Pretty impressed with the performance for such a “lite” device.
Personally I would not have made the casing as it is now (simple box form without the frills would have been better, I think. Would fit better in a backpack).
I don’t really fancy the greenish side (or whatever that color is) but that’s purely personal preference (OTOH makes it a bit harder to miss when looking for it).
For now pretty default config.
Ethernet - ethernet no problems reaching close to Gb speed using iperf3 (950-ish Mpbs, processor hovers around 50% when doing so)
Station mode connecting to nearby AX3 and laptop on ethernet I can reach close to and sometimes just over 400Mbps using iperf3 to local server. Keep in mind, that’s using 2GHz Wifi !
Processor hovers around 41-43% here.
Nice thing. I will use it as replacement for my road warrior mAP, I think.
One remark: the label with the default password for admin and wifi is STUPIDLY small !! There is plenty of room on that label to print things a bit bigger.
Thank God for x3 zoom mode on my smartphone.
Setup:
ISP modem
RB5009 connected to it, that one using VLAN trunk to AX3, that one to NAS with iperf server (already verified using PC to RB5009 I get close to Gb speed on iperf, hence I conclude my internal LAN will not be the bottleneck)
AX Lite connected to ISP modem so no direct connection between RB5009 and AX Lite apart from Wireguard tunnel.
Verified by killing tunnel, no connection anymore to the subnets to be tested.
PC connected to AX Lite
Iperf-results from PC:
TCP: 195 Mbps down, 211Mbps up and CPU on AX Lite maxing out. Profile tool show wireguard, unclassified and networking as the huge performance hogs (wg only 15%, the other 2 around 28-30%).
UDP: 405 Mbps down, 400 Mbps up
remark: bandwith limited on down to 410M and 400M on up because above that, the counter for lost datagrams shoots WAY UP to +50% and more on several runs so useless. With limits applied I usually see 0% loss (not always, most of the times). Using this limit CPU stays around 20% so more then enough spare capacity but it’s the network part which kills things here.
If these tests should be done differently to obtain more reliable results, let me know.
That fifth port amounts to a big difference if you need PoE. The ax² has PoE in and out, while the “lite” has no PoE at all.
Lack of PoE output falls out of the very different powering options between these two: the USB-C input on the “lite” is clever, and it’ll make it a great travel router since you’ve likely got a laptop charger in the bag already, but without a 5 to 48V boost converter, it could never do PoE. The extra components needed to do that doubtless would’ve added too much cost to keep it “lite.”
(There’s such a thing as 48V USB PD, but then you couldn’t use any random USB-C power brick you had sitting around, nullifying one of this design’s advantages.)
Lack of PoE input is less easy to explain. This class of device makes a great router-on-a-stick. PoE isn’t essential to that mission, but if you’re trying to do this to solve problems with old infrastructure by turning a single Ethernet trunk line into a VLAN-per-port fan-out by adding one of these, you might be running out of power points as well. That might push you to the ax² all by itself.
Having POE (in or out) or 1 port more (4 i.s.o. 5) is not related to the original remark made by Buckeye.
Completely different topic.
On AX devices (the ones I have seen: AX2 and AX3) POE in and out are on the same port now. And that’s something I don’t really understand.
I can get it from a cost/design point of view but functionally … not so much. I much preferred how it was on AC2/AC3 (in on eth1, out on eth5).
I understood however for ISP purposes it might be better this way. Alas, nothing we can change about that.
I see AX Lite in the same category as mAP and mAP lite.
mAP also has POE out but personally, I have never used it (except once for a quick lab-setup with cap).
mAP Lite I have powered more then enough using that POE option so that was indeed an added value there. But having USB-C as power-input on AXLite makes it not that much of a problem nowadays.
For a lot of ISP devices it might require POE feeding, so the same cable as the one connected to eth1 = powering ISP device AND is configured as Internet In.
[quote=tangent post_id=996806 time=1681724917 user_id=188567]
Yes, I must have mistaken this thread as about “hAP ax lite”, its limitations, and its usages, and not a private conversation between you two. There is no possibility that someone might have found my observation interesting or helpful.
[/quote]
I’m not sure I understand why you react this way …
There is no problem with the comment you made w.r.t. POE (not at all ! It is a very valid comment), it is however not related to the quote you used and the context of that part of the conversation where you pulled it from.
If 802.3 PoE is supported by Mikrotik devices, they support either af or at variants. 802.3af goes up to 15.4W (at PSE, less than that on PD depending on power losses on UTP cable) and 802.3at goes up to 30W (same considerations apply). If PSE is 802.3at, then device could support one 802.3af client (if we take away own device consumption) and even then it could violate 802.3 standard regarding output voltage (device doesn’t regulate voltage and if voltage drop between original PSE and device is high, it would fall out of standard range (which is at least 42V on PoE out port).
Now, hAP ax2 and hAP ax3 only support passive PoE with input range of 18-28V. Both have PoE out current limitation at around 0.6A (0.625A for hAP ax3) and that happens to be in line of many Mikrotik PoE out devices. I’ve yet to see Mikrotik document which would mention PoE in maximum current, but given that at least in this case PoE in both directions shares same port, I think it’s safe to assume that PoE in current should likely stay below 1A (long term) to be on safe side (not to cook some sensitive element of ethernet port).
Now, most passive PoE devices come with 24V power brick (and the two devices mentioned even can’t take higher voltage), which means something around 24W of total power consumption. hAP ax2 is rated at 12W and hAP ax3 is rated at 15W.
Which leaves pretty small power budget for PoE out when device is powered via PoE in (true both in passive and IEE 802.3 PoE cases). And that is, I strongly believe, rationale behind making one port a PoE in/out combo.
That was definitely not the intention.
If that was the impression you got from my response, then there is a nuance I missed (English is not my native language).
My point about having 4 vs 5 ports on a vlan-aware switch was that with 5 ports, you can have internet on one port, and two “two port switches”, where ports 2 and 3 on vlan x ports 4 and 5 on vlan y. That can be useful if you have dumb switches connected to ports 3 and 4 both going different areas e.g. guest/iot on one switch and home on the other, which still having convenient access to each of the two vlans at the router itself. Of course that “problem” can be solved with an extra vlan aware switch near the router.
It would have been nice if it could have been at least powered by PoE, but then what type? It would be most useful if it coule be powered by 24V passive as well as standard 802.3af, but that probably would have changed the price point and perhaps there would have been sourcing issues on some parts as well.
I am surprised that no one has been able to confirm whether the hap ac lite will work with zerotier, and if so, how it performs.
If you point me to a cookie-cutter setup guide, I’m willing to try
(again: haven’t spend much time on it since I do not believe (yet) in that route since I have more confidence in my own wireguard setups for my purposes. But always eager to learn …who knows …)
