Electronics > Projects, Designs, and Technical Stuff

Oscilloscope ASIC Kickstarter instead of Open scope.

<< < (27/37) > >>

Mechatrommer:
there are top down type design, and there are called bottom up design. what i think nctnico trying to say is we start with FPGA design, let it receive some input, store it somewhere, do the triggering and output some data, and see how fast can it go... but then (to my knowledge) we also will need either to simulate the IO, or provide the suitable device to send and receive the data from the "FW". assuming the FPGA now can get 20Gbps throughput, then what? sooner or later the HW will expand to the real thing. ok maybe the GSps ADC can be replaced with Gbps FPGA to output some simulated or random data to the main FW/FPGA, and then stored in array of RAM etc, and output the 500KWfm/s data to the display. this we can call a demo platform to show that it can be done, that is phase one. costing of the complete system BOM can comes later in phase two,  but it should be realistic depending on how cheap they can be mass ordered. now we can share knowledges if someone have a chap or factory next door that can source the parts at lowest price, and the logistics and all. then goes to the Kickstarter.. but i bet, the high end GHz GSps spec is not going to be cheap, maybe say 1-5K$, depending on how you see it, some people may say its better than 45K$ of mainstream brands... without the base (demo) platform to show, we can talk empty wind forever.

nctnico:

--- Quote from: imo on January 14, 2020, 09:28:52 am ---
--- Quote ---It really doesn't matter at this stage. First get software going and then look at which platforms it can actually run on.
--- End quote ---

Hmm, I doubt so. Nobody will start to mess with fw/sw development unless he/she/it sees the hw first. Unless you are getting $$ for such an effort you will not do..

--- End quote ---
That is why a Kickstarter campaign to pay for software and FPGA engineers makes so much sense. There is no use to develop hardware without software. There are plenty of FPGA development kits out there to do the software development on.


--- Quote from: Mechatrommer on January 14, 2020, 10:11:21 am ---there are top down type design, and there are called bottom up design. what i think nctnico trying to say is we start with FPGA design,

--- End quote ---
No. Do not start with the FPGA design. Start with the software. Design the user interface framework and determine requirements for software signal processing. From there you get to the stage where you can see what makes sense to put inside the FPGA and what to keep in software (likely this leads to a more or less modular approach where you can implement features in software first and then move to the FPGA for better performance).

In an oscilloscope everything hinges on how to present a signal to the user and what kind of operations may take place on the signal. The actual acquisition part is 1% of the work. Remember that most DSO manufacturers went through a complete rewrite of their software at some point.

Mechatrommer:

--- Quote from: nctnico on January 14, 2020, 10:21:07 am ---No. Do not start with the FPGA design. Start with the software. Design the user interface framework and determine requirements for software signal processing.

--- End quote ---
a person with good enough knowledge of overall system (or several people expert on each area) can do this on drawing table at $0 cost. SW guy with his part and FPGA guy with his part. its when they start the work is where the $ (workhour) is.


--- Quote from: nctnico on January 14, 2020, 10:21:07 am ---The actual acquisition part is 1% of the work

--- End quote ---
you said that maybe because you are the HW guy. maybe its 1% the work, but 99% the cost. some hobbiests can treat their free part time as $0 developing SW, but not with buying parts for R&D. let alone tools to verify them.

nctnico:

--- Quote from: Mechatrommer on January 14, 2020, 11:03:38 am ---
--- Quote from: nctnico on January 14, 2020, 10:21:07 am ---No. Do not start with the FPGA design. Start with the software. Design the user interface framework and determine requirements for software signal processing.

--- End quote ---
a person with good enough knowledge of overall system (or several people expert on each area) can do this on drawing table at $0 cost. SW guy with his part and FPGA guy with his part. its when they start the work is where the $ (workhour) is.


--- Quote from: nctnico on January 14, 2020, 10:21:07 am ---The actual acquisition part is 1% of the work

--- End quote ---
you said that maybe because you are the HW guy. maybe its 1% the work, but 99% the cost. some hobbiests can treat their free part time as $0 developing SW, but not with buying parts for R&D. let alone tools to verify them.

--- End quote ---
It is not about cost. Developing oscilloscope software is far beyond hobbyist level. I estimate it will take a few experienced software engineers 3000 to 5000 hours to get a reasonable basic oscilloscope firmware done. So basically it will take a US $500k Kickstarter campaign to get somewhere. IMHO the idea that the software development for such a huge project like oscilloscope firmware can be done for free is wrong. If you look at open source projects then you'll see most of the contributions are made by companies.

OwO:
The hardware, FPGA firmware, and software must all be designed together because they are tightly inter-related. If this is to be a USB based scope the device must do most of the work compressing the waveform into a bar graph for display. No, the FPGA is NOT an accelerator. The FPGA handles data too fast to be ever presented to the system bus. It hosts a pipeline that must be designed a certain way for data to not be lost and for triggers to work reliably. You can not design it from a software perspective and then "accelerate" the slow parts later.

I can tackle most of the software + firmware work myself (my current roles are firstly RF and FPGA engineer, followed by software engineer). If the ADCs were available for cheap, I see no problem designing the product end-to-end with a UI that has the basic functionality. People (including on this forum) already go great lengths to develop improved software when they see attractive hardware, so it would work even better if there was officially supported open source software to begin with. The precondition of course is that the hardware is of sufficiently good bang for buck for it to be considered worth it.

Navigation

[0] Message Index

[#] Next page

[*] Previous page

There was an error while thanking
Thanking...
Go to full version
Powered by SMFPacks Advanced Attachments Uploader Mod