home-built DSO (digital oscilloscope) ideas

[3v0]

Beebop what do you not like about the bitscope?

Bill, a while back you showed the Unicorn working as a scope. The fellow who authored the code was going to make some mods for the Unicorn. Where is that at?

 

[BeeBop]

Beebop what do you not like about the bitscope?

I've had myriad problems with the PC side of it; crashes, driver corruption, etc.
Also, perhaps due more to my own lack, than the instrument, I've never been able to use both channels at the same time, for, say, phase angle measurement.

I guess if it was all I had, I'd figure out this, and live with updating the software until it finally works well, but I have this old analog scope.

I much prefer analog, any way. I have used Tek's digital scopes, and they are very nice indeed, however when using them, never relied on the menu readouts, as they diverged with the reading using the cursors.

 

[petrv]

Yes, I am aware of them (bitscope) .... but they are expensive. The addition of logic analyzer is nice but I already have one.

Unicorn - sure you can make a "toy" scope with built-in A/D converter in the PIC, or better one when using external ADC connected to one of the ports, but still with sampling speed limited by the speed of the MCU (and also available RAM in the micro but the sampling speed is more limiting)

Petr

 

[BeeBop]

Yes, I am aware of them (bitscope) .... but they are expensive. The addition of logic analyzer is nice but I already have one.

Unicorn - sure you can make a "toy" scope with built-in A/D converter in the PIC, or better one when using external ADC connected to one of the ports, but still with sampling speed limited by the speed of the MCU (and also available RAM in the micro but the sampling speed is more limiting)

Petr

I WASN'T suggesting you buy one! I posted the link, because there is information there which could help you to do what you want. Or is what you are looking for, a complete project, already done for you, and you just take the credit for a job well done?

I think that the closest you will get to a complete project is what they are offering on Bitscope's site.

 

[petrv]

What are you talking about Beebop, complete project to take over ? Where did you get the idea ?

I have already a design in my mind, the only thing I am not sure about is the analog part (front-end) the digital part is clear to me. I will use Xilinx FPGA (not Lattice like Bitscope) simply because I have already all the development tools for Xilinx and also have some experience with Xilinx FPGA.

A.F.E -> ADC -> FPGA -> MCU with USB (PIC) -> connection to PC

I am going to write a design in VHDL for the FPGA and run simulation to see how fast I can sample with the Xilinx XC3S100E FPGA.

Petr

 

[BeeBop]

Oh, OK, sorry, misunderstanding.

I thought, that even though they are using Lattice, and you are using Xilinx, they are both Gate Arrays, and you could get some hints there.

I'm not very experienced with gate arrays, but can't you also implement a micro controller in your FPGA?

You said:

I was thinking about using even faster ADCs - they are available from National or other companies but the challenge is to read the data fast enough from those chips and store them in RAM...

There are much faster micros than PIC. How about an ARM core?

 

[Papabravo]

Buying a display ? Dedicated PC ? Of course not, nothing of that. I think (almost) everyone today has a PC and no need to have a dedicated one and no need to have a display if you have your PC -> display cost equals 0. For example I have bought an USB logic analyzer for a fraction of the price of a standalone logic analyzer (Intronix) and I don't need a dedicated PC (which you would argue bring the price to the same level as the standalone L.A.)

Anyway I started the thread to discuss technical ideas how to build such an device and not if or why yes or no.

And even if you would be correct with the price, we do things for fun,
this is our hobby and not a business.

Petr

You are of course entitled to your opinions and predelictions when it comes to spending your money. If your time is worth nothing (ie there is no opportunity cost to pursuing this project) then by all means knock yourself out. As I see it your problems are going to be the analog front end, the triggering circuit, and the delayed sweep. What were you planning to use for probes? Are they going to be compensated? Are they going to be x1 or x10? How will you couple them to the scope? What kind of user interface did you have in mind for setting the sensitivity of the vertical amplifier, and the sampling rate?

I think it is an ambitious project, and I wish you luck.

 

[petrv]

reading 60 MSPS directly to a micro would require as a minimum 60 MIPS and that is only in the case the MCU will read port in every instruction and also store it into memory, so you need to do with only *one* instruction
- read the port
- store the value in RAM
- increment the address pointer
- repeat until loop counter is 0
if you can do this with 2 instructions you need 120 MIPS and so on ... not easy even with more advanced MCUs.

I could do a MCU in the FPGA but for several reasons it is more practical to use external one.

My first attempt on synthesizing the configuration for the FPGA looks promising (results from the synthesis - Xilinx ISE):

   Minimum period: 3.934ns (Maximum Frequency: 254.223MHz)
   Minimum input arrival time before clock: 4.734ns
   Maximum output required time after clock: 4.519ns

 

[3v0]

Unicorn - sure you can make a "toy" scope with built-in A/D converter in the PIC, or better one when using external ADC connected to one of the ports, but still with sampling speed limited by the speed of the MCU (and also available RAM in the micro but the sampling speed is more limiting)
Petr

What a grouch! I never suggested that you should go with one.

For most of what my students do a toy that worked would be enough.

 

[Papabravo]

You might be able to make this project happen, but I don't think you can convince anybody that you can do this on the cheap.

 

[petrv]

As you think wrong, I don't need to convince you but I can convince many people - many test and development equipment is overpriced and is possible to build for much less. For example original Microchip ICD 2 sells for $160 but clones are available for less than $50. Same for ATMEL JTAG ICE. Or another example - Scotty's Spectrum Analyzer project had design goal to cost less than $200. Try to find a RF spectrum analyzer working till 1 GHz for that price ...

And last thing: your posts are not constructive and off-topic here. If you wish to talk about economic aspects of home-built equipment, please start your own thread.

Petr

 

[Papabravo]

petrv

I challenge you to substantiate your boasts about the availability of low priced test equipment with concrete examples. I leave it to the other members to judge the validity of your claims. Did Scotty succeed in making a spectrum analyzer for under $200. What are the performance characteristics? Specifically I think you are after a DSO for under $100.00. Do I have that correct? My claim is that it can't be done.

I disagree with your characterization of my posts. They are germaine and relevant to the question that you posed in a public forum. Until and unless you become a moderator on this forum it is not for you to decide anything about them. You are free to ignore them as are all of the other members. If you don't like that situation, then by all means take your complaint to the moderators. They have a pretty good handle on what is and is not appropriate and I trust their judgement. I believe that you are obligated to treat all members including myself in a more respectful way.

 

[petrv]

Possible design upgrade: I have constrained my FPGA design to 100 MHz clock and no problem at all, I believe I can get even higher speed - so I think I may upgrade the ADC to ADC08100 (100 MSPS or even) or even ADC08200 (200 MSPS) they are still reasonably priced, also the FPGA can be upgraded to XC3S250E (same pinout and 3 times bigger storage memory).

Update: Even 200 MHz passed ok but that is really close to the limit of the capabilities of the Spartan 3E FPGA. But at least I know I have the possibility to go until 200 MSPS

AFE: National suggest using their high-speed op-amp LMH6702 (there is even circuit showing how to iterface it to the ADC), I wonder if anyone here has experience with high-speed op-amps like this one - any suggestion for an alternate device for the A.F.E. ?

Thanks,

Petr

 

[Papabravo]

The supply rails are limited to ±6.75 volts, so to have useful ranges you're going to need some attenuators in front of this amplifier. With such a wideband amplifier you're going to need very careful layout to control the impedance and eliminate noise on the input.

The CMRR and PSRR are unremarkable at about 50 dB nominal. That's only a factor of 300 or so. That means 8 bits will be OK, but you can forget about 10 or 12 bit conversions.

The working gain should be somewhere between 1 and 2 to avoid excessive attenuation above 150 MHz and to avoid the peaking at a gain of +1 if you are going to use the non-inverting configuration.

I would paln on using the SOT23-5 package to minimize the effect of bond wires.

I don't think this amplifier is a great choice, but it may work for you since you really haven't defined your requirements yet.