Broadcom's SiPM data sheets detail key performance parameters of their silicon photomultipliers - a key aspect being the linearity of response to increasing signal and the point at which the SiPM becomes saturated and cannot deliver more signal regardless of the number of incident photons.
The main parameters influencing linearity and saturation are the number of available SPADs; the recovery time of those SPADs and the Photon Detection Efficiency.
Pulse length is also an important issue and this is divided into two cases:-
In the case of very fast pulses whose duration is significantly shorter than the recovery time the primary limit will be the number of microcells - as the light level approaches and then exceeds the number of available detection sites (SPADs) so non-linearity emerges. Note however that this model assumes a uniform distribution of photons over the surface of the SiPM, one per SPAD. However this idealised beam profile is rarely seen in practice and so non-linearity occurs at lower signal levels when two or more photons are incident on the same SPAD with only a single charge released.
For relatively long pulses, say 1µs or longer, it is also important that the cells recover as fast as possible in order to be able to respond to the next incident photon. As the number of incident photons approaches ~70% of the saturation level the SiPM begins to demonstrate non-linearity as photons are detected by partially recharged SPADs. These partially charged SPADs will therefore release a lower charge per photon and as with fast pulses beam uniformity and the distribution of incident photons also has an effect.
The easiest attribute to understand is the number of SPADs - in the case of Broadcom's NUV-MT technology with a standard 40µm SPAD pitch a larger area equates to more SPADs.
The plot shows ABBR-S4N44P014M with 3.72 x 3.72mm² active area and 8622 SPADs together with the AFBR-S4N66P014M with an active area 6.14 x 6.14mm² and 22428 SPADs.
As expected the maximum number of detected photons for the two SiPMs is very close to the number of SPADs but it can be observed that whilst both exhibit linear behaviour at lower light levels (inset graph) the smaller part starts to show non-linearity at lower incident light levels.
As mentioned above the PDE affects the linear range as higher PDE means more of the incident photons will be detected. With this in mind users might opt for a lower Overbias to reduce PDE and increase Dynamic Range. Bear in mind that this approach is not always advisable since it can result in inferior accuracy of energy measurements due to increased statistical fluctuations. It is necessary for users to evaluate and optimise device performance for their specific application.
Broadcom have produced a number of helpful application notes including SiPM Dynamic Range, Linearity and Saturation which discusses the issues highlighted in this article. This and other application notes can be accessed below
 |
Broadcom's SiPM Family overview |
 |
Product Brief, NUV-MT SiPM range |
|
A Brief Introduction to SiPMs |
|
Working with Broadcom SiPMs |
|
SiPM Dynamic Range, Linearity and Saturation |
|
NUV-MT Performance Correlation |
|
NUV-MT Single-Photon Measurements |
|
SiPM Characteristics for PMT Users |
For more detail please visit the Support pages using the links in this article and on the right hand side of this page.
