HomeSupportKnowledge Base → FAQ about MCS and Embedded

FAQ about MCS and Embedded vision



FAQ

Introduction


Q: Which XIMEA cameras are suitable for either Embedded vision or MCS?

There are several camera lines offered by XIMEA that are fit for use in various Multiple camera systems (MCS) or Embedded vision setups.
The advised camera versions to use would be a combination of xiX line and xPlatform.


Q: What are the different ways to connect the XIMEA embedded cameras?

The main option is to use FFC (flexible flat cables) and either connect it to the computer with the help of adapter or interconnect it with xSwitch which multiplexes many cameras and transfers the synchronized data to computer over single cable with minimal latency possible.
Another possibility is to utilize these cameras with various NVIDIA Jetson modules (TX2, Xavier AGX or Nano) through carrier board like xEC2. This allows to process the data with the help of GPU.


Latency


Q: What is the Glass to Glass latency of these cameras?

There is no simple way how to answer the question regarding latency, as the term is differently interpreted.
Overall, latency depends on many factors and occurs in different steps of image acquisition:

1. Time until exposure start:
The required time until exposure start highly depends on the trigger type being used.
For hardware triggering, the overall latency between trigger signal and exposure start is less than 10 us.
For software triggering, it highly depends on the operating system. Windows/Linux are no real-time systems.
That means you can never estimate or influence, when an operation is being executed.
Therefore, latency is (from experience) > 1ms for software triggering.

2. Exposure time:
Latency here depends on exposure time and mode of camera (free-run vs. triggered acquisition).
All parameters regarding exposure time can be set by using XIMEA API.

3. Readout and data transfer time:
The readout and data transfer time can be calculated by dividing the size of the image by the available transfer speed.
The latter depends on the technology being used.
PCIe Gen2x2 enables 10Gbit/s, Gen2x4 enables 20Gbit/s and PCIe Gen3x8 up to 64Gbit/s.

4. API:
The final latency depends on the functions and parameters being used in XIMEA API for image processing.
For instance, if you use de-bayering and white balance, the processing time increases.
This also depends on the hardware used, especially on the processor unit.

After step four, the image can be retrieved from XIMEA API and used for your application. More about it HERE.
In summary, PCIe and the xPlatform have high influence on step 3, readout and data transfer time.


Firefly connection version


Q: Can Firefly connector models be connected to PC?

While primarily these camera versions are meant to connect to the xSwitch platform, there is a way how to attach them directly to PC and it is described HERE


Q: Does the flex ribbon or Firefly cable transfer power?

Yes, flexible flat ribbon cables (FFC) and some Firefly cables do transfer the power and synchronization/triggering signals up until the adapter board where a standard GPIO is located.
This is the reason why most xiX models do not have a GPIO interface on the backside of the camera, except Firefly which looks like THIS.


Other


Q: Is it possible to sort multiple cameras in XIMEA software?

There is an option how to do it in simple way in XIMEA CamTool, described HERE.


Q: What are the combinations for Multi USB3 cameras interconnection with PC?

There are simple commercial setup options allowing to Embedded single or few XIMEA USB3 (or USB 2.0) cameras with a mini PC as well as more professional and complex ways to multiplex a high number of cameras with compact industrial PC platforms. Below are just several examples which were suggested in various situations:

Imago VisionBox:
AGE-X (tested: AGE-X10: 4 x USB3.0, Z77, USB3.0 interfaces share 5 GBit / s bandwidth)
Multi camera setup with 4x USB3 cameras (tested: AGE-X Tiger+)
Multi camera setup with 12x USB3 cameras (tested: AGE-X Serval+)
Pokini i / i2 (tested: 4 x MQ013xG-E2 at 60 fps)

Intel NUC

VIA EPIA-P910-10Q Pico-ITX