diff --git a/src/raman_fitting/deconvolution_models/base_peak.py b/src/raman_fitting/deconvolution_models/base_peak.py
index 1b44f03..297ba0c 100644
--- a/src/raman_fitting/deconvolution_models/base_peak.py
+++ b/src/raman_fitting/deconvolution_models/base_peak.py
@@ -248,7 +248,7 @@ def _main():
         )
         models[model_name] = lmfit_comp_model
         print(lmfit_comp_model)
-    breakpoint()
+    # breakpoint()
 
 
 if __name__ == "__main__":
diff --git a/src/raman_fitting/processing/baseline_subtraction.py b/src/raman_fitting/processing/baseline_subtraction.py
index 0857580..926bcf3 100644
--- a/src/raman_fitting/processing/baseline_subtraction.py
+++ b/src/raman_fitting/processing/baseline_subtraction.py
@@ -1,91 +1,108 @@
 import logging
+from typing import Optional
 
 import numpy as np
 from scipy.stats import linregress
 
-from .splitter import get_default_spectrum_windows
+from .splitter import get_default_spectrum_windows, split_spectrum_data_in_windows
+from .spectrum_template import SpectrumData
 
 logger = logging.getLogger(__name__)
 
 
+def subtract_baseline_per_window(spec, label, windows_data, window_limits):
+    ramanshift = spec.ramanshift
+    intensity = spec.intensity
+    window_name = spec.window_name
+    if not ramanshift.any():
+        return intensity, (0, 0)
+    # breakpoint()
+    lbl_1st_order = list(filter(lambda x: "1st_order" in x, windows_data.keys()))
+    if any(i in label for i in ("full", "norm")) and lbl_1st_order:
+        i_fltrd_dspkd_fit = windows_data.get(lbl_1st_order[0]).intensity
+    else:
+        i_fltrd_dspkd_fit = intensity
+    window_config = window_limits.get(window_name)
+
+    bl_linear = linregress(
+        ramanshift[[0, -1]],
+        [
+            np.mean(i_fltrd_dspkd_fit[0 : window_config.min]),
+            np.mean(i_fltrd_dspkd_fit[window_config.max : :]),
+        ],
+    )
+    i_blcor = intensity - (bl_linear[0] * ramanshift + bl_linear[1])
+    return i_blcor, bl_linear
+
+
+def get_normalization_factor(data, norm_method="simple") -> float:
+    try:
+        if norm_method == "simple":
+            normalization_intensity = np.nanmax(data["normalization"].intensity)
+        elif norm_method == "fit":
+            raise NotImplementedError("NormalizeFit not yet implemented")
+            # IDEA not implemented
+            # normalization = NormalizeFit(
+            #     self.blcorr_data["1st_order"], plotprint=False
+            # )  # IDEA still implement this NormalizeFit
+            # normalization_intensity = normalization["IG"]
+        else:
+            logger.warning(f"unknown normalization method {norm_method}")
+            normalization_intensity = 1
+    except Exception as exc:
+        logger.error(f"normalization error {exc}")
+        normalization_intensity = 1
+
+    return normalization_intensity
+
+
+def normalize_data(windows_data, norm_factor, label: Optional[str] = None) -> dict:
+    ret = {}
+    for window_name, spec in windows_data.items():
+        norm_label = f"norm_blcorr_{window_name}"
+        if label:
+            norm_label = f"{label}_{norm_label}"
+        _data = SpectrumData(
+            spec.ramanshift, spec.intensity * norm_factor, norm_label, window_name
+        )
+        ret.update(**{window_name: _data})
+    return ret
+
+
+def subtract_loop(windows_data: dict, window_limits: dict, label=None):
+    _blcorr = {}
+    _info = {}
+    for window_name, spec in windows_data.items():
+        blcorr_int, blcorr_lin = subtract_baseline_per_window(
+            spec, window_name, windows_data, window_limits
+        )
+        label = f"blcorr_{window_name}"
+        if label:
+            label = f"{label}_{label}"
+        _data = SpectrumData(spec.ramanshift, blcorr_int, label, window_name)
+        _blcorr.update(**{window_name: _data})
+        _info.update(**{window_name: blcorr_lin})
+    return _blcorr, _info
+
+
 class BaselineSubtractorNormalizer:
     """
     For baseline subtraction as well as normalization of a spectrum
     """
 
-    def __init__(self, *args, **kws):
-        self.split_spectrum_data_in_windows()
-        self.windowlimits = get_default_spectrum_windows()
-        blcorr_data, blcorr_info = self.subtract_loop()
+    def __init__(self, ramanshift: np.array, intensity: np.array, label: str = None):
+        self._ramanshift = ramanshift
+        self._intensity = intensity
+        self._label = label
+        self.windows_data = split_spectrum_data_in_windows(
+            ramanshift=ramanshift, intensity=intensity, label=label
+        )
+        self.window_limits = get_default_spectrum_windows()
+        blcorr_data, blcorr_info = subtract_loop(
+            self.windows_data, self.window_limits, label=self._label
+        )
         self.blcorr_data = blcorr_data
         self.blcorr_info = blcorr_info
-        normalization_intensity = self.get_normalization_factor()
+        normalization_intensity = get_normalization_factor(self.blcorr_data)
         self.norm_factor = 1 / normalization_intensity
-        self.norm_data = self.normalize_data(self.blcorr_data, self.norm_factor)
-
-    def subtract_loop(self):
-        _blcorr = {}
-        _info = {}
-        for windowname, spec in self.windows_data.items():
-            blcorr_int, blcorr_lin = self.subtract_baseline_per_window(windowname, spec)
-            label = f"blcorr_{windowname}"
-            if self.label:
-                label = f"{self.label}_{label}"
-            _data = self.data(spec.ramanshift, blcorr_int, label)
-            _blcorr.update(**{windowname: _data})
-            _info.update(**{windowname: blcorr_lin})
-        return _blcorr, _info
-
-    def subtract_baseline_per_window(self, windowname, spec):
-        rs = spec.ramanshift
-        if not rs.any():
-            return spec.intensity, (0, 0)
-
-        if windowname[0:4] in ("full", "norm"):
-            i_fltrd_dspkd_fit = self.windows_data.get("1st_order").intensity
-        else:
-            i_fltrd_dspkd_fit = spec.intensity
-        _limits = self.windowlimits.get(windowname)
-
-        bl_linear = linregress(
-            rs[[0, -1]],
-            [
-                np.mean(i_fltrd_dspkd_fit[0 : _limits[0]]),
-                np.mean(i_fltrd_dspkd_fit[_limits[1] : :]),
-            ],
-        )
-        i_blcor = spec.intensity - (bl_linear[0] * rs + bl_linear[1])
-        return i_blcor, bl_linear
-
-    def get_normalization_factor(self, norm_method="simple") -> float:
-        try:
-            if norm_method == "simple":
-                normalization_intensity = np.nanmax(
-                    self.blcorr_data["normalization"].intensity
-                )
-            elif norm_method == "fit":
-                raise NotImplementedError("NormalizeFit not yet implemented")
-                # IDEA not implemented
-                # normalization = NormalizeFit(
-                #     self.blcorr_data["1st_order"], plotprint=False
-                # )  # IDEA still implement this NormalizeFit
-                # normalization_intensity = normalization["IG"]
-            else:
-                logger.warning(f"unknown normalization method {norm_method}")
-                normalization_intensity = 1
-        except Exception as exc:
-            logger.error(f"normalization error {exc}")
-            normalization_intensity = 1
-
-        return normalization_intensity
-
-    def normalize_data(self, data, norm_factor) -> dict:
-        ret = {}
-        for windowname, spec in data.items():
-            label = f"norm_blcorr_{windowname}"
-            if self.label:
-                label = f"{self.label}_{label}"
-
-            _data = self.data(spec.ramanshift, spec.intensity * self.norm_factor, label)
-            ret.update(**{windowname: _data})
-        return ret
+        self.norm_data = normalize_data(self.blcorr_data, self.norm_factor)
diff --git a/src/raman_fitting/processing/post_processing.py b/src/raman_fitting/processing/post_processing.py
index 218b3b9..9d38f78 100644
--- a/src/raman_fitting/processing/post_processing.py
+++ b/src/raman_fitting/processing/post_processing.py
@@ -27,12 +27,13 @@ def split_data(self, on_lbl="filtered"):
     def spectrum_methods_delegator(self):
         for method, on_lbl, out_lbl in POST_PROCESS_METHODS:
             try:
-                breakpoint()
+                # breakpoint()
                 getattr(self, method)(on_lbl=on_lbl, out_lbl=out_lbl)
             except Exception as exc:
                 logger.error(
                     f"spectrum_methods_delegator, {self._qcnm} {method} failed for {self.file} with {exc}"
                 )
+                raise exc from exc
 
         self.set_clean_data_df()
         self.set_df_from_register()
@@ -54,7 +55,10 @@ def baseline_correction(self, on_lbl="despiked", out_lbl="clean_data"):
         _r, _int, _lbl = self.register.get(on_lbl)
         _baseline_corrected = BaselineSubtractorNormalizer(_r, _int, label="despiked")
         self._baseline_corrected = _baseline_corrected
-
-        _fullspec = _baseline_corrected.norm_data["full"]
-        self.register_spectrum(_fullspec.ramanshift, _fullspec.intensity, out_lbl)
+        full_keys = list(
+            filter(lambda x: x.endswith("full"), _baseline_corrected.norm_data)
+        )
+        if full_keys:
+            _fullspec = _baseline_corrected.norm_data[full_keys[0]]
+            self.register_spectrum(_fullspec.ramanshift, _fullspec.intensity, out_lbl)
         self.clean_data = _baseline_corrected.norm_data
diff --git a/src/raman_fitting/processing/spectrum_constructor.py b/src/raman_fitting/processing/spectrum_constructor.py
index 89bdcb8..44f2566 100644
--- a/src/raman_fitting/processing/spectrum_constructor.py
+++ b/src/raman_fitting/processing/spectrum_constructor.py
@@ -93,6 +93,9 @@ def load_data_delegator(self):
         self.info = {**self.info, **self.run_kwargs}
 
     def set_clean_data_df(self):
+        if self.clean_data is None:
+            return
+
         self.clean_df = {
             k: pd.DataFrame(
                 {"ramanshift": val.ramanshift, f"int_{self.SamplePos}": val.intensity}
diff --git a/src/raman_fitting/processing/spectrum_template.py b/src/raman_fitting/processing/spectrum_template.py
index dbd2d01..2952c58 100644
--- a/src/raman_fitting/processing/spectrum_template.py
+++ b/src/raman_fitting/processing/spectrum_template.py
@@ -1,11 +1,21 @@
 #!/usr/bin/env python3
 # -*- coding: utf-8 -*-
 from collections import namedtuple
+from dataclasses import dataclass
 
+import numpy as np
 
 SpecTemplate = namedtuple("Spectrum", "ramanshift intensity label")
 
 
+@dataclass
+class SpectrumData:
+    ramanshift: np.array
+    intensity: np.array
+    label: str
+    window_name: str
+
+
 class SpectrumTemplate:
     def __init__(self, spec_name="spectrum_info"):
         self.spec_name = spec_name
diff --git a/src/raman_fitting/processing/splitter.py b/src/raman_fitting/processing/splitter.py
index 1cd2c95..d2229d7 100644
--- a/src/raman_fitting/processing/splitter.py
+++ b/src/raman_fitting/processing/splitter.py
@@ -2,16 +2,17 @@
 import numpy as np
 
 from pydantic import BaseModel
+from .spectrum_template import SpectrumData
 
-SPECTRUM_WINDOWS = [
-    {"name": "full", "min": 200, "max": 3600},
-    {"name": "full_1st_2nd", "min": 800, "max": 3500},
-    {"name": "low", "min": 150, "max": 850, "extra_margin": 10},
-    {"name": "1st_order", "min": 900, "max": 2000},
-    {"name": "mid", "min": 1850, "max": 2150, "extra_margin": 10},
-    {"name": "2nd_order", "min": 2150, "max": 3380},
-    {"name": "normalization", "min": 1500, "max": 1675, "extra_margin": 10},
-]
+SPECTRUM_WINDOWS = {
+    "full": {"min": 200, "max": 3600},
+    "full_1st_2nd": {"min": 800, "max": 3500},
+    "low": {"min": 150, "max": 850, "extra_margin": 10},
+    "1st_order": {"min": 900, "max": 2000},
+    "mid": {"min": 1850, "max": 2150, "extra_margin": 10},
+    "2nd_order": {"min": 2150, "max": 3380},
+    "normalization": {"min": 1500, "max": 1675, "extra_margin": 10},
+}
 
 
 class SpectrumWindow(BaseModel):
@@ -23,14 +24,14 @@ class SpectrumWindow(BaseModel):
 
 def get_default_spectrum_windows() -> Dict[str, SpectrumWindow]:
     windows = {}
-    for window_config in SPECTRUM_WINDOWS:
-        windows[window_config["name"]] = SpectrumWindow(**window_config)
+    for window_name, window_config in SPECTRUM_WINDOWS.items():
+        windows[window_name] = SpectrumWindow(name=window_name, **window_config)
     return windows
 
 
 def split_spectrum_data_in_windows(
     ramanshift, intensity, spec_windows=None, label=None
-) -> Dict:
+) -> Dict[str, SpectrumData]:
     """
     For splitting of spectra into the several SpectrumWindows,
     the names of the windows are taken from SpectrumWindows
@@ -40,12 +41,17 @@ def split_spectrum_data_in_windows(
     if spec_windows is None:
         spec_windows = get_default_spectrum_windows()
     windows_data = {}
-    for windowname, window in spec_windows.items():
+    for window_name, window in spec_windows.items():
         # find indices of window in ramanshift array
         ind = (ramanshift >= np.min(window.min)) & (ramanshift <= np.max(window.max))
-        window_lbl = f"window_{windowname}"
+        window_lbl = f"window_{window_name}"
         if label is not None:
             window_lbl = f"{label}_{window_lbl}"
-        _data = {"ramanshift": ramanshift[ind], "intensity": intensity[ind]}
-        windows_data[window_lbl] = _data
+        _data = {
+            "ramanshift": ramanshift[ind],
+            "intensity": intensity[ind],
+            "label": window_lbl,
+            "window_name": window_name,
+        }
+        windows_data[window_lbl] = SpectrumData(**_data)
     return windows_data