diff --git a/src/builder2ibek/dbcompare.py b/src/builder2ibek/dbcompare.py
index b8dd44e..2efc0fc 100644
--- a/src/builder2ibek/dbcompare.py
+++ b/src/builder2ibek/dbcompare.py
@@ -1,11 +1,34 @@
 import re
+from dataclasses import dataclass, field
 from pathlib import Path
 
-regex_record = re.compile(r'record *\( *([^,]*), *"?([^"]*)"? *\)[\s\S]{')
+# https://regex101.com/r/M5dmVh/1
+# extract record type, name and fields from an EPICS database file
+regex_record = re.compile(r'record *\( *([^,]*), *"?([^"]*)"? *\)[\s\S]*?{([\s\S]*?)}')
+regex_field = re.compile(r'field *\( *([^,]*), *"?([^"]*)"? *\)')
+
+
+@dataclass
+class EpicsDb:
+    path: Path
+    text: str = ""
+    records: set[str] = field(default_factory=lambda: set())
+    fields: dict[str, set[str]] = field(default_factory=lambda: {})
+
+
+def normalize_float_records(s: set[str]) -> set[str]:
+    normalised = set()
+    for item in s:
+        try:
+            name, value = item.split(maxsplit=1)
+            normalised.add(f"{name} {float(value)}")
+        except (ValueError, IndexError):
+            normalised.add(item)
+    return normalised
 
 
 def compare_dbs(
-    original: Path, new: Path, ignore: list[str], output: Path | None = None
+    old_path: Path, new_path: Path, ignore: list[str], output: Path | None = None
 ):
     """
     validate that two DBs have the same set of records
@@ -13,29 +36,32 @@ def compare_dbs(
     used to ensure that an IOC converted to epics-containers has the same
     records as the original builder IOC
     """
-    old_text = original.read_text()
-    new_text = new.read_text()
+    old = EpicsDb(old_path)
+    new = EpicsDb(new_path)
 
-    old_set: set[str] = set()
-    for record in regex_record.finditer(old_text):
-        old_set.add(f"{record.group(1)} {record.group(2)}")
-    new_set: set[str] = set()
-    for record in regex_record.finditer(new_text):
-        new_set.add(f"{record.group(1)} {record.group(2)}")
+    for db in [old, new]:
+        db.text = db.path.read_text()
+        for record in regex_record.finditer(db.text):
+            rec_str = f"{record.group(1)} {record.group(2)}"
+            db.records.add(rec_str)
+            db.fields[rec_str] = set()
+            for rec_field in regex_field.finditer(record.group(3)):
+                db.fields[rec_str].add(f"{rec_field.group(1)} {rec_field.group(2)}")
 
-    old_only = sorted(old_set - new_set)
-    new_only = sorted(new_set - old_set)
+    old_only = sorted(old.records - new.records)
+    new_only = sorted(new.records - old.records)
+    both = sorted(old.records & new.records)
 
     old_only_filtered = old_only.copy()
     new_only_filtered = new_only.copy()
-    for rec in old_only:
-        for s in ignore:
-            if s in rec:
-                old_only_filtered.remove(rec)
-    for rec in new_only:
-        for s in ignore:
-            if s in rec:
-                new_only_filtered.remove(rec)
+    for filtered, unfiltered in [
+        (old_only_filtered, old_only),
+        (new_only_filtered, new_only),
+    ]:
+        for record_str in unfiltered:
+            for s in ignore:
+                if s in record_str:
+                    filtered.remove(record_str)
 
     result = (
         "*******************************************************************\n"
@@ -47,11 +73,20 @@ def compare_dbs(
         + "\n".join(new_only_filtered)
         + "\n\n"
         + "*******************************************************************\n"
-        + f"records in original:    {len(old_set)}\n"
-        f"  records in new:         {len(new_set)}\n"
+        + f"records in original:    {len(old.records)}\n"
+        f"  records in new:         {len(new.records)}\n"
         f"  records missing in new: {len(old_only_filtered)}\n"
         f"  records extra in new:   {len(new_only_filtered)}\n"
+        + "*******************************************************************\n"
     )
+
+    for record_str in both:
+        # validate the fields are the same
+        old_norm = normalize_float_records(old.fields[record_str])
+        new_norm = normalize_float_records(new.fields[record_str])
+        if old_norm != new_norm:
+            result += f"\nfields for '{record_str}' are different between the two DBs"
+
     if not output:
         print(result)
     else: