I always try to think so far out of the box that people tend to think I am crazy, or maybe I have just shocked myself way too much. However, do you think 30 years from now we will have any discrete components? Will we just be using small- and large-scale integration devices (ICs and modules)? What about the reactive components as part of noise suppression and in power applications?
Could you imagine opening up your application and not seeing any PC boards or maybe just a PCB with a couple chips? I mean, come on, if we believe the Disney cartoons from 50 years ago, we'll have cars and buildings that float off the ground. So in the next 50 years, why can't we bend the electronic limits and pack everything from the 120V input, through the supply rails, and on to the microprocessor into a nice small module?
Hmm… I guess we already do all this today. I mean, we do have wall warts with transformers and electrolytic caps, obviously we have micros and nice power modules, ASICs, etc… What if you could just place your entire circuit together just like a few Lego pieces? What if your schematic now is just a block diagram? Could you imagine buying your design from a catalogue, flipping through the power supply pages… Maybe I am just way too overboard on this one, but it's nice to think about the time you would save.
However, if it were that simple, that would take away from our world of competitiveness and innovation. Maybe electronics would become boring, albeit simplified. Maybe new ideas would lessen, and creative designs would diminish. Maybe it is just not possible, no matter where our knowledge takes us.
I just go back to the day where I was considering putting an entire analog sensor circuit into an ASIC. I think my circuit, including the small power supply, contained about 40 components. I had an ASIC supplier review the design, because I wanted this entire sensor circuit in a chip. I think the circuit could have been condensed just slightly. The ASIC would have brought my BOM count down from 40 to about 27 components.
What's the point in that? Obviously the metal polyfilm cap, the electrolytics, the diodes, and a few other large parts could not be incorporated into the ASIC and had to stay discrete. Just daydream for a second — thinking if all 40 of those components could have been jammed into a 1 inch by 1 inch block with maybe three connector ports: AC and sensor inputs, outputs for LEDs. I could mount that little block anywhere and not really have to worry about ESD or wire routing or managing a large BOM, purchasing individual components, etc.
I have come to realize that each application is different — even circuits in the same application one year to the next. Due to the thousands of different designs in the world, it may not be even close to possible to catalogue designs. Maybe our only miniscule chance would be to catalogue or modularize standard circuits based on the industry segment: medical, military, telecom, industrial, etc. Each industry would have design modules to choose from and piece together.
Maybe I should come back to reality based on our actual electronic component capabilities and not wonder too far out of the box. Although, maybe I am on to something, but just need to get there one small step at a time.
Let me know your thoughts.
Related posts:
- Medical ASICs: What’s Next in Silicon Integration?
- Where Is Analog Integration Going?
- Medical Device ASIC Integration: Optimize, Don’t Maximize
- Between Discrete & Integrated Circuits
- ASICs vs. Semi-Discrete Design
- 4 Misconceptions to Consider When Embarking on an Analog ASIC
- My ASIC+SoC Reassessment
Hi Jason,
My experience has been that two issues need to be addressed when contemplating an ASIC conversion:
(1) Willingness to look at an integrated approach that might be radically different than the discrete approach rather than simply moving the same design from Platform A (PCB) to Platform B (ASIC).
If one simply moves a design across platforms, the performance will invariably be worse or the solution will cost much more or both. To understand this, think about a 12-opamp discrete design.
It is cheap to buy 12 discrete op amps with offsets of 2mV or less. This is because the manufacturer's yield might be 0.98 per amplifier. But if that same 0.98 yield amplifier is placed 12 times on a single die, the yield will drop from 0.98 to 0.78. So either the ASIC price goes up to reflect the ASIC yield loss or the performance specification is lowered so the ASIC yield is back to 0.98.
(2) There has to be a financial incentive to take on the risk. This doesn't always mean a BOM cost reduction. It can also be a product footprint shrink when the discrete solution no longer fits in the next-generation space allocation. Think smartphone power circuits, for example.
I doubt that the industry will move to a total discrete-less generation of electronics for a long long time, but the market will certainly continue to shrink. A chief reason for this is the necessity to protect sensitive ICs from the ugliness of power delivery (AC line power, motors, etc.) using off-chip devices such as clamp diodes, transzorbs, etc.
Maybe in the future we will succeed in integrating many large parts, getting integrated metal polyfilm capacitances, electrolytics, diodes and so on … , having the same electrical perfomance of the discrete larger parts.
“Maybe electronics would become boring, albeit simplified. Maybe new ideas would lessen, and creative designs would diminish”
I don't think that Electronics will become boring, because there's no limit to human creativity and so the ways to combine the simple blocks will be really a wide number and the applications will be new , up to date and responding to the needs of the future generations.
” Maybe our only miniscule chance would be to catalogue or modularize standard circuits based on the industry segment: medical, military, telecom, industrial, etc. Each industry would have design modules to choose from and piece together.”
I wonder if in the near future the industry segments you cited will be separated one from each other. Just today the difference between two segments is very low, for example the medical and the telecom industries are dealing with devices to realize a cryptographic telecommunication, enhancing the effectiveness of the connection between two places.
I think that in the future the military segment will coincide with the telecom segment and so on.
I believe that thinking out of the box is really a good thing. Only a vision of the future roadmaps of the electronics and more in general , of technology , can give us the chance to realize what we need, satisfying our priorities as human beings. A good engineer has the responsability of think out of the box , because this is the only way to find out something that no one else have done before.
@etnapowers, I would agree with you on the creative things. There is no chance of new ideas would lessen and creative design to diminish. Because in fact if all the major electronics circuits will be available in the module form, it would be even easier to combine all the things together and create something new very fast. The scenario will be rapid prototyping.
This will even decrease the chance of silly errors that are committed by engineers in prototyping a concept. Because modules for specific circuits would have been already tested and prooven.
The question of optimization might arrive when designing final product from prototype. Because at the time of optimization of power you can not change the power consumption of perticular chip.
@Scott, I would completely agree on your first point that desingn constraints for the descreate design is much different than that of the chip design. So, there would definitely be some different approach for the same function.
The second point is also valid as more and more devices are being integrated inside the chip which makes the final footprint shrink day by day. So, this will continuue. But whatever technology we have today, has been optimised and integration in the chip level has been saturated. But to integrate inductors and electrolytic capacitors kind of things might take time. I won't be surprised if it needs completely different innovation in the chip manufacturing technology.
@Vishal: I'm happy for you agreeing my vision. As concerns optimization I can say that the small power consumption has to be achieved on each of the single blocks, then the overall system will be optimized.
“Just daydream for a second — thinking if all 40 of those components could have been jammed into a 1 inch by 1 inch block with maybe three connector ports: AC and sensor inputs, outputs for LEDs.”
@Jason,
It is the shortest way to get many of us out of business. Those who are making a living designing PCB assemblies for a larger product would have to think about changing their field.
Coming to the point, you are absolutely right. An integrated device would relieve us from those dreaded ESD and other tests. May be, as you said, we only will have a block instead of a circuit topology in future.
@jason, I completely agree with you. Now a days the total chip design is done by the tools ,there is little for human to interfere to solve small small errors ,as you said there is no chance to innovate for young engineers who are totally habituated to work with tools. But advantage is that these tools are reason for this rapid development and advancement in technology.
I don't think that Electronics will become boring, because there's no limit to human creativity
@etnapowers, I totally agree with you. I think we will find new way to innovate. I am sure designers will try finding new architectures to make existing systems more optimal.
Only a vision of the future roadmaps of the electronics and more in general , of technology , can give us the chance to realize what we need
@etnapowers, I agree with you. But the most challenging part is to visualize the future roadmaps. So many new technologies are evolving it would be hard to visualize the future roadmap. It would be interesting to see what new developments will take place in nanotechnology because many believe nanotechnology will eventually replace silicon transistorsand completely change the IC industry
Because in fact if all the major electronics circuits will be available in the module form, it would be even easier to combine all the things together and create something new very fast.
@Vishal, true combining modules to create something new will help us to create new products faster but sometimes such designs may not create optimal designs because we are forced to use existing system rather than create a new system which meets the spec defined.
But whatever technology we have today, has been optimised and integration in the chip level has been saturated.
@Vishal, true but lets not forget that we are still able to shrink transistor sizes which means we can optimize the existing modules for power and delay numbers.
@etnapowers, How much optimization can we achieve totally if optimize small blocks? I think now a days all standard circuits are optimized in area , power, speed .so even if we try for further optimization it will not effect that much.
@yalanand, the same concern I have also expressed in my comment also. If we take the example of power, it can not be optimised further in the final product, as the modules will have its own design and specs which we can not change.
Even the voltage level compatibility can also be a concern. Though it can be achived by some or the other way by using voltage translators or voltage shifters.
If we take the example of power, it can not be optimised further in the final product, as the modules will have its own design and specs which we can not change.
@Vishal, agreed. So if we want to build a system which has optimal power, delay and area numbers then its always better to implement the system in the same process.
Thank you yalanand, this is the only way for innovation, otherwise the engineers would not have a reason to exist. The human creativity is the key for the innovation, an engineer will be always necessary to create and innovate the architectures and the applications of electronics.
@yalanand, nanotechnology is a technology that holds terrific promises of development, I think that it will be successful only if it will be reliable and moreover , cheap. The reliability and the price of an electronic device are two key factors for a choise to buy it or not.
@yalanand, this is true but the optimization can be intended also as conversion of a block, performing a specified function, to a new technology. For example think to the conversion of a multiplexer module to the new technology photonics on silicon . You have many options to decrease area occupation, increase speed and so on.