We ran into [Paul Allen] at CES. He was showing off Sigzig, a super-low noise data logger which his company is just rolling out.
A couple of years ago he worked on a standalone chemical sensor and had a few extra boards sitting around after the project was done. As any resourceful hacker will do, he reached for them as the closest and easiest solution when needing to log data as a quick test. It wasn’t for quite some time that he went back to try out commercially available loggers and found a problem in doing so.
The performance of off-the-shelf data loggers wasn’t doing it for [Paul’s] team. They kept having issues with the noise level found in the samples. Since he had been patching into the chemical sensor PCBs and getting better results, the impetus for a new product appeared.
The flagship 24-bit 8-channel Sigzig samples 0-5v with less than 1uV of noise. A less expensive 4-channel differential unit offers 18-bit with 10-12 uV of noise. They are targeting $199 and $399 price points for the two units. We asked about the sample rate in the video below. The smaller version shown here captures up to 240 samples per second. The big guy has the hardware potential to sample 30,000 times per second but since the data is continuously streaming over USB that rate is currently limited to much less.
Update: It has been pointed out in the comments that USB may not be the choke point for sample rate.
30,000 24 bit samples per second doesn’t even come close to saturating USB1.0, never mind USB 2 or 3. They’re doing something nasty with the USB interface somewhere.
I suspect they have some sort of USB to serial converter, that is where the limit comes from.
Also, it takes quite a beefy microcontroller with usb to push a lot of data, even if the wire speed is 480Mbps.
Also, if you watch the video, he states that he is not a software guy and there is a software limitation on the protocols he is using.
Real FTDI chips handle 250Kilobaud so that would be the 30,000 samples per second if they are only 8 bit.
I talked to them about it. They should push an update out in the future that works in conjunction w/ their software to do high sampling rates.
You’re right, it’s not limited by USB it is limited by software. When working on the prototypes we had to make some decisions about communication with the computer. Native USB is faster but much harder for anyone else to write their own programs around. I talked with a lot of people who wanted to be able to connect up a number of sensors and then have their own custom UI for the setup. We provide our own software for monitoring all channels and general data collection but we are not in the business of custom software for everyone. So we decided to go with a virtual COM port and a UART interface. The downside is definitely speed, but on the upside it is super easy to grab the data or write your own UI with only a basic serial library. In the end these are our very first models, we already plan to add others in the future including one specifically for very high speed. Even so, temperature, humidity, do you really want 1k samples per second for weeks? Remember this came from the guts of long term chemical vapor sensing.
I once did a test where a host PC running Windows XP polled a homemade USB CDC device. The device responded by immediately returning a few bytes of sampled data. On receipt of this data the host immediately sent the next poll, so this proceeded in a loop, intended to test the limitations of sending many small packets bidirectionally over USB CDC. I expected the throughput to be low and limited, but not that the limit would be the host CPU on a decent machine; it pegged at 100% with only about 8K/sec transferred!
Without polling, the same USB device was capable of sending 300K/sec continuously, with only 15% CPU use on the host. Seems the Windows USB stack can be very inefficient in some cases. Just one of the many USB pitfalls the Sigzig folks might have run into.
I haven’t done the same test with newer versions of Windows, or Linux. Since I needed rapid polling in the project I was working towards, I abandoned USB in favor of Ethernet, which worked much better.
The ADC is most likely the other limiting factor for the sampling rate. Not likely you would get much faster speeds without losing usable bits.
In reality, loose piece of unshielded/single ended wiring and breadboard isn’t exactly what you should do for low noise measurements. Hope there is some ground isolation in that USB interface.
Does really a “weekendhacker” really want to spend 200$ on this? Maybe I am blind but I did not find any technical infomation besides voltage lebels and samples/s. Like what ADC is used etc…
I dont know, did you spend 300-1000 on 8bit oscilloscope by any chance? Do you need more precision measurements over long periods of time?
Can someone tell my why I want that device over a saleae logic analyzer?
Edit: I kinda like the idea of having the pinout on top, so my jumper wire ave smooth curves without bending.
I have one of the new Saleae Logic 8 data loggers. I really enjoy the added analog, great for seeing the actual waveform that generated the digital signal. It’s also very fast, however you capture a chunk of time and then review the data afterwords. The SiGZiG is a real time data logger used to continuously capture readings from sensors over hours, days, weeks. Also the Saleae is more about speed than low noise whereas the SiGZiG is the inverse. It’s another tool for different jobs.
By asking that question you pretty much declared you dont want this, or salea or any other device at all, as you dont know what it does in the first place despite clear description in the article. To sum it up its for tech stuffs, go back to the kitchen.
Hmm, actually I might be the one going back to the kitchen, just realized Salea makes analog loggers now, but still calls them ‘logic analyzers’ …
I’m having a difficult time following your logic: what does the kitchen have to do with knowing technical stuff? Did I miss something in the article?
Reading this on HaD… erm, is this open-sourced project? Or at least, is there hacked alternaive firmware that allows playing Zork on it?
My soldering iron isn’t open source either… :-(
Don’t worry, my soldering iron is closed source too, but I think of releasing schematics/source code for it. You know, personal projects with no documentation…
Looks like a pretty sweet tool for the box. The big guy’s price point is a little bit high, but 24 bit ADC, wow! If the sample rate was higher (like a lot higher), I’d consider buying it for seismic physical modeling experiments.
The price point for the 24 bit is high for me too. That’s why I made the 4 channel version. However even then I don’t set the price, I am just “The Engineer.”
Calm down guys. There are no doubt hackers who’d want to incorporate these devices into larger and more complex systems. And personally, I like to hear news from CES. Ideas that never leave someone’s garage shop do little to change the world. Quite a few ideas need to be commercialized and enter production.
i wonder how it compares to the labjacks
poor mans hacked version would be USB sound card capable of 24bit sampling and moddable to remove ac coupling
They are probably using ADS1298 or 1299 for the pro version which max sample rate is 32K and has 8 simultaneous 24 bit Delta Sigma ADCs… That is their limitation probably.