Author Topic: What type of interface is used in PC-based scopes to communicate with ACQ board?  (Read 5035 times)

0 Members and 1 Guest are viewing this topic.

Offline MykeilTopic starter

  • Contributor
  • Posts: 26
  • Country: pl
Hi,

I am looking for information how many lanes is used in pc-based scopes to communicate with acquisition board and what is the type of interface.
Is it PCI or PCIe? 1.0, 2.0, 3.0? Lanes x1 x2 x8 x16?

I am looking for high data transfer rate scope to my multi-channel measuring system. I think the fastest will be, which uses PCIe with more lanes. The data will be read on Labview installed on the scope, so no physical medium will be used to transfer the data.

If you have pc-based scope please put information what is the version of PCI/PCIe used to communicate with acquisition board and how many lanes.

Mostly I am interested in LeCroy scopes:
- LeCroy WaveSurfer Xs-A
- LeCroy WaveSurfer 400
- LeCroy WaveRunner 6000/6000A
- LeCroy SDA 3000-6000
- LeCroy WaveMaster 8000A



What I know:
- Tektronix TDS5000: PCI 2.0
- Tektronix TDS7000: PCI 2.0
- Agilent 5483x: PCI 2.0
- LeCroy DDA 3000-5000: PCI
- LeCroy WavePro 7000A: PCI
- LeCroy WaveSurfer Xs: PCI

Thank you. Have a nice day,
Mykeil
« Last Edit: April 22, 2017, 04:13:23 pm by Mykeil »
 

Offline MykeilTopic starter

  • Contributor
  • Posts: 26
  • Country: pl
I am looking for cheap Windows-based oscilloscope which uses PCI Express interface to communicate its PC motherboard with acquisition board.
Anyone with any?
« Last Edit: April 22, 2017, 04:58:27 pm by Mykeil »
 

Offline ebclr

  • Super Contributor
  • ***
  • Posts: 2328
  • Country: 00
With scopes with ethernet and Usb with lcd  priced at 299 USD, this can be considered legacy stuff, Usb scopes still there but old PCI exp cards hard to find in now days
 

Offline MykeilTopic starter

  • Contributor
  • Posts: 26
  • Country: pl
Thanks for reply.
PCIe digitizer cards are good option but not cheap. I am looking for standalone Windows-based oscilloscope which uses PCIe as interface to communicate its pc builded board and acquisition board.
« Last Edit: April 22, 2017, 05:48:22 pm by Mykeil »
 

Offline nctnico

  • Super Contributor
  • ***
  • Posts: 26755
  • Country: nl
    • NCT Developments
IMHO you are going at this the wrong way. The PCI version is not going to tell you anything about how fast the data transfer actually is. This depends greatly on the acquisition hardware, PCI bus utilisation and speed of the post processing software. If you want guaranteed speeds then you'll be better off with dedicated hardware. What are your requirements (number of channels, bandwidth, acquisition length, cycle time)? Perhaps some of us know a better way to achieve your goal.
There are small lies, big lies and then there is what is on the screen of your oscilloscope.
 

Offline MykeilTopic starter

  • Contributor
  • Posts: 26
  • Country: pl
I want to make multi-channel acquisition system for scanning voltage on pcbs. The amount of channels will depend how fast is ADC in the scope. I wan to have 10-100MS per channel. If the scope will have 5GS/s it will be 50-500 channels. I will use multiplexer. The sample rate is not a problem. The problem is with the record memory. If I want to record 500ms with 5GS/s sample rate I need 2.5 Giga points record. I will not have, so I am trying to solve this issue by continuously grabbing recorded data by Labview. Labview will be installed on oscilloscope, so there will be no physical medium between Labview and oscilloscope.

If the pc board will be too slow I can change it into new one or use PCIe link adapter to connect it to another more powerful motherboard, so it is not a problem. The problem is slow data rate between acquisition board and pc motherboard.

Thanks for reply.
« Last Edit: April 22, 2017, 06:26:12 pm by Mykeil »
 

Offline nctnico

  • Super Contributor
  • ***
  • Posts: 26755
  • Country: nl
    • NCT Developments
Using a multiplexer isn't going to work to slice 5Gs/s into seperate channels. Forget about that. What you need is a stack of cheap acquisition modules. The cheapest are the USB (maybe ethernet) ones like from Hantek and others. An oscilloscope like you listed before will cost at least 4000 euro in good condition used. Add to that the price of a multiplexer (which can't really do what you want) and you'll be in the >6000 euro ballpark.
« Last Edit: April 22, 2017, 06:37:09 pm by nctnico »
There are small lies, big lies and then there is what is on the screen of your oscilloscope.
 

Offline ebclr

  • Super Contributor
  • ***
  • Posts: 2328
  • Country: 00
 

Offline MykeilTopic starter

  • Contributor
  • Posts: 26
  • Country: pl
An oscilloscope like you listed before will cost at least 4000 euro in good condition used.
I am ready for that.

I did not found PCI/PXI  30-channels acquisition card with 100MS/s/ch sampling rate and 300MHz bandwidth.

http://www.ti.com/tool/adc12j2700evm
Nice piece of hardware. USB 3.0 output. It should give 5Gbit/s. It has 32GB SRAM and can store 2 Giga points of 16-bit records.
I found 4GS/s model http://www.ti.com/tool/adc12j4000evm
« Last Edit: April 22, 2017, 07:10:42 pm by Mykeil »
 

Offline abraxa

  • Frequent Contributor
  • **
  • Posts: 377
  • Country: de
  • Sigrok associate
Quote
I did not found PCI/PXI  30-channels acquisition card with 100MS/s/ch sampling rate and 300MHz bandwidth.
What's the point of 300 MHz bandwidth if you have 100 MS/s?

Anyway, to give you inspiration to look also in other directions:
http://tmi.yokogawa.com/ea/products/data-acquisition-equipment/high-speed-data-acquisition/sl1000-high-speed-data-acquisition-unit/

16 channels per unit at 100 MS/s, so up to 128 channels when synchronizing 8 units. Real-time streaming via optional gigabit ethernet or onto optional built-in harddrive, so may fit your bill if you can live with the 20 MHz bandwidth (which imo is reasonable for 100 MS/s).
 

Offline MykeilTopic starter

  • Contributor
  • Posts: 26
  • Country: pl
Maybe 100MS/s is not too much for 300MHz bw, but after a full scan I plan to make deeper scanning with higher sample rate and lower channels amount.
That is an advantage between idea of acquisition cards and one big adc.

For now the best is Texas Instruments evaluation board http://www.ti.com/tool/tsw14j56evm with http://www.ti.com/tool/adc12j4000evm. All for $3250.
It has 4GS/s, 4GB record memory and USB 3.0 for data transfer.

I need to look for PCIe digitizers for this job.

It don't have to have high data output transfer rate if it has about 4GB record memory.
« Last Edit: April 22, 2017, 09:42:43 pm by Mykeil »
 

Online T3sl4co1l

  • Super Contributor
  • ***
  • Posts: 21609
  • Country: us
  • Expert, Analog Electronics, PCB Layout, EMC
    • Seven Transistor Labs
Quote
I did not found PCI/PXI  30-channels acquisition card with 100MS/s/ch sampling rate and 300MHz bandwidth.
What's the point of 300 MHz bandwidth if you have 100 MS/s?

I've looked at very few signals that are non-periodic while demanding that much bandwidth.

Tim
Seven Transistor Labs, LLC
Electronic design, from concept to prototype.
Bringing a project to life?  Send me a message!
 

Offline MykeilTopic starter

  • Contributor
  • Posts: 26
  • Country: pl
Only one Windows-based scope uses PCIe to transfer data from acquisition board. It is LeCroy WaveSurfer Xs-A. It uses PCIe 1.0 x1 which means it can transfer data with 250MB/s speed and theoretically I can reach 250MS/s sample rate with unlimited memory. It is too low for my needs. This solution not satisfies my needs.

Now I am looking for the cheapest scope with Giga points record memory. Anyone?

I made a research and I think the cheapest is Agilent/Keysight DSO9104A. 1GHz bw, 40 GS/s (20GS/s/2ch) and 1Gpts record memory.
« Last Edit: April 23, 2017, 06:12:56 pm by Mykeil »
 

Offline MykeilTopic starter

  • Contributor
  • Posts: 26
  • Country: pl
https://www.xilinx.com/products/boards-and-kits/1-f31o3b.html

56 GSPS/channel
4 channel
24 GB of DDR-4 Memory (3 banks of 64-bit)
8-bit ADC at up to 56 GSPS
Xilinx UltraScale™ XCKU115 FPGA
« Last Edit: April 23, 2017, 07:37:09 pm by Mykeil »
 

Offline nctnico

  • Super Contributor
  • ***
  • Posts: 26755
  • Country: nl
    • NCT Developments
How do you expect to interleave the signals using a multiplexer?
There are small lies, big lies and then there is what is on the screen of your oscilloscope.
 

Offline MykeilTopic starter

  • Contributor
  • Posts: 26
  • Country: pl
How do you expect to interleave the signals using a multiplexer?
The plan is to control multiplexers by NI FPGA card. When all data will be captured it will be cut it into separate traces in Labview after measurements. I will not separate the data in real time.
« Last Edit: April 23, 2017, 08:12:50 pm by Mykeil »
 

Offline nctnico

  • Super Contributor
  • ***
  • Posts: 26755
  • Country: nl
    • NCT Developments
How do you expect to interleave the signals using a multiplexer?
The plan is to control multiplexers by NI FPGA card. When all data will be captured it will be cut it into separate traces in Labview after measurements. I will not separate the data in real time.
But how do you plan to correlate the data? With a multiplexer you'll get a piece of data from time X on channel 1, a piece of data from time X+1 on channel 2. Much like a video monitor multiplexer you sometimes see to watch multiple cameras on one monitor. You can see what happens everywhere but you can't see what happens at the same time.
There are small lies, big lies and then there is what is on the screen of your oscilloscope.
 

Offline MrW0lf

  • Frequent Contributor
  • **
  • Posts: 922
  • Country: ee
    • lab!fyi
Now I am looking for the cheapest scope with Giga points record memory. Anyone?

https://www.picotech.com/oscilloscope/6000/picoscope-6000-specifications
PicoScope 6404D
500MHz analog, 2GSa memory, 5GSa/s single ch, 1.25GSa/s 4ch.
€6,295
This is actual scope. Not RF ADC.
« Last Edit: April 23, 2017, 08:54:56 pm by MrW0lf »
 
The following users thanked this post: Mykeil

Offline MykeilTopic starter

  • Contributor
  • Posts: 26
  • Country: pl
But how do you plan to correlate the data? With a multiplexer you'll get a piece of data from time X on channel 1, a piece of data from time X+1 on channel 2. Much like a video monitor multiplexer you sometimes see to watch multiple cameras on one monitor. You can see what happens everywhere but you can't see what happens at the same time.
Samples will be in their original time moment. I will not correlate the data in the time and move samples.

Lets assume that I have 1GS/s scope and 3-channel multiplexer. I will see on recorded data:

1. ch. sample  |  mux switch to 2. ch.  |  2. ch. sample  |  mux switch to 3. ch  |  3. ch sample  |  mux switch to 1. ch.  | ...

0 - 1ns       : 1. channel sample
1ns - 10ns  : mux switch to 2. ch.
10ns - 11ns: 2. channel sample
11ns - 21ns: mux switch to 3. ch.
21ns - 22ns: 3. channel sample
22ns - 32ns: mux switch to 1. ch.

32ns - 33ns: 1. channel sample
33ns - 43ns  : mux switch to 2. ch.
43ns - 44ns: 2. channel sample
44ns - 54ns: mux switch to 3. ch.
54ns - 55ns: 3. channel sample
55ns - 65ns: mux switch to 1. ch.

65ns - 66ns: 1. channel sample
66ns - 76ns  : mux switch to 2. ch.
76ns - 77ns: 2. channel sample
77ns - 87ns: mux switch to 3. ch.
87ns - 88ns: 3. channel sample
88ns - 98ns: mux switch to 1. ch.

It is always 33ns sample rate.

I assumed 10 ns time for mux switching operation and stabilization, but I think it may be longer. I hope not. I could have higher sample rate if I would have faster multiplexer.

Now I have 30 MSamples 3-channel oscilloscope with accuracy in time 33ns.

I don't need to know did the scope recorded the sample. If 1GS/s is set I will switch the multiplexer after 1ns and that is all. If I would be not synchronized with adc clock I would increase this time to 2ns, but I will loose sample rate or just synchronize with adc.

So, I also need fast 1xN multiplexer with lot of channels :D
« Last Edit: April 23, 2017, 09:31:23 pm by Mykeil »
 

Offline nctnico

  • Super Contributor
  • ***
  • Posts: 26755
  • Country: nl
    • NCT Developments
This seems rather wasteful to me. For one of my customers I did an FPGA design for a distributed data acquisition system based on a time synchronisation system developed by CERN using similar acquisition modules like this one (100Ms/s 30MHz 14 bit):

http://www.ohwr.org/projects/fmc-adc-100m14b4cha/wiki

This module goes onto a base board which can be used standalone or in a PC.

The beauty of this system is that it is all open source/open hardware so you can build whatever you like with it. The time synchronisation system enables to synchronise two nodes in the ps range.
« Last Edit: April 23, 2017, 09:32:58 pm by nctnico »
There are small lies, big lies and then there is what is on the screen of your oscilloscope.
 
The following users thanked this post: Mykeil

Offline MykeilTopic starter

  • Contributor
  • Posts: 26
  • Country: pl
Did you had to use multiplexer?
 

Offline nctnico

  • Super Contributor
  • ***
  • Posts: 26755
  • Country: nl
    • NCT Developments
No, each signal has it's own input. Also nodes can be far away (10km) from eachother without losing time correlation.
There are small lies, big lies and then there is what is on the screen of your oscilloscope.
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf