I have tried to connect 3 cAP ac after each other on one PoE port on a hEX PoE. P1 in on #1 P2 out, P1 in and p2 out on #2 plus P1 in on #3
The two first work fine but number 3 seem to get only power. Not visible in WinBox.
Isn’t there enough power at the end? Cable length between the cAP’s are 12-15m
cAP and cAP ac have significantly different power requirements. On the product drawing, you never see a cAP ac behind another cAP ac, it’s always only cAP that is powered from cAP ac’s PoE-out port.
A cAP ac powered from the PoE port of another cAP ac may work if the traffic is low, but once you connect more active wireless clients, you will exceed the power capacity of the hEX and get reboots or traffic loss. The design apparently assumes that a gigabit Ethernet connection is enough for a single cAP ac but would become a bottleneck for two, so there is no reason to provide enough power for the second one.
Put it otherwise:
Hex POE uses a DC/DC of 24V
Max 2A per port in that voltage range.
That’s 48W max per port.
cap AC uses 13W, can go up to 24W.
Times 3 when you chain one after the other and you’re already pretty close to the limit for doing nothing … when they are active, they will go over that total limit.
Secondly (and more importantly): output current of POE Out on capAC is only 500mA. Assume that same 24V that’s only 12W.
You can (should) not connect another cAP AC after a cAP AC.
cAP only uses 4W, that’s no problem to connect after cAP AC (and then it stops since cAP only has 1 ETH port).
Thanks for the information.
Would it work if I added a PoE injector on cAP ac 3 and up?
Can only pull one cable.
/thomas
No.
See my second remark.
It would be needed for each cap ac after the first one.
OK.
thank’s all
Back to the drawingboard
/thomas
The PoE power is limited for fire safety - over a wire of a given cross-section, you can only allow a certain amount of current to prevent overheating. Many factors come into play, such as number of cables running in parallel and whether they are in a solid wall, free running, or in some sort of thermal insulation; of course, the design relies on worst case and hence the 802.1af only allows 15 W @ 48V which is some 0.3 A. So even the 2A permitted on hEX PoE port are way above that.
Having said that, if your installation permits that (a single cable cooled down by mortar/plaster), you can use some dirty tricks, like extracting the power using external passive hardware. But as the cAP ac has no barrel plug, you need three extractors/injectors next to it - one to split the power and data coming from the hEX PoE, another one to re-join them on the path to the next cAP ac, and a third one to re-join them for the local cAP. So the solution will be quite bulky. To reduce the risk of fire, I’d highly recommend to connect the hEX PoE to a 48 V power supply - this way, you will deliver the same amount of power using less current, so the loss on the cables will be much lower (the power loss is proportional to square of the current).
Hello Sindy,
Thank you for your ponits. Well taken. We will see if there might be room for one cable to each cAP ac.
Will be a job for the constructor.
/thomas
Hi,
I am running at home 2x CAPac (ap2+ap3), daisy chained like this: PC–firewall—switch—hEXpoe–ap2–ap3
There is a hEXs hanging on the switch and running CAPSman, ap2+ap3 are configured with local forwarding. Everything is running ROS 6.49.7. The frequency of the CPUs of the APs is configured as auto. The hEXpoe is connected to it’s original 1.2A@24V power supply.
They are not really highly loaded, just used by my family of four persons and I did not see any problems with them. (After some weeks of pain spent on the initial tuning…)
You have made me think however and I ran a test: I have enabled on my PC and ap2+ap3 a bwtest server. And I am running TCP bidirectional tests between PC-ap2+PC-ap3
I didn’t want to measure the throughput, but I wanted to generate a lot of CPU usage on both CAPs, which I have successfully reached (85-100% on both).
On the HEXpoe port statistics 24Vs are shown with 420mA max, power is 10.1W max.
If the cAPs are not running any tests, power consumption goes back to ~335mA@24V, roughly 8W.
Running the tests also put some load on the switch chips as well. The only thing I did not put under more load with the tests, are the WLAN chips, but they are active anyway and have something like 15 Devices connected permanently to both 2.4GHz&5GHz bands.
So I conclude, the 13W max power consumption given by MT is a very conservative rating, probably also taking it into the account that the CAPs do not necessarily get the optimum voltage and some safety overhead.
It looks like to me, that it’s perfectly fine to daisy chain 2 CAPac devices. Probably daisy chaining 3 of them is a bit too much.
Also with 802.11ac using both bands simultaneously I never got near to 1G speeds. I have lot’s of interference from many neighbors on both bands.
I am only using 40MHz channels on 5G and 20MHz on 2.4G. The highest bit rate I see is 324Mbps. The end to end throughput is even lower.
This is still perfectly satisfactory for streaming videos, video conferencing and surfing the web.
For bigger bandwidth, there are RJ45 sockets or ax and hopefully the 6GHz band in the future…
Thank you Woland.
Your ponits could help a lot and reduce the cabling.
Well worth trying
/thomas