"Universal" Data logger

geo_leeman

I often run into the issue of needing a "high sample rate" (1-1000Hz) high resolution (16-bit or better) data logger that is low power, has SD storage, and sometimes needs GPS timing. This is generally for environmental monitoring sensors and applications. I've been thinking about making a "universal" data logger to avoid rebuilding the same functionality into many of my projects and wanted to get some feedback from the forum about features you think are useful and/or features you think should be canned.

Hardware Feature List:
* 4 channel 16-bit ADC (is switchable input filtering attractive?)
* RTC w/ backup supercap
* GPS module w/ external antenna connector, exposed PPS
* SD Card for data storage (and maybe flash with a download feature for smaller data storage?)
* Serial data in, exposed I2C, and GPIO pins
* Possibly on-board switching regulator so you can provide 3.3 or ~12 VDC?

Firmware:
* Serial configuration of the logger
* Serial data transfer

I'm thinking that with the GPS, etc. off (actually unpowered, not just idle) this would be very low power, but with everything in use it could be a very capable data logger.

Peter Jakacki

I'm running one straight from a LiPo through an LDO with nA shutdown current. The RTC or auxiliary "PIC" chip wakes up the system as necessary but in another mode I have it gate the MEMs oscillator, effectively stopping the clock without it having to reboot. I normally use a very small supercap for the RTC as I only need to keep it running for days at the most since it is almost always being powered and the LiPo feeds the RTC as well. Any RTC that gets rundown is obviously not in use and so setting the clock again is not a problem. But where possible I try to implement digital sensors so there is no actual need for a central A/D. Such is the case with all kinds of sensors these days anyway and the odd voltage input can have its own special little module to suit the voltage or current and in these cases I just use a small 8-pin TSSOP MCU for A/D and asynch serial out, just 1 wire.

Since the GPS "antenna" can be quite large I find it just as easy to place the GPS module with integral antenna into its own little case as the "antenna" external to the logger and and that way I don't need coax, just power and serial.

As for configuration I just talk to Tachyon with simple text strings via Bluetooth and I can even reprogram it the same way.

geo_leeman

I end up interfacing with a lot of strain gauge bridges and other analog output sensors (high precision differential pressure transducers, etc). I'm thinking of having many footprints on the board and only populating what was needed for a given project. A 70x70 mm 2 layer should fit most things?

Tracy Allen

George, I'm with you on that, I empathize with the notion or dream of a "universal" data logger for environmental science. Think companies like Campbell or Onset, many more. The problem as I see it is the ubiquitous outlier. As soon as you get it done, the next job is something else, one that needs a glue interface for the signal, the power supply, communications, storage, galvanic isolation, acquisition timing, packaging, or whatever. And cost. That is sort of how I make my living, the special interfaces. It can be an interface or a standalone logger using the Prop. I think of it more as a platform than as a logger per se. In the end when cost is involved it often comes down more to a template and a specialized build.

geo_leeman

I agree, there is always that "one more thing" to add to a "universal" tool. I was thinking of a small prototyping area maybe? I'm hoping that this would reduce the cost of developing hardware for scientific applications once I have it made - the cost of doing the initial design is amortized over many projects (hopefully). It's starting to look like a dev board with a lot of extra peripherals.

Peter Jakacki

Rather than a small prototyping area I find it is better to allow for a small prototyping module to be plugged in. That way not only can you replace the circuit, you can also come along later and plug in a proper pcb. Usually the prototype module is nothing more than a plated through matrix board cut to size and fitted with sockets as I prefer to use pins on the main board. I have my own 46x20mm "Puppy" format I use with a 8x2 connector on one end from the Prop including I2C and then the other end is the external interface. I could never understand why Parallax don't do something similar as it makes the board far more reusable than one which has a ratsnest modified many times over.

Tracy Allen

For prototyping I favor a "universal" core module that connects via a header to what is usually a larger breakout board. The breakout at the very least has the outer level of i/o pin protection and power supplies for external devices, as well as the off-board connections, which are often Phoenix screw-down terminals. I may have custom interfaces hanging off of that. But if a contract requires more than one or two copies, then it is going to get a custom breakout board. Isn't OshPark great! As it needs more copies, it always becomes more cost conscious, needs to shed the fat, and the universal core and the breakout have to merge.

I like small simple projects better than big complex ones. Who wouldn't?! The effort required grows O(n^2) with the number of parts and functions and O(n^3) with generality.

geo_leeman

Agreed that smaller and simpler is always better. I'd think a minimum compliment on the "core module" would be the prop, RTC, SD card, and input protection?

Cluso99

geo_leeman wrote: »

Agreed that smaller and simpler is always better. I'd think a minimum compliment on the "core module" would be the prop, RTC, SD card, and input protection?

That is pretty much what my P8XBlade2 does. Just missing RTC, which BTW can also be done by the Prop but using more power.

Tracy Allen

I include an ADC, 3V power with wide input range, low dropout, and low quiescent current, and a couple of power supply switches on the board, one for an external SD card and one for external sensors. The issue with things like SD cards and USB interfaces is that packaging is often an important pre-consideration for a data logger. User access to those parts needs to be flexible so that slots can be cut or the parts can be better oriented in relation to a panel. On the other hand, when it has to be small and cost is a major issue, then a single board is better. But at that point, it is no longer general purpose. Another thing is how it gets connected to external sensors. I like to use green Phoenix terminals, but in some instances there has to be some kind of header or direct solder connection to a panel. There may need to be an additional level of i/o protection on a breakout board. Another thing about the power supply, it may be wise to use a voltage regulator that can handle reverse connections and load dumps.

The RTC has a heartbeat output and ample battery backed RAM. The one second heartbeat is pretty much essential for a data logger. Another necessity on any battery powered data logger is means to check the battery voltage, to indicate its condition with an led or as a check on data integrity. It is true that ADCs are less important these days, but certainly not gone. They have a lot of advantages over smart sensors for field instrumentation, because they give a check point that a farm hand with a voltmeter can check to help with remote troubleshooting. So many problems with field data have to do with external cabling or conditions.

geo_leeman

Tracy - I completely agree about analog instruments. It's also nice to have more control over the sampling/filtering that way. Have you had much experience using GPS time bases on your equipment?

Tracy Allen

About the ADC, there are so many different kinds, ask, "for what purpose?". You've mentioned in your other threads the magnetometer and a requirement for relatively high speed acquisition and analog filtering. That narrows things down to be sure. What are you using? I'm usually concerned more with DC accuracy and resolution, things like thermocouples, so that narrows things down in a different direction. I include ADS1115, 16 bit with PGA. I've also done higher speed things, using sigma delta POSDET mode, but the PCB has to be built with that in mind. Generality applies only within preset boundaries.

I've included GPS on loggers but more as a time and place reference than as a time base per se.

geo_leeman

Well, for the "general" (again, the problem with a universal style of logger I suppose) instrumentation applications, the ADS1115 would be a good choice. I've used it on several projects in the past. I do have a few special projects from time to time that require 24-bit ADCs or really high sample rates, but 1-10Hz at 16-bits is generally a good "universal" bet for geoscience problems.