Troubleshooting
Downloading a Vijeo Designer Project to a Remote HMI Over VPN
Short answer
The greyed-out NAT fields are not the blocker: they only activate for a target and download method that supports address translation. When a VPN already routes your PC into the remote LAN, leave them alone, download straight to the panel's private address, and give the HMI a default gateway so its replies can leave the subnet.
On this page
- Why are the Use NAT checkbox and the router IP fields greyed out?
- Does the download go through the VPN or through NAT?
- What settings does the HMI need before the download?
- How do I prove the PC can reach the panel before downloading?
- Ping works, so why will the download still not start?
- Why does a large project abort halfway through the transfer?
- When is NAT with port forwarding the right answer?
- Questions asked
The greyed-out NAT fields are not your roadblock. When the VPN already routes your engineering PC into the remote LAN, the download goes straight to the panel's private address and the NAT block stays untouched; the one panel-side setting that decides success is the default gateway. The case behind this question was a Vijeo Designer Basic v1.1 project aimed at a GXU3512 panel at a remote site, with the Use NAT checkbox and the router IP and TCP port fields inactive — a limit of the target's download configuration, not a fault in the transfer path. One illustrative network runs through this article: panel 192.168.1.50, mask 255.255.255.0, default gateway 192.168.1.1. Substitute your own values; the decisions do not change.
Why are the Use NAT checkbox and the router IP fields greyed out? #
Because the dialog only offers NAT for the target and download method that supports address translation — in the mode your project is in, the fields are locked by design, and over a VPN that lock works in your favor because you should not use them anyway.
The NAT block exists for exactly one situation: no tunnel, a panel behind a router, and a port-forwarding rule that rewrites an incoming WAN connection to the panel's LAN address. The router IP and TCP port fields appear only for the target and download configuration that supports that translation; in the configuration you have, they stay inactive. Do not spend an afternoon forcing the checkbox alive. If you ever reach a site with no tunnel, re-select the target or change the download method so the Ethernet-with-NAT configuration is active and the fields return. With the tunnel up, greyed out is the correct state.
Does the download go through the VPN or through NAT? #
Through the VPN: with your PC routed into the remote LAN you address the panel at its private address, 192.168.1.50 in the example, and NAT stays out of the path entirely — the two are alternative transports, never a combination. A VPN and a NAT rule are two answers to the same question: how does my packet reach a private address at a remote site? The tunnel answers by routing — your PC gets a route into the remote subnet, so the panel's private address is an ordinary destination. NAT answers by translation at the edge — you aim at the router's public address and the router rewrites the destination. Pick one. Stacking both puts a second translation on a path that already works, and the session breaks.
| Where the engineering PC sits | Address you enter for the target | NAT block | HMI default gateway |
|---|---|---|---|
| VPN tunnel into the remote LAN | Panel's private IP (192.168.1.50 in the example) | Leave untouched | Required |
| Same flat LAN as the panel | Panel's private IP | Unused | Not required |
| Internet only, router forwarding ports | Router's public WAN IP plus the forwarded TCP ports | Required: ticked and filled | Required |
| Agent-based remote-access appliance | The address the appliance's client presents | Unused | Required in routed mode; bridged mode needs none |
If the tunnel itself is what you still have to build, treat the remote HMI access with dynamic DNS and a VPN setup as the separate job to finish first.
What settings does the HMI need before the download? #
Three values: its own IP address, its subnet mask, and a default gateway pointing at the site router's LAN interface — 192.168.1.1 in the example — because without a gateway the panel accepts your packet but cannot answer it. Your PC arrives at the panel carrying a tunnel-side source address, 10.8.0.5 in the example. The mask puts that source outside the panel's own subnet, so the reply cannot be delivered directly; the panel hands it to the default gateway. With the gateway field empty, the panel's stack drops the reply, your ping times out and the download never opens. Set the gateway in the panel's network settings screen and the return path exists.
This is the most common cause of a panel that stays silent over an otherwise healthy VPN, and it is a panel setting, not a network fault: the router was always there, the panel just was never told to use it.
How do I prove the PC can reach the panel before downloading? #
Run three checks from the engineering PC: ipconfig /all to confirm the VPN adapter holds an address, route print to confirm the remote subnet points into the tunnel, then ping the panel ten times — a pass is ten replies out of ten.
ipconfig /all— the VPN adapter must show an address.route print— the remote subnet, 192.168.1.0 with mask 255.255.255.0, must route via the VPN interface, not via your local default gateway.ping 192.168.1.50 -n 10— every request must come back; recurring loss means you fix routing before opening the software.
The ping is not a formality: why HMI programming software needs a ping reply to download is that the transfer session opens over the same path your ICMP test uses, so a failed ping predicts a failed download exactly.
Worked example — subnet overlap (illustrative addresses): your office PC is 192.168.1.34/24 and the panel is 192.168.1.50/24. Apply the mask to each end: 192.168.1.34 AND 255.255.255.0 = 192.168.1.0, and 192.168.1.50 AND 255.255.255.0 = 192.168.1.0. Both ends resolve to the same network, so your PC treats the panel as on-link and asks the office LAN for its MAC address instead of sending anything into the tunnel — the VPN can be perfect and the ping still dies locally. Decision: renumber one side, for instance the plant to 192.168.10.0/24 with the panel at 192.168.10.50 and gateway 192.168.10.1, then retest. A VPN concentrator that 1:1 translates the remote range is the alternative; address the panel through the translated range.
One routing trap remains: a split-tunnel client only routes the subnets the VPN pushes to it. If the remote subnet is missing from that list, everything except the tunnel endpoint fails. Have the panel's subnet added to the pushed networks and retest.
Ping works, so why will the download still not start? #
Two causes cover most of it: the PC firewall treating the VPN adapter as a public network and filtering the transfer, or a gate on the panel itself — check the firewall first, it is the more common one. Windows Defender Firewall, or an endpoint suite, classifies the VPN adapter as a public network and blocks the session on it. Change the adapter's profile to private, or add an allow rule for the Vijeo Designer executable on all profiles, and the transfer goes through.
On the panel side, vendor documentation adds one gate: HMI units manufactured from June 2020 onward carry a switch in their network settings that enables or disables Ethernet downloads. On a newer panel with the switch off, the download will not start no matter how clean the ping is — turn it on. If both are clear and the transfer still refuses to connect, the fault can sit in the software-to-panel path rather than the network, the same class of failure as when an HMI is detected but the project download fails; rule that out before touching routing again.
Why does a large project abort halfway through the transfer? #
Because the tunnel is losing packets, clamping the MTU, or both: confirm with a 20-ping test where twenty replies out of twenty is the pass, and an MTU probe at a 1400-byte payload, then retry — two or three attempts to complete a big file on a slow link is normal.
Run ping 192.168.1.50 -n 20 and read the loss: recurring loss on a long ping is a link problem, not a software problem, so fix the tunnel before retrying. Probe the path with ping 192.168.1.50 -f -l 1400; if the 1400-byte payload fails while a smaller one passes, the tunnel is clamping the MTU — lower the VPN interface's MTU below the failing size or enable TCP MSS clamping on the router. Round-trip times of a few hundred milliseconds do not block the transfer, they only stretch it, so budget the time instead of hunting a fault.
When is NAT with port forwarding the right answer? #
Only when no tunnel exists and the site router can forward a WAN port to the panel: then you tick Use NAT, enter the router's public IP and the forwarded TCP ports, and the router rewrites the destination to the panel's private address. The port numbers are specific to your Vijeo Designer version; take them from the product documentation for that version. The panel still needs its default gateway in this mode, because its replies leave the subnet through the translating router. Verify the forwarded path from outside the site network — a mobile tether, not the office LAN — with a TCP test against the public address and external port; an ICMP ping to the public address proves nothing about the forwarded TCP path.
One safety note: restrict the forwarding rule to your source address and remove it once the transfer is done — a permanently forwarded port leaves the panel's transfer service reachable from the internet. The decision in one line: tunnel up — private IP, gateway set, NAT untouched; no tunnel — public IP, forwarded ports, NAT filled; never both at once.
Questions asked
Can I tick Use NAT and stay on the VPN at the same time?
No. The two are alternative transports, and stacking them puts a second address translation into a path the tunnel already completes. With the VPN up, your packets already arrive on the panel's subnet; tick nothing in the NAT block and aim the download at the panel's private address. Reserve NAT for the day you have no tunnel at all.
Which TCP ports do I forward when NAT is the only option?
The numbers are specific to your Vijeo Designer version, so take them from the product documentation for that version rather than from another project. Whatever they are, the external ports in the router's forwarding rule must match what you type in the dialog's TCP port fields, and the rule must map them to the panel's private address. A wrong port fails as a silent timeout, not an error.
Is downloading through port forwarding safe?
It is workable but weaker than a tunnel: a forwarded port leaves the panel's transfer service reachable from the whole internet for as long as the rule exists. If you must forward, restrict the rule to your source address and delete it as soon as the transfer completes. For routine remote work, keep the VPN and leave forwarding off.
Can I load the project without using the network at all?
Yes. The Download field in the target's Property Inspector offers USB, Serial and File system alongside Ethernet, so a USB transfer sidesteps routing, gateways, firewalls and MTU entirely. For a one-off visit or a tunnel that will not behave, loading from memory media is the pragmatic route; the network diagnosis can wait.
Why does the download work from the office but fail from home?
Almost always split tunneling. A split-tunnel client only routes the subnets the VPN pushes to it; if the panel's subnet is missing from that list, your ping leaves through the home router and dies on the internet. Ask the VPN administrator to add the remote subnet to the pushed networks — the AllowedIPs setting on WireGuard — and retest.