RE: Are USB3 Vision cameras any better than GigE Vision cameras?

[Mark Rivers via Tech-talk <tech-talk@aps.anl.gov>]

> Has the simple fact of using switches shown you a different behavior or is that really related to sharing the bandwidth (so that the camera has to compete for it)?

I have used both direct connect and switches, and I have never seen problems with disconnects or other instability like that with our cameras.  I have seen it with other devices like power supplies, Modbus pumps, etc.  Putting those on a dedicated NIC solved the problem.  I attribute this to some broadcast traffic causing problems with the TCP stack in those devices.

In general, my practice has been that if a camera is almost aways going to be viewed from a specific beamline workstation then I connect it via a dedicated NIC on that workstation.  The camera IOC runs on that workstation.  That way the traffic is generally never on the shared Ethernet at all.  If someone wants to view that camera from another machine then the IOC output images (typically using PVA), are transported across the network.  Those can be compressed or rate-limited in the PVA plugin.  If the camera is intended to be generally viewed from multiple workstations then generally the camera to IOC computer is via a switch.  In this case the IOC output images always go across the network.

Mark


-----Original Message-----
From: Henrique F. Simoes <henrique.simoes@lnls.br> 
Sent: Wednesday, August 27, 2025 12:29 PM
To: Mark Rivers <rivers@cars.uchicago.edu>
Cc: Ana Clara de Souza Oliveira <ana.clara@lnls.br>; tech-talk@aps.anl.gov
Subject: Re: Are USB3 Vision cameras any better than GigE Vision cameras?

Hi, Mark.

[CC'ing everybody on this one. Sorry for the duplicate message.]

On Mon Aug 25, 2025 at 12:46 PM -03, Mark Rivers wrote:
> Hi Henrique,
>
> I have been using both USB3 Vision and GigE Vision cameras for many 
> years.  In general, I have found both to be very stable, though GigE 
> Vision cameras sometimes have occasional dropped frames.

Interesting.

> I have used USB3 Vision for cameras that require greater bandwidth 
> than GigE can support, for example this Grasshopper3 camera
>
> https://urldefense.us/v3/__https://www.teledynevisionsolutions.com/products/grasshopper3-usb3/?mo__;!!G_uCfscf7eWS!ZrBMaOvTauGj1pV6BQ2dEqImS0LJnuSAW0cc4Td2HmLzNfUWuzS7cKbm2XJ5dvh3yLTWvMgFMxK7M6y2cP5XOB2ve_U$ 
> del=GS3-U3-23S6M-C&vertical=machine%20vision&segment=iis

Nice. That's really an advantage I didn't consider before.

>> To me, they have the drawback of requiring a computer right next to 
>> it,
>
> That is not really true.  I use this active USB3 extender cable, and 
> the distance between the camera and the computer can be more than 25 
> meters (20m for the extender cable plus 3m for the normal USB-3 
> cable).  The above camera runs at the full USB3 rate with no problems.
>
> https://urldefense.us/v3/__https://www.amazon.com/SIIG-Active-Repeater-20-Meters-JU-CB0811-S1/dp/__;!!G_uCfscf7eWS!ZrBMaOvTauGj1pV6BQ2dEqImS0LJnuSAW0cc4Td2HmLzNfUWuzS7cKbm2XJ5dvh3yLTWvMgFMxK7M6y2cP5XDnQfzR4$ 
> B0053YLYLC?th=1

Oh, sure. I should've been more explicit about what I meant by "right next to it". With USB cameras, we would probably need a computer per beamline. With Ethernet cameras, we are moving to setup where multiple cameras are controlled by a data-center server with a 100Gbps network card. This means possibly having a single or just a few more powerful machines for cameras from all beamlines, placed in a proper room (with restricted access, appropriate cooling, etc). Of course, we will be relying on multiple switches to get the traffic to the data center. As of now, a single beamline is using this setup, though.

Even for that beamline, I've heard at least once about disconnections as well, so this migration doesn't seem to solve anything in this regard.

> GigE cameras can have issues if they are sharing network bandwidth 
> with many devices, but in general I have found them to be quite 
> reliable.  You can eliminate virtually all issues by using a dedicated 
> NIC and direct connection to the camera.

That's something different from our general setup. I only know of a single FLIR camera using point-to-point link. The beamline staff reports they often need to power cycle it before using nevertheless.

Most of them have their traffic routed by the beamline switch to a rather weak computer with a 1Gpbs NIC, which can share the bandwidth with multiple cameras. However, users acquire with one camera at a time (as it is used for beam alignment) with one-eighth of the maximum frame rate (which turns out to deliver 100Mbps of data). With such a low bandwidth usage, I wouldn't expect this to be causing instabilities such as sudden disconnections, and so on.

Has the simple fact of using switches shown you a different behavior or is that really related to sharing the bandwidth (so that the camera has to compete for it)?

> That is less convenient than using switches, but it allows greater 
> distances than USB3.  I am running my tomography data collection with 
> a 10G Ethernet camera on a dedicated NIC and have no problems with 
> dropped packets or dropped frames.
>
> https://urldefense.us/v3/__https://www.teledynevisionsolutions.com/products/oryx-10gige/?model=OR__;!!G_uCfscf7eWS!ZrBMaOvTauGj1pV6BQ2dEqImS0LJnuSAW0cc4Td2HmLzNfUWuzS7cKbm2XJ5dvh3yLTWvMgFMxK7M6y2cP5XiF1rxKU$ 
> X-10GS-51S5M-C&vertical=machine%20vision&segment=iis

Thanks for sharing your experience.

--
Henrique F. Simões
Control Software Group (SwC)
Brazilian Synchrotron Light Laboratory (LNLS) Brazilian Center for Research in Energy and Materials (CNPEM)

Aviso Legal: Esta mensagem e seus anexos podem conter informações confidenciais e/ou de uso restrito. Observe atentamente seu conteúdo e considere eventual consulta ao remetente antes de copiá-la, divulgá-la ou distribuí-la. Se você recebeu esta mensagem por engano, por favor avise o remetente e apague-a imediatamente.

Disclaimer: This email and its attachments may contain confidential and/or privileged information. Observe its content carefully and consider possible querying to the sender before copying, disclosing or distributing it. If you have received this email by mistake, please notify the sender and delete it immediately.