Reply to Thread

Thanks Guys, I ordered 2 to start. Looking forward to giving them a try!

Same, right off their web site.

Been getting them directly from their website. Normally have them in about a week.

Hey Guys, where are you guys buying these from? Is there a recommended reseller?

Thank you all.


Sent from my iPad using Tapatalk

Originally Posted by JSLLC4Life

On the flip side, I have had supervisors with BACnet drives with 400 or so devices that were unusable without splitting the devices among multiple ports.

Wow, I was seeing the network was a mess at 1100 devices. And I was pretty happy with the tuning and response I got. But I haven't seen anything in between around 175 devices and that 1100 site.
I guess I got lucky it was usable at all.... it was clearly still not zippy. But I was able to get it usable and not too noticeable.

The downside to multiple BACnet/IP ports is the JACE acts a router so if the JACE goes down, you lose comm between the networks. I have a site I moved from the RS-485 expansion cards to to 6 routers, all using 47808. The JACE really perked up and no comm issues. There was around 250 devices.

On the flip side, I have had supervisors with BACnet drives with 400 or so devices that were unusable without splitting the devices among multiple ports.

In general, using the local RS-485 ports on a JACE will take up more cpu cycles than if you let a router handle the token passing.
Tridium says they improved this in Niagara 4.12 somewhat by offloading some of that to a co-processor on the JACE, though I haven't personally evaluated this.

Now if you have multiple IP to ms/tp routers, then separating them onto individual UDP ports (e.g: ipPort 47808, ipPort1 47809, ipPort2 47810) does indeed improve polling cycle times/

Here's a response from Tridium's developer on this issue:

Your observations are correct. Every NetworkPort has a single PollService with two poll threads. Your workaround isolates each MS/TP trunk and adds poll threads, significantly increasing overall throughput. Instead of capping overall throughput by the slowest (and second slowest) MS/TP device in the entire network, in your scenario overall throughput is determined by the sum of the total available MS/TP throughput. There is currently no “out of the box” solution better than your optimization.

Here's the thread on niagara-community:
https://www.niagara-community.com/s/...er-performance

Originally Posted by lin

Of course 3 ports and routing function will use more resources on JACE, that is a tradeoff.

It will take more CPU, no way around that. Mo data, mo cpu cycles. It doesn't seem to impact CPU on a J8000 a huge amount, but the BACnet performance boost is very noticeable. Don't hesitate much to set it up this was if I have multiple loaded up MS/TP trunks pulling in a single Jace.

I have at least a few J8000s with 4x IP ports dedicated to each MS/TP router with more than 200 BACnet devices on it and its not even close to CPU bound. Its pretty much required if you want to push the number of MS/TP devices to that number.

The vanilla bacnet driver has a limit of 2(?) Threads per port. The Niagara Bacnet AWS/OWS driver allows one to manually assign a number of threads.
I was on a site with 1100 Bacnet devices. Well.... maybe 10ish Alerton BCMs, that had 30 to 100 mstp devices each, maybe 7 or 8 JACES with MSTP devices, and a bunch of other Bacnet IP devices (like energy vamlves and meters.... 1100 to 1200 total devices. All done on the vanilla BacnetIP driver. Lots of traffic. On a server licensed for devices directly.
That was the first time I REALLY noticed this limitation. I had to really lay on the tuning policies, throw everyone into slow policy, and then manage things to get some important (read:customer was watching these things constantly) things to update in a timely manner.
I would have loved to divide it into 3rds or quaters - each on different ports, but it was not going to happen for a variety of political reasons....
Hopefully that real world experience helps quantify it a little. I have also done 3 routers on 1 JACE - each on successive UDP port numbers. Didn't hurt at all and got the job done. My uneducated opinion is that multiple UDP ports are not useful/helpful/noticeable if there is fewer than 100 devices per port. But that is my guesstimate. Someone will pull out a Niagara document proving me wrong. I have seen a network of around 100 devices plenty of times with no significant lag all on the same UDP port.

I have bare experience on JACE.

Someone here had mentioned there is a polling thread limitation for each port on JACE.
Setup 3 BACnet networks will speed up polling, but you have to enable routing function on JACE if you need devices to communicate between different BACnet networks.

Of course 3 ports and routing function will use more resources on JACE, that is a tradeoff.

Looking forward to the opinions of others

Gents, I just setup three of these BACrouters on a site. Lin these were very easy to setup and came right in as soon as I had the IP/instances right. My question revolves around having multiple BACnet/IP routers. Each router will have 60 devices on it, I'm wondering if I should use different BACnet ports for each (for example router 1 47808, router 2 47809, etc.)? I believe some of you smart guys mentioned there is a limit to the number of threads available to one port?

Wondering if there is a best practice here?
Followup question. Will changing ports impose more "load" on the JACE than leaving at one port? Because I will need to setup three BACnet/IP networks if I change the ports.

Details Honeywell JACE at N4.11. Almost all devices are VAVs with a few VFDs thrown in. I will have tuning policies applied.

Originally Posted by jschulze

The RS-485 ports on the router are isolated and it uses a 2-position terminal block with only + and -. The device does not expose its RS-485 transceiver's reference signal

Correct. Would add that both 485 ports are also isolated from each other.

They are solid and have more features than I have seen in any other mstp router on the market. Factor in the cost, seems some OEMs should up their game.

Originally Posted by numbawunfela

I like them. I would do it

I ordered 6 of them. Build quality seems good.

This thread has been GREAT. We have been looking at cost effective ways to expand our BACnet network across our property where we have ethernet/network from building to building and will likely run MSTP throughout. We primarily use Reliable Controls equipment but were looking for a cheaper solution for the initial transition from Ethernet IP to MSTP. Really great to see all of the feedback and Lin's hard work in response.

I like them. I would do it

Thinking about ordering some of these to offload the CPU from doing the heavy lifting for MSTP on a JACE running close to the limit. Any reason to give some of these a go?

The RS-485 ports on the router are isolated and it uses a 2-position terminal block with only + and -. The device does not expose its RS-485 transceiver's reference signal, therefore, there is no place to connect the reference wire to the router for 3-wire comm setups (unless, of course, you want to open the device up and solder the ref wire onto the RS-485 transceiver's ground pin).

Perhaps the idea with this design is that because it's isolated, you only need to connect the + and - and since the 0V reference of the transceiver is floating, it should tend to settle at the halfway point between the + and - signals. I can't speak as to why this design decision was made, but the recommended practice for isolated RS-485 devices is to expose the reference signal on the terminal block.

I only land the 3rd wire if it measures significantly different than 24v common as measured with a voltmeter. Since the 3rd wire is the electrical reference for the rs485 comms on the 3 wire side, and so is the 24v common on the 2 wire side.
If I land the 3rd wire, it is through a resistor. I use a 120ohm termination resistor. The rs485 spec calls for 100ohm, but the termination resistor is handy and works fine.