Electronics > Projects, Designs, and Technical Stuff

Oscilloscope ASIC Kickstarter instead of Open scope.

<< < (5/37) > >>

KE5FX:
Yep, the RFSoC really does sound like the right way to go.   Analog Devices seems pretty upset about them, for some reason.   :popcorn:

excitedbox:
Please read thoroughly before posting it seems a lot of replies are based on misconceptions of what these asics do and what it is I am suggesting.

In Short. ASIC IS THE FRONT END. ADC, DSP, Memory Controller, Clock, Triggering, All those things. Then whoever wants to use it buys an ARM soc with Arm cores, LCD driver, Uart, IO, all those things you need for the displaying of the signals etc. for a few $ from ST and connects it to the memory.


Also keep in mind that I am not proposing designing or producing a single scope. I want to produce a cheaper solution to making scopes and it is a FACT that these asics do just that. If you are a manufacturer buying FPGAs and ADCs right now, you can just as well purchase an ASIC to combine those things. With free access to the basic framework for the software your development costs drop even more.

How much they sell the product for in the end is still up to them in the end. Any speculation of what is achievable price wise is just speculation and WAY to early to say what is or isn`t feasible beyond Keysight using the same chip in a $450 and using 4 of them in their $45k scope. If each company right now can afford to make their own ASIC and hoard it for themselves the economies are there even for low production. Then multiple companies using the same chip must be worth it even if it just gives them cheap access to the IP to produce the chips they need.

A 300mm wafer produces hundreds of CPUs and even 7nm costs 8k per wafer. Yes they make higher numbers but 7nm is also WAY more expensive than 130nm or 350nm.

This is not a problem of can it bring results it is a problem of can we get it financed. This levels the playing field because it removes the large upfront nonrecoverable expense and spreads it across multiple companies over time.


I think people misunderstood me on the price example. I meant if they all pooled that 1/3 of the cost of those projects and put it towards investing in the development of an asic you would have raised enough money for the development of the chip. Then the other 400-500 would have paid for a much better scope using that chip.

Nobody is gonna use our chip for a 100GHz scope. Be realistic. This doesn´t have to be an all or nothing thing. Just as there are budget CPUs and Server chips.

If we make a chip for up to 1Ghz and 4-5Gsps using interleaving on 1 channel that would bring the cost of the scopes used by over 50% of the market down to something affordable. These scopes right now are in the 40-50k range because of software licenses. If you use 4 chips you can have 4 channels with 4Gsps or you can use 1 chip and have 1Gsps.

A chip like that would bring a 1GHz 2 ch scope down a ton in price.

You would not want to put the processor in the asic. The asic would contain the analog front end and signal processing and memory interface. Everything beyond that can be handled by off the shelf ARM SOCs for pennies. Most of the features that make a scope desirable such as UI, LCD, USB interface, Ethernet, Wifi, Displaying a waveform or inputting values for a waveform to output, that can be done using a few ARM cores. The only part that needs to be on the asic is the Front-end because of the speed at which signals need to be captured, processed and stored. 

What makes it expensive right now is that there are only few analog FPGAs and those that do have some analog functions only have slow ADCs because they are not specialized for the test equipment market. That means you need a fast ADC and an FPGA to process the signal

Also keep in mind the concept of SOCs has only really taken off in the last few years. Before that parts count in electronics was much higher because the development cost was higher. The technology, ecosystems and way of doing things just hasn´t been around that long to say this is a missed opportunity because as we see the companies that are at the tip of the spear are now starting to implement this. I am sure that in a few years some company with more of a application focused product portfolio will make such a thing.

Piecing together IP from many companies has become much simpler and there are now basic building blocks that are combined just like we build a circuit on a PCB but smaller. Combining a DSP core, DDR3 controller, ADC, DAC and attenuation. etc. onto a single chip is much easier and the Open Core Project, and a few others make it possible for anyone to build a chip.

Keep in mind that Keysight and the crew design the chip from the ground up because they want to go for huge bandwidth and maximum savings an so on. We can use IP cores because we wouldn't need 100GHz signal capture. Nobody is gonna use our chip for a million dollar Oscilloscope.

If Feelelec can make a signal generator and sell it for under $100 why is it a $1200 upgrade for a 25Mhz gen on a Rigol scope? Obviously the cost is just an arbitrary number and if you can bring the parts cost down and provide an open source framework of tools each company can customize their UI and features to their hearts content and drop prices to a reasonable level.

There are millions of scopes sold world wide and it would be more if they were more affordable. Don´t kid yourself and believe that because of small production runs they can´t afford to make them cheaper. If this was the case they wouldn`t give you the Hardware for a $4000 scope for $999 and charge the rest in upgrade fees.


If the chip has a modular design meaning you can slice off a channel on the ADC etc. it should be possible to make a scope for 250 or 300 or 500 or 1000 by putting in a whole chip or half chip or multiple chips. Just like AMD has done with their chiplet architecture. The 3970x sits on the same module as the 3990x they just solder on 4 chiplets less.

Anyone that is interested in working on this send me a PM and lets see if we can at least do a market analysis to see if it is possible. I assume everyone has discord maybe we can move the discussion there. I made a discord server. OScopeAsic is the server name.  https://discord.gg/DVmNC3

My dad is a retired chip engineer (Zeng Labs, Original AMD Radeon GPU, Iphone 3 to 4 A processors at Apple, and SOCs at LG) and although he wont want to work on this full time he would be a valuable asset for any questions that might arise.



Silicon Wizard, I know there are budget scopes in that price range. I have a Hantek 5000 series myself. Those scopes could be even better if they used a custom asic. Right now they are buying ADCs that cost about as much as an ASIC would. The budget scopes are great and I would never have been able to afford a scope if it wasn`t for them but as soon as you go past what they offer you are stuck buying a scope that costs as much as a car.



My post has gotten long enough. I ask people to please read thoroughly before posting and informing themselves of what these chips do. Half my post is just correcting misconceptions. It doesn´t help you or the project to get off topic discussing things that are irrelevant.

excitedbox:
By the way the Oscilloscope market is about to reach $1.7 Billion in the next few years. So don`t tell me that the market size is to small and the numbers are not there. That is probably close to 1 million scopes per year being shipped by about 70 companies.

https://www.marketwatch.com/press-release/oscilloscope-market-size-to-cross-1720-million-by-2024-2019-11-01

daqq:

--- Quote ---ASIC IS THE FRONT END. ADC, DSP, Memory Controller, Clock, Triggering, All those things.
--- End quote ---
My guess: there's no way you'd fill the budget for a competitive IC that could handle all of that, that includes:
- paying a team that is able to pull it off for a reasonable time frame
- NRE for prototypes and initial batch
- development and validation
- IP (you either license a load of IP or develop everything from scratch, ADC included in such a way that it does not step on any patents)
- tools
from kickstarter, given that the repayment is pretty much a thank you and the possibility of cheaper scopes in the distant future. The kinds of cash you'd need is in the serious venture capitalist area. Good luck and all, but I'm skeptical.

--- Quote ---I assume everyone has discord maybe we can move the discussion there. I made a discord server.
--- End quote ---
Why? Aside from the fact that not everyone has discord, the forum is a much better platform for this sort of discussion.

tautech:

--- Quote from: excitedbox on January 09, 2020, 08:06:28 pm ---
.......................

If we make a chip for up to 1Ghz and 4-5Gsps using interleaving on 1 channel that would bring the cost of the scopes used by over 50% of the market down to something affordable.

................

--- End quote ---
Happening already:
https://siglentna.com/product/sds5104x-1-ghz-4-ch/


--- Quote ---These scopes right now are in the 40-50k range because of software licenses.
--- End quote ---
Rubbish, may a few are but the marketplace is rapidly changing.  :popcorn:

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