Author Topic: NI Vector Signal Transceiver  (Read 8506 times)

0 Members and 1 Guest are viewing this topic.

Offline MarsupilamiTopic starter

  • Frequent Contributor
  • **
  • Posts: 262
  • Country: us
NI Vector Signal Transceiver
« on: May 22, 2023, 01:07:27 pm »
UPDATE:
TLDR, it's kinda done.
https://hackaday.io/project/192236-industrial-6ghz-rf-analyzergenerator-in-a-desktop


------

Original post:


Hi Folks,

Pinky promise I'm not in any way affiliated, I'm just running a search once in a while on ebay for PXIe instruments for my collection, and I noticed that the running price for used PXIe-5644R VSTs dropped to around $350, which is insane. I'm guessing these are being phased out from certain manufacturing lines and hitting the used market in mass.
I was contemplating a while ago to make an adapter board and 3d print a bracket to be able to put this in a double 5.25" disk bay in a PC and wire it to a PCIe slot on the motherboard. Otherwise a used PXIe chassis / controller now costs more than the instrument itself.
Software out of the box is quite simplified, but if one doesn't mind a bit of coding this is a very capable piece of RF hardware for the price.


« Last Edit: October 01, 2023, 06:16:29 pm by Marsupilami »
 
The following users thanked this post: zrq

Offline rfclown

  • Frequent Contributor
  • **
  • Posts: 407
  • Country: us
Re: NI Vector Signal Transceiver
« Reply #1 on: May 22, 2023, 06:00:37 pm »
Thanks for sharing this. I'll have to look into this unit when I have time. I'm always interested in SDR transceiver hardware. Crazy what you can get for the money once something starts getting dumped on the used market. My first was the original Agilent VXI VSA hardware. Then it was the e4406A. Those are Rx only. Lately I've been using ADALM Plutos since they are Tx and Rx, and are wider bandwidth. The limitations of Pluto are frustrating, but it's hard to complain for the price.
 

Offline MarsupilamiTopic starter

  • Frequent Contributor
  • **
  • Posts: 262
  • Country: us
Re: NI Vector Signal Transceiver
« Reply #2 on: May 27, 2023, 06:14:19 am »
That's some good stuff. I use a e4440a at work now. I like that series.

Meanwhile I got very excited about putting a PXI VST into a regular PC and started a challenge project on it.



I might say not bad since Monday. I'm only 72% that it will work and even if it does there are a bunch of mechanical integration issues, but I'll deal with that if I get there.
 

Online RoGeorge

  • Super Contributor
  • ***
  • Posts: 6050
  • Country: ro
Re: NI Vector Signal Transceiver
« Reply #3 on: May 27, 2023, 08:51:38 am »
Might be cheaper in both time and money to look for a second hand PXI chassis, either NI or Agilent.  If you do so, make sure the chassis comes with a CPU and a power supply modules, or else you'll need to buy them later.  AFAIK you'll still need proprietary software to fully control the hardware (unless the chassis+control unit you bought came with software, too), but I've heard NI has now a free to use license for their LabVIEW.

Either DIY or ebay buy, I'm curious how the project will go, good luck with it.  :-+
« Last Edit: May 27, 2023, 08:58:35 am by RoGeorge »
 
The following users thanked this post: Marsupilami

Offline switchabl

  • Frequent Contributor
  • **
  • Posts: 389
  • Country: de
Re: NI Vector Signal Transceiver
« Reply #4 on: May 27, 2023, 11:14:37 am »
It is relatively easy to find a used PXI chassis and controller (or MXI interface). But PXIe stuff was still quite expensive last time I looked.
 
The following users thanked this post: Marsupilami

Offline MarsupilamiTopic starter

  • Frequent Contributor
  • **
  • Posts: 262
  • Country: us
Re: NI Vector Signal Transceiver
« Reply #5 on: May 27, 2023, 11:18:27 am »
I have all that stuff, this is not a practical matter. :D

It is possible to find chassis on ebay for around $500, like a PXIE-1062Q or 1071, but it's hit or miss, offered as parts since, I'm guessing, these sellers don't have reliable ways to test functionality. Tested, known good units are still above a thousand.
Controllers (CPU) are an even bigger problem. The cheapest solution is the PXIe-8379 thunderbolt remote controller. You can find these around $300. As far as I know these were never released fully to the public and they use Thunderbolt 2, which needs an adapter (Apple ~$50) to modern USB Type-C Thunderbolt 3 hosts and actually not many PCs work with them due to some legacy security feature. I bought an Intel NUC just for this purpose, and it sort of sees the device but not doing anything else. Out of 4 my old work Lenovo laptop was the only one that run just fine with it connected.
Embedded controllers are again go for above $1000 usually.

So if you add that together it's closer to 2 grand which is cost of doing business if the instrument itself is also 2 grand, but now you can buy those for 300 freaking dollars! And there are a bunch from multiple sellers.

The software is easy. The driver includes what NI calls "Soft Front Panels". These provide basic SA / SG functionality. The screenshot in my original post is from that.
Programming libraries are available freely for LabVIEW, CVI, Visual C\C++ and .NET.
The two things that are not possible without expensive, paid software is to have wireless standard measurements (e.g. LTE etc.) and to program the FPGA, because that needs a LabVIEW FPGA license. Latter one is a cool feature, but for general analyzer / generator use the default FPGA image has everything included.

What I'm trying to get to here is to have the board made cheap. Connectors are ~$20 apiece, so let's say total BOM cost is $100. Another $50-$100 is the PCIe extension cable, then the only other thing needed is a crappy old-ish PC to put this whole thing into. I'm planning on buying a refurbished Dell Optiplex for about $100, that has a full size 5.25" drive bay.

FYI: https://www.ni.com/en-us/support/documentation/supplemental/12/the-ni-vector-signal-transceiver-hardware-architecture.html

« Last Edit: May 28, 2023, 03:34:29 am by Marsupilami »
 
The following users thanked this post: switchabl

Offline zrq

  • Frequent Contributor
  • **
  • Posts: 259
  • Country: 00
Re: NI Vector Signal Transceiver
« Reply #6 on: May 29, 2023, 10:43:33 am »
This looks very cool. I didn't know it's possible to build a PXIe to PCIe bridge without a serious amount of hacking.

I'm very tempted to grab one and pray/prey for PXIe chassis later. Even plus the price of a new PXIe-1090 controller it makes a competitive offer as a modern RTSA.
 

Offline samofab

  • Regular Contributor
  • *
  • Posts: 148
  • Country: si
Re: NI Vector Signal Transceiver
« Reply #7 on: May 29, 2023, 04:15:24 pm »
Quote
Meanwhile I got very excited about putting a PXI VST into a regular PC and started a challenge project on it.

@Marsupilami please do share the design, even if not finished. It sounds like a great idea.
 

Online KE5FX

  • Super Contributor
  • ***
  • Posts: 1839
  • Country: us
    • KE5FX.COM
Re: NI Vector Signal Transceiver
« Reply #8 on: May 29, 2023, 04:25:46 pm »
Would be super interested in hearing how this goes as well.  Modular T&M gear is underrated.
 

Offline MarsupilamiTopic starter

  • Frequent Contributor
  • **
  • Posts: 262
  • Country: us
Re: NI Vector Signal Transceiver
« Reply #9 on: May 29, 2023, 08:43:03 pm »
Thanks guys, interest keeps me going. :D

I started a thread in the projects section, I think it belongs there more.
https://www.eevblog.com/forum/projects/pxie-instrument-in-a-regular-pc/
 
The following users thanked this post: switchabl

Online jhenderson0107

  • Regular Contributor
  • *
  • Posts: 133
  • Country: us
    • Elk Engineering
Re: NI Vector Signal Transceiver
« Reply #10 on: May 30, 2023, 03:26:50 pm »
I assembled a NI PXIe-5663E VSA and PXIe-5673E VSG piecemeal from eBay a few years ago.  They are installed in a PXIe-1078 chassis which includes a PXIe-8370 MXI adapter to allow a 15' cabled PCI Express connection to a desktop PC with a PCIe-8371 MXI adapter.  NI labels cabled PCI Express 'MXI'.  I almost exclusively use the soft front panel (SFP) turnkey software. 

Some positives:
  Full-featured and instantaneous remote control via mouse and keyboard
  SFP user interface very similar to stand-alone instruments
  GUI much more responsive than my vintage spectrum analyzer (E4440A)
  Very compact compared to stand-alone devices. 

Some negatives:
  Low frequency range restrictions
  SFP instruments can't be controlled using VISA.
  Performance is good, but not exceptional.

I am planning to sell it.  I prefer stand-alone instruments with Visa control. 
 
 

 
« Last Edit: May 30, 2023, 03:50:40 pm by jhenderson0107 »
 

Offline MarsupilamiTopic starter

  • Frequent Contributor
  • **
  • Posts: 262
  • Country: us
Re: NI Vector Signal Transceiver
« Reply #11 on: May 30, 2023, 05:01:36 pm »
Gathering that piece by piece sounds tedious. :D

I never really looked at these as proper replacement for boxed instruments. When doing measurements manually a physical UI will always beat the remote software imho. Also performance wise it's important to realize these devices are targeted for high-volume production test first and everything else second so their performance is tuned to that. The whole architecture excels at tasks where there's significant data to be moved and processed on the host PC.

All that being said 6GHz SA/SG combo for $350 makes it worth.
 

Online jhenderson0107

  • Regular Contributor
  • *
  • Posts: 133
  • Country: us
    • Elk Engineering
Re: NI Vector Signal Transceiver
« Reply #12 on: May 30, 2023, 05:42:39 pm »
I agree and wouldn't hesitate for $350.  I also have two PXIe-1073 chassis (see attachment) which are even smaller and would be well-suited to hosting this VST.  They connect via single-lane MXI adapters to a PC also.  For a virtual instrument, single-lane is fine. 
 

Offline MarsupilamiTopic starter

  • Frequent Contributor
  • **
  • Posts: 262
  • Country: us
Re: NI Vector Signal Transceiver
« Reply #13 on: May 31, 2023, 01:00:17 am »
I agree and wouldn't hesitate for $350.  I also have two PXIe-1073 chassis (see attachment) which are even smaller and would be well-suited to hosting this VST.  They connect via single-lane MXI adapters to a PC also.  For a virtual instrument, single-lane is fine.

Yep, that'd be a good setup. What's that module in your chassis? I can't see the model number.

I did a quick search to see how much the 1073s are going used, and ohh eBay... I don't think this was ever released publicly either:


How cool is that 2 in 1!
 

Online jhenderson0107

  • Regular Contributor
  • *
  • Posts: 133
  • Country: us
    • Elk Engineering
Re: NI Vector Signal Transceiver
« Reply #14 on: May 31, 2023, 02:36:12 pm »
...What's that module in your chassis? I can't see the model number...

That module is the PXI-4461 which is a 24-bit audio interface.  It's useful for accelerometer and audio testing. 

Regarding that 1078D chassis:  I have seen that before, but never in the wild.  Given its cost premium, I'd prefer to inject a decent 10 MHz reference clock onto the backplane via a PXI timing card. 

One cool thing about these NI chassis is that some slots support both PXI and PXIe connections ('H' -designated hybrid slots), so either type of 3U card can be used.  For example, the PXI-4461 is a legacy PXI card (not PCI Express).  It was cheaper and more than sufficient for audio transfer speeds. 
« Last Edit: May 31, 2023, 02:56:09 pm by jhenderson0107 »
 

Online jhenderson0107

  • Regular Contributor
  • *
  • Posts: 133
  • Country: us
    • Elk Engineering
Re: NI Vector Signal Transceiver
« Reply #15 on: June 04, 2023, 12:10:33 am »
...I agree and wouldn't hesitate for $350. 
I didn't hesitate and purchased one.  It runs the standard RF spectrum analyzer and RF signal generator virtual instruments.  Works well and is very space-efficient. 
 

Offline MarsupilamiTopic starter

  • Frequent Contributor
  • **
  • Posts: 262
  • Country: us
Re: NI Vector Signal Transceiver
« Reply #16 on: June 04, 2023, 04:05:19 am »
I didn't hesitate and purchased one.  It runs the standard RF spectrum analyzer and RF signal generator virtual instruments.  Works well and is very space-efficient.

Nice! :)
 

Offline MarsupilamiTopic starter

  • Frequent Contributor
  • **
  • Posts: 262
  • Country: us
Re: NI Vector Signal Transceiver
« Reply #17 on: June 27, 2023, 04:06:14 am »
I've been posting about my trepidations in the Projects forum, but now that it's working I'll throw it in here too:


 
The following users thanked this post: TiN, KE5FX

Offline MarsupilamiTopic starter

  • Frequent Contributor
  • **
  • Posts: 262
  • Country: us
Re: NI Vector Signal Transceiver
« Reply #18 on: October 01, 2023, 06:14:36 pm »
For anyone that might care this is as done as probably it ever will be.
PCB files and info on hackaday.

I also started writing GNU Radio blocks for it and got the analyzer part working, but I kinda gave up when I realized that a no double copy custom buffer implementation is beyond my time budget. It was still fun. 


 
The following users thanked this post: KE5FX, zrq

Online KE5FX

  • Super Contributor
  • ***
  • Posts: 1839
  • Country: us
    • KE5FX.COM
Re: NI Vector Signal Transceiver
« Reply #19 on: December 06, 2023, 01:22:46 am »
Does this look like the same Mini-SAS adapter you used? 

Edit: also, just to confirm, is the "VST PCIe MiniSAS 20230831A for hackaday.io.zip" file still the best version to send to JLCPCB?

I wrote a pretty handy console app to drive the bladeRF and VSG60A (which I keep meaning to release one of these days), and was thinking it might be nice to use it with these NI modules as well.
« Last Edit: December 06, 2023, 01:40:07 am by KE5FX »
 

Offline Minki

  • Contributor
  • Posts: 28
  • Country: jp
Re: NI Vector Signal Transceiver
« Reply #20 on: December 07, 2023, 07:25:02 am »
1946841-0
Try to start first step.....
Marsupilami would you like to share the housing step file...?I want  to 3D print a housing for it :-[

Thank You

I wrote a pretty handy console app to drive the bladeRF and VSG60A (which I keep meaning to release one of these days), and was thinking it might be nice to use it with these NI modules as well.
Sorry a off topic question that do you know how to turn bladeRF to a wide band spectrum...?Very little information in combine the many time FFT result to pot out full spectrum.
 

Online KE5FX

  • Super Contributor
  • ***
  • Posts: 1839
  • Country: us
    • KE5FX.COM
Re: NI Vector Signal Transceiver
« Reply #21 on: December 07, 2023, 02:06:29 pm »
(Attachment Link)
Try to start first step.....
Marsupilami would you like to share the housing step file...?I want  to 3D print a housing for it :-[

Thank You

I wrote a pretty handy console app to drive the bladeRF and VSG60A (which I keep meaning to release one of these days), and was thinking it might be nice to use it with these NI modules as well.
Sorry a off topic question that do you know how to turn bladeRF to a wide band spectrum...?Very little information in combine the many time FFT result to pot out full spectrum.

With receivers that rely on host-based signal processing, you're always going to be limited by how quickly the device can stream data to the PC over USB or Ethernet.  It really just comes down to setting the frequency at either a high level (using existing control software) or low level (writing your own programs), taking a snapshot, and stitching the transformed 'periodograms' together.  That can be done by concatenating FFT bins logically before plotting them, or by plotting them and combining the images.

 

Offline MarsupilamiTopic starter

  • Frequent Contributor
  • **
  • Posts: 262
  • Country: us
Re: NI Vector Signal Transceiver
« Reply #22 on: December 07, 2023, 04:25:00 pm »
Does this look like the same Mini-SAS adapter you used? 

Edit: also, just to confirm, is the "VST PCIe MiniSAS 20230831A for hackaday.io.zip" file still the best version to send to JLCPCB?

I wrote a pretty handy console app to drive the bladeRF and VSG60A (which I keep meaning to release one of these days), and was thinking it might be nice to use it with these NI modules as well.

I got this one, but it seems similar enough. (I had to patch a resistor on mine though, I don't know how others are configured but it all seems to be the same basic design.)
https://a.co/d/8VI1OF7

File sounds correct. I asked JLCPCB if they have a better format to share, but apparently they don't. I think PCBWAY let's you share a design on their website, JLC should have that function too.

Make sure you post if you get it to work!
 
The following users thanked this post: KE5FX

Offline MarsupilamiTopic starter

  • Frequent Contributor
  • **
  • Posts: 262
  • Country: us
Re: NI Vector Signal Transceiver
« Reply #23 on: December 07, 2023, 04:43:37 pm »
(Attachment Link)
Try to start first step.....
Marsupilami would you like to share the housing step file...?I want  to 3D print a housing for it :-[

Oh wow, look at those boards!  :-+

I decided not to upload the mechanical design to hackaday, as I used manufacturer models in there and I didn't want to figure out the licensing implications.
Also my design it not very useful if you're not using this PC case I believe, but I can send the files to you nevertheless.

The most important part I used is this: https://www.digikey.com/short/d5m828vt
(Please not that this cage doesn't have the front lip needed for the card ejector to work.)
You can also go to the Schroff website and download the solid model for it.
Unfortunately the card guides are a bigger problem. This would be the proper part: https://schroff.nvent.com/en-gb/products/enc64568-089
There are a few iterations of this from Schroff, but none of them are recently stocked at DigiKey or similar distributors.
I found a bag of these https://schroff.nvent.com/en-gb/products/enc64560-005 on ebay.
Simplifying and 3D printing these parts could be an option, but consider the heat generated by the device.



 

Offline Minki

  • Contributor
  • Posts: 28
  • Country: jp
Re: NI Vector Signal Transceiver
« Reply #24 on: December 08, 2023, 02:24:28 pm »
(Attachment Link)
Try to start first step.....
Marsupilami would you like to share the housing step file...?I want  to 3D print a housing for it :-[

Oh wow, look at those boards!  :-+

I decided not to upload the mechanical design to hackaday, as I used manufacturer models in there and I didn't want to figure out the licensing implications.
Also my design it not very useful if you're not using this PC case I believe, but I can send the files to you nevertheless.

The most important part I used is this: https://www.digikey.com/short/d5m828vt
(Please not that this cage doesn't have the front lip needed for the card ejector to work.)
You can also go to the Schroff website and download the solid model for it.
Unfortunately the card guides are a bigger problem. This would be the proper part: https://schroff.nvent.com/en-gb/products/enc64568-089
There are a few iterations of this from Schroff, but none of them are recently stocked at DigiKey or similar distributors.
I found a bag of these https://schroff.nvent.com/en-gb/products/enc64560-005 on ebay.
Simplifying and 3D printing these parts could be an option, but consider the heat generated by the device.
Thank for your share i will try to using PC to print the card guides.
Current the tricky part is this bar(?)
1948065-0
Since the plane will done by laser cut acrylic board
1948059-1
So no plane to buy the SUBRACK(For save some cost)
And how do i test the PCI-E to SFF-8643 card had the reset signal?Just probe the RST# pin when powwer on the PC?
And ...does those connector no need to solder?I see you didn't solder it in your photo.
1948077-2

Many Thank :-[ :-[ :-[ :-*


 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf