Open Source Oscilloscope, can it be done?

[ogden]

There's brute force and there's brute force.  Don't underestimate the impact of known plaintext.

Firmware that may have large blocks of 0's or 1's or any other known plaintext shall be (and usually are) encrypted using AES-CTR (counter mode) which is safe against known plaintext brute force attack.

[jebivetar]

Someone out there does seem to be working on an open source scope.

The (wonderful) Interactive HTML BOM Kicad plug-in https://github.com/openscopeproject/InteractiveHtmlBom has a couple of interactive demos "of (soon to be published) in house designs" one of which OSPx201 which looks like a pretty-far-along oscilloscope.

https://openscopeproject.org/InteractiveHtmlBomDemo/OSPx201/ibom.html

A two-sided <10x10cm open source oscilloscope (cheaply fabbed) with a ~$100 BOM capable of 10Mhz or so would be revolutionary. 

[ogden]

Someone out there does seem to be working on an open source scope.

https://openscopeproject.org/InteractiveHtmlBomDemo/OSPx201/ibom.html
A two-sided <10x10cm open source oscilloscope (cheaply fabbed) with a ~$100 BOM capable of 10Mhz or so would be revolutionary.

I struggle to find ADC on board. If built-in 5Msps ADC's of stm32f303 are used, then that thing does not seem very appealing, especially knowing that ADC's of stm32 shall be run from PLL which have inherent jitter. I found 12MHz ceramic resonator for stm32 instead [facepalm].

[2N3055]

LOL jebivetar....

[bloguetronica]

Hi,

I found this open source scope design that looks serious: https://www.scopefun.com/. Although it is PC based, it seems to have a decent analog front-end and ADC. The layout is very well designed and the hardware very professionally made. This is way above the average DIY kit, IMHO.

They will launch a second version, but I don't know how soon. Kudos for who developed this amazing project that, IMO, should be in the spotlight.

Kind regards, Samuel Lourenço

[janoc]

A two-sided <10x10cm open source oscilloscope (cheaply fabbed) with a ~$100 BOM capable of 10Mhz or so would be revolutionary.

What would be revolutionary on it, exactly?

E.g. this Owon is more capable and for less than your $100 BOM.
https://www.sainsmart.com/products/owon-vds1022-25mhz-2-1-channel-pc-usb-oscilloscope

Or get one of the Bitscopes, those have open source software support, I believe:
http://my.bitscope.com/store/

Or just get a used analog scope which will likely be still much more capable than anything you can build for $100. And those tend to be as "open source" as it gets, especially if you get a classic scope with no fancy digital stuff.

It is not two sides board and certainly not 10x10cm, but that is a totally unrealistic requirement for a 10MHz scope. Keep in mind that you need at least ~100Msps to actually see a 10MHz square wave as a square wave - that's not going to work on a tiny two sided board. Four layers would be a minimum.

Open source is cool, but seriously, what do you exactly want to hack on a device like this? No matter how clever software you design, you will be always limited by the necessarily crappy/low cost hardware. And if you think it will be somehow cheaper than the "regular" options, then you are completely wrong, as shown above.

[ogden]

Keep in mind that you need at least ~100Msps to actually see a 10MHz square wave as a square wave - that's not going to work on a tiny two sided board. Four layers would be a minimum.

To avoid any confusion here - scope bandwidth does not mean that you will be able to see square wave of particular frequency on it:

https://www.tek.com/support/faqs/how-bandwidth-defined-oscilloscope-0

On the bright side, manufacturers try to put significant margins here. It is reported that some quite popular entry-level Rigol 100MHz scope have actual 150MHz bandwidth or so. I may stand corrected. If interested - seek EEVblog forums threads about Rigol scopes.

[rhb]

In the end it comes down to what's your  budget and what's your skill level.  There is no way to beat the price-performance of COTS DSOs.

If what you want is a FOSS FW DSO, then get a Zynq based DSO (non-Instek. I have that already) and send me a PM. Also if you are interested in sub $1K 3 GHz DSOs.  There's a LeCroy DDA-120 HW modification that would give 3 GHz on a single channel.  A lot more could be done if one replaced the PowerPC board with an FPGA/ARM board.

[ogden]

In the end it comes down to what's your  budget and what's your skill level.  There is no way to beat the price-performance of COTS DSOs.

It's already discussed earlier in this thread - that building oscilloscope for purpose is not feasible. Only reason some shall consider building scope - 1) just education/fun 2) specific, custom DSP function which is available only in hi-end scopes or not available at all. Example: hopeless FFT implementation in nearly all the scopes except R&S RTO series.

[janoc]

To avoid any confusion here - scope bandwidth does not mean that you will be able to see square wave of particular frequency on it:

https://www.tek.com/support/faqs/how-bandwidth-defined-oscilloscope-0

Of course, I do know that. However, he wasn't speaking about bandwidth but about it being "10MHz capable" (whatever that means).

From a practical point of view, being able to display a 10MHz square wave signal today is pretty much at the low end of what one should ask from an oscilloscope (even Arduinos run at higher clocks).

If someone wants to only do audio, then a regular sound card will do a better job than such gizmo anyway.

[bloguetronica]

...
From a practical point of view, being able to display a 10MHz square wave signal today is pretty much at the low end of what one should ask from an oscilloscope (even Arduinos run at higher clocks).
...

If you want to see a 10MHz square wave, you should use a 100MHz oscilloscope, or, at least, a 50MHz one. Now, let me tell you that designing a 50MHz oscilloscope is not easy. The "50MHz" only refers to the cutt-off frequency of the steep low pass filter that you'll have to employ in your AFE. Plus, it has to be a Bessel aligned one, to avoid ringing. That means you will have to use a DAC capable of, at least 500MSa/s of sampling (because Bessel filters may be steep, but are not very selective by any standards, even very high order ones - and you will have to avoid aliasing, hence the extra sampling rate). Although that is low end from a professional standpoint, and I agree on that part, it is not easy to design. No DIY kit scope will do, sorry.

But that is just my two cents.

Kind regards, Samuel Lourenço

[rhb]

Amazingly, if you want >10 GSa/S  it's a COTS dev board build.  The 1 GSa/S DSO is where it gets really hard.  But hijacking a COTS DSO is so much easier than designing and building one there is no comparison.

The Zynq based scopes are generally more money, but it gives a well defined infrastructure imposed by the low parts count.

[janoc]

Dave has just tweeted this:

https://www.crowdsupply.com/andy-haas/haasoscope

That should pretty much satisfy the OPs needs - small, $99, 4 channels, reasonable specs and open source. However, it is not two sided board (that's just not realistic) and you better do read the manual, there are plenty of gotchas for the unwary (like the bandwidth/sensitivity and voltage limits).

If you need a general purpose oscilloscope you would do better buying a normal desktop instrument than this (even used analog one). However, this gadget can be very useful for some specific applications.

[mbest]

Dave has just tweeted this:

https://www.crowdsupply.com/andy-haas/haasoscope

That should pretty much satisfy the OPs needs - small, $99, 4 channels, reasonable specs and open source. However, it is not two sided board (that's just not realistic) and you better do read the manual, there are plenty of gotchas for the unwary (like the bandwidth/sensitivity and voltage limits).

Only if you don't watch the mailbag video or previous mention in the live stream.

If you need a general purpose oscilloscope you would do better buying a normal desktop instrument than this (even used analog one). However, this gadget can be very useful for some specific applications.

As a newbie, I'm convinced to stay away from USB scopes

-Mike

[janoc]

Only if you don't watch the mailbag video or previous mention in the live stream.

Yeah, I have watched the video afterwards. However, the most of the problem seems to be a really crappy software - that's fixable. The rest is kinda par for the course, with pennies being pinched everywhere. Except maybe for the ridiculous issue with the BNC connectors being too close to each other.

However, if the OP wants an open source scope, it is probably the best contender available. Which doesn't mean it is any good only that there aren't really any other open source scopes around.

[abraxa]

Yeah, I have watched the video afterwards. However, the most of the problem seems to be a really crappy software - that's fixable. The rest is kinda par for the course, with pennies being pinched everywhere. Except maybe for the ridiculous issue with the BNC connectors being too close to each other.

However, if the OP wants an open source scope, it is probably the best contender available. Which doesn't mean it is any good only that there aren't really any other open source scopes around.

I'll bite and politely disagree - 8k sample buffer/channel is a joke. I know why that limitation exists but that doesn't change the fact that it makes it a bad device to do actual work with. The Scopefun project is not much better: 10k sample buffer per channel. From my point of view, that makes both really only suitable for basic signal analysis, unfortunately.

However, since both are open, it's possible to extend them and add DDR RAM support. If someone would do that with the Scopefun and make actual kits to buy, I'd bite.

[bloguetronica]

Yeah, I have watched the video afterwards. However, the most of the problem seems to be a really crappy software - that's fixable. The rest is kinda par for the course, with pennies being pinched everywhere. Except maybe for the ridiculous issue with the BNC connectors being too close to each other.

However, if the OP wants an open source scope, it is probably the best contender available. Which doesn't mean it is any good only that there aren't really any other open source scopes around.

I'll bite and politely disagree - 8k sample buffer/channel is a joke. I know why that limitation exists but that doesn't change the fact that it makes it a bad device to do actual work with. The Scopefun project is not much better: 10k sample buffer per channel. From my point of view, that makes both really only suitable for basic signal analysis, unfortunately.

However, since both are open, it's possible to extend them and add DDR RAM support. If someone would do that with the Scopefun and make actual kits to buy, I'd bite.

I don't consider a 8KSa buffer a joke. It is quite useful, and pretty standard (actually, 10KSa is more standard). If you said that 1KSa or 2KSa was a joke, I would agree. Anyway, you can't expect to pay less for an advanced scope, even if it is OSH (if that exists).

How do you expect to raise the buffer limit? I'm just curious. By the way, if I recall correctly, there will be a new version of the ScopeFun oscilloscope.

Kind regards, Samuel Lourenço

[janoc]

I'll bite and politely disagree - 8k sample buffer/channel is a joke. I know why that limitation exists but that doesn't change the fact that it makes it a bad device to do actual work with. The Scopefun project is not much better: 10k sample buffer per channel. From my point of view, that makes both really only suitable for basic signal analysis, unfortunately.

DDR3 support likely wouldn't fit on that size of the board.

And re 8k - well, I guess the Tektronix TDS2024C bench (!) scopes with 2.5kB are also a joke then? A $2700 joke, btw (https://www.newark.com/tektronix/tds2024c/oscilloscope-digital-storage-4/dp/86R8482 )

The small sample memory is really the least of the problem with this device.

From my point of view, that makes both really only suitable for basic signal analysis, unfortunately.

And what exactly did you expect to use a $100 devboard for? If you need more than the basics, you probably would buy a proper scope anyway.

I don't think that anyone (except maybe the author of the gadget) thinks this is a viable general purpose instrument. What this "Haasoscope" could be useful for is data acquisition where you need many channels and similar jobs (that's why I was talking about specific niche uses), where you will likely write your own software to go with it too. That's where it being open source is actually the most valuable.

[amy]

The technology is not great under the premise of standing on the shoulders of giants in the reality is not open source, play to his greatest advantage, to complete their own projects.

[prasimix]

It seems that ScopeFun is coming: https://www.crowdsupply.com/scopefun/open-source-instrumentation

Memory depth: 128,000,000 samples per channel

[rhb]

FWIW I let the magic smoke out of the NAND chip on my Instek GDS-2072E, so I have to fix it before resuming work on it. That will require learning to remove and replace the chip.  Hopefully I didn't kill the Zynq as replacing a BGA would be a lot harder.

Meanwhile I'm waiting on a book on FPGA based DSP to show up.  I bought 3 recent DSP texts and the one I want most is taking the longest.

I've laid out the architecture for the ADC to DDR section which will have LeCroy style stackable math functions, but have not started beating my head against Vivado yet.  I had another TEA relapse and bought a Tek 11801 and a pair of SD-22 sampling heads.

[OwO]

I've laid out the architecture for the ADC to DDR section which will have LeCroy style stackable math functions, but have not started beating my head against Vivado yet.  I had another TEA relapse and bought a Tek 11801 and a pair of SD-22 sampling heads.

Maybe sketch a rough block diagram and post it here. I have some experience with FPGA DSP and Zynq SoC design and might be able to help with the design.

[rhb]

Thanks.  Main intent is to reserve most of the DSP blocks for partial reconfiguration so the user can perform whatever operations they want before dumping the stream into DDR for display processing.  That's a pretty high bar.

There will be a slight delay though. 

My copy of "FPGAS-based Implementation of Signal Processing Systems" , by Roger Woods et al, 2nd ed finally arrived so I want to read it first.