diff --git a/ambed/cagc.cpp b/ambed/cagc.cpp
index 721d69f..71dd28d 100644
--- a/ambed/cagc.cpp
+++ b/ambed/cagc.cpp
@@ -36,17 +36,25 @@
 CAGC::CAGC(float initialLeveldB)
 {
     // set internal gain appropriately
-    m_g = powf(10.0f, -initialLeveldB/20.0f);
+    m_g = pow(10.0f, initialLeveldB/20.0f);
 
     // ensure resulting gain is not arbitrarily low
     if (m_g < 1e-16f)
         m_g = 1e-16f;
 
-    m_scale = 1.0f;
-    m_bandwidth = 1e-2f;
+    m_scale = (float)0xFFFF;
+    m_bandwidth = 1e-2f;                              //TODO : Move to parameter ?
+    m_gMax = pow(10.0f, initialLeveldB + 10.0f/20.0f);//+- 10dB Margin, TODO Move margin to constant
+    m_gMin = pow(10.0f, initialLeveldB - 10.0f/20.0f);
+    m_alpha = m_bandwidth;
     m_y2_prime = 1.0f;
 }
 
+float CAGC::GetGain()
+{
+    return 20.0f*log10(m_g);
+}
+
 void CAGC::Apply(uint8 * voice, int size)
 {
     for (int i = 0; i < size; i+=2)
@@ -54,6 +62,11 @@ void CAGC::Apply(uint8 * voice, int size)
         //Get the sample
         float _x = (float)(short)MAKEWORD(voice[i+1], voice[i]);
 
+        //This AGC tries to smooth energy so it does not exceed 1
+        //Therefore divide by our max supposed value
+        //Maybe we could also change y2 prime calculation below, but for now stick with this
+        _x /= m_scale;
+
         //apply AGC
         // apply gain to input sample
         float _y = _x * m_g;
@@ -66,16 +79,18 @@ void CAGC::Apply(uint8 * voice, int size)
 
         // update gain according to output energy
         if (m_y2_prime > 1e-6f)
-            m_g *= exp( -0.5f * m_alpha * log(m_y2_prime) );
+            m_g *= exp( -0.5f * m_alpha * log(m_y2_prime) ); 
 
-        // clamp to 120 dB gain
-        if (m_g > 1e6f)
-            m_g = 1e6f;
+        // clamp gain
+        if (m_g > m_gMax)
+            m_g = m_gMax;
+        else if(m_g < m_gMin)
+            m_g = m_gMin;
 
         // apply output scale
         _y *= m_scale;
 
-        //write processed sample back to it
+        //write processed sample back
         voice[i] = HIBYTE((short)_y);
         voice[i+1] = LOBYTE((short)_y);
     }    
diff --git a/ambed/cagc.h b/ambed/cagc.h
index b549197..ad8aa6e 100644
--- a/ambed/cagc.h
+++ b/ambed/cagc.h
@@ -21,7 +21,7 @@
 //    You should have received a copy of the GNU General Public License
 //    along with Foobar.  If not, see <http://www.gnu.org/licenses/>.
 // ----------------------------------------------------------------------------
-// Geoffrey Merck F4FXL / KC3FRA AGC code borrowed from Liquid DSP
+// Geoffrey Merck F4FXL / KC3FRA AGC code largely inspired by Liquid DSP
 // Only took the parts we need qnd recoeded it to be close the XLX coding style
 // https://github.com/jgaeddert/liquid-dsp/blob/master/src/agc/src/agc.c
 
@@ -29,7 +29,6 @@
 #define cagc_h
 
 #include "main.h"
-#include "math.h"
 
 class CAGC
 {
@@ -39,11 +38,12 @@ public:
 
     //methods
     void Apply(uint8 * voice, int size);
-    float GetGain(){ return -20.0f*log10(m_g); }//gets current gain 
+    float GetGain();//gets current gain 
     
 private:
     float m_g;            // current gain value
-    float m_scale;        // output scale value
+    float m_gMax, m_gMin;       //gain clamping
+    float m_scale;        // scale value
 
     // gain control loop filter parameters
     float m_bandwidth;    // bandwidth-time constant