The Analog Swiss Army Knife

11300

While FPGAs get all the credit for being the hip new thing, they are inherently digital devices. Without a proper ADC and DAC, you won’t be delving into the analog domain with your programmable logic. Maxim has just put out a chip that does just that: an analog swiss army knife with 20 pins that are configurable as analog to digital converter, digital to analog converters, GPIO, or any mix of the above.

The MAX11300 includes twenty IO ports, each capable of becoming an ADC, DAC, or GPIO, with pairs of ports capable of being configured as a logic level translator or an analog switch. The ADCs and DACs are 12-bit, with input and output ranges from -10V to +10V.

As a nice little bonus, the chip is controlled over SPI, making this an interesting device for a small “do anything analog” tool we’re sure will hit Tindie or Seeed Studio before the year is out. Luckily for whoever would create such a device, Maxim has a nice GUI for configuring each of the 20 pins on their chip, Of course Maxim already offers an evaluation kit for the MAX11300. It’s $100 USD and is Windows only.

The MAX11300 is available in either 40-pin TQFN or 48-pin TQFP packages (with the larger, easier to solder TQFP shipping later) for about $5.80 USD in quantity 1000, or $11.37 in quantity one.Video below showing off the MAX11300 reading and writing analog values to a few pins, and a good look at the configuration software.

33 thoughts on “The Analog Swiss Army Knife

  1. At first glance it looks like an Atmel XMEGAE5 with out the core. Seriously, it’s got an many pins as the XMEGA, and no more ADC/DAC resolution. Tell me again why I need this if I still need a uC to control it. Why not just choose a uC with most of thise analog functions built in? Maybe I missed something.

    1. In some cases, on-board ADCs and DACs are a better option.

      But having them on a separate IC allows you to delegate priority. You don’t need to operate as much in interrupt contexts on your main microcontroller.

      Whether you should use a chip like this is a subtle choice, and very much application dependent.

  2. I’m sorry, but I can’t think of any product that would benefit from this. I really like Maxim, but what is the purpose of this thing? And for the price, they could have included an ARM core in there as well… PSOC?

    1. The advantage I can see is 20 DAC’s in addition to the 20 ADC, and also the -10 to 0, 0 to +10, or -5 to +5 voltage swing (driving 25mA) directly from the DAC (if I got it right).

      Many PSOC’s are a bit low on resources on the DAC end, however since they support I2S with a pre-built module (http://www.cypress.com/?rID=46464) you can easily add more DACS if you need.

  3. Really not sure what the market this is aiming for. You very rarely have to change the pin configurations in a product. Unless the input/output level scaling is *exactly* as the Maxim chip, one would need some external signal conditioning (scaling/level shifting with amplifier or anti-alias filters etc) which tends to required fixed function and direction.

    The flexibility is only useful when the part is dirt cheap so you can make variations of your product line or stock one part and use it for different things with different programming.

    For cost sensitive high volume use, ARM chips with built-in ADC/DAC are plenty, cheap and you have to pair this chip with a micro anyways. With ARM chips moving into the main stream, you got to wonder if this chip is needed or how it fits.

    1. “You very rarely have to change the pin configurations in a product. ”

      Why don’t you go visit the linux web site for motherboard hardware sensors, read about how these chips are used in hundreds and hundreds of different motherboards. Are you going to tell me you can use the same pin configuration for different motherboards?

      “Unless the input/output level scaling is *exactly* as the Maxim chip, one would need some external signal conditioning (scaling/level shifting with amplifier or anti-alias filters etc) which tends to required fixed function and direction. ”

      Yeah that’s why this Maxim chip has PROGRAMMABLE VOLTAGE RANGES and SAMPLE AVERAGING.

      “you have to pair this chip with a micro anyways.” Well I guess you could call a Xeon server processor a “micro” if you want to.

      “you got to wonder if this chip is needed or how it fits.”

      WOW JUST IMAGINE having a single linux kernel driver that can work with many different motherboards.

      1. I don’t really see why this would ever appear on a motherboard. There are dedicated, low-cost, single-purpose sensors to measure voltage and temperature (the only things I know of that motherboard sensors do), and SMBus (i2C) is basically the universal standard for interconnecting that stuff. Motherboard designers simply pick and choose the type and quantity of each sensor they need, toss them on the i2C bus, then wire the address pins in the PCB design to be able to talk to them individually.

        This Maxim chip is extremely expensive (5-10 times the cost of single-purpose sensors), and uses SPI. SPI is much faster than i2C, but if you compare SPI-based Linux drivers to i2C-based drivers, you’ll quickly realize that SPI is much more convoluted to configure. This chip may have some usefulness in PLC industrial control systems, but it certainly doesn’t belong on a motherboard.

    2. First off these chips are about 10X more expensive the those “PC health” monitoring chips. So they are not used in a price sensitive PC motherboard. I am very familiar with those. Those chips runs off a slow SM Bus at 100kbps and not SPI. They are not dynamically reconfiguring pin out with an analog switch for the ADC/DAC. You don’t need that if it sits on a PC motherboard with predetermined pinouts of all the voltages.

      Like I said it is fine and dandy unless you live in the real world and have to deal with funny sensors that is outside the range of Maxim part. Either much lower FS range or higher or require additonal off chip signal conditioning, then all of this fancy stuff are useless. e.g. raw signals from RTD, load cells, thermal couple etc. May be if Maxim throw in a good PGA and not just mux select voltage dividers. ADC/DAC means that you would want anti-alias filter, buffering and protection if you are driving real world stuff.

      Not sure why they want to big crosspoint mux. Just put the blocks inside a TQFP100 and let me use all the I/O.

      TQFP is certainly not for the typical HaD readers here. Add the 0.5mm pitch (0.020″) and exposed thermal pad under the chip. With reflow, it is doable.

      1. Totally agree. A *true* analog “Swiss Army knife” would contain PGAs, VCAs, RMS detectors, switched-capacitor filters, and window comparators. This is nothing more than a modern Cortex-M4 microcontroller without the core, sold at three times the price. Totally unimpressed.

  4. This product appears to be aimed squarely at the industrial IO market. You can get configurable digital in and out IO cards and machine mount hardware, but analog is generally fixed. This hardware should help to make the analog IO configurable on the fly, too.

    1. Exactly what I was thinking .. programmable, universal I/O is incredibly handy.

      For most industrial applications, 12 bits is plenty .. And the DACs are a nice addition (all too rare in mcus).

    2. I agree… The industrial I/O market is where this device would shine. My concern is product availability. One too many times I designed in a Maxim part, only to get knackered with silly lead times and/or sporadic availability… and so I typically do not design in a specific (i.e. single source) Maxim part.

  5. I guess I don’t see this being used in any final designs, but as a part for the hobbyist prototyping crowd? Imagine this as a shield for any of the popular microcontrollers, it would be an amazing asset to have in your toolkit.

    1. the look-and-feel seems to be pretty much what LabView users would expect. LabView and associated data acquisition hardware is not unpopular but kind of on the expensive side. I can see open source hardware modules with this chip become handy in the lab / research sector.

  6. I was really hoping this chip would have a configurable analog signal processing core, something like what Anadigm offers. As mentioned above, most mid level MCU’s come with 12 bit D-A and A-D and they’re far more cost competitive. Frequently though, its the processing and filtering that can really eat up resources and be tedious to code.

    1. I also like to hope that products would come with features that are not necessary in the product’s target market. Yes indeed it is necessary to perform sophisticated signal analysis of my computer’s power supply voltages.

  7. Well, I am now offended. Not by Maxim, but all the users talking about 12-bit ADCs and DACs without once mentioning SAMPLE RATE…
    Had to go to theweb page, tech specs, to find: 400K sample /sec via SPI. Not even parallel data I/O.

Leave a Reply to @Cortez PittmanCancel reply

Please be kind and respectful to help make the comments section excellent. (Comment Policy)

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