Analog Integration Can Be Like Playing With Fire

All the recent talk on this site about analog/digital/mixed-signal integration has me thinking. In the hands of skilled, responsible practitioners, these powerful tools can produce great results. In the wrong hands, it's like playing with fire, and the results may be tragic.

If you have a high-enough-volume application for a specific function with a known sensor, known signals, known loads, etc., then integration makes sense. Yeah, the cost of a mask set (plus revisions — and with analog content, there will be revisions) is high, so it’s a big bet, but the cost per unit will definitely be lower. Each application has its crossover point in terms of volume versus non-recurring expenses.

On the other hand, if your application is not very high-volume (or you really don’t know, regardless of the marketing department’s really cool PowerPoint deck), or you’re in a startup weighing the alternatives of using the cash in the bank to make payroll this month or blowing it all on a mask set, perhaps the choice is not quite so clear.

The decision on digital integration is a lot more straightforward. If there is a programmable processor in the mix, many bugs can be patched in software — that’s sometimes hard to do if there is an “analog bug” or shortfall in accuracy. If you’re thinking about integrating an analog input, get very nervous if your designer (or your own internal voice) says something like: “I do 100,000-transistor chip designs all the time. Even though I’ve never done an analog design, I think I can knock out one of these sigma-delta converters. I’ve read a few papers, and think I get it. No big deal, couple of amplifiers, maybe a total of 100 more transistors… piece of cake.”

Recipe for disaster.

I recall an incident when the first samples of a custom power-management chip got to the customer, who found that the startup sequence got scrambled under certain unforeseen circumstances. There were two choices: an all-levels change to the chip (read: months and tens of kilobucks), or… wait for it… a patch to the system software. The customer’s engineering team developed a workaround in a few hours, and software solved an analog problem.

Incidentally, I notice that most of the discussion has revolved around integrating “classical analog” on a mostly-digital chip. Nobody has raised the issue of the “fringe” analog functions like RF, power management, and sensors (e.g., MEMS). But there’s a huge push to integrate that stuff, too.

I recall visiting a customer once to introduce a line of wireless chipsets. Our product for the then-current mainstream standard was single-chip, while the next generation, with much higher performance needed in the RF and mixed-signal sections, was three chips. The customer asked why we didn’t go directly to a single-chip for the new standard. I explained the advantages of partitioning the system into technologies that best fit each part of the system — digital processor in the best available fine-line CMOS logic process; mixed-signal in the most well characterized analog CMOS process; and RF in a BiCMOS process for lowest possible noise.

He thought a moment than asked how soon we could get it all onto one chip.

8 comments on “Analog Integration Can Be Like Playing With Fire

  1. Lee H Goldberg
    January 18, 2013

    You make some excellent points here – I appreciate them even though most of my design experience is in the digital side of embedded systems. You are indeed correct that the realities of the techical universe and the business universe are nearly orthogonal to each other at times and that the pressure from marketing types to integrate can have terrible consequences. 

    I spent many years covering the beginnnigs of the 802.11 standard (WiFi to all you latecomers) and the race to deliver products to support it. Back in the late 90's I watched several comapnies breathlessly announce their “single-chip solutions” for dual-band (2.4 & 5 GHz) radios but refuse to discuss the technical details of how they managed to get any kind of performance using the bulky, slow CMOS processes availiable to them at the time. In fact, they really were not able to and those products were quite terrible compared to the 2- and 3-chip solutions available at the time.

    But these one-chip wonders were cheap, and an entre generation of laptop adapters and base stations emerged with worse range and throughput than the previous genration. The crappy chips were so prevalnet that I advised my friends to hold off on buying anyWiFi gear for a year unless they could be assured that the chip set they used was not made by one of the three comapnies who integrated too soon and too much.

    I also suspect that the failure to meet customer expectations was a primary contributo to the demise of two out of those three companies.




  2. eafpres
    January 18, 2013

    I was selling antenna solutions for things like PCMCIA WiFi cards at about the same time, even a little earlier.  Everybody was doing reference designs and needed an antenna solution and none of them had ever actually done any RF.  It was a wild few years.  When the dual band stuff started appearing we would get “interesting” requests, such as could we reduce the bandwidth (yes, reduce) of the antenna in certain areas of the spectrum?  They wanted the antenna to be a filter becuase their front ends weren't right and stuff between bands would add to noise.  

  3. Lee H Goldberg
    January 18, 2013

    Ouch! Even though I'm somewhat “RF-challenged” I feel your pain. Your comments about letting digital types like me attmept to design Wi-Fi gear  finally explain why several early adapter cards from leading brands were prone to tricks such as being horribly uni-directional because they allowed the laptop to completely shield one side of the antenna. I also suspect that poor RF grounding schemes contributd to their wonky behavior.

    And your comment about “interesting” non-sequetur questions reminds me of a gullable marketing person I knew when i built spacecradt for GE. I was helping them out by leading a tour of our plant to some VIPs. We toured the design and subsystem assembly areas as well as the hgh bays where we assembeld the spacecraft and then proceeded tothe thermal-vac facility where we had a 65 foot vacuum chamber that simulated space conditions. After explaining how we used a mechanical “roughing” pump” to pull most of the air out of the chamber before we engaged the mollecular pumps which would take enough remaining molecules out to simulate a 350-mile orbit, I jokingly explained that we usually did not engage the third set of pumps because pumping he gravity out of the chamber used too much electricity.

    That got a good chuckle out of my guests and I didn't think anything more of it until a couple of months later when a buddy of mine who'd helped me run the tour told me that he'd been down at the thermal-vac chamber when the marketing person came through with another tour. Apparently, she took the joke I made seriously and was solemnly telling her guests how the test engineers “simulated space conditions by pumping all the gravity out of the chamber”.

    Whether it's reducing bandwidth in antennas or pumping gravity, giving a layperson a little technical information can be a dangerous practice.


  4. Via Monger
    January 21, 2013

    Very excellent points regarding analog/mixed-signal integration. No one should go off and design and fabrication a mixed-signal chip just for the fun of it. There are usually very compelling reasons why a designer considers mixed-signal ASIC integration:


    • Cost reduction
    • Size, Weight, Power (SWaP) improvement (sometimes SWAP+C)
    • Form-Fit-Function replacement of an obsolete part
    • Improved manufacturability
    • Enhanced reliability
    • IP protection (although not impossible it is harder to reverse engineer a chip compared to a parts on board design)

    Now, before you read any further I am the VP of Marketing for Triad Semiconductor and our goal is to make mixed-signal chip design available to everyone.

    How do you reduce fabrication cost of mixed-signal ICs?

    Full-custom IP developers at Triad assemble die with analog and digital IP. This IP includes: resistor arrays, capacitor arrays, switch arrays, op-amps (many styles), power transistor arrays, high-voltage tiles, band-gaps, current steering DAC sections, configurable I/O, digital tiles, ARM Cortex-M0 processors, SRAM, NVM memory,…

    The IP is arranged in tiles that are mixed-and-matched throughout the die. These resources are then overlaid with a patented global routing fabric and wafers contaning these die are partially processed and staged at the foundry awaiting a single mask layer change to customize or configure a user's design onto the via configurable array (VCA).

    Since only a single mask layer needs to be changed to configure a VCA the fabrication costs are a tiny fraction of traditional full-custom IC development costs.

    What do you do when the silicon doesn't match the simulation?

    Since via-configurable arrays (VCAs) are built from silicon-proven IP blocks there is a high degree of coorelation between VCA simulation models and actual silicon results.

    But seriously, this is mixed-signal design and if somebody ever tells you that you should expect first-time success with a mixed-signal design then calmly but quickly leave that PowerPoint presentation.

    Unlike a digital ASIC where you can often fix problems with metal-only changes and “spare logic”, a mixed-signal circuit is more-often-than-not going to require all layer changes. These all layer changes require the expense of a completely new mask set and you have to wait three months to get new parts back.

    Since VCAs are single mask configurable, the expense to make a fix or change to a mixed-signal VCA is minimized.

    VCAs require only a single mask fabrication step at the foundry meaning that new parts can be fabricated, packaged, tested and delivered to you in as little as four weeks. So, when a problem happens (and even if the chip is right the marketing guy will be asking for one more feature) you can fix it quickly and inexpensively.

    How can a non-IC designer design mixed signal chips?

    Historically, if you wanted to design your own mixed-signal IC you needed to be an expert in full-custom analog IC design and layout and you needed to own all the expensive tools associated with that kind of design. Our mission is to bring the FPGA business model to analog IC design.

    (bit of an aside here) Note: I didn't say that we wanted to make an analog FPGA because we don't think field-programmable analog circuits have the right mix of performance and affordablity that most of the market needs. That's why we focus of high-performance, cost-effective VIA CONFIGURABLE analog & mixed-signal.

    To bring the FPGA business model to configurable analog IC design we need to drive the Non-Recurring Engineering (NRE) costs of making your own VCA down to zero. So far, Triad has radically reduced the cost of developing a mixed-signal ASIC with our VCA technology.

    To reduce the outsourced NRE charges for IC development down to zero, you will need to do the VCA design yourself. But what if you are a systems designer and not an IC designer.

    ViaDesignerTM – Mixed-Signal IC Design Tools for Systems Designers

    We've developed a high-level mixed-signal design and simulation environment that combines: Schematics, VHDL-entry, Verilog-entry, SPICE simulation, digital simulation, VHDL-AMS modeling and simulation and a true mixed-signal simulation engine. To simplify the design process the ViaDesigner software comes with a library of mixed-signal design wizards. The wizards allow you to specify major block requirements and the wizard generates high-fidelity models and circuits for blocks including: op-amps, prog gain op-amps, fully differential op-amps, TIAs, high-voltage circuits, integrators, filters, sigma delta modulators, ADCs, DACs, power management blocks, voltage references, …


    One-chip fever can be fatal or it just might spread like wildfire.

    I know – none of this is possible. And yet, VCA technology is already being used in mission critical defense applications, protects workers from noxious gases, is utilized in FDA Class II medical devices, and is shipping in the millions for commercial applications.


  5. goafrit2
    January 22, 2013

    Sure, analog integration is challenging and could be difficult. But I do not see it as  playing with fire. One of the things people do not understand is that for a great product to be built, there needs to be a process. Process here is not the fabrication process but business process. I have designed, integrated MEMS devices which require both mechanical and electrical parts under different technologies and they worked fine. The secrete is seeing the big picture and getting guys to understand how the small pieces make real impacts in the final project.

  6. Brad Albing
    March 21, 2013

    Geeze – “Hey it's just a radio receiver, how hard can it be?” So, all you had to do then would be to figure out a way to make that monopole antenna look like a muliti-pole, multi-zero filter. “Hey it's just RF, how hard can it be?”

  7. eafpres
    March 21, 2013

    @Brad–If I had nickel for everytime somebody said “it's easy”!

  8. Brad Albing
    March 28, 2013

    We can just hand-wind some inductors, add a few shunt capacitors – just try some different configurations until we get one that works.

    That was my design technique when I was 14, but it's hardly suitable as a manufacturing strategy.

Leave a Reply

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