Rick's b.log - 2020/03/29
It is the 21st of November 2024 You are 18.118.255.51, pleased to meet you!

DSO Shell Oscilloscope

I decided that today (windy, grey, cold) would be when I'd put it together.

The hard part was sorting out all of the components. Thankfully I didn't have to try to read the colour bands on the resistors, just get the multimeter to tell me.

Soldering was not so hard. The difficult part was soldering the big BNC socket with a 30W iron. It took a while to heat it up enough. Otherwise, it was mostly just resistors. A few capacitors (both types), some headers and switches. Nothing too complicated. Certainly nothing like the original JYE DSO that I built that was _complicated_.

All in all, I think it took about an hour and a half and two cups of tea.

But, wait... Why do I have another?

Simple. They're cheap, they're fun to make (and not too difficult), and at the end you have something useful.
Okay, granted, it's an oscilloscope using the ADC of an STM32 microcontroller (ARM Cortex M3), so all input needs to be scaled into something that fits into the 0-3.3V (centre point at +1.65V) range of the ADC and a 12 bit sampling rate of 0-200kHz with a maximum timebase of 10µS/div; to output the information on a 320×240 2.4" LCD. This is all to be expected given it is using a microcontroller to perform all of the functions of a traditional oscilloscope. Unlike many models you'll find on-line, this one comes with a smart case. Really, all it is missing is a LiPo cell inside.
It's certainly not perfect by any means, the small input range of the ADC means signals can be noisy, and this is actually a 1024 point storage scope (yes, it can work as trigger-and-hold) but the problem with this is that when using it as a normal scope, it's sampling is across the entire memory and not just what is visible on-screen. That might be worth remembering if it isn't behaving quite as you expect. Also the input stage doesn't use voltage regulators, just a 3.0V Zener diode for bias voltage, and resistors around an ICL7660 (voltage invertor) to drop the voltage to around ~4.5V. This means that the setup and behaviour of the analogue parts of the board will depend upon the input supply voltage. Clearly this isn't the best, but then given that this entire kit including post came to under €20, it's actually quite impressive what you get for the price. And for the purists, there ought to be places to patch in a VREG if necessary, such as a 78L05/79L05 so AV+/AV- can be a solid and stable 5V, with the bias voltage taken from this. That ought to tidy up the noise, but then you'd need to ask if it's worth the trouble - are you likely to be using the device in a context where such accuracy (from a microcontroller) is desired?
To put this into context, here is a photo from 2017 showing a direct comparison between what a CRT 'scope can resolve from a composite video signal, and what the STM32 device can make of it:

Your comments:

David Pilling, 30th March 2020, 15:23

Looks like fun. One thing you can do which a real scope cannot (assuming battery power) is get away from ground potential. Your Tube is full of videos of people connecting ground on their scope inputs to some high voltage with disastrous consequences.  From my playing with the STM32 max sample rate was 1MHz. Maybe next development will be to offer WiFi or blue tooth connectivity. But then again many software scopes which have come to grief on the software bit.  I'm always sat indoors with a scope, loads of 50Hz hum, wonder what the background signal looks like stood in the middle of a field - nothing is probably not that interesting. Definitely the place for electronic experiments or setting up HiFi (battery powered).

Rick, 30th March 2020, 18:53

Yes, the datasheet says it has a "1MSa/s" sample rate. Obviously it will need to sample multiple times to get a proper waveform...Nyquist.    I don't imagine either WiFi or Bluetooth will appear. The STM32 that is used doesn't support either, and I believe there's quite a bit of software support needed to implement either (which is why RISC OS is still without).    It can store a current waveform to EEPROM and recall it later. It is also able to transmit the waveform (I presume the visible one, not the saved one) as a series of textual messages to the serial port (J5 internally) at 115kbps 8N1.    Outdoors? Well, the CRT scope (mains powered) gave quite an interesting result when the probes were stuck into the ground. See https://heyrick.eu/blog/index.php?diary=20140902 for a partial attempt at photographing it.    I'd go out now and try, but I don't know where I put the earlier DSO Shell, and that's where the battery adaptor and proper probe are. Something to try later on...    Electronic experiments in the middle of a field. It'd probably get the neighbours talking. But, then, the first thing that comes to mind is TESLA COIL! (obligatory capital letters) :-)

David Pilling, 31st March 2020, 18:29

Yeah there is a cheap ebay kit for TESLA COIL. I built it. Interesting results. If you got away from all the domestic electrical noise, my guess is that the first thing you'd hear would be the noise from the scope.

Rick, 31st March 2020, 21:00

Oh, no no no. I wouldn't hook an oscilloscope to a Tesla coil. That would be reckless. If you're going to pump a billion volts into something, it should be something deserving, like a ZX81.

David Pilling, 1st April 2020, 15:38

Tesla is high voltage + high frequency. I'd not suggest putting it through anything. The interesting thing is that it discharges into empty air. The kits have an audio input and you can make out sound caused by the discharge. It is akin to those plasma globes you can get - small ones of those are only 10 quid on ebay, USB power, but again the core of them is high frequency + voltage. The standard circuit is "slayer".