diff --git a/libtascar/include/acousticmodel.h b/libtascar/include/acousticmodel.h
index 8aa2c3fe..e7753753 100644
--- a/libtascar/include/acousticmodel.h
+++ b/libtascar/include/acousticmodel.h
@@ -173,6 +173,7 @@ namespace TASCAR {
       uint32_t layers;
       float maxdist;
       float minlevel;
+      float nearfieldlimit;
       uint32_t sincorder;
       gainmodel_t gainmodel;
       bool airabsorption;
@@ -211,7 +212,7 @@ namespace TASCAR {
       void update_refpoint(const pos_t& psrc_physical,
                            const pos_t& psrc_virtual, pos_t& prel,
                            float& distance, float& gain, bool b_img,
-                           gainmodel_t gainmodel);
+                           gainmodel_t gainmodel, float& nearfieldlimit);
       void set_next_gain(float gain);
       void set_fade(float targetgain, float duration, float start = -1);
       void apply_gain();
diff --git a/libtascar/src/acousticmodel.cc b/libtascar/src/acousticmodel.cc
index 1fc871f3..52e621e1 100644
--- a/libtascar/src/acousticmodel.cc
+++ b/libtascar/src/acousticmodel.cc
@@ -120,7 +120,7 @@ acoustic_model_t::acoustic_model_t(float c,float fs,uint32_t chunksize,
   ismorder(getorder())
 {
   pos_t prel;
-  receiver_->update_refpoint(src_->position,src_->position,prel,distance,gain,false,src_->gainmodel);
+  receiver_->update_refpoint(src_->position,src_->position,prel,distance,gain,false,src_->gainmodel, src_->nearfieldlimit);
   gain = 1.0;
   vstate.resize(obstacles_.size());
   if(receiver_->layers & src_->layers)
@@ -196,7 +196,7 @@ uint32_t acoustic_model_t::process(const TASCAR::transport_t& tp)
           // end obstacles
           receiver_->update_refpoint(primary->position, position, prel,
                                      nextdistance, nextgain, ismorder > 0,
-                                     src_->gainmodel);
+                                     src_->gainmodel, src_->nearfieldlimit);
           if(nextdistance > src_->maxdist)
             return 0;
           nextgain *= srcgainmod;
@@ -474,7 +474,8 @@ diffuse_acoustic_model_t::diffuse_acoustic_model_t(float fs,uint32_t chunksize,d
   pos_t prel;
   float d(1.0);
   float gain;
-  receiver_->update_refpoint(src_->center,src_->center,prel,d,gain,false,GAIN_INVR);
+  float nflimit(0.1);
+  receiver_->update_refpoint(src_->center,src_->center,prel,d,gain,false,GAIN_INVR,nflimit);
 }
 
 diffuse_acoustic_model_t::~diffuse_acoustic_model_t()
@@ -491,9 +492,10 @@ uint32_t diffuse_acoustic_model_t::process(const TASCAR::transport_t&)
   pos_t prel;
   float d(0.0);
   float nextgain(1.0);
+  float nflimit(0.1);
   // calculate relative geometry between source and receiver:
   receiver_->update_refpoint(src_->center, src_->center, prel, d, nextgain,
-                             false, GAIN_INVR);
+                             false, GAIN_INVR, nflimit);
   shoebox_t box(*src_);
   // box.size = src_->size;
   box.center = pos_t();
@@ -692,7 +694,7 @@ void receiver_t::add_diffuse_sound_field_rec(const amb1wave_t& chunk, receivermo
 void receiver_t::update_refpoint(const pos_t& psrc_physical,
                                  const pos_t& psrc_virtual, pos_t& prel,
                                  float& distance, float& gain, bool b_img,
-                                 gainmodel_t gainmodel)
+                                 gainmodel_t gainmodel, float& nearfieldlimit)
 {
 
   if(volumetric.has_volume()) {
@@ -705,7 +707,7 @@ void receiver_t::update_refpoint(const pos_t& psrc_physical,
     float d(box.nextpoint(prel).normf());
     if(falloff > 0)
       gain = (0.5f + 0.5f * cosf(TASCAR_PIf * std::min(1.0f, d / falloff))) /
-             std::max(0.1f, avgdist);
+             std::max(nearfieldlimit, avgdist);
     else {
       switch(gainmodel) {
       case GAIN_INVR:
@@ -723,7 +725,7 @@ void receiver_t::update_refpoint(const pos_t& psrc_physical,
     distance = prel.normf();
     switch(gainmodel) {
     case GAIN_INVR:
-      gain = 1.0f / std::max(0.1f, distance);
+      gain = 1.0f / std::max(nearfieldlimit, distance);
       break;
     case GAIN_UNITY:
       gain = 1.0f;
@@ -856,7 +858,7 @@ source_t::source_t(tsccfg::node_t xmlsrc, const std::string& name,
                    const std::string& parentname)
     : sourcemod_t(xmlsrc), licensed_component_t(typeid(*this).name()),
       ismmin(0), ismmax(2147483647), layers(0xffffffff), maxdist(3700),
-      minlevel(0), sincorder(0), gainmodel(GAIN_INVR), airabsorption(true),
+      minlevel(0), nearfieldlimit(0.1), sincorder(0), gainmodel(GAIN_INVR), airabsorption(true),
       delayline(true), size(0), active(true),
       // is_prepared(false),
       plugins(xmlsrc, name, parentname)
@@ -866,6 +868,8 @@ source_t::source_t(tsccfg::node_t xmlsrc, const std::string& name,
       "physical size of sound source (effect depends on rendering method)");
   GET_ATTRIBUTE(maxdist, "m", "maximum distance to be used in delay lines");
   GET_ATTRIBUTE_DBSPL(minlevel, "Level threshold for rendering");
+  GET_ATTRIBUTE(nearfieldlimit, "m",
+                "distance arond 1/r source where the gain is constant");
   GET_ATTRIBUTE_BOOL(airabsorption, "apply air absorption filter");
   GET_ATTRIBUTE_BOOL(delayline, "use delayline");
   std::string gr("1/r");