diff --git a/cmProc5.c b/cmProc5.c
index eabdd35..1cfc16d 100644
--- a/cmProc5.c
+++ b/cmProc5.c
@@ -918,8 +918,11 @@ cmNlmsEc_t* cmNlmsEcAlloc( cmCtx* ctx, cmNlmsEc_t* ap, float mu, unsigned hN, un
 {
   cmNlmsEc_t* p = cmObjAlloc(cmNlmsEc_t,ctx,ap);
 
-  // allocate the vect array
+  // allocate the vect array's
+  p->uVa = cmVectArrayAlloc(ctx, kFloatVaFl );
+  p->fVa = cmVectArrayAlloc(ctx, kFloatVaFl );
   p->eVa = cmVectArrayAlloc(ctx, kFloatVaFl );
+   
   
   if( mu != 0 )  
     if( cmNlmsEcInit(p,mu,hN,delayN) != cmOkRC )
@@ -1024,13 +1027,27 @@ cmRC_t      cmNlmsEcExec( cmNlmsEc_t* p, const cmSample_t* xV, const cmSample_t*
 
   }
 
+  cmVectArrayAppendS(p->uVa,xV,xyN);
+  cmVectArrayAppendS(p->fVa,fV,xyN);
+  cmVectArrayAppendS(p->eVa,yV,xyN);
+   
+
   return cmOkRC;
 }
 
 
 cmRC_t      cmNlmsEcWrite( cmNlmsEc_t* p, const cmChar_t* dirStr )
 {
+
+  if( p->uVa != NULL )
+    cmVectArrayWriteDirFn(p->uVa, dirStr, "nlms_unfiltered.va");
+
+  if( p->fVa != NULL )
+    cmVectArrayWriteDirFn(p->fVa, dirStr, "nlms_filtered.va");
+
   if( p->eVa != NULL )
-    cmVectArrayWriteDirFn(p->eVa,dirStr, "nlms_err.va");
+    cmVectArrayWriteDirFn(p->eVa, dirStr, "nlms_out.va");
+
+  
   return cmOkRC;
 }
diff --git a/cmProc5.h b/cmProc5.h
index 050a07d..ad8a7b6 100644
--- a/cmProc5.h
+++ b/cmProc5.h
@@ -235,10 +235,11 @@ extern "C" {
     unsigned       hN;          // filter length
     unsigned       delayN;      // fixed delay to apply to align xV with fV.
     double*        wV;          // wV[hN] filter weights
-    double*        hV;          // hV[hN] filter delay line
-    
-    unsigned       w0i;
-    
+    double*        hV;          // hV[hN] filter delay line    
+    unsigned       w0i;         // The index into hV[] of the start of the delay line.
+
+    cmVectArray_t* uVa;
+    cmVectArray_t* fVa;
     cmVectArray_t* eVa;
   } cmNlmsEc_t;