-
#25 Reply
Posted by
JPortici
on 08 Mar, 2017 07:30
-
I think for a bit more money and most likely not that much time to invest, it's possible to build the analog frontend out of ICs.
When looking at all the teardowns of new scopes, it is noteworthy that many manufacturers go the extra mile and build at least part of their frontends out of discrete components to save on the components.
ehr... i believe it's few component -> less noise / more bandwidth
-
#26 Reply
Posted by
paulcav
on 08 Mar, 2017 20:38
-
-
#27 Reply
Posted by
paulcav
on 09 Mar, 2017 03:18
-
I found this DIY DSO here
http://www.jyetech.com/Products/LcdScope/e150.phpthat may be of interest
DSO Shell has similar bandwitch as DSO138 though it contains significant improvements in structure, mechanical, and operation.
Display and MCU are now mounted on the same board (mainboard) to avoid using inter-board pin-headers. Feedback from DSO138 usres have shown that soldering on the pin-headers are often problematic for many beginners. Analog channel is placed on a separated board which contains most user install parts. This brings in better separation between analog and digital circuits. Without the analog portion the mainboard is actually a general purpose control board that can be used in many other applications. The mainboard is fully installed (except a few through-hold components) and factory tested before packing.
And a viddeo of someone building it
-
#28 Reply
Posted by
CCitizenTO
on 09 Mar, 2017 16:45
-
I think the biggest hurdle in building an open source oscilloscope is running afoul of patents and patent lawyers at big name test gear manufacturers like Rigol and Tektronics. When you go from a one off build to production you start cutting into their marketshare and profits that's when you may start getting legalese nastygrams sent your way even if you just offered it up in kit form.
The other aspect that I see in most of Dave's teardowns is that to get high bandwidth oscilloscopes requires a good number of ADCs because it seems like most ADCs he finds in these oscilloscopes are 20-100MHz and they use like 5 or 6 of them to process the required bandwidth. I suspect they use the most reliable and cheapest chips they can source which is why on a 100MHz bandwidth scope you'll find 5x 20MHz chips as opposed to a single 100MHz chip. From a system engineering standpoint it would make sense if you could use 5x $0.20 part as opposed to a $3 part you'd improve your profit margin by $2.
-
#29 Reply
Posted by
james_s
on 09 Mar, 2017 20:06
-
I don't see an open source scope being much of a threat to the big companies, although lawyers might see things differently. A company looking to purchase a DSO is not likely to even consider a hobbyist grade open source/DIY model. I still think that with current tech anyway no open source scope is going to come close to the price to performance ratio of something like the Rigols.
-
#30 Reply
Posted by
paulcav
on 10 Mar, 2017 00:16
-
I think this is an interesting topic however, i think we should define what is the goal of the project. Should it be a low cost item, what are the parameters such as speed, number of channels, screen size, screen resolution, touch screen, number of settings and many many more.
With regards to the legality side, if it was made as a NFP or even sponsored by a tech company then it would be considered legal and above board.
Consider this if we would get sponsorship from Rigol and/or Tektronics or some other brand that could assist in some way then they could guide us away from the patented issues. They could provide support in finding or suggesting parts, testing solutions, issues to avoid that they have encountered, we could all learn from the big guys.
-
#31 Reply
Posted by
CCitizenTO
on 10 Mar, 2017 01:04
-
I think this is an interesting topic however, i think we should define what is the goal of the project. Should it be a low cost item, what are the parameters such as speed, number of channels, screen size, screen resolution, touch screen, number of settings and many many more.
I think low cost is an important parameter if you are making an open source/hobby level oscilloscope. Or you could offer a range of options like the big guys offer.
Speed: I'm guessing you are referencing the bandwidth. 20MHz would be the bare minimum acceptable for hobby electronics use. 100MHz would be better if it doesn't cost much more.
Channels: Needs at least 2 channels, ideally 4 channels.
Screen Size: Likely the single most expensive component so cost is likely the biggest consideration in choice. A 5-7" LCD display would be like $50-100
Touch Screen: Not necessary, likely increases expense when you can just use knobs and buttons for adjusting settings for cheaper than buying a touchscreen.
-
#32 Reply
Posted by
james_s
on 10 Mar, 2017 01:43
-
Screens are dirt cheap, you can get 7" TFT monitors for under $30. Yeah they're not super high res but I bet the panel would be adequate for a scope. The most expensive part by far will be the capture hardware, ADC followed by the front end. I'd say 50MHz is about the minimum threshold between useful scope and toy, but yeah 100MHz would be better.
-
#33 Reply
Posted by
ebclr
on 10 Mar, 2017 02:08
-
-
#34 Reply
Posted by
Andrey_irk
on 20 Mar, 2017 08:46
-
There is another problem nobody noticed: US has export restrictions on some high speed ADCs. For example, you can't buy 8-bit 1GS/s ADC if it has more than 1 channel. And there are different speed limits for each resolution. The more resolution is the lower speed you can get.
I can be wrong, but I think Red Pitaya chose 125MHz 14-bit ADC for their board because it was the fastest 14-bit ADC they could get without licence and they didn't want to use multiple chips.
-
#35 Reply
Posted by
ebclr
on 20 Mar, 2017 08:51
-
This export restriction apply only to what they call "hostile destinations", and is easily worked out with a form fill supposing you are not on the target restrict area, also everything is easily available in China, where the unique restriction is to have money.
-
#36 Reply
Posted by
paulcav
on 20 Mar, 2017 08:58
-
-
#37 Reply
Posted by
Andrey_irk
on 21 Mar, 2017 00:41
-
This export restriction apply only to what they call "hostile destinations", and is easily worked out with a form fill supposing you are not on the target restrict area, also everything is easily available in China, where the unique restriction is to have money.
Maybe, but here in Russia, for example, its not that easy. And it was so even before sanctions. I was talking about buying a 250MS/s 14bit ADC with a local Analog distributor about a year ago and realized that it was possible to buy them one way or another if you were a company, but for a hobbyist it wasn't worth it.
So, you can't use such stuff in a real open source project because people won't be able to build the thing.
-
#38 Reply
Posted by
Andrey_irk
on 21 Mar, 2017 01:01
-
There are some chip manufactures like TI will sell 14 bit ADCs for a small fortune
http://www.ti.com/product/ADC32RF45/samplebuy
Then if you actually try to order samples of the "forbidden" stuff they will say that it is impossible to ship them to you.
-
#39 Reply
Posted by
amspire
on 21 Mar, 2017 02:33
-
There are a few types of specialized scopes that I would be interested. One is a sampling scope, with perhaps a 10GHz analog bandwidth. The hardware for this can be very simple and cheap except for the sampling head and the trigger circuit. The A/D can be something like a cheap 1MS/s 16 bit - speed doesn't matter. One of the idea's I saw somewhere was a 50 ohm transmission line on the sampling board with, say, 20 sets of sampling diodes about every cm or so. All the diodes are hit with the sampling pulse at exactly the same time (pulse clock in the centre, transmission line circling around it so that all sampling pulse lines exactly the same length), so you end up with a sample consisting of 20 realtime samples with a 50ps spacing. You can still use it as a sampling scope, but you also have enough consecutive realtime samples to see an edge or a one time glitch. Wouldn't have a clue about doing this myself, but it would be great if a GHz guru did something brilliant.
The second one would be something extremely cheap, low featured box (probably pretty slow speed) but with very smart triggering, big memory and continuous capturing ability. Something you could perhaps leave connected to the mains or anything you want to monitor and it can be set to catch any anomaly. Something with a parts cost of $10 to $40 that you control via PC, phone, etc.
-
#40 Reply
Posted by
daybyter
on 21 Mar, 2017 03:56
-
-
#41 Reply
Posted by
Andrey_irk
on 21 Mar, 2017 04:25
-
The A/D can be something like a cheap 1MS/s 16 bit - speed doesn't matter. One of the idea's I saw somewhere was a 50 ohm transmission line on the sampling board with, say, 20 sets of sampling diodes about every cm or so. All the diodes are hit with the sampling pulse at exactly the same time (pulse clock in the centre, transmission line circling around it so that all sampling pulse lines exactly the same length), so you end up with a sample consisting of 20 realtime samples with a 50ps spacing. You can still use it as a sampling scope, but you also have enough consecutive realtime samples to see an edge or a one time glitch. Wouldn't have a clue about doing this myself, but it would be great if a GHz guru did something brilliant.
This is interesting. So the trick here seems to be in building these sample-and-hold circuits. And you'll need very high speed diodes or something like that.
-
#42 Reply
Posted by
amspire
on 21 Mar, 2017 05:00
-
The A/D can be something like a cheap 1MS/s 16 bit - speed doesn't matter. One of the idea's I saw somewhere was a 50 ohm transmission line on the sampling board with, say, 20 sets of sampling diodes about every cm or so....
This is interesting. So the trick here seems to be in building these sample-and-hold circuits. And you'll need very high speed diodes or something like that.
I did mean 20 sets of sampling diodes - one every cm.
The sample and hold circuits are usually made from 2 or 4 high speed diodes.
There was an article that came out about 2000 with a simple scope circuit:
http://www.redrok.com/sampscope.htmIt has some nice ideas - like using the input capacitance of a common low frequency fet opamp as the first stage of a sub nanosecond aperture sample and hold.
-
#43 Reply
Posted by
ebclr
on 21 Mar, 2017 08:33
-
Andrey_irk
Forget USA and Europe if you live in Russia
China, is Putins best friend, there is your place to buy, Near everything real and Fake can be found in China
www.taobao.comwww.aliespress.com
-
#44 Reply
Posted by
ohdsp
on 22 Mar, 2017 23:28
-
That STM32 $10 scope looks interesting.
Considering a faster design; if you could use some cheap fast 8 bit ADCs and interleave them with some clever manipulation on an FPGA interface that would seem the cheapest (although more complex to develop) option. If you can do this then maybe you could do things such as using some ADCs to do a rough scan of the waveform based on time base, but then use the extra ADCs to look at a much smaller ranges by adjusting reference voltages/sampling times based on the active user display.
If possible it would remove the need for an expensive single fast/high resolution ADC and allow for some neat features. Have found myself many times on scopes with the megazoom type functions where I can see the overall waveform but with not enough resolution to see the detail in the section I want; and where you can't set the trigger to capture that smaller section of data without seeing all the waveform. Would be nice if you could capture a waveform and then just point the scope to a timebase position and tell it "I want more detail here please!".
I like crazy ideas....
-
#45 Reply
Posted by
daybyter
on 23 Mar, 2017 01:19
-
I think the 10$ scope needs a simple ui. Then overclock the mcu maybe.
Another option could be one of those 13$ fpga boards + some old pc ram dimm?
-
#46 Reply
Posted by
james_s
on 23 Mar, 2017 15:56
-
The problem with the $13 FPGA boards (which I'm a big fan of) is the number of IO pins. Once you start trying to connect parallel memory you quickly go through a lot of IO pins. IIRC there are about 75 general purpose pins and that has to cover everything. Memory, display, ADC, user input, etc.
-
#47 Reply
Posted by
Andrey_irk
on 24 Mar, 2017 05:49
-
Andrey_irk
Forget USA and Europe if you live in Russia
China, is Putins best friend, there is your place to buy, Near everything real and Fake can be found in China
www.taobao.com
www.aliespress.com
I was talking about availability of high speed ADCs for hobbyists around the world. Russia was just an example.
And I can't imagine a hobbyist buying 400$ worth IC at Aliexpress or Taobao.
-
#48 Reply
Posted by
ebclr
on 24 Mar, 2017 06:29
-
In fact you can buy even a million USD of components specially from taobao, As many you know , more you know what you don't know
-
#49 Reply
Posted by
qu1ck
on 30 Apr, 2017 00:55
-
This guy collected lots of opensource(ish) scope projects, starting from dumbest bare mcu variants to advanced ones with decent sampling rate external ADC, FPGA, waveform memory and the whole shebang.
http://stm32duino.com/viewtopic.php?f=45&t=1710I particularly like the STM32F303 variants since it has better built-in ADC than most mainstream MCUs.
Check out GFXscope. The firmware for it isn't free though.
Also FusionScope, with free firmware. Author of that one is working on new version with external ADC and opamps.
An example of more advanced device is Neil Scope 3. 20 Mhz analog frontend, 100 msps ADC
http://radiokot.ru/circuit/digital/measure/107/Gotta learn Russian or suffer through google translate though