I know how a 3 Transistor CMOS camera pixel works. As I understand, 4T CMOS pixels have one additional transistor that connects the photodiode to a so-called floating node.

What is that floating node and what is it good for? Seems like a storage for photo-generated electrons, but I thought CMOS pixels give you voltage, not charges (light CCD's...).

Maybe it's good for multi-integration to reduce readout noise (think I was reading that somewhere)?

Any help is welcome, including good hints for further reading on the principle of camera pixels.

  • \$\begingroup\$ Could you update your question with schematics of both versions for comparison? \$\endgroup\$
    – Theodore
    May 19, 2021 at 21:13
  • \$\begingroup\$ @Theodore: In <ieeexplore.ieee.org/document/6742594> Fig. 3 the circuit of a 4T pixel is shown. \$\endgroup\$ May 20, 2021 at 8:02

1 Answer 1


A charge transmission phenomenon is not a feature that differentiates CCD and APS devices. A photodetector used in CCD and 4T APS image sensors is the pinned photodiode (PPD), and in PPD, incident photons do generate charge in the pixel photodiode that is later converted to a voltage signal.

In 3T APS cell the photodiode is not pinned and connected directly to the output buffer (source follower). The Wikipedia article explains the advantages of photodiode pinning:

The pinned photodiode, when coupled with the transfer gate, allows complete charge transfer from the pinned photodiode to the floating diffusion (which is further connected to the gate of the read-out transistor) eliminating lag. The use of intrapixel charge transfer can offer lower noise by enabling the use of correlated double sampling (CDS).

In a 4T APS cell, the charge from photodiode is transmitted, via a TG transistor, to the source follower transistor's gate and only there the charge is "converted" to the cell's output voltage \$V_{out}\$ to be read by the readout electronics when the cell's row and column are selected. The transfer physics of carriers from the PPD is similar to CCD charge transfer (page 38; 6/11 in pdf doc).

Consequently, the further improvements of fixed-pattern noise (FPN) figures are connected mostly with PPD design solutions, as a fabrication of a p+ layer surrounding the shallow trench isolation sidewalls in order to isolate n-type buried signal charge storage well ("floating node"). You can read about the development of PPD structures that resulted in low noise, high quantum efficiency and low dark current image sensors, both CCD and APS, in the IEEE Journal article A Review of the Pinned Photodiode for CCD and CMOS Image Sensors by Eric Fossum, a creator of APS device.

  • \$\begingroup\$ <ieeexplore.ieee.org/document/6742594> shines a lot of light of my problem, thanks! However, the FD's purpose is still not really clear for me since I was reading it's a storage used in multi-integration. Is that true, or does it have any other purpose? \$\endgroup\$ May 20, 2021 at 7:44
  • \$\begingroup\$ To ask more precise: Fig. 4 in <ieeexplore.ieee.org/document/6742594>: Photo-electrons are produced at the P-N-junction and stored in the N layer (SW) until transfered to FD. This would allow many integrations of the PPD and subsequent transfer to FD accumulating electrons until one final read-out via source follower is done. Is that the purpose of FD? \$\endgroup\$ May 20, 2021 at 8:00
  • \$\begingroup\$ 1) Eric Fossum's article explains the floating diffusion and what purpose it serves at length, for example, in Section D, page 35 and elsewhere. I won't do it any better. 2) What do you mean, "multi-integration"? you better explain it at length or provide the reference. \$\endgroup\$
    – V.V.T
    May 20, 2021 at 11:57
  • \$\begingroup\$ .: The article decribes that all electrons are transfered to the FD (wasn't possible without PPD). But why is it an advantage to have the source follower connected to the FD instead of a photodiode? (I'm a physicist, not an electrical engineer... maybe the FD is a common concept for you, but not for me. To explain in your words would definitely help). Thanks a lot! \$\endgroup\$ May 20, 2021 at 13:30
  • 1
    \$\begingroup\$ With the direct contact between the PD and SF gate that 3T APS uses to convert the charges into output voltage, you acquire an additional capacitance and kTC noise (en.wikipedia.org/wiki/…) of the pn junction into the bargain. \$\endgroup\$
    – V.V.T
    May 21, 2021 at 0:29

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.