We are anxious to see the finished product of [Mark Omo’s] entry into our one square inch project. It is a 20 megasample per second oscilloscope that fits the form factor and includes a tiny OLED screen. We will confess that we started thinking if you could use these as replacements for panel meters or find some other excuse for it to exist. We finally realized, though, that it might not be very practical but it is undeniably cool.
There are some mockup PCB layouts, but the design appears feasible. A PIC32MZ provides the horsepower. [Mark] plans to use an interleaved mode in the chip’s converters to get 20 megasamples per second and a bandwidth of 10 MHz. It appears he’ll use DMA to drive the OLED. In addition to the OLED and the PIC, there’s a termination network and a variable gain stage and that’s about it.
The tiny board won’t allow for a standard BNC, so the input is an SMA. We’ve seen that before on other tiny scopes. From the PCB mockups, it looks like there will be some buttons along the edges, too, so expect a user interface on the diminutive screen.
This is the second square inch contest we’ve had and if you want to enter, the deadline’s been extended, but not for much longer. There’s real money on the table, including $500 for the winner. Test equipment seems to be a popular theme, although if last time is any predictor, there are plenty of other options.
Some high end audio amps use oscilloscopes. McIntosh, for one. They came in handy for tuning FM receivers or just checking stereo separation or phase. The rest of the time they’re just cool. Of course you don’t need 10 MHz of BW.
Also, on-chip logic analyzers and oscilloscopes are sometimes implemented as part of a self-testing chip design. They are quite tiny in comparison, but of course there’s a big difference between examining a signal on the same silicon die, and a traditional oscilloscope!
On-chip logic analyzer for self-testing – sounds like printf debugging for hw-makers…
More like built-in unit testing.
The printf of hardware design is the scan chain; unfortunately, you end up having to dump the contents of every single register on the chip, so it isn’t quite so efficient…
For the parts of random logic that are covered by scan, that’s all well and good. Memories tend to be surrounded by BIST-shell, some hard IPs do their own thing, and anything hooked up to the analog (especially high speed CML chains) is often hard to add any DFT for, and you’re lucky if you get some spot voltage measurements. Built-in functional test seems rarely sensible to me, but I’ve seen it done a few times. In my experience defect orientated methods (e.g. scan stuck-at, delay fault, IDDQ etc) are the bulk of it.
I recall a “quadraphonic” receiver in the 1970’s that had an oscilloscope type of display. Maybe it was just a light bulb behind a circular green disk. I only saw it in print ads, so I’m not sure. It may have been a Marantz.
GIYF,
Marantz 4400
Pioneer QX-747, SD-1100
SAE Full Digital Tuner (bonus, it has a Nixie readout!)
Fun but ancient fact regarding the Pioneer QX-747:
On vinyl, that sizzling cymble(?)that sort of panned and spread out,in Ted Nugent, “Stranglehold” would trigger the discrete quad decoder, IF the unit was set to the discrete quad position. I can’t recall the cartridge, but Still have the Ortofon 212 turntable (green light model).
I never knew if it was simply a distortion~artifact of the cartridge & album or if the album possibly had any material that feel into in the frequency range of the quad carrier signal.
The album was a pretty hot mix.
Considering that we had some vinyl that was mixed to carry 4 channels, I expect the coincidence could have occurred.
Still have the QX-747 packed away in storage.
Probably has that one dead channel, again, memory is a bit fuzzy now.
Don’t lose track of it, those suckers (working) are getting about $300!
OTOH, if you’re tired of having it take up space, I’ll dispose of it properly.
B^)
https://www.hifishark.com/search?q=Pioneer+QX-747
That reminds me, I have a RadioShack ProbeScope somewhere. Its display is about 1×2.5 cm, but it is longer as it is a hand held DMM. The scope function is quite limited, IIRC. Mine in particular has a loose connection somewhere inside, it needs to have the right “flex” to work.
http://eksfiles.net/2016/12/revisiting-the-radio-shack-probescope/
It’s not 20 megasamples/second, it’s 20 millisamples/second!
Optimistic if he thinks he’ll get 10MHz bandwidth out of 20Msamples/sec. That’s right at the edge of Nyquists limit and assumes a perfect low pass filter. It’s more likely to alias like crazy. The limit is more like 2 – 5MHz at that rate.
2MHz is still useful. It just needs to be made more clear that 20Msps != 10 MHz scope.
Since we don’t have the schematics I assumed he meant he would do the input stage with the 3dB point at 10 MHz. You are right, anything at 10 MHz isn’t going to look great, but you can design your analog bandwidth to be anything regardless of the sample rate. It doesn’t make sense to go over 1/2 but you can do it.
You could even subsample something higher frequency if you wanted; you just have design your input stage accordingly.
This reminds me of the Gabotrinics XProtolab. Slightly bigger and lower sample rate (2Msps) but commercially availabe for a reasonable price and they are made to fit on a breadboard, so you can easily stow them together with your project.
But why would the silly size limit of 645.16mm be of any importance?
645.16mm^2
or is that 645.16^2 mm…
Oops, typo. It’s 25.4^2 mm^2
This one?
http://www.gabotronics.com/development-boards/xmega-xprotolab.htm
Oh, and kit form for $25!
http://www.gabotronics.com/electronic-kits-modules/kit-to-build-xprotolab.htm
I’m still confused how one can claim 10 MHz of bandwidth at 20 MS/s? Even if you have a phase lock… Sample a 10 MHz sine wave at 20 MS/s and you get anything from a DC line to a rough square wave. When then SNR < 1, you can't really call it accurate.
He doesn’t claim that it’s accurate anywhere I can see.
20 Msps and 10 MHz of analog bandwidth will get you a pretty accurate reading of a 2MHz square wave. While the usable range is ~ 2MHz, you still need those analog 10MHz for that square wave to appear square. So it kind of adds up.
You can’t get a faithful representation of a single event if there are components in the input signal higher than the Nyquist limit, BUT, you can have a ‘scope that can measure signals at far higher frequencies by using creative triggering of the sample and hold circuit on repetitive signals. An instrument that does this is called a sampling scope, and there were oscillographs produced before there were CRTs that used this technique. https://hackaday.com/2018/09/11/the-pre-crt-oscilloscope/ I’ve also seen this done on automated test equipment to capture waveforms above the frequency range of the ADCs. Even in the absence of proper sampling techniques, a digital scope can gather useful information on signals well above its sampling rate, as long as the analog path has sufficient bandwidth. You can discern that a signal is present, and its approximate amplitude, for example.
One of my first jobs in a Calibration Lab was calibrating a sampling plug-in for a Tektronix oscilloscope.
I had no idea what a Sampling plug-in was supposed to do!
I think it was on my 3rd day that while adjusting the main knob the “eye” appeared on the screen!
WHOA! this thing can display a waveform higher than the ‘scopes bandwidth!
The 555 Piano has a good chance though
Do you have to actually build the project to win the prize money? If it’s just half-baked ideas, I could win millions.
presumably you have to have a 3d render of the half-baked idea as well.