would like thrown in. The per-unit cost is quite reasonable. You would be able to build your audio device on a tiny PC board and would just need to add connectors, speakers, and a power supply.

But wait! There's a significant NRE (non-recurring engineering) cost. That's the one-time cost that the IC company is going to charge you to design and test your circuitry on a chunk of silicon. If you were planning on building 1k or 10k units, that NRE amortized over a production run that small would make the per-unit cost exorbitant.

So, in an example such as this, you may be better off picking your op-amps and class-D amplifier ICs, laying out a larger PC board, and building your product the old-fashioned (i.e., pre-21st century) way. There is an added benefit: Trouble-shooting your device (both at the design and manufacturing stages) is far easier and less costly.

Have you encountered a problem like this before? At what manufacturing quantity did you hit the break-even point (integrated version vs. individual ICs)?

— Brad Albing, Editor-in-Chief, Planet Analog and Integration Nation Circle me on Google+

Related posts:

24 comments on “Tradeoffs: Should I Use an Integrated Analog IC or Individual ICs?

  1. SunitaT
    December 31, 2013

    @Brad, one more factor which determines whether we should go for integrated Analog IC or Individual IC is the amount of space we have. If the product is compact and space is constraint then its better to choose Integrated Analog IC.

  2. Vishal Prajapati
    January 2, 2014

    @Sunita, Ya obviously the portable device with space constraints have different consideration all toghether. It needs to consider power, efficiency, space, aesthetics, portability, connectivity, etc.

     

    I think Brad is talking about general scenario where no such restrictions are there. You can have two options amongst which you can choose, a single IC or individual IC. The arguments in the blog stands very true to my openion.

  3. RedDerek
    January 2, 2014

    At what manufacturing quantity did you hit the break-even point (integrated version vs. individual ICs)?

    @BradA – This can be answered by the semi manufacturing company. Find out the cost of 1000, 5000, 10,000, 50,000, etc volume. Then the designer would have to have a real market analysis of what the sales volume would be and at what price. Then run the numbers as a marketing person would.

    I did this for dozens of products that I pushed for when I was Technical Marketing Manager at Vishay-Siliconix. It was the only way to see if I could get a product started in the development process. I guess it could be used for a blog in the future – now noted.

  4. Davidled
    January 2, 2014

    Determination for using either an Integrated Analog IC or Individual IC could be depended on how to design it and production line in my view.

  5. Brad Albing
    January 2, 2014

    @VP – yes, I had intended that to be in regard to the more general scenario. Thanks for that clarification. Sometimes in my haste to make a point, I omit a few details. So for space and power constrained situations, it makes sense to use a more integrated approach. Still, even then, if you are only building small production quantities, it may just not be cost effective and practical.

  6. Brad Albing
    January 2, 2014

    @RedDerek – nicely summarized. You're right about that approach – that's what the Marketing guys do with their spreadsheets. And yes, sounds like that would make for a good blog topic. Let us know if you want to prepare such a blog.

  7. RedDerek
    January 2, 2014

    I put it on my blog list. Just will have to sanitize a bit to protect some confidential stuff. But the concept will be there.

  8. David Maciel Silva
    January 3, 2014

    To justify the investment would be necessary to analyze the cost of sale is the final customer, in case if this product becomes a fad, it will be really great, because using a single ci, we can master the manufacture and have a CI, which is unique summarizing the company so you can ask for more of this product ICs, will be dedicated.

    On the other hand if none of these hypotheses are true, and size is not an impediment can simply sell a lot of 1000 pieces and never see the client.

    Will depend on the (ROI) return on investment.

  9. goafrit2
    January 5, 2014

    Largely, a monolithic IC will be the better strategy. However, it turns out that we cannot make all things inside one IT. What I do is to pack as many things I can inside an IC so that I have fewer needs of individual ICs. It is a really challenging strategy as if the IC fails, everything is done. But for cost, you are in a great competitive advantage provided you have volume to justify it.

  10. goafrit2
    January 5, 2014

    Determination for using either an Integrated Analog IC or Individual IC could be depended on how to design it and production line in my view.

    I do not think that should be the factors. The key is the size of the market. If you are making a product that few will buy, it may not be helpful to make all in one IC. But if you break the units, people can use them in different other applications. Volume should drive the decision making. It is never a good idea to go full monolithic when you do not have volume.

  11. samicksha
    January 6, 2014

    I agree you Maciel on your point about ROI, yes it keeps relevance before you give a practical outlook to your decision and design. But before proceeding with any such decision design makes a difference here, i.e. what kind of program i design keeping the fact in mind that analog signal amplification and filtering is usually critical which can consume a good space in design.

  12. David Maciel Silva
    January 6, 2014

    Analyzing the point “physical space”, we have a point other than the cost, it will have to be small ..

    For this angle without counting costs, and thinking only of the physical space, is the best way of all to be miniaturized.

  13. samicksha
    January 7, 2014

    Analyzing the point “physical space”, we have a point other than the cost, it will have to be small ..

    I am little confused,  is it you are suggesting to use individual IC..

  14. fasmicro
    January 8, 2014

    The decision to go completely monolithic or otherwise cannot be universal. It has to depend on the product or project. Anyone that has worked in some HV designs will conclude that there are still many things that can be done off-chip. It costs silicon to have charge pumps to bump 1.8V to 42V. Some of those big caps used in MENS analog front ends could be better off-chip.

  15. fasmicro
    January 8, 2014

    >> The key is the size of the market. If you are making a product that few will buy, it may not be helpful to make all in one IC.

    Remeber that there are some designs which can only be done off-chip. Floating gates are being used to make specia types of anaog memories. But to get that, one has to have a special production line. If a designer thinks he can get everything inside a chip, the person will be thinking we do not have PCB anymore. PCB is there to bail out for those units that are better down as individula ICs than being morphed onto one big one.

  16. Victor Lorenzo
    January 9, 2014

    I agree on the implications of production volume on deciding whether to go 'almost-everything-on-one-chip' or not, and also on the need to keep some types of components and ICs as individuals. There's another mixed solution I've seen around for quite long time now, it consists on placing the individual components on the PCB (SMD first), glue the dies for the ICs with the bumped pads upside and then wire-bond them to tracks on the PCB. Cost is reduced on one side as the individual ICs are acquired on wafer (encapsulation cost reduction) but on the other side cost is rised as specialized mounting and wire-bonding machinery is required.

    On some cases, using flip-chip technologies may also help.

  17. Bob F.
    January 9, 2014

    Generally speaking, it is relatively easy to determine from a cost standpoint if an Analog ASIC is beneficial over using a collection of standard analog ICs. Simply sum up the cost of all the analog chips (maybe a little digital too) and estimate the LIFETIME volume of your product…them plot your point on the graph below.  If you are above the line, it's a no brainer…do the ASIC… This analysis incorporates all front end NRE and mask tooling costs….can't get much easier.

     

     

    Of course, there are many other reasons to do an Analog ASIC and they have been well vetted in another discussion on this site.

  18. fasmicro
    January 13, 2014

    @Victor >> On some cases, using flip-chip technologies may also help.

    What is a flip-chip technology? It is a kind of hybrid manufacturing process that encapsulates monolithic and individual IC production?

  19. fasmicro
    January 13, 2014

    >> If you are above the line, it's a no brainer…do the ASIC…

    This is very fascinating. Do they have any available framework one can use. Like an Excel sheet one can edit for this. There is always this battle in companies to go all monolithic of the individual strategy but most times, it is based on gut-feelings and speculations. A more objective process like this graph can help. Is this graph editable and any source?

  20. Bob F.
    January 13, 2014

    FasMicro,

    The graphs were constructed from real integrations we've done. For example, one customer had a very simple requirement. Volume was about 100K/ yr. Total NRE and tooling costs associated with this development were $ 117K USD…that included everything… product design, mask fab, one wafer run, and test system development… to get to production. Another more complicated design from an automotive company had volume at 1,000,000 per year and development was $460K. We used dozens of examples to compile the curves.

    Generally speaking, the higher the cost of the existing “discrete” implementation, the higher the total NRE & tooling will be…Typical is $250K to $300K.

  21. Brad Albing
    January 13, 2014

    @Bob — thanks Bob – good background info for anyone considering a custom device.

  22. Victor Lorenzo
    January 14, 2014

    @fasmicro. There are several variants, including the ones used in 3D silicon arrays, but this technology basically consists in creating metallizations in the silicon connection points (pads) and making them “grow” (solder bumps) so they can serve as direct PCB connections.

    They call it “flip” because these pads are placed in the “top”, not in buttom substrate so it is necessary to flip the silicon die before placing it in the connection substrate (eg. PCB).

    In some implementations solder paste is used for soldering the chip to the PCB (chips with large silicon die and fairly large pads pitch); in some other cases an anisotropic conductive adhesive is used.

    I've seen uses of this technology in most HF RFID tags, many high end CPUs (desktop and servers), low cost toy controllers and also in low cost OLED displays that integrate the controller/framer/frame buffer IC in the same FPC cable.

    I have two OLED models that under the microscope (x10) one can see the bumped pads and the FPC tracks when looking at the back side of the FPC cable. This image below shows more or less one silicon die placed over the FPC cable. It is the rectangular piece of silicon laying over the flat cable which is painted green.

     

    This next image shows a curious combination, FPC cable, SMD components, flip-chip display controller and FPC-to-FPC soldering.

  23. goafrit2
    January 17, 2014

    >> Generally speaking, the higher the cost of the existing “discrete” implementation, the higher the total NRE & tooling will be…Typical is $250K to $300K.

    Amazing details. Thank you. This is the kind of analysis when put in engineering management journals get many citations. It is backed by hard data and not some wishy economists imagining theoritical things. Thanks

  24. goafrit2
    January 17, 2014

    >> They call it “flip” because these pads are placed in the “top”, not in buttom substrate so it is necessary to flip the silicon die before placing it in the connection substrate

    They used that in the latest Tegra processor which was unveiled by Nvidia. By having the bump on substrate, they reduce the heat that need to be applied to connect the PCB. Practically, you can use better lighter materials for the board. Great explanation

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.