These USB oscilloscopes and analyzers offer an astonishing amount of performance for little cost, and they're extremely portable.
Though traditional oscilloscopes and logic analyzers are large bench-consuming beasts we all hate to lug onto an airplane, a new generation of devices, some smaller than a pack of cigarettes (remember those?) offer amazing capability at reasonable prices. Last month I reviewed several small oscilloscopes; in this article I'll look at one unit that's half scope and half analyzer, and then examine a few others that are pure logic analyzers. Figure 1 shows the units I've evaluated.
Figure 1: Clockwise from top: BitScope, DigiView, Ant8, and USBee ZX
Though all were pretty intuitive to use, the units I'm covering this month all desperately need decent manuals and help files.
BitScope (www.bitscope.com) offers products ranging from $295 to $1,495. I looked at model 310, which sells for $495.
This instrument is a hermaphrodite, a mixed-signal oscilloscope (MSO) that's both 8-channel logic analyzer and two-input digital storage oscilloscope (DSO). MSOs shine in the embedded world where digital and analog signals interact in strange and mysterious ways. An MSO will display both domains on the same screen and can trigger on either a digital condition or at some analog voltage level. There's no better way to see how the logic and analog relate.
The BitScope comes with a real printed users' manual that, while lacking detail—actually an awful lot of detail—includes the unit's schematics. There's no help file.
BitScope's model 310 requires 12V from an external power supply, which isn't a wall wart but is the preferable in-line design with a cord that plugs into the wall. There's no power switch, unfortunately. The unit is housed in a solid aluminum powder-coated case, which feels very substantial for such a small (6 x 4 x 2 inches) device.
The specs are impressive. Analog bandwidth is 100MHz, though the A/D converter's is 75MHz. A note explicitly claims "[the] BS300 includes compensation for this convertor [sic] to achieve 100MHz bandwidth throughout." I'm not sure what that means. Max sample rate is 40Msps (million samples per second), a departure from usual scope specifications where the analog bandwidth is several times lower than the sample rate. But the BitScope uses a single flash A/D converter multiplexed between the two channels dumping data into a 128Ks (kilosample) buffer. Eight digital channels go into their own 128Ks buffer.
The A/D converter has 6.8 to 7.6 effective bits of resolution. At lower speeds the software averages data boosting resolution to either 9.5 bits (according to the manual) or 11.5 (from the web site). Max input voltage is 5.5V with a *1 probe. Get a decent *10 and you'll be able to probe most embedded systems without loading the signals unduly.
Use caution when comparing the BitScope's resolution to that of other units. I can't help but wonder if these conservative figures result from very careful characterization of the A/D converter and analog front end. The schematics suggest that the company put a lot of care into the analog design, borne out by the unit's performance. Linearity and distortion were nearly perfect even at high sweep rates.
Installation was trivial, and an uninstall program is included.
There's no way to save waveform data to a file yet, though I'm told that's coming in version 1.3 of the software, due out about when this article appears.
The unit's variable 0 to 50s trigger holdoff is a very useful feature that prevents early retriggering. And the delayed sweep feature is a boon for people searching for events that occur long after the trigger. When enabled, the sweep doesn't start till some user-entered time elapses after the trigger. In the olden days delayed sweep came with dual timebases and triggering systems; this seems less common today even on bench scopes, which is a pity. But no one would expect this level of capability on a $495 scope.
Triggering is quite powerful, and a zoom/pan mechanism makes it easy to slide around in the deep data buffer. I couldn't figure out how to center the trigger, though, so it appears there's no way to look at pre-trigger events. The trigger level is displayed in a small version of the scope window (upper left corner of Figure 2) and is hard for these 51-year-old eyes to see. I'd prefer a line drawn on the main screen, as on most conventional DSOs. There's no explicit external analog trigger, though it's easy to use one of the digital inputs as such.
Figure 2: BitScope's mixed display, showing an analog channel and the eight digital inputs
The GUI is moderately pretty, though not quite as intuitive or as simple as described for the scopes evaluated last month. Bench-type knobs are replaced with typical Windows controls, which work well but break the oscilloscope paradigm. Coupled with the poor manual, it's sometimes hard to figure out how to use the deep set of features embedded in the hardware.
Screen updates are very, very fast, running up to 50 frames/second. But then occasionally unpredictable and annoying delays hang things up. Windows' system monitor shows the code isn't burning CPU time so perhaps there's some communications glitch.
Changing settings sometimes maddeningly resets the vertical gain and time base settings to their default values. Anything that increases the engineer's workload is a bad idea. All in all, the software has the feel of being not quite done yet.
In analog alternate mode the maximum sweep rate is 10ns/division, which falls to 50ns in chop. When displaying mixed signals (digital and analog at the same time) the maximum is 25ns/division.
I was a little disappointed to find the BitScope can only show one of its two analog channels when the digital signals are on the screen as shown in Figure 2.
A very nice spectrum-analyzer display shown in Figure 3 looks almost too uncluttered at first; there are no labels on the axis. Yet moving the vertical cursor displays mdB clearly.
Figure 3: BitScope's spectrum analyzer display
Interestingly, the digital inputs go to a replaceable pod, which is a small circuit board you can position near the system you're probing. The pod ensures the data that's sent to the BitScope doesn't suffer from crosstalk or other signal degradation. As supplied the pod samples 5V signals; a different pod will handle 3V logic.
An arbitrary-waveform generator can turn analog channel B into an output for sine, square, or ramps. Below 1MHz the signal integrity is reasonable; above that frequency the distortion is severe.
There are no analog probes, not surprising for such an inexpensive unit. The digital clips are of medium quality; you'll never get them onto an SMT pin, though with a bit of cunning I managed to connect to through-hole IC leads.
BitScope model 310 is a well-made hardy unit. The analog design received a lot of attention. It's almost instrument-quality. I'd really like to see pretrigger data, and the occasional data drop-outs were annoying, but if the new software fixes most of those problems I'd love to have this unit in my toolbox.
Techtools' (http://tech-tools.com) $499 DigiView logic analyzer samples 18 channels into a 128Ks buffer. It's a palm-sized unit in a simple but rugged plastic box that will slip into your shirt pocket. Power comes from the USB connection, so don't run this from an unpowered hub. The company claims it will work with 2V to 5V logic, though I didn't test this.
Software installation was trivial, though no uninstall was included. Help is quite poor; you'll have to fiddle with the thing to figure it out. That's unfortunate as it's a quite innovative design that could benefit from a bit more descriptive information.
Figure 4: DigiView's main display
The DigiView doesn't have a sweep control of any sort, giving an uncluttered display (shown in Figure 4) that accentuates the acquired data. The unit always sucks in data at 100MHz, which would swamp the buffer in a mere millisecond in a conventional unit. But hardware compresses the data in real time so that 128Ks buffer goes a lot further. My test setup generated eight channels of staggered 10s pulses. The DigiView managed to acquire 79 million samples of this, or 0.8 of a second. That's an awful lot of data, but of course your mileage will vary depending on the input signals. With the fixed 10ns sampling rate, there's no external clock.
The device took 1.5 seconds to upload the compressed buffer. One of the DigiView's strengths is its extensive search features, needed to manage such a deep virtual buffer.
The supplied clip leads are of good quality and are fine for working with large-pitch SMT devices. But the colors of the wires are off—channel 0 is brown, instead of black, as is standard on most instruments, since black corresponds to zero in the resistor color code.
Figure 5: DigiView configuration screen
Figure 5 shows the unit's configuration menu. In this example I've enabled only eight of the possible 18 channels and have configured them as individual signals as well as a bus. The screen shot in Figure 4 shows data acquired as both the bus and as separate signals.
Although the GUI will save data in standard CSV comma-delimited text files, such as used by many spreadsheet programs, it won't snag a graphical screen shot. Me, I like the screen shots to illustrate technical documents though plenty of third-party programs will capture Windows screen images.
Overall I'd call the DigiView an interesting, well-built unit with good mechanical engineering (box, wires, clips). It sucks in an awful lot of data both in width (18 channels) and depth. The GUI is pretty easy to navigate.
USB Instruments' (www.usb-instruments.com) ANT8 logic analyzer is the eight-channel baby brother to the company's ANT16 word-sized unit. It's sold in the USA by Saelig (www.saelig.com) for a mere $222.
Taking a cue from Apple the device has a see-through case that doesn't seem to quite fit correctly around the USB cable. This unit will fit inside your pocket protector and is powered from the USB host. I peered through the plastic and found a half-dozen ICs, one of which is a Xilinx Spartan FPGA. Presumably many of these analyzers have new FPGA equations downloaded to configure triggers, which is pretty cool engineering.
The datasheet seems to suggest that the device is tailored for 5V logic.
I was never able to successfully install the software on my desktop PC despite numerous attempts and reboots, though it worked flawlessly on my laptop. The documentation is for a Windows 98 environment, though was easy enough to adapt the instructions to XP.
Where some of the other units have deep buffers, the ANT8's buffer is a fairly tiny 3,072 samples, which uploads instantaneously to the PC. The unit's upside is speed: it will acquire at up to 500MHz, a breathtaking rate for such a tiny and inexpensive device.
If there's a provision for an external clock, I couldn't find it.
Figure 6: ANT8 main display
Figure 7: ANT8 triggering menu
The exceedingly easy-to-use user interface (shown in Figure 6) was very intuitive and almost made up for the woeful help file. You can save data (in CSV or a proprietary ASCII format) but not screen images.
Triggering is astonishingly versatile. It's simple to set basic triggers on the main screen, but a tab pops up the advanced triggering screen (shown in Figure 7) that can look for durations, number of events (with a 10-bit counter), and interactions between two patterns. This screen, too, is well designed and immediately updates to gray out choices no longer valid as you configure the trigger.
The clip leads simply couldn't be worse. Sure, this is a very inexpensive device and decent clips are expensive. A few inches of cheap ribbon cable sprout from the logic analyzer and are terminated by a cauliflower-head of clips that appear to come from the local Radio Shack. Happily they'll surely break with just a little use, giving you the chance to build a better cable with real clips. Whereas the DigiView had colors that were shifted by one from the resistor code, on the ANT8 they're moved two positions, so red is bit 0.
With all of the leads in a single ribbon cable, and no grounds between wires, I worry that at high speeds crosstalk could possibly corrupt the signals enough to cause incorrect displays or even effect the operation of the system being tested.
My gut feel: the unit looks like a toy, sucks in data like an Oreck vacuum, has a buffer as shallow as a tidal pool, and is as easy to use as an iPod. None of the other units came close in triggering capability. If you replaced the confounded probes it would make a nice travel tool.
USBee's (www.usbee.com) USBee ZX is the high-end of a family of eight-channel logic analyzers ranging in price from $295 to $895. It's hard to compare this unit to the others as it has some philosophical differences, starting with the two 15-minute videos that step you through the device's operation. There's no help file in the conventional sense; help only—and annoyingly—calls up snippets of the video you've already seen. Some of the features are, near as I could find, undocumented, though most are pretty intuitive.
Though it's usually powered through the USB cable, provision has been made to attach an external 3.3V or 5V supply. The documentation doesn't state logic levels, but it's reasonable to assume that when powered by 3.3V it will correctly handle those signal levels. It does require USB 2.0 and will not work with USB1.1.
The unit will acquire at rates to 24Msps and stores, near as I could understand, a million samples in the on-board buffer. But it fires data to the PC at USB 2.0's blistering rates so can stack samples on the PC in real time. My fast PC gave a virtual buffer of almost 500Ms (million of samples). And this all happens seamlessly with no noticeable delays.
Figure 8: USBee screen
The user interface, shown in Figure 8 is simple and reasonably easy to use. On the left side, the four trigger levels offer plenty of capability but aren't nearly as extensive as, say, those on the ANT8. In the picture note that two of channel 7's conditions are combined to trigger on a high-going edge.
This is the only unit reviewed with a dedicated external trigger input, and it has an external clock as well. The supplied clips were the best of all of the analyzers.
The USBee isn't really a logic analyzer per se; it's more a collection of tools. There's a 75KHz frequency counter, a 1MHz frequency generator, an I2C controller, and much more. The signal generator is particularly interesting as you can save waveforms captured by the logic analyzer, edit them, and then replay the data out the clip leads. A ToolBuilder suite lets the user create new tools altogether, which is a pretty cool idea.
The USBee has a simple user interface, makes it easy to configure triggers, is ultra-portable, and most importantly offers an interesting degree of extensibility.
Useful mini tools for EEs
The old-timer in me still yearns for knobs, whether a scope's time base control or a sextant's knurled micrometer. There's something ineffably satisfying about reaching for a physical control that virtual instruments can't beat. When you're buried in the bowels of a big PCB it's easier to twist a knob than mouse around a screen.
But these USB instruments offer an astonishing amount of performance for little cost. Some will satisfy the lab requirements for many real embedded systems development efforts. And for traveling I'd grab one of these in a heartbeat rather than a conventional 20-pound all-in-one scope or logic analyzer.
Which one is best? That's impossible to say. I'd sure like the Picoscope's 12-bit A/D combined with the BitScope's MSO, the DigiView's 18 channels, and the ANT8's triggering. For training purposes it's hard to beat Parallax's $129 scope or ANT8's $222 analyzer.
Jack G. Ganssle is a lecturer and consultant on embedded development issues. He conducts seminars on embedded systems and helps companies with their embedded challenges. Contact him at firstname.lastname@example.org.
Another great article! I SOLD my high-quality oscilloscope after designing a very, very complex 5-processor board with the ANT-16. I figured, why do I need 40lb of beast with two-channel capability when I just want a 1-bit A/D that can show me all my control lines, the bus states, etc.? I shaped each of my strobes so that I had enough overlap for everything to work, and it actually DID work--all thanks to that ANT-16.
You are right about the probes being short--you have to rip apart the ANT-16's ridiculously short probes to connect any two points. But, because they are so short, EMI is not an issue, and you can always arrange for uncorrelated signals to be in proximity so that crosstalk is minimized.
The software is fine for the ANT series and the whole thing just works. I can't say the other units are bad, but the ANT's ultra high speed means that you get very fine resolution of the pulses for very little money. It's just amazing how well this little ANT can help you solve embedded control problems.
Again, thanks for your article, keep up the good work!
Knapp Engineering Services
I have greatly enjoyed the two articles on the USB scopes/analyzers. This is obviously an area that is developing, but has great potential for education as well as "low budget" embedded systems design. I think there is a lot of promise with products like the BitScope since it's an "open" design.
I have greatly enjoyed the two articles on the USB scopes/analyzers. This is obviously an area that is developing, but has great potential for education as well as "low budget" embedded design. I think there is a lot of promise with products like the BitScope since it's an "open" design.
- Mike Braden
Interesting article. I just wanted to mention a product which you apparently didn't find. The LogicPort logic analyzer from Intronix has 34 channels, 500MHz sample rate, +6v to -6v adjustable threshold, and is USB powered. It has many of the features you found lacking in some of the reviewed products. Sells for $379.00. http://www.pcTestInstruments.com
- Harrison Young