XMOSCOPE

[octal]

Version: 0.01
Status: Under development
License: LGPL

This project is aimed at creating a low cost Digital Storage Oscilloscope based on an XMOS device. The XMOS XS1-L1 or L2 chip will be used to drive a fast ADC and store sampled data to a fast memory.

All data will be uploaded later to a PC/MAC and analysed by a software (cross platform software written in C++/QT).
The main concern is to have a LOW COST and good performance oscilloscope. From an older experience I got with another chip, I think that 25MHz band widht could be achieved easily.

The total price of the whole made project should stay less than a total of 100 euros.

[Vaati]

I've been looking into building scopes with low-cost microcontrollers/microprocessors for a while. I am definitely going to bookmark this...

[TonyD]

Corin's NetScope is a great project to get ideas from.

Web Based Ethernet Oscilloscope
http://archive.xmoslinkers.org/node/284

Here's a video of it:

http://www.youtube.com/watch?v=6_rxEcsORyU

RussF has a similar idea perhaps you could combine projects
https://www.xcore.com/projects/xmos-web ... peanalyser

Another interesting open source scope to check out is the PropScope at www.parallax.com, it has a bandwidth of 25MHz.

[octal]

Hello,
thank you for the links. I already have visited those links and printed Corin's thesis ;)
Corin used the G4 chip mainly because the need for speed and more cores to handle all the high speed ethernet interface. The problem with G4 is that it exists only in BGA package, which makes it impossible to be soldered by hand and makes it a bit expensive to got it made at a factory in low quantities. I want the XMOSCOPE to be cheap and yet powerful tool.

I studied deeply PropScope (2 months ago), and I has been a lot inspired by it. While the analog part of XMOSCOPE can be based on the same chip as PropScope, on the digital part I dont like PropScope choices.
XMOS chip is really truly deterministic, and has by far better IO caracteristics than propeller. PropScope send live data throught USB to PC, this leads to a lot of limitations. What I want to do is to have a true DSO, a scope that has its own high speed memory to do sampling. yesturday I spent a lot of time looking at and comparing various memory products that can go under 10ns. I'll post some details and the project webpage here in few days.

[octal]

well,
I checked out various choices of fast ADC for sampling analog data. I need to check the datasheets deeply to do final choice. The choice will also depend upon the input scaling amplifier. The best (and easy to use) adc I found are:

ADC08060 60 MSPS
ADS830 60Msps
105Msps: LTC2282 (12-Bit), LTC2280 (10-Bit)
80Msps: LTC2294 (12-Bit), LTC2289 (10-Bit)
65Msps: LTC2293 (12-Bit), LTC2288 (10-Bit)
40Msps: LTC2292 (12-Bit), LTC2287 (10-Bit)
25Msps: LTC2291 (12-Bit), LTC2286 (10-Bit)
Maxim also seem to offer some nice ADC, but their sampling rate is either too slow, or they send data via a serial interface.
I'm still looking for a high speed memory (a 256 or 512KBytes). I cannot find one priced moderately (I need to keep this scope low price).

[octal]

Well, still working on the DSO.
For data storage, the best I could find at acceptable price are memories with 512KByte (4Mbit) at 10ns, from Cypress, ALLIANCE MEMORY INC, or Samsung.
CY7C1049DV33 Series 4-Mbit (512 K x 8) 3.3 V 10 ns Static RAM - TSOP II-44
AS7C34096A Series 4-Mbit (512 K x 8) 3.3 V 10 ns CMOS Static RAM - TSOP 11-44
K6R4008V1D (512 K x 8) 512Kx8 Bit High Speed Static RAM(3.3V Operating)

The main problem when working with such RAMs, is the parallel access. Parallel access means speed, but it means also more IO pins on the main chip that will control the RAM. To control such memories I need at least about 30 signals (19 for adr, 8 for data, 3 or 4 for control signals).
the XMOS device that will control this memory need also to have more io capabilities to control the ADC, and more IO pins to do eventual other kind of control operation and communication stuff with PC or other hardware.
I think that I'll need to switch to use XS1-L1 128TQFP version (68 IO instead of 36 on the 64TQFP).
Well, some users will for sure say that even with the 64TQFP version I'll still have enough IO to control either an IO expander or any other device. This can be considered, but I prefer the design to be ready for future evolutions, and sincerly the price between the two chips is not too high (less than adding an external additional device with all IO speed penalities of external IO expanders).
Second constraint that is driving me to go for a 128 pin version is the possible preparation of an XLINKed version of the DSO, that mean the possible creation of a DSO where users can stack modules to make it 2, 3, ..n input DSO, each input having its dedicated 512 (or more) KBytes of ram.
I'm also thinking at having this DSO eventually control its own GLCD display and thus having its own GUi instead of relaying on a PC. To control an external (good) display with touch capabilities we need more IO.

... stay in touch ... more on that in near future.

[JohnR]

In case you haven't chosen an amplifier for your scope, there is the National LMH6518 which is a DC-coupled digitally controlled 900 Mhz amplifier. In the Applications section, its use as an oscilloscope is mentioned.

I recently bought a couple of these devices for a different application, but have not yet had time to evaluate them.

JohnR

[octal]

Thank you John for pointing this out. I'll check the datasheet :)

[DrMario]

Neat. It will be even more useful to me. I have a serious need for the best oscilloscope as I work on everything - from repairs to hobbyist jobs.

And, you also should try to shield the amplifier circuitry and ADC, and use very good decoupling capacitors as every tiny details do really count. Amplifier sometimes act like an antenna reciever (one thing to watch out for).

And, be aware that even the blindingly-fast fancy ADC chip can be quite expensive, but will make up for extremely fast Analog to Digital conversion - very useful for 100 - 400MHz electrical/signal waveform scoping.

[octal]

Well, just an update.
Well, today I ordered most of the LOGIC part of the XMOSCOPE. I need to finish the main board havinng the XMOS chip + Fast ram + buffering/protection circuitery + FT2232 chip. I'll start work on the analog part of the project right after having this main part working.
I hope to have something working (on LOGIC part only) before the end of this february (I'll start making the PCB this weekend).
As you may have noted, I added an FT2232 chip to do communication with computer side. This chip does not add a lot to final cost, and have a lot of advatages:
- Speed (damn speed)
- No need to have our own USB VID/PID to sell final product
- No wast of precious MIPS (I'm using only L1 chip ;) )
- the drivers are already available for this chip to do USB com on most platforms (at least Win/Linux/Mac-OS/X)
- (don't know if even possible) I also want to check the possibility to make this onboard chip to act as a JTAG to do onboard debug/prog using the same platform (without the need for the JTAG2).

As you may notice, this LOGIC part of the circuitery will also let me do tests and will be the base for a dev board I want to prepare for XMOS devices. (more on that will be published in few days on the XDevTools group (don't miss that, subsribe and give your suggestions).