Open Source "Lab instrument system" Idea

[Bloch]

Where have before been talk of a

Community Bench Meter and Open Source Multimeter


And that has been mentioned several things that makes it's impossible or at least why it will not happen.

When designing equipment for example a multimeter. It is normal to try meet everyone's needs but that is almost always impossible.

• Price
• function
• accuracy
• Size
• etc

And this will in a open source “forum” be a never ending discussion .
This is a proposal for usable open source diy lab equipment. "lab instrument system" / "bench instrument" Still working on a god name   

 
Instead of building a multi-function instrument. So, the system consists of simple modules.

• Volt meter.
• Ampere meter
• Ohm meter
• Digital power supply
• Digital input/output
• etc

Every module is divided in a front end and a output end.
The idea is that we don't care about the front end but have a very easy text protocol out of the module.
Every module can the be expanded with a

• logger cpu.
• Controls
• display

There can also be a "main" module to collect all modules to a big display and or a COM /  ETH / bluetooth etc. to controle it from a PC/MAC/Phone

[David_AVD]

I've been thinking of a similar thing for a while.  So far, I've been looking into the PC software side of things and was looking to use the SCPI protocol.  Unfortunately it's fallen to the back burner for now due to lack of hours in the day!

[FenderBender]

Would be a nice idea, though I don't know if it could really compete with any production model. I mean even autoranging might be a challenge, though off the top of my head I think I can figure out how to do it...

However, what something like this might offer is a good all around solution for the hobbyist or shallow-pocketed.

When you proposed the idea, it reminded me of the Tek 7xxx series that has a mainframe and then plug-in units.


photo by tekfan

If you were to sell it like that though, you'd need a custom enclosure of some sort.

Well who knows. Maybe it can happen.

[jucole]

Hi,  after reading through the open-source multimeter comments I got the impression the 'module' approach was quite popular.   So why not use perhaps an old flat screen monitor like the one Dave brought back from the dead in his video as a dedicated display for a multi-test tool - this is very rough etc but something along the lines of...



Afterwards I realised that a separate remote box unit with the inputs / buttons on it would work better.  just a thought anyway! ;-)

[FenderBender]

You'd have to figure out how to drive it. I guess you'd essentially need to have a PC. Maybe running Unix of some sort.

[gxti]

You can drive a VGA display using a microcontroller with some difficulty, but FPGAs are generally a better fit and can do digital formats as well. But I do think an OS (probably Linux) would be smart because more people will be comfortable contributing to it. You can also combine the features and run Linux on the FPGA :-) OSes are rather common on higher-end instruments, especially scopes.

[FenderBender]

Real big ambitions here guys. It might be more worth it just to use a small like 4" TFT and drive it from a MCU or FPGA. Or you could use a Rasberry pi.

[pickle9000]

The Maximite is an excellent example of how to drive a VGA monitor.

http://geoffg.net/maximite.html

...mike

[FenderBender]

What would be the cost of the total machine? Would this be a product to rival Agilent? Or would it be something small and accessible to hobbyists and the like?

[jucole]

To me such a project is not about trying to rival commercial projects; It's all about the learning process itself.  If you use a Rasberry pi or FPGA Linux you might as well just use a laptop.  If you want high specs - you might as well buy some solid used test gear from Ebay.

It doesn't have to be over-cooked as well; For example the display driver could be an fpga, some ram and a couple of resistors for a video-out sync to either a tv or monitor at 800 x 600 with 3 shades of black.  The capture module might be re-chargable battery powered, with bluetooth to send tiny update display packets to the display module. The display module would have a uC to manage the display layout.

The attraction for me is the "all-in-one",  the remote capture module and also you can make use of old displays.

[kripton2035]

Real big ambitions here guys. It might be more worth it just to use a small like 4" TFT and drive it from a MCU or FPGA. Or you could use a Rasberry pi.

+1 for the rasberry pi. ideal candidate for this job.
and far less quick obsolete than a laptop ...

[David_AVD]

+1 for the rasberry pi. ideal candidate for this job.
and far less quick obsolete than a laptop ...

Until the Pi is replaced with the next new model / best thing...

[rolycat]

+1 for the rasberry pi. ideal candidate for this job.
and far less quick obsolete than a laptop ...

Until the Pi is replaced with the next new model / best thing...

The Pi is cheap and extremely capable, has an accessible and well specified GPIO connector, offers built-in USB and Ethernet, and boasts a huge and rapidly growing user base.

Would 'the next new model' invalidate all of those advantages? I doubt it.

There seems to be a school of thought which believes that the popularity of the Raspberry Pi with newcomers to programming and/or electronics somehow makes it unworthy for serious consideration. 

[hlavac]

Open source hardware/software would be extremely beneficial for something that would replace multichannel oscilloscope, signal generator, spectrum analyzer, frequency counter etc in a single cheap, expansible universal instrument for a hobbyist.
In modern instruments the hardware is not really expensive,
it's the software options that cost arm and leg...

[Bloch]

The VGA / HDMI / DVI "Controler" module is a great option.

But the priority must be to build good "Front End" modules first.

>>though I don't know if it could really compete with any production model<<
We cant compete on every things. Especial on price.


That we can do is make a


easy to build
easy expandable
good quality components/layout/schematic
And maybe most important is to make a good/fun modules. That from the noob to a "Dave" have need to have / build look in the eyes    


>>What would be the cost of the total machine? Would this be a product to rival Agilent? Or would it be something small and accessible to hobbyists and the like?<<


The nice thing about this is it open/free. Next you build that you want/need. That way you will learn how it works and will be posible to fix it if it dont work.


But if you work on a very important product or need some important measurement then i hope you buy the Agilent one. We will never get all the safety approvals / EMC on the modules.


So i can only see the modules as a Enthusiasts / school / learn thing.

[jucole]

The form factor of the modules could look something along the lines of one of these..



Then all you need is a "caddy" to plug them into.. 





(edit, i tweaked the picture in case of copyright issues.)

[Bloch]

Then all you need is a "caddy" to plug them into..

Well that is not off the shell box.... At least not cheap. Correct me if i am wrong.


In the bench thread Dave did mentioned http://www.polycase.com/zn2 I must say it is a nice box.

[jucole]

Then all you need is a "caddy" to plug them into..

Well that is not off the shell box.... At least not cheap. Correct me if i am wrong.


In the bench thread Dave did mentioned http://www.polycase.com/zn2 I must say it is a nice box.

+1  nice!

[poptones]

I bought tone of the pis. After looking over the docs I can't really say it's all that.

The GPIO is only a partial bit of what's available form the chip, and is not well documented. They don't even tell you how fast the lines can change state with any certainty. The network is just a usb dongle; the "camera" interface (that DOES allow high speed input) is proprietary, you're not going to be able to develop for it without ndas and binary blobs.

The pi should not be the basis of ANY "open hardware" system - it can't, because it's not open hardware. You'd be way better off picking an intel motherboard as the basis for an "open" system. HD3000/4000 video is fine for any application like this, and the drivers are fairly robust. You'll get way more computing power, low price, and a well documented and free software development platform.

[David_AVD]

I have a few items (power supply, multi-channel logic probe, colour bar generator) that I built as Eurocard modules a few years ago.  They plug into a frame.  Nice, but not super cheap.  You can get varying width modules as just card fronts or fully enclosed cases.

[gxti]

Agreed, Pi has no place in open-source hardware. Even ignoring the ideological issues, they might stop making it in a year, or change it enough that the OSHW stuff has to change, too. Plus it's nothing special -- there are Pi-like boards coming out already, some from reputable designers like Olimex that will probably stick around for a little while.

[senso]

Probably an Cortex M4 has the raw power to do an multimeter, a lab supply and to some extents a signal generator and logic analyzer.

[Bored@Work]

You are discussing irrelevant issues. These days computing power is readily available in all shapes and forms. Lack of computing power is not a risk for such a project and does not dictate other design decisions.

Enclosures are readily available, too. Although most in ugly shapes and forms. Still lack of an enclosure is not a risk for such a project. You can always go larger and uglier if needed.An enclosure dictates design decisions in the sense that it dictates size, shape, safety, etc. But you currently have not even a rough estimate about the required size, safety, etc.

Instead of wasting time with these trivial issues now, what about the showstoppers that will kill your project? E.g. where is your analog meter front-end? Ups. Are you waiting for a white night or a miracle? Good luck.

[Rerouter]

(focusing on bench type as i see that more feasable)

being how the interface is meant to be modular, perhpas look at it in a modular way, e.g. what type of interface would have to be used to get the most capability out of the least dollars, what voltages, signals and communication protocols are likely to be better fed from a single point than from each module, and the golden question how will it all connect?

e.g. this modular design will always have to have a power supply or atleast power source in it somewhere, and here the most common voltages can be sourced from, the currents involved would depend a little on what is to be tied off of the thing, but the point is what voltages, i myself am left thinking 3.3V, 5V, -5V, and possibly 12V for the possibility of a power supply output on the thing,

that in my mind covers all digital and near all analog design requirements, removing what would otherwise be regulators on each module, and leaves the possibility of specialty voltages to be generated from them

next would be signals, are there any signals such as 16Mhz, 100Khz or even 1Hz that would simplify the design of the possible modules in any significant fashion over having a seperate crystal per module?

and finally communication, would there be an I2C bus? would there be some SPI loop arrangement? high/low status lines or interrupts, and any secondary module to module only busses? and even the possibility of taking those busses and piggybacking into a second one? would the various busses rebroadcast or be seperate and how would that be decided? will level shifters be provided or even footprints that can be bridged, or straight signals?

and for the connections i feel like the tek series plugins should be a backplane based solution, with a common and cheap connector, for this would think a 0.1" SIL connector, but this can be furthur discussed, and also what spacing should be used, to allow larger things to use up 2-3-4-x wide slots,

what i am getting at is before we try and make it into a product perhaps we should work out the mechanisms to build upon, such as what will be provided to try and spur some community involvement with possible modules, and solutions, as i myself wouldnt mind remaking a few things for a good system

[jucole]

I'm really tempted to have a go myself because I'd learn quite a bit, also it would be a lesson for others on how NOT to design test gear! ;-)

(edit - a quick visual of the stacker box approach)