Lets discuss how to make OSHW more successful (inspiration by FOSS design cycle)

[svenskelektronik]

Let me introduce myself (first post)
I got my eyes on technology 15 years ago when I was creating a website for my elementary school band. Without any prior knowledge I started looking at my favorite band website (Necrophagist at that time) and really wanted my own band website to look like that. Not long after, I found out that it was possible to see the underlying code for the design of any webpage. After a quick webdesign tutorial I started copying cool features from different webpages to finally build my own band website. This was my first tech project of many to come, without realizing it at the time that project introduced me to the hacking mentality of FOSS and OSHW which brings me to the thread topic...

Why FOSS is ahead of OSHW
In my introduction I gave you an example of how FOSS contributed to my fist tech project by the sharing of the source code and how I then contributed by the sharing my altered version of that code. My first FOSS project shares one shortcoming with many OSHW projects of today:  The inability to contribute back to the original project. I belive the success of FOSS is closely linked to the ability for many to work and contribute to a project.

The collaboration is easly done in software development by push and pull request and both the original auther and the contributor can easly verify the result by compiling the source code. We can already see the use of push pull request in OSHW development, but its failing because implementing the contribution to the master branch requires the original auther to verify the changes. The verification of hardware is done by producing a physical prototype of the new version and test it, this is both expensive and time consuming which is causing resistance to collaborate in OSHW.

Applying FOSS DNA to OSHW projects
I belive we need remove friction in the verification process of a push pull request in OSHW development. In order to remove this friction, I propose that the PCBA fabs and OSHW projects should cooperate more closely.

My offer to the OSHW community
I do run a PCBA factory www.svenskelektronikproduktion.se/en and we want to cooperate with OSHW projects that we find interesting by offering free prototype and verification services to help them launch,maintain and update their product. In return we want to produce the product for the original auther so that the original auther can sell the product for a profit. We can also assist with disribution of the product and ofcourse share the profit with the auther. Please send your request to fredrik@svenskelektronikproduktion.se .

Please provide suggestions start a discussion in this thread on how to increase the popularity of OSHW. What can the fabs, the community and OSHW projects authers improve on?

[rhb]

The key is successful designs like the $20 LCR tester prototype.  The Chinese have been going like mad on  that with excellent results. For things like DSOs, the manufacturing process works against OSHW. Too much test gear required even if you have a finished design.  The complexity threshold moves over time as test gear becomes cheaper.

Aside from assisting in the development, you're in a very good position to popularize designs by making small runs and selling them online.  The way to measure success is the number of copies being sold on AliExpress and eBay.  When things become available there, drop that device from the current active product list.  Many of the Chinese merchants have very little money, so they have to be very careful about what they choose to manufacture.  By providing sales figures on OSHW designs you will greatly increase the likelihood that a design will be copied.

I think that making money from designing OSHW  is a pipe dream.  However, someone who has designed a successful OSHW device can command better compensation in the employment market whether as a regular employee or as a contractor.  IMHO the real benefit to OSHW is the reduction of the barrier to entry into serious EE work by reducing the cost of test gear for a personal lab.  This results from giving all the profits to the manufacturers who all have to compete producing the same basic design.

Sun, IBM, SGI, DEC and HP spent a fortune on the workstation wars.  Then IBM committed a billion dollars a year to Linux development and wiped them all out.  Instead of $30-50K, workstations are under $5K, and the meager profits go to the hardware manufacturers.  Yes, there are still $50K workstations. Microway sells s a very nice one with 1 TB of DRAM, 48 Xeon cores and several Tesla GPUs.  It runs Linux.

[madires]

Since the Transistortester is already mentioned you could manufacture and distribute the ATmega644/1284 based version including a few  hardware options. Nearly all Chinese clones have just an ATmega328 and the only 324 based clone has a poor pin assignment and minimal functionality. This could be an opportunity. Karl-Heinz (k-firmware) an I (m-firmware) aren't interested in manufacturing and distribution because it would involve too much hassle for a OSHW project (prepaid disposal, accounting, taxes and so on - it's insane over here).

[svenskelektronik]

rhb:
Thank you for your informative reply. We will prefere quite specialized products that wont sell in big quantities as we are quite competitive on low volume production, even compared to Asian manufactures.
The only requirement I have to keep this initiative going is that I can cover my costs in the long run. I will prefere to run these products when time is available in the assembly line instead of planning the production, which will reduce cost. My motivation in this is my own interest in OSHW and the promotion of my company.

Madires:
I will have a look at the ATmega644/1284 version of the Transistortester. I there is a particular reason why one would preffere the ATmega644/1284 version over the ATmega328 version?

[madires]

The ATmega644/1284 version offers hardware SPI (faster display output), hardware USART, hardware I2C, more hardware options and more Flash/EEPROM/RAM of course. The ATmega328 is maxed out with all features enabled.

[rhb]

There is a vast array of things that won't sell in large quantity, but of necessity must be done on multilayer PCBs.  Many of which are designed by professional engineers for their own use.  They don't want to go into producing a hobby project for resale.  I'd like to suggest that you take QEX from ARRL and look for what you think are suitable candidates.

Unfortunately I'm an old guy who did much of his design work before SMD parts became common and then did nothing in hardware for many years.  So I haven't learned any PCB design tools.  I've got a slew of them installed, but I'm going to have my hands full with Vivado and Verilog for a good while.

Something that came up in the MDS-2000E thread is affordable calibration kit.  A simple RC circuit with precision voltage reference, switchable dividers,  proper JFET input voltage followers and a sparse L1 pursuit code would provide an excellent general purpose tool for testing  and quantifying ADC errors.  The circuit for this is trivial and the software not difficult, at least for me.  The hard part is PCB layout good to 1 GHz or better.

Another big need is ADC and comparator boards for use with FPGA boards.  There are lots of these in the very expensive range, but none I know of that I can buy for $100.  I bought a scope for less than a 1 GS/S ADC board would cost.  The probes to implement  a proper logic analyzer seem not to be available except at over $500.  The Digilent Discovery has *no* front end, just bare FPGA pins.  A modest amount of logic for setting thresholds and 4 ADCMP561s at ~$4 each are all that is needed to provide fast 40 V tolerant probes.  The Zynq makes a vast array of test instrumentation relatively easy to construct.  But you still need the analog front end.

I mention these because they are integral requirements for my Zynq project.  My goal is FOSS FW for commercial instruments, but having a Zynq board with a fast ADC and/or a set of fast comparators or low cost mezzanine boards would enable lots of other things.  There is only so much time available to any human.  Most things that can be done entirely on one's own have been done.

[T3sl4co1l]

Large FOSS projects have negative incremental cost.

It seems unlikely that production costs could ever possibly be low enough to achieve the same in OSHW.

Simple economics!

I've not seen any large OSHW projects, and none that are more than toys.  In other words, the development cost is negligible, or already sunk (such as an education cost -- a project done to familiarize oneself with the tools or techniques, say), and the production cost is up to the seller or end user.

Tim

[awallin]

I've not seen any large OSHW projects, and none that are more than toys.  In other words, the development cost is negligible, or already sunk (such as an education cost -- a project done to familiarize oneself with the tools or techniques, say), and the production cost is up to the seller or end user.

define 'large'.. depending on your viewpoint CERN hosts a number of smaller and bigger projects on ohwr.org
some designs have multiple commercial manufacturers and now a life-span of 5..7+ years with design-revisions etc.
Nothing is ofcourse free, and the bigger professional looking desings are done by people working full time on them - paid by a host institution who believes in open hardware.

Now that universities are mostly required to publish research in open-access journals - should we require that government funded schools and institutions release hardware also under open licenses? 

[svenskelektronik]

I think instruments and production equipment is a great segment to start applying OSHW and also FOSS FW. Looking back at the FOSS movement, we know that Richard Stallman and others built the Emacs editor, the GNU compiler, GNU debugger and also many usefull UNIX tools that later became the GNU/Linux operating system which you all know of today. My point here is that OSHW projects that targets hardware developers will attract more contributors and gain more success. Another key feature is to make products able to communicate/compatible with each other, this is especially beneficial to production equipment but also handy in mesurement equipment.

I can see a positive trend in schools of today, some of them are transitioning to OSHW and FOSS products and environment in their hardware and computer labs. For example I have been contacted by a teacher from the university I graduated from regarding FOSS, this resulted in the use of KiCAD in the electronics lab and in some of his courses.

From my experience it usually requires really expensive IC:s to build high end instruments. The problem here is you'll need to buy several thousands to get the prices down at a competitive level which will require a lot of cash and commitment as some IC:s may cost up to 50$. If the intrument has a special unique feature, for example FOSS FW, it may be possible to start up small. Another way would be to start out with entery level instruments as the part cost of a couple of 1K unis is managable, and we do not need to fear competition so much as parts do not drop much in price after past a couple of 1K units.

[rhb]

I think that you're in the best position to decide what projects to produce.  It will take a good bit of searching and separating the wheat from the chaff, but once you have brought a few things to market successfully, the projects will come to you.

IMHO things at the level of an Instek GDS-2074E are best dealt with by developing FOSS software to run on them which is why I'm pursuing that.  A precision voltage reference such as the project in the Metrology thread would be my suggestion of something that would greatly benefit from what you're suggesting.  I bought Leo Bodnar's pulser and his GPSDO.  I'd love to be able to buy a voltage reference in a similar price range.  I suspect there's a pretty good market for a $200-300 multi-decade precision voltage reference.

Consider this:

A precision voltage reference with an ADC and a DAC connected to an MCU.  Have some switchable  R and C values.   Set  the desired output voltage using the ADC and a timer referenced to a GPSDO.  Use the ADC value to address a lookup table that feeds the DAC.  The curve shape is known analytically.   By switching the R & C values  and measuring the change in time constant against a GPSDO one can correct for all the errors including the ADC and DAC.  Aside from the voltage reference it does not require precision parts. A sparse, minimum summed absolute error  L1 pursuit will handle the self calibration process very well.  Much better than a traditional least squared error L2 solution.  I can write the software for such a device in less time than I can learn KiCAD or any other PCB design tool.

I've never seen or even considered such a design.  It's is entirely the result of replying to your statements.  I've examined the mathematics of using the RC curve and an L1 solution to correct for ADC and DAC errors, but  had not considered applying it to the problem of constructing a precision divider.  At the price of needing to run the self calibration code on a PC, it's really quite easy to implement relative to the basic voltage reference itself.  I'm going to head over to Metrology and see what the voltnuts think.

Have Fun!
Reg

[kasbah]

Hi Frederik, I replied to you on Reddit as well so sorry for the cross posting (but you started it  ). My project Kitspace is trying to reduce this friction in OSHW electronics collaboration as well by providing 1-click orders for components and PCBs.

It's a growing repository so you can probably find a something on there that is worthy of your support. The fact that detailing parts in a bill of materials is a requirement to be on Kitspace should make it easier on your side to source the components and assemble the PCBs.

I have been helping document some microscopy projects [1] [2] and wouldn't mind getting some PCBs for these and try and put them together myself.

[rhb]

I should like to suggest a 1 ppm voltage calibrator based on the LTZ1000 and the AD5791.  Maurice Egan of AD has designed a circuit for a 1 ppm voltage divider using the AD5791. As LT has been acquired by AD they might consider contributing such a design.  I've contacted Mr. Egan to see if AD can make board files available, but reference schematics are available and there has been a long running thread in Metrology on the LTZ1000.

This is clearly much better than what I suggested earlier. Based on quantity one prices a BOM cost of about $150 seems possible for small runs.  At  around $300 it would be a nice complement to a GPSDO and pulser for calibration and testing in a personal lab setting.

[svenskelektronik]

kasbah:
I'm familiar with your services and I'm glad to stumble upon you here. I have sent you an email to the registered mail of your github account, I look forward to have your reply.

rhb:
Thats great information! Even if Mr. Egan wont support the idea the shematics, datacheet and the Metrology forum will provide usefull information. But lets wait and see what Mr. Egan thinks about it!

[rhb]

I've not heard from Egan, but there is an eval board for the DAC for $150.  It takes a daughter board with an LTZ1000, but that is $450 from Mouser.

In any case, AC & DC voltage calibration tools seem to me a very good fit for what you want to do.  It's a niche, low volume product for which there are no convenient and economical alternatives to building your own and a number of designs are available.  I certainly don't need 1 ppm voltage.  Even 10 ppm would be a huge step up from nothing.  I've learned that there are lower cost parts.

There is the problem of calibrating the calibration tools.  So one factor which may be helpful would be to offer measurements using traceable equipment.  No adjustments, just measure and report the performance.  For devices controlled over USB or ethernet that's very quick and easy to dowith a modest amount of programming.  If the design incorporated a flash memory the calibration data could be stored on the device.  So all that would required was connecting the device and running the test program.  Receiving and shipping would probably be the bulk of the work.

[svenskelektronik]

The design should obviously start from these eval boards. What program do you use? I may help with the board layout in KiCAD if there is an opening in my schedule.

[rhb]

AD supplies some software.  There was a video link  posted to a thread I started which shows it.  I don't know anything about it though.    Personally I'd rather use an MCU and some decade switches I have to set the voltage rather than require a PC.  I'd provide USB SCPI control at a minimum, but with the ability to control as an independent instrument.

https://www.eevblog.com/forum/metrology/an-arbitrary-output-voltage-divider-for-a-precision-voltage-reference/msg1455753/#msg1455753

An 18 digit AD5781 is about 1/2 the price of a 20 bit AD5791.  I'm told by @cellularmitosis that an LM399 voltage reference is much cheaper (<1/10th $)  and easier to construct.  That would only offer 10 ppm, but if it could be sold for $100 - 150 I think it would be a good  way to test the idea.  AD uses the same PCB for the eval board for both and charges the same. An LTZ1000/AD5791 with controller board totals $760 US from AD which puts it beyond what I can justify for my needs.

Most of yesterday was consumed by tracking down the cause of kernel panics on my OpenSolaris Hipster system running Windows and Debian in VMs under VirtualBox.  A particular SanDisk USB flash drive proved to be the culprit.  Hipster is fine with the drive and other drives work fine with the VMs.

I've not heard from Egan.  I don't find board documentation such as BoM and Gerbers on the AD website.

Temperature control and compensation apparently get much harder going from 10 ppm to 1 ppm.  I'll put together a BoM and pricing for the parts in the net few days.  @cellularmitosis is sending me a bare PCB for an LM399 reference.  I'll also dig through the Metrology section threads.

[svenskelektronik]

Great update. I support your suggestion to start with a 10 ppm meter first, that will help alot in every step in the project.

[rhb]

I've pointed @cellularmitosis at this thread.  He's a major contributor to the Metrolgy group. He's sending me an LM399 PCB.  He also recently bought an AD5781 chip.

He also sent me this, though I've not had time to check into it.

I like your idea of the price-conscious calibrator.  In fact, Ian Johnson (a member of the EEVblog forum) sells exactly such a device (LM399, but the DAC might be a different part from TI), but it is about $400.  However, he has gone very in-depth with the design.  Definitely worth looking into.

(he initially started off the design with a different reference, so some of this is out of date)

http://www.ianjohnston.com/index.php/onlineshop/handheld-precision-digital-voltage-source-v2-detail

http://www.ianjohnston.com/index.php/videos/20-video-blog-022-handheld-precision-digital-voltage-source

$400 seems quite reasonable for a packaged product.

Another item to consider  is a good quality diode noise source with a calibration curve stored in flash.  The eBay BG7TBL unit is not very flat and has a lot of coherent noise below 2 MHz.

http://www.vhfcomm.co.uk/Noise%20diode.pdf

https://www.rf-microwave.com/app/resources/uploads/diodes/VHFComm_NW303.pdf

One could use hydraulic forming to make an enclosure very cheaply from aluminum sheet.

Commercial units are quite expensive.

[svenskelektronik]

Have you heard anything more form Cellularmitosis, do you think he is interested to contribute in this projet?

Regarding enclosure it may be best to start out with a mouldless process to cut some start costs. Sheet metal cutting + bending would be a good option in this case.
I made a simple model of the enclusure so you get the idea. The enclosure consist of two parts which are bended 2 times each and then put together. Opening for the display, buttons and connectors will be added when the PCB is done, its really cheap to punch out these features. The sourcing price of the completed enclosure in anodized aluminium would be around 5$ each if ordering more than 100 units.

[rhb]

I have heard more from @celluarmitosis both PM and in the thread I started in Metroogy. 

If you read the 2nd of the articles you'll see that box is much too big to give good performance.  Typically the commercial units are milled from round bar.  The cavity housing the circuit needs to be small so that resonances are above the frequency range of interest.  Have a look here:

https://www.eevblog.com/forum/metrology/diode-noise-sources-and-calibration-club/

I wrote some more about the concept I have in mind.  There's also a link to an affordable series of noise sources I had not heard of.  I'd want to significantly outperform those.  I'm not interested in competing with those.

I've got a pretty complete metal shop, lathe, mill, 3 in 1 shear, press brake and slip roll, a heavy press brake die for use in my 20 ton press, welding equipment, etc.  the only thing missing is proper HVAC  and power and organization.

[svenskelektronik]

I replied your PM.

[svenskelektronik]

Hi.

The Voltage Reference Project is now being discussed in this thread:
https://www.eevblog.com/forum/metrology/a-low-cost-oshw-voltage-calibration-reference-project/

Please feel free to keep the topic alive by suggesting your ideas and OSHW projects.

[rhb]

I should like to add that we have here a person who is in a position to make things happen.  I suggested a diode noise source and then when @cellularmitosis sent me 3 LM399 voltage reference kits,  I suggested doing something along that line as I now had prototype hardware in hand.

I've gotten a bit of flack in the metrology group for what most view as overly ambitious goals.  I don't disagree except that I have enough experience with sparse L1 pursuits to think it might make a significant difference.  I think it's worth seeing how good a standard can be made at the inflection point before it starts getting expensive.

I think the test for a good OSHW project is that it fill a need that is not being met at a price that the majority of those who need it can afford.

So I'd like to ask a direct question.  What do *you* think would be useful to a significant number of enthusiastic amateurs whether their interest is electronics or other things.

[b_force]

The key to 'success' is actually make a finished product.
I like the whole idea of open source products, but most of them are just simply not finished products (very far from it)
To many (major) bugs, incompatibilities, horrendous interface and practically no service.

You basically are being send into the woods with half finished gear and good luck with it.
I think the mean reason is that most open source projects are being created by tech heads and engineers and mostly only serve a very specific niche.

How you organize your code, bug tracking, meetings etc etc are all just very minor details and formalities.
Unfortunately these things are also what people tend to focus on the most.
I have tried collaborating in a few projects, but quit for exactly these reasons.
A lot of times people also don't have time enough or aren't dedicated enough to really finish an idea.

[svenskelektronik]

What you say is very true for any free/open source project. One thing that helps to predict this behaviour is to understand the motivation behind each contributor.