Cool Android Oscilloscope Project

[kudzo]

Not sure if any one has already posted anything about it here, could not find it via the search.. just stumbled upon a very cool project:

http://www.osciprime.com/

here is a short description from their webpage:
"OsciPrime is an Open Source Android high speed data acquisition hardware platform and Android oscilloscope application. All layers of the project - from hardware, to firmware, up to software - are well documented in great detail."

[mtdoc]

Um - 6 Msps  and 3.3 - 8 MHz bandwidth?  I don't think so.

$ 249 Swiss Francs?  Fail.

[BravoV]

Um - 6 Msps  and 3.3 - 8 MHz bandwidth?  I don't think so.

$ 249 Swiss Francs?  Fail.

+1 , don't know why all these many recent toy scope makers look like never aware of those cheap Chinese DSO price ? 

[mtdoc]

Ya gotta love their promotional picture:  Ooo! Ooo! look. - It has "analog switches" and a "dual op amp!"  even optional LEDs!  What wonders!

I guess there's one born every minute.....

[Fsck]

Love how they even flaunt the "best bachelor thesis of the technical faculty" thing for this project..
In the report, they even say "the usability of Android in an industrial environment has been demonstrated".


this is less than a toy at 6Msamples, more like a novelty item - like an abacus.

[taemun]

Sick of seeing "open source" slapped on something, as if that somehow makes it better engineered.

That said, the layout looks mostly ok, but they clearly weren't ambitious enough - particularly at that pricepoint. It costs about as much as a DS1052E.

Not sure why it needs a TO-220 3.3V LDO.

I'd have to guess the 2x 6MSPS limitation is probably for them shepherding the data through the Java, and onto the screen at reasonable speed.

Then again, in the market of "Android-specific Scopes" this is probably the best one - the only competition is this homebrew Bluetooth unit from 2010:
http://projectproto.blogspot.com.au/2010/09/android-bluetooth-oscilloscope.html

[Pasky]

These are the kind of things that are great for school projects, but commercially stupid.

[andreas.rudolf]

Hi

I am one of the creators of the OsciPrime, I have received an e-mail that somebody created a post about it, so thanks

We definitely understand that the price of 249 CHF is rather high, however you would be surprised if you knew how low our margin per item sold is. We started this project in 2010 as our EE Bachelor's thesis, and at that time we would have never thought that we would eventually make a product of it. We have continuously posted our progress to our blog (http://android.serverbox.ch for those interested). It was only after many requests from people around the world interested in re-building/purchasing this oscilloscope that we have taken the step to make a product of it. The simple truth is, we can not produce in the thousands and therefore we can not compete with low-price hardware. We still wanted to give those interested in the project a possibility to get the hardware though.

What we are very proud of is that this product is open-source from top to bottom. Does open source make a product more interesting? Decide for yourself. It is also built completely from scratch. We did not use any system-on-chip or "analog" building blocks, instead the whole analog and digital front-end is created using only basic components. Could you build a cheaper/better oscilloscope with a ready-to-use SoC? Probably, but that was never our goal during the initial creation of this project. We wanted to build something self-contained and also share our experience with everybody (at that time there was almost no information available online about the internals of an oscilloscope for instance). Also, the Android software is freely available on www.osciprime.com (or for a small fee in the Google Play Store), and it also works for audio signals through the microphone input of your tablet/smartphone.

However, this product is by no means a toy. Of course, if you compare it to a real-size desktop oscilloscope, then it is, but so is every other USB oscilloscope. We have used the OsciPrime productively ever since. It works very well for measuring signals for any micro controller projects you might have or to reverse-engineer how your workout device is controlled etc. Also, our customers have reported that they like to use it abroad when traveling by plane with only carry-on-luggage allowed. Its main purpose is that it is very mobile/portable, no external power is needed, since it gets powered via USB (therefore USB host mode is required on the tablet/smartphone).

About the specifications, I don't want to name any specific brands, but most portable devices have similar specs, also they are comparable in price or cost more, and they are certainly not open source. Some of the other manufacturers don't even mention the analog bandwidth in their spec sheet. Anyhow, the whole analog bandwidth specification is kind of pointless (that is for low-end devices, I am by no way talking about high-end products!). Why would you need a bandwidth of say 200Mhz if you only sample at 20Mhz? If you sample with 20Mhz, a signal with a frequency of around 1Mhz will be practically visible on the screen, that is, you would see 20 samples per signal period. By the way, the OsciPrime may sample continuously with a frequency of up to 24Mhz but it does not work fluently on all devices, so we did not promote it. (You can be sure that other manufactures would have put this on their front page).

And yes, mtdoc, we love our promotional pictures :-D I think it gives a quick and nice overview not only of the product, but also on oscilloscope internals as well.

I hope this somehow debunks the image of us sitting in our golden chairs staring at our golden screens laughing about our stupid customers we continuously try to rip off. On the flip side, we try to be as transparent as possible. What you see is what you get

If you are really interested, I would be happy to answer any questions you might have. Though I can not promise I will reply immediately because it is already pretty late here in Switzerland.

@Fsck The thesis was called Using Android in Industrial Automation.. The main question was, whether Android could continuously process an incoming data stream of around 10MB per second. This has been achieved and no samples are lost. But of course, you could argue that you need to consider other factors as well.

@taemun Actually the 2x6 MSPS limitation is from USB 2.0.. even though USB 2.0 could theoretically support 480Mbit/s or 60MB/s, in reality most controllers only allow about 30MB/s, so we wanted to be on the save side. Some tablets like the Nexus 7 have no problem with single channel 24Mhz (or 24MB/s) sampling. The samples are processed in native C, hence why it is possible to "shepherd" at those speeds. The samples of interest are then transferred via java native interface.

Best Regards
andreas

[Fsck]

Picoscope has a 5GSa scope on USB3, and their USB2 scopes max at 1G if I remember correctly.

[andreas.rudolf]

Picoscope has a 5GSa scope on USB3, and their USB2 scopes max at 1G if I remember correctly.

I don't think it is fair to compare a 3300+ USD scope with a 280 USD one. Also the OsciPrime does not need a 12V external power supply but can run multiple hours being powered through USB by a tablet. Oh, and of course you can use the touch screen of the tablet ;-)

That being said, I think Picoscope have really great products.

[Fsck]

well, we could compare it to the 100MHz hacked DS1052E for basically the same cost.
or a 10MHz, 100M sampling Picoscope which costs 10$ CAD less.

[taemun]

About the specifications, I don't want to name any specific brands, but most portable devices have similar specs, also they are comparable in price or cost more, and they are certainly not open source. Some of the other manufacturers don't even mention the analog bandwidth in their spec sheet. Anyhow, the whole analog bandwidth specification is kind of pointless (that is for low-end devices, I am by no way talking about high-end products!). Why would you need a bandwidth of say 200Mhz if you only sample at 20Mhz? If you sample with 20Mhz, a signal with a frequency of around 1Mhz will be practically visible on the screen, that is, you would see 20 samples per signal period. By the way, the OsciPrime may sample continuously with a frequency of up to 24Mhz but it does not work fluently on all devices, so we did not promote it. (You can be sure that other manufactures would have put this on their front page).

...snip...

@taemun Actually the 2x6 MSPS limitation is from USB 2.0.. even though USB 2.0 could theoretically support 480Mbit/s or 60MB/s, in reality most controllers only allow about 30MB/s, so we wanted to be on the save side. Some tablets like the Nexus 7 have no problem with single channel 24Mhz (or 24MB/s) sampling. The samples are processed in native C, hence why it is possible to "shepherd" at those speeds. The samples of interest are then transferred via java native interface.

I'd rather that you actually stated the capabilities of your device, letting us know that if we had a better tablet/phone/whatever that we could get better performance. If you merely state the lowest common denominator, you're selling yourself short.

As for the 2x6MB/s USB2 limitation, you've just put the counterpoint to that, which is that you can push 24MB/s to a suitable endpoint. It isn't a USB2 problem, it's an tablet processing/USB host problem.

I'm glad that you've written proper NDK code for all that stuff

I suspect that you're making these in low quantity, which is why your costs are so high. If you did a production run of a thousand units, I'd think the sale price could be half what it is.

[andreas.rudolf]

well, we could compare it to the 100MHz hacked DS1052E for basically the same cost.
or a 10MHz, 100M sampling Picoscope which costs 10$ CAD less.

The DS1052E is a desktop oscilloscope, right? I would agree on the 100M Picoscope.. although you will also need a laptop/computer to operate it.
If you're looking for a nice portable alternative, I guess the DSO Quad is quite cool, unfortunately I don't own one though.

[BravoV]

Bitscope

[andreas.rudolf]

I'd rather that you actually stated the capabilities of your device, letting us know that if we had a better tablet/phone/whatever that we could get better performance. If you merely state the lowest common denominator, you're selling yourself short.

Hmm we never thought about it this way. I personally would probably be pissed if I read 24Mhz sample rate and then my tablet would only run reliably at 6Mhz. But I see what you mean. That being said, the hardware can support higher sample rates. The Android software has a "debug" tab, where you can already change the sampling rate configuration, however, this is rather experimental at the moment, and not yet seamlessly integrated into the application. Once this is implemented, we will update the spec sheets as well.

As for the 2x6MB/s USB2 limitation, you've just put the counterpoint to that, which is that you can push 24MB/s to a suitable endpoint. It isn't a USB2 problem, it's an tablet processing/USB host problem.

Exactly, some smartphones just don't have the processing power to handle this constant data throughput.
Then on the other hand, there are tablets which could handle the data throughput but are limited by the 30MB/s USB 2.0 throughput (we are pushing all samples continuously from the hardware to the Android device; another common approach is to temporarily store the samples on the hardware and then send only a sub-set of them, in which case USB throughput speed would not matter as much). This 30MB/s USB limitation is not only seen in tablets, we have also tried with a Cypress USB development board and a desktop Linux computer, no processing, just plain data shoveling. The maximum data throughput was around 33MB iirc. So it is most likely a combination of USB host controller and USB device controller of some sort, and it's possible that some are a bit faster/slower. Generally I am quite astonished that USB 2.0 can deliver half of its theoretical maximum speed. Just think of all the "300" or "450" Mbit/s Wireless N devices, in reality you rarely get more than 30Mbit/s :-)

I'm glad that you've written proper NDK code for all that stuff

Thanks :-D

I suspect that you're making these in low quantity, which is why your costs are so high. If you did a production run of a thousand units, I'd think the sale price could be half what it is.

Quite possibly, but a thousand+ units is quite an investment unfortunately.

[EEVblog]

Picoscope has a 5GSa scope on USB3, and their USB2 scopes max at 1G if I remember correctly.

That's a grossy unfair comparison.
This is a much better comparison:
https://www.kickstarter.com/projects/751733865/smartscope-reinventing-the-oscilloscope?ref=live

[EEVblog]

To me this isn't really an oscilloscope, because it doesn't have a proper vertical attenuation in the familiar scales.
It's more like the description on the page - "OsciPrime is an Open Source Android high speed data acquisition hardware platform"
It's essentially a real time USB data logger, which happens to have an oscilloscope-like application.