How-to
Connecting a Yokogawa ExaOPC server to Canary Historian
Short answer
Yes. Canary Historian's OPC client collector reads tags straight from ExaOPC, Yokogawa's OPC server for CENTUM, over DA or UA with no custom driver. Run the DA collector on the ExaOPC host or a same-domain host: cross-host DCOM is where these links break.
On this page
Yes — Canary Historian collects from Yokogawa ExaOPC with its own OPC client collector. No custom driver and no third-party gateway sit between them: ExaOPC is Yokogawa's OPC server for CENTUM, Canary's collector consumes OPC DA and UA servers, and the two meet directly. The data path is CENTUM station → ExaOPC server → OPC session → Canary collector → Canary Historian. The job comes down to three decisions: which interface, which host runs the collector, and how fast you poll. DCOM is where these links fail, so the host decision carries the most weight.
The data path and where it fails #
Canary never talks to the controllers. ExaOPC does, and it presents three documented OPC interfaces: Data Access, which reads and writes live process data; Alarms and Events; and Historical Data Access. Live collection into a historian is a DA job. Pulling history already stored on the control-system side is a different exercise — the same client-side distinction as reading Proficy Historian OPC-HDA tags into Ignition — and alarms ride the A&E interface that display clients use, as in bringing CENTUM alarms onto a PC display through ExaOPC.
Every fault on the DA path lands in one of two places, and you separate them first:
- Binding faults — the collector cannot open a session with the server. Causes: DCOM access or launch permission denied, wrong server name, unreachable host, firewall.
- Tag faults — the session opens and individual items come back bad quality, stale, or not found. Causes: point inactive on the DCS side, comm fault between ExaOPC and the controller, mistyped item ID.
Fix binding before tags; no tag-level adjustment helps while the session cannot open.
Choose the path before you configure anything #
| Path | Use it when | What it costs | Where it breaks |
|---|---|---|---|
| DA collector on the ExaOPC host | ExaOPC exposes DA, the classic case | nothing crosses the network | one host carries both ExaOPC and the collector load |
| UA collector to a UA endpoint | your ExaOPC version or an added UA layer serves UA | one TCP port, certificate trust, user tokens | certificate trust and endpoint security policy mismatch |
| DA-to-UA gateway on the ExaOPC host | policy forbids DCOM across that firewall and only DA exists | one more component to license, patch and monitor | the gateway becomes its own failure point |
Check the interface list in your ExaOPC documentation or its configuration utility — Yokogawa documents DA, A&E and HDA; whether a UA endpoint exists depends on your version and any UA layer installed with it. The decision: if a UA endpoint is exposed, use it, because one port and a certificate beat a dynamic RPC range. If only DA exists, run the Canary collector on the ExaOPC host, or on a host in the same domain with DCOM fully configured. Add a gateway only when DCOM cannot cross that firewall at all.
Configure the Canary OPC DA collector #
- Prove ExaOPC first. On the ExaOPC host, read one tag with any OPC DA test client. If that fails, the fault is on the Yokogawa side; fix it before Canary enters the picture.
- Install the OPC Core Components. A Windows OPC DA client needs the OPC Foundation's OPC Core Components on the collector host, taken from the OPC Foundation.
- Pick the collector host. The same host as ExaOPC removes network DCOM entirely. A separate host needs matching user accounts, DCOM launch and access permission for the ExaOPC application, and port 135 plus a fixed, restricted RPC port range open through the firewall. If you are specifying a new host for the job, a fanless mini PC in the cabinet beside the DCS gateway keeps the collector inside the same zone.
- Create the collector. Current builds do this in the OPC collector tile — Canary documents the tile and the DA collector procedure for versions 24 and 25, and the UA collector walkthrough from version 19. Builds around Canary Admin 13 did the same job in the Logger Administrator application. Enter the server host and the ExaOPC server name (its ProgID) exactly as registered on that host.
- Bring in the tags. Browse the address space, add one tag, copy it out to a spreadsheet to learn the column layout, build the full list, and paste it back — that is the documented bulk-import route.
- Set the update rate per group to the speed of the process, never faster than the underlying CENTUM point update.
- Run the collector under a read-only account. ExaOPC's DA server writes process data as well as reading it, and a historian must never be able to move a setpoint. This is the one access decision you do not compromise on.
- Start it and watch the connection state first. Tag values come second; a green session with bad-quality tags is a tag fault, not a success.
Size the poll rate with real arithmetic #
Illustrative inputs — replace them with your own tag list: 40 fast analog points at 1 s, 260 slow analog points at 10 s, 1200 discrete points at 30 s, 1500 tags total.
- fast group: 40 tags × 1 read / 1 s = 40 reads/s
- slow group: 260 tags × 1 read / 10 s = 26 reads/s
- discretes: 1200 tags × 1 read / 30 s = 40 reads/s
- total: 106 item reads per second against ExaOPC
Polling all 1500 tags at 1 s instead would be 1500 reads/s — about 14 times the load (1500 ÷ 106 ≈ 14.2) — to re-read points that mostly did not change between polls. Three rate groups, each held at or below the CENTUM point's own update rate, is the design. Commission while watching the ExaOPC host CPU, and re-run this arithmetic whenever the tag list grows.
Prove the data before you trust the trend #
The end-state check is the same one you apply after moving Wonderware Historian data into Canary Historian: the stored value must match the source. Work it in this order:
- Compare three views of one tag: the test-client value on the ExaOPC host, the collector's value, and the stored historian value. All three must agree within one update interval, with good quality in each.
- Cover the data types: one analog, one digital, one counter or totalizer.
- Change the value on the process side — if your procedure allows it — and confirm the change lands in Canary within the configured update interval.
- Confirm which timestamp Canary stores, the server timestamp or collection time, and sync the DCS stations and the Canary host to one time source so the choice cannot hide controller timing.
- In a maintenance window, stop and restart the ExaOPC service. Tags must go to bad quality and recover on their own; values that stay good through the outage mean you are reading a cache, not the plant.
What breaks this link over time #
- Bitness. A 32-bit OPC client against a 64-bit server registration, or the reverse, fails to bind. Match them or share the host.
- Windows change. DCOM hardening updates and changed service accounts silently remove access. Re-test the link after every patch cycle on both hosts.
- Tag drift. Renamed or deleted CENTUM points leave the collector polling dead items. Reconcile the Canary tag list after every DCS change.
- Polling creep. New tags added at the fastest group rate. Re-run the reads-per-second arithmetic at every change.
- Access creep. Someone grants the collector write permission for a one-off fix and never rolls it back. Audit the account with the DCS review.
- Segmentation. IEC 62443 builds a plant from zones joined by defined, controlled conduits; a collector belongs inside the DCS's allowed zone or behind a documented conduit. A wide-open dynamic RPC range through the firewall is not a conduit.
Questions asked
Does Canary need a special driver for ExaOPC?
No. Canary's OPC client collector speaks the OPC DA and UA interfaces and ExaOPC serves them, so the connection is native. What the collector host does need for DA is the OPC Foundation's OPC Core Components, plus a working DCOM setup if it runs remotely. The UA path replaces both with a single TCP port and certificate trust.
Can this collector write setpoints back to CENTUM?
The DA interface technically allows writes, and that is exactly why the collector account must stay read-only. A historian exists to log values, not to move them. If an operator needs setpoint access, give it to a controlled client with its own account and its own audit trail, never to the collection path.
How do I get CENTUM alarms into Canary too?
The DA collector configured here logs process values. CENTUM alarms ride ExaOPC's separate alarms and events interface, which display clients consume. If you want alarms historized on this path, expose alarm states as DA points, for example a latched status tag per alarm, and log them like any other digital point.
How do I know whether my ExaOPC offers a UA endpoint?
Read the interface list in your ExaOPC documentation or in the configuration utility on the server host. Yokogawa documents Data Access, Alarms and Events, and HDA for ExaOPC; whether a UA endpoint exists depends on the version and any UA layer installed with it. Do not infer it from the Canary side, which only shows servers the host actually exposes.
What is the practical tag limit for one DA collector?
The documented Canary material gives no fixed tag ceiling. The real constraint is item reads per second against the ExaOPC host: 40 tags at 1 s, 260 at 10 s and 1200 at 30 s total 106 reads per second. Group tags by process speed, hold each group at or below the CENTUM point update rate, and watch host CPU during commissioning.