New firmware for scopes. Open or otherwise.

[Ed.Kloonk]

I had a chat today with a code rat that flashes consumer hardware for poops and giggles. And for beer if a willing customer is willing. 

Turns out many of these cheap DSOs are prime for some full-tilt firmware tweaks like the common router or even the android phones.

What I am interested to know is the implication of any major firmware tweaks.

We are talking stepping away from the manufactures' code in the way that you can flash a router with dd-wrt or tomato.

My questions.

Would you use it?
What features would you like to see?
1054z looks like the perfect initial target.

Bottom line, I am trying to gauge the investment value of this idea.

[tautech]

    
You make out like it's not being done around here big time already.

70 MHz to 150, 100 to 200
100 MHz to 350 and all options
350 MHz to 1 GHz " " "

And that's just scopes, then there's PSU's, AWG's, Spectrum analyzers, RF gens and so on.

[Ed.Kloonk]

   
You make out like it's not being done around here big time already.

70 MHz to 150, 100 to 200
100 MHz to 350 and all options
350 MHz to 1 GHz " " "

And that's just scopes, then there's PSU's, AWG's, Spectrum analyzers, RF gens and so on.

I'm talking walking away from the base firmware. Where the interface is somewhat standardized across several models and makes.

I'm talking completely nuking the standard firmware. What would your ideal Frankenstein creation look like?

Again, just starting out with a couple of cheap scopes. Not taking over the world.

[2N3055]

   
You make out like it's not being done around here big time already.

70 MHz to 150, 100 to 200
100 MHz to 350 and all options
350 MHz to 1 GHz " " "

And that's just scopes, then there's PSU's, AWG's, Spectrum analyzers, RF gens and so on.

I'm talking walking away from the base firmware. Where the interface is somewhat standardized across several models and makes.

I'm talking completely nuking the standard firmware. What would your ideal Frankenstein creation look like?

Every few months there is someone with that idea.. And they find a way to flash new empty linux on the scope (mostly reference,base distribution from chip manufacturer). Basically taking perfectly functioning instrument that took 100-200 engineer years to develop, and make it useless linux computer with crappy processos, not much memory and small screen.
An to prove how smart they are they run Doom on it to prove they did it.
And then, when they realize they have no clue how do real work (the scope part) it dies silent death...

Fact that DS1054Z is 300-400 USD makes people think it is simple. It's not. It cannot be done in makers, hobby or even open source community. Because there is no skillset to do it, among people that would want to do it. Those that know how, do it for (good) money. And it is a hard, coordinated work, that is not compatible with people hacking and twiddling some code up.

It is not possible, not by people that are "code rat that flashes consumer hardware for poops and giggles. And for beer if a willing customer is willing."

[tautech]

I'm talking walking away from the base firmware. Where the interface is somewhat standardized across several models and makes.

I'm talking completely nuking the standard firmware. What would your ideal Frankenstein creation look like?

Again, just starting out with a couple of cheap scopes. Not taking over the world.

IMO it's a much bigger task than you might imagine and for what real gain ? 
Even cheap scopes are pretty complex these days.

[Ed.Kloonk]

IMO it's a much bigger task than you might imagine and for what real gain ? 

I don't know. It's why I'm asking. 

I'm looking for the silver bullet. What is missing from the software? When you move around your scope or even from scope to scope, what is the biggest thing that futzes with your productivity?

Even cheap scopes are pretty complex these days.

Yeah nar. As with a mature software stack such as dd-wrt, you spin off model-specific modules. 

The scope business is on a race to the bottom like the router industry was a decade ago.

I think there might be a buck in it.

[2N3055]

IMO it's a much bigger task than you might imagine and for what real gain ? 

I don't know. It's why I'm asking. 

I'm looking for the silver bullet. What is missing from the software? When you move around your scope or even from scope to scope, what is the biggest thing that futzes with your productivity?

Even cheap scopes are pretty complex these days.

Yeah nar. As with a mature software stack such as dd-wrt, you spin off model-specific modules. 

The scope business is on a race to the bottom like the router industry was a decade ago.

I think there might be a buck in it.

To quote you , yeah nar...

Routers ARE simple. Scopes are not.

I already explained everything. Read again.
I know my English is horrible, but it was clear enough on this, I think.

Or to put it Apple like simple:  Waste of time.  That includes even discussing it. Just do something else, that is useful.

[tautech]

IMO it's a much bigger task than you might imagine and for what real gain ? 

I don't know. It's why I'm asking. 

I'm looking for the silver bullet. What is missing from the software? When you move around your scope or even from scope to scope, what is the biggest thing that futzes with your productivity?

Lack of familiarity.
Simple to fix if we take the time.

This stems mainly from the many and varied tasks that one wants to do with a scope are vastly different and with modern scopes being quite complex it takes a mo to get it all as you like for each task at hand. Pro users are better positioned in this regard as they mostly know their scopes inside out and back to front yet we each have our own usage 'style'.
That alone makes one users methodology a nightmare for another which is why scopes have evolved to attempt to suit a variety of usage styles. For that reason you can often find the same settings in multiple menus.

Experienced users generally have little trouble using any decent scope whereas the novice has a lot to learn about just driving a scope let alone knowing what one should expect to see before even connecting a probe.

The scope business is on a race to the bottom like the router industry was a decade ago.

Ya think so ?
From where I sit much advanced functionality has been added into scopes than might have even been expected just a few years ago and for even less cost !

As a good few have said in the last year or two, we live in exciting times that so much good TE is so affordable.

[Elasia]

if you really want to horse around then pick up some siglent scopes that are already linux based 

[Ed.Kloonk]

Ya think so ?
From where I sit much advanced functionality has been added into scopes than might have even been expected just a few years ago and for even less cost !

Sorry, but I don't understand how that isn't precisely a race to the bottom.

One preliminary use case I saw is a substitute for an Arduino data logger with really nice front end. Live updater.

Of course DSO's are not as simple as a ordinary router - at the moment.

The router analogy is prolly flawed. But..
It's a whole paradigm shift. I know that 95% of people won't see it, but 99% of people use their ordinary router without needing anything more.

I'm not asking for how the software can do what it can do better, I'm asking what you could use the hardware for and what the standard software prevents you from doing so.

[Elasia]

The scope business is on a race to the bottom like the router industry was a decade ago.

Ya think so ?
From where I sit much advanced functionality has been added into scopes than might have even been expected just a few years ago and for even less cost !

As a good few have said in the last year or two, we like in exciting times that so much good TE is so affordable.

WTB 8 channel  analog + 32 channel digital and all touchscreen for... 500 bucks!   

[tautech]

The scope business is on a race to the bottom like the router industry was a decade ago.

Ya think so ?
From where I sit much advanced functionality has been added into scopes than might have even been expected just a few years ago and for even less cost !

As a good few have said in the last year or two, we live in exciting times that so much good TE is so affordable.

WTB 8 channel  analog + 32 channel digital and all touchscreen for... 500 bucks!   

Only 8 + 32 ? 
 

[Ed.Kloonk]

The scope business is on a race to the bottom like the router industry was a decade ago.

Ya think so ?
From where I sit much advanced functionality has been added into scopes than might have even been expected just a few years ago and for even less cost !

As a good few have said in the last year or two, we like in exciting times that so much good TE is so affordable.

WTB 8 channel  analog + 32 channel digital and all touchscreen for... 500 bucks!   

I know, but if you're young and broke and you do have a basic scope sitting on the shelf. I understand for you $500 bucks is a couple of hours at work. 

But if you're a hacker and you've got nothing but perhaps something you got given for your birthday I can see a use case.

[tautech]

Here's one for ya Ed;
Cascading DSO's for multi channel work with an I/O to I/O on another and some method of nulling delay times so scope displays can be aligned in time for 6/8/12/16 channel operation.....not MSO.
 

Edit to add
First DSO in the line assumed as master as it's triggered on a trace of interest with any further DSO's displays time aligned.

[tmbinc]

I've worked with the Rigol DS10x4Z (and similar) and Siglent 1104X-E (and similar).

The Rigol has a CPU that's not a lot of fun to work with (ARMv5T still). I've ported Linux with framebuffer, and then couldn't figure out how they actually stream the data from the ADCs, and overall it was painful.

I've then joined an existing effort (https://github.com/360nosc0pe/fpga) to re-write the bitstream and software for the Siglent. It's a much nicer hardware - Zynq-based, so a standard ARM linux binaries work - and we've reverse engineered almost all of the pinout, frontend control, clock etc., sufficient to make it receive data from both ADCs. In theory, at that point it's already useful, for example if you want to stream data and don't need a UI.

Then again nobody ever wrote a UI or more code so now it just bit-rots. But if I ever need a fast ADC on an FPGA, I'll likely pull that board out and continue that work. Or if LiteX zynq support gets better.

I don't understand the hostility here though. There are things that the manufacturers firmware can do incredibly well. But every user is different, and every use case is different. Fact is that freely-programmable hardware that is competitive to a cheap scope in terms of specs is significantly more expensive (like NI PXIe stuff), and that there are some low-hanging fruits (for example a realtime UART decoder - how useful would that be if we'd have a /dev/ttyCH1..4 on the scope that just works? That's not even a lot of effort - just use a threshold output to drive an instantiated UART).

I love OpenWRT, but at the same time it's not what I'd recommend if someone just asks for a nice router. But there are enough use cases that justify the existence, and in these cases OpenWRT can be just the right tool for the job.

[Ed.Kloonk]

Here's one for ya Ed;
Cascading DSO's for multi channel work with an I/O to I/O on another and some method of nulling delay times so scope displays can be aligned in time for 6/8/12/16 channel operation.....not MSO.
 

Holy cow.

[Elasia]

Only 8 + 32 ? 
 

Physical limitations... That makes a nice bottom row, give it a decade to hit sub 2k

[tautech]

Here's one for ya Ed;
Cascading DSO's for multi channel work with an I/O to I/O on another and some method of nulling delay times so scope displays can be aligned in time for 6/8/12/16 channel operation.....not MSO.
 

Holy cow.

Yep but very useful if needed.
You could pull the rug from right under 8ch scope manufacturers. 

[macaba]

I've then joined an existing effort (https://github.com/360nosc0pe/fpga) to re-write the bitstream and software for the Siglent. It's a much nicer hardware - Zynq-based, so a standard ARM linux binaries work - and we've reverse engineered almost all of the pinout, frontend control, clock etc., sufficient to make it receive data from both ADCs. In theory, at that point it's already useful, for example if you want to stream data and don't need a UI.

I started writing a UI for the MSO5000 (also Zynq) and got a simple example to run. The issue is that I didn't want to touch the FPGA firmware (I know my limitations) and therefore didn't have enough reverse engineering documentation to control the capture subsystem. Without that, wasn't much point in continuing for the sake of a dummy UI.

Thanks for that link... looks like an interesting potential avenue to explore.

[Ed.Kloonk]

I've worked with the Rigol DS10x4Z (and similar) and Siglent 1104X-E (and similar).

The Rigol has a CPU that's not a lot of fun to work with (ARMv5T still). I've ported Linux with framebuffer, and then couldn't figure out how they actually stream the data from the ADCs, and overall it was painful.

I've then joined an existing effort (https://github.com/360nosc0pe/fpga) to re-write the bitstream and software for the Siglent. It's a much nicer hardware - Zynq-based, so a standard ARM linux binaries work - and we've reverse engineered almost all of the pinout, frontend control, clock etc., sufficient to make it receive data from both ADCs. In theory, at that point it's already useful, for example if you want to stream data and don't need a UI.

Then again nobody ever wrote a UI or more code so now it just bit-rots. But if I ever need a fast ADC on an FPGA, I'll likely pull that board out and continue that work. Or if LiteX zynq support gets better.

I don't understand the hostility here though. There are things that the manufacturers firmware can do incredibly well. But every user is different, and every use case is different. Fact is that freely-programmable hardware that is competitive to a cheap scope in terms of specs is significantly more expensive (like NI PXIe stuff), and that there are some low-hanging fruits (for example a realtime UART decoder - how useful would that be if we'd have a /dev/ttyCH1..4 on the scope that just works? That's not even a lot of effort - just use a threshold output to drive an instantiated UART).

I love OpenWRT, but at the same time it's not what I'd recommend if someone just asks for a nice router. But there are enough use cases that justify the existence, and in these cases OpenWRT can be just the right tool for the job.

Thanks.

I forgot about OpenWRT. The last time I rolled up my sleeves, there was DD-WRT and tomato if you were adventurous. 

I just use routers out of the box, so I'm the worst.  If I'm in a good mood they might get an admin password change.

Thanks again for the information. I'll pass it on.

 

[nctnico]

Every few months there is someone with that idea.. And they find a way to flash new empty linux on the scope (mostly reference,base distribution from chip manufacturer). Basically taking perfectly functioning instrument that took 100-200 engineer years to develop, and make it useless linux computer with crappy processos, not much memory and small screen.

I agree. And the existing oscilloscope hardware isn't such a good start anyway because it is very limited. Also the hardware isn't the hard part; the software is 100 times more complex. If an open source oscilloscope is ever feasible the software must be as easy to develop as possible and the use of FPGA must be minimal. One of the ideas I have been toying with is to take an extremely simple FPGA design which can do basic triggering and buffering. After acquiring data the FPGA just streams the data into a SOC (with a beefy GPU) using a PCI express bus. From their the GPU which is A) easy to program and B) has a massive amount of processing power can crunch the data into a trace on the screen (which can be a 4k or ultra-wide monitor for all I care). But for me this needs to be a paid project.

[Ed.Kloonk]

Here's one for ya Ed;
Cascading DSO's for multi channel work with an I/O to I/O on another and some method of nulling delay times so scope displays can be aligned in time for 6/8/12/16 channel operation.....not MSO.
 

Holy cow.

Yep but very useful if needed.
You could pull the rug from right under 8ch scope manufacturers. 

Tomorrow's thread: Tablet DSO's and why MS are A holes.

 

[Ed.Kloonk]

Every few months there is someone with that idea.. And they find a way to flash new empty linux on the scope (mostly reference,base distribution from chip manufacturer). Basically taking perfectly functioning instrument that took 100-200 engineer years to develop, and make it useless linux computer with crappy processos, not much memory and small screen.

I agree. And the existing oscilloscope hardware isn't such a good start anyway because it is very limited. Also the hardware isn't the hard part; the software is 100 times more complex. If an open source oscilloscope is ever feasible the software must be as easy to develop as possible and the use of FPGA must be minimal. One of the ideas I have been toying with is to take an extremely simple FPGA design which can do basic triggering and buffering. After acquiring data the FPGA just streams the data into a SOC (with a beefy GPU) using a PCI express bus. From their the GPU which is A) easy to program and B) has a massive amount of processing power can crunch the data into a trace on the screen. But for me this needs to be a paid project.

I'm glad that you can explain that it is hard instead of just saying it's impossible.

[Fungus]

1054z looks like the perfect initial target.

Why? It's about the only one in the pack that doesn't run Linux.

[Ed.Kloonk]

1054z looks like the perfect initial target.

Why? It's about the only one in the pack that doesn't run Linux.

Yeah. Pity.