diff --git a/docs/advanced/elec_properties/dos.md b/docs/advanced/elec_properties/dos.md
index 02d17b106c..c89e9d47d8 100644
--- a/docs/advanced/elec_properties/dos.md
+++ b/docs/advanced/elec_properties/dos.md
@@ -33,11 +33,6 @@ pw_diag_thr 1.0e-7
 init_chg file
 out_dos 1
 dos_sigma 0.07
-
-#Parameters (Smearing)
-smearing_method gaussian
-smearing_sigma 0.02
-
 ```
 
 Some parameters in the INPUT file are explained:
diff --git a/docs/community/faq.md b/docs/community/faq.md
index c59ab84eb7..2109c5ee13 100644
--- a/docs/community/faq.md
+++ b/docs/community/faq.md
@@ -52,6 +52,11 @@ If you train the DeePKS model with force labels, then the DeePKS model can provi
 
 Run `/usr/bin/time -v mpirun -n 4 abacus`, and locate "Maximum resident set size" in the output log at the end. Please note that this value is the peak memory size of the main MPI process.
 
+**10. Why there are two sigma (smearing_sigma and dos_sigma) in some examples for dos calculation?**
+
+ The tag `smearing_sigma` is used for SCF calculation, and does not affect NSCF calculation. The tag `dos_smearing` is only used for plotting density of states, which does affect SCF or NSCF results. So `smearing_sigma` should not be set in dos calculation.
+
+
 [back to top](#frequently-asked-questions)
 
 ## Failed jobs
diff --git a/examples/dos/lcao_Si2/INPUT2 b/examples/dos/lcao_Si2/INPUT2
index aa3896d768..8f02b8355c 100644
--- a/examples/dos/lcao_Si2/INPUT2
+++ b/examples/dos/lcao_Si2/INPUT2
@@ -13,3 +13,4 @@ ecutwfc			50
 #Parameters (File)
 init_chg		file
 out_dos			1
+dos_sigma 0.07