AD 2 100MSPS Data Logger with Raspberry?

[pascaljackson]

Hi guys,

I would like to find out if it is possible to log an analog signal at 100MSPS for about 5 to 10 minutes? I red that the buffer of the AD2 is very limited so it wont safe a continuously file without any gaps. But this would be fundamental. If there are these gaps, is it possible to estimate the length of these gaps?

If this would work, what proper solutions are there to safe the data? I thought a Raspberry would be perfect for this. But I red that the AD2 has some problems with it. How would the software waveforms safe the data? Is it easy to export it to excel?

Would be great if you can help me.

 

Pascal

[av_disp]

Hello @pascaljackson,

>>I would like to find out if it is possible to log an analog signal at 100MSPS for about 5 to 10 minutes?

I am 99% sure that this is not achievable with AD2 or any other low priced device what I am aware of.

Let's see how big should be onboard buffer for storing the data for 5 minutes. You are talking about an analog signal I will assume a digital signal (i.e. each state can be described by 1 bit only, in case of an analog signal it will be far more than 1 bit (8 or more if we are talking about some precision). So, 100M * 5 * 60 =  30 000 Mb i.e. 30 Gb.

As far as I know, the best what you can find among $100-$400 devices (logic analyzers) is equipped with 2Gb RAM (Digital Discovery and such). And that's a lot of memory! Other LA which dont have big buffers rely on real time data streaming - you will not get the required speed from those devices (it has been already discussed on the forum in great details).

But may be you are trying to solve the problem not in the most optimal way? Do you want to explain why do you think you need to collect the data for such long time?

thanks.

 

 

[pascaljackson]

Hi @av_disp,

 

thanks for your replay.

Ok i see the problem. 

I'll get an analog volt signal from an AD8302 (phase detector). That one is build into an rc helicopter(lets say it is planed to be in one). So I am looking for a solution that can log the data for one flight and being small and light at the same time. I will check the math again if i really need the 100MSPS. 
But lets imagine I want to use the AD2 as a logger. And I want to use the 100MSPS. How long are the periods that it would log the data(or how long are the gaps;))? I guess then the raspberry would be a limitation in terms of transferring the data out of the buffer onto the sd card?    

[av_disp]

19 hours ago, pascaljackson said:

Hi @av_disp,

But lets imagine I want to use the AD2 as a logger. And I want to use the 100MSPS. How long are the periods that it would log the data(or how long are the gaps;))? I guess then the raspberry would be a limitation in terms of transferring the data out of the buffer onto the sd card?    

Regarding the speed of real time streaming, please check the following:

I ran some tests on different computers and published the results there.

Regarding using Pi - I don;t know if you know about the difference between write speed of  SD card / conventional HDD / SSD but if you think about real time streaming then you would be better off just writing your data on SSD.

[attila]

Hi @pascaljackson

To store samples at 100MHz you need 200-400MiBps (1-2 channels of 16bit/sample). In this case SD card is not an option you need DDR or USB3 and SSD.
For 5-10 minutes you also would need 60-240GiB of storage.

With AD2 the gap between capture depends on the processing that needs to be done on the data, minimum is about 10ms. On SD/MMC based systems this might be more.
With continuous recording you could get about 1-2MHz.

Unfortunately the AD2 is not working with RPi. See:

 

[pascaljackson]

Thank you guys. @attila @av_disp

This sounds not that promising.

But I checked my calculations. So now 5Mhz of samples would be great enough, or 2,5Mhz for less precision. 
So if I understood you right, I'll have a chain of data which is a specific period of time long and then it takes about 10ms to get the buffer cleared and start the data loggin again at 5Mhz? I would need the volt values in a format to bring it afterwards into excel. Would there be a time value to every volt value? 

Yes i red about the RPi problems. Since the RPi does not have any USB3.0, it wouldn't be usable at all.
So for the beginning I would forget about that and for the first step just test it with a pc. And if that could work I can search for another solution to safe the data mobile. Something like that maybe: http://www.hardkernel.com/main/products/prdt_info.php?g_code=G143452239825. It has a USB3.0 port. In combination with a fast USB Stick.

 

 

[attila]

Hi @pascaljackson

What I wanted to say is that you can take many acquisition at up to 100MHz with gaps of about 10ms or perform one long acquisition (record for minutes or hours) at 1-2MHz sample rate.
The acquisition or recording is taken at a constant sample rate. You get time (1/sample rate) and voltage units. The acquisitions don't have exact timestamp, just system level timing at ms precision.

The high speed communication would be needed between the ADC and storage, for storing large number of samples at high rate (100MSsec).
Like a device with large onboard DDR option or a device with USB3/PCI connection to PC to store the samples.