diff --git a/tests/node_allp.rs b/tests/node_allp.rs
new file mode 100644
index 0000000..7a03bdf
--- /dev/null
+++ b/tests/node_allp.rs
@@ -0,0 +1,60 @@
+mod common;
+use common::*;
+
+#[test]
+fn check_node_allp() {
+    let (node_conf, mut node_exec) = new_node_engine();
+    let mut matrix = Matrix::new(node_conf, 4, 4);
+
+    let test = NodeId::Test(0);
+    let ap   = NodeId::AllP(0);
+    let out  = NodeId::Out(0);
+    matrix.place(0, 0, Cell::empty(test)
+                       .out(None, None, test.out("tsig")));
+    matrix.place(0, 1, Cell::empty(ap)
+                       .input(ap.inp("inp"), None, None)
+                       .out(None, None, ap.out("sig")));
+    matrix.place(0, 2, Cell::empty(out)
+                       .input(out.inp("ch1"), None, None)
+                       .out(None, None, None));
+    matrix.place(1, 0, Cell::empty(test)
+                       .out(None, None, test.out("tsig")));
+    matrix.place(1, 1, Cell::empty(out)
+                       .input(out.inp("ch2"), None, None)
+                       .out(None, None, None));
+    pset_d(&mut matrix, ap, "time", 3.0);
+    matrix.sync().unwrap();
+
+    pset_s(&mut matrix, test, "trig", 1);
+
+    let res = run_for_ms(&mut node_exec, 20.0);
+
+    // the original signal on ch2: 2ms trigger up:
+    let mut v = vec![1.0; (2.0 * 44.1_f32).ceil() as usize];
+    v.append(&mut vec![0.0; (18.0 * 44.1_f32).ceil() as usize]);
+    assert_vec_feq!(res.1, v);
+
+    // now signal on ch1 from the allpass:
+    // starts with original signal * -0.7
+    let mut v = vec![-0.7; (2.0 * 44.1_f32).ceil() as usize];
+    // silence for 1ms, which is the internal delay of the allpass
+    v.append(&mut vec![0.0; (1.0 * 44.1_f32).floor() as usize - 3]);
+
+    // allpass feedback of the original signal for 2ms:
+    // XXX: the smearing before and after the allpass is due to the
+    // cubic interpolation!
+    v.append(&mut vec![-0.03150302, 0.25802, 1.0735]);
+    v.append(&mut vec![1.0; (2.0 * 44.1_f32).ceil() as usize - 3]);
+    v.append(&mut vec![1.0315, 0.7419, -0.0735]);
+    // 1ms allpass silence like before:
+    v.append(&mut vec![0.0; (1.0 * 44.1_f32).floor() as usize - 6]);
+
+    // 2ms the previous 1.0 * 0.7 fed back into the filter,
+    // including even more smearing due to cubic interpolation:
+    v.append(&mut vec![0.0006, -0.0120, 0.0106, 0.3444, 0.7801, 0.6962]);
+    v.append(&mut vec![0.7; (2.0 * 44.1_f32).floor() as usize - 5]);
+    v.append(&mut vec![0.6993, 0.712, 0.6893, 0.3555, -0.0801, 0.0037]);
+
+    //d// println!("res={:?}", res.1);
+    assert_vec_feq!(res.0, v);
+}