Do not understand

[sirhc]

Well we are always sorry when something slips past us like the 3.3V CAPs. We do fully test each switch before we serialize them but in this case they would pass our testing procedure. We did test all Ethernet ports but we did not test all ports active at once. We have since changed our testing procedures to catch something like this in the future.

With the C51 CAP on the 250 watt power supplies it only showed if using less than 40-60 watts and the switch had to be running anywhere from a couple hours to as many as 12 hours before it exhibited the issue. We have since had the manufacturer change their testing procedure of the power supplies plus they changed their BOM (spec a better CAP). We also now do random spot 24 hour burn in tests to help catch something like this in the future.

The Ground MOD is not a defect just something we learned over time examining RMA units that would greatly reduce damage to units from poor grounding. Mind you with this MOD the Ground Current is still traversing the Ethernet Cable as you can not stop this no matter what we did because some radio manufacturers are bonding DC negative to Earth Ground instead of using isolated circuits. Ground current traversing the Ethernet cable is a bad thing that can cause other issues like Ethernet errors. This is why people should do better grounding, we are making the switch handle the current much better but now this current is still traversing the radio to get to the tower so will it handle this ground current????

To test if a radio is bonding DC negative to Earth Ground use an Ohm meter and measure Ohms between a DC negative PIN such as PIN 7 and 8 of the radio and the outside of the antenna connector (SMA or N-Type). If you see near 0 Ohms then it is bonding DC negative to Earth ground and ground current will flow across your Ethernet cable if the electrical service ground rods are not bonded properly to the tower ground rods.

If the unit is an all in one then test Ohms between PINs 7 and 8 and their mounting bracket if it is metal.

Radio like a NanoStation is all in a plastic casing so it is isolated and not a problem.

Now the ground MOD will also help you against Static or ESD charges at the radio/Antenna is you are not running a dedicated ground wire / path as that charge will follow the Ethernet cable down to ground. Static charges build up from wind blowing across the tower/radio/antenna and the humidity level is just eight which often occurs in the Spring as sudden rain event s often follow a sudden temperature change which cycles through the humidity levels just before it rains.

You can RMA the WS-8-150-AC units and we will fix them, sorry.

We could also put up a how to for those that would rather just replace the CAPs themselves? It's not any harder to do then the Ground Current MOD.

We try to be honest about everything and we also try to explain what the issues is which is not the normal way manufacturers handle things like this. The industry standard is to simply deny everything under most circumstances and not many provide details on how to make the fix yourself if you want.

If you want to enact the Ground MOD then simply program one up as a replacement then swap it out. Once you get that unit back to your office you can apply the ground modification and use it to swap out the next unit and so on. This is what we are doing at my WISP. Currently our goal is to have them all swapped out before the rainy season starts when rain is the enemy as it messes with ground rod potentials. Currently we are about half way through the swap out. We are doing the swap out after 10-11PM and total down time per tower is about 5 minutes. You can also use this maintenance window to test and swap out any batteries.

[sirhc]

I do realize there is added cost for RMA when international shipping is involved. Keep in mind that Netonix bears half of the added cost as well with their return shipping on warranty RMA.

Yea we really do feel bad about this shipping incurred. If we get big enough we are thinking about opening RMA centers in countries. Like one in EU and Latin America.

For now they can ship it back via postal mail which is not too bad and for now we always return via UPS or DHL which is costly for us so we want to avoid this too.

We also offer that if an international customer sends in an RMA(s) either for warranty or repair they can then place an order from the web store for as many switches as in their RMA and select "Warehouse pickup" and put in the comments "Ship with RMA ####" at checkout and we will then ship that order back with their RMA(s) at no extra shipping charges.

So if they RMA 4 switches they could buy up to 4 items from the web store to have shipped back with their RMA.

They are only required to pay for the return shipping of those units not covered under warranty the rest ship free with those units at our cost.

[ste]

Due to this problem we looked thru all our netonixes and found 2 additional WS-6-Mini with fluctuation voltages shown at the status page.

Shipping cost is one thing. But it is neglectable to the problems caused. A dead Site-Switch is an outage which has to be immediately fixed (day or night) and which causes us problems with our customers. So someone (often me) has to drop everything. Pick up a replacement switch, configure it, drive, install, ...
Someone has to handle the customer calls while the outage occurs. So a day full of work is easily gone.

WIth the CAP problem we should have gotten a warning. Switches sent within the problematic timeframe had to be called back and tested for bad cap.
So you let us run into problems without notice. May be a common behaviour but nothing I have to like.

[sirhc]

So you let us run into problems without notice. May be a common behaviour but nothing I have to like.

Maybe if you ignored all the posts, discussions, and firmware release on the forums regarding this you could say we did not tell/warn you?

The fluctuating voltage is a sign the CAP is not in the desired tolerance range but this does not mean it will fail, there is an acceptable range of fluctuation that can occur before it causes an issue. It will however cause the false voltage fluctuations to be reported on a larger percentage of units which the firmware will fix. Only a much smaller percentage will ever exhibit the reboot issue which is caused but the 3.3V dipping below 3.1V which triggers the reset circuit. The vast majority will simply see the 3.3V fluctuate or ripple 800+/- mV which messes up the reported 24V and 48V voltages not the actual voltages be put out. It will not get worse over time, it is what it is as the MTBF of those CAPs is far longer than the expected life of the switch.

If you want to find out if it will fail simply upgrade the firmware to v1.4.7rc14 and put it on your bench and link up all ports, they do not need to be a POE device just a link.

Let it run overnight (this is overkill as it should lockup relatively quickly), if it does not lock up it never will.

But we did tell everyone as soon as we knew about the issue, here is one such post:
viewtopic.php?f=17&t=2101&p=15646&hilit=+defective+voltage+cap#p15646

Now this recent defective CAP issues effecting the WS-6-MINI and WS-8-150-XX models is excluded from the above numbers and is covered under warranty as a free repair and we simply remove the CAPs and replace them and send back. Symptoms voltage fluctuations and in rare instances rebooting can occur. The percentage of the units afflicted with this is less than 1% and most units can be fixed by upgrading to the latest v1.4.5rcX firmware as we can correct the issue to a limited extend in software. This issues has been corrected on all units being built as of 9/7/2016

If we knew exact serial #s that were defective we would tell people but since the CAPs have a +/- tolerance the best we could do was a wide sample test which we did and estimated a percentage that may be affected.

We then told people this issue existed and will be covered for the life of the switch. We changed our BOM, and we changed our testing procedure to catch it in the future.

We did release a firmware patch to compensate for the fluctuations so that false voltage alerts would not be sent as the 3.3V is used as a reference voltage for 24V and 48V readings and a little fluctuation of 3.3V makes the reading for 24V and 48V swing wildly. The 24V and 48V are not really swinging just our reading is off.

v1.4.5

FIXED/CHANGED
- Fixed voltage fluctuation that affected a small percentage of WS-6-MINI, WS-8-150-AC, WS-8-150-DC

v1.4.7rc14

FIXED/CHANGED
- Fixed switch rebooting if multiple ports lose link too quickly - RC14

Look, we are sorry something slipped past us;
- We absolutely did tell people it existed.
- We told people how to check for it.
- We extended the warranty on this issue for the life of the switch.
- We offered creative deals with free shipping on new orders for international customers to offset the cost to RMA them back.

We also offer that if an international customer sends in an RMA(s) either for warranty or repair they can then place an order from the web store for as many switches as in their RMA and select "Warehouse pickup" and put in the comments "Ship with RMA ####" at checkout and we will then ship that order back with their RMA(s) at no extra shipping charges.

PLACE THE ORDER AFTER YOU GET YOUR RAM #

So if they RMA 4 switches they could buy up to 4 items from the web store to have shipped back with their RMA.

They are only required to pay for the return shipping of those units not covered under warranty the rest ship free with those units at our cost.

All we can do is try harder in the future to catch things like this which we are. I think we do things differently than most manufacturers in that we try to be honest about things like this. We even offer people information on how to test for and fix the issue themselves which is almost never done in this industry.

When people find an issue that we feel is something we may be doing wrong we jump right on it and come out with a solution as fast as possible. In this case it took awhile as it only affect a small percentage of units and our testing at that time did not expose it. As as I said we pretty much do an autopsy on all RMA units and after a period of time something clicked and the issue was exposed. Then we did a WIDE sample testing of hundred of units which took days. Then we discussed internally to find solutions and we did what we could. All boards already built (thousands) in packaging ready for assembly were hand modified by Julian. All units in the warehouse that were built and boxed were unboxed and had their boards removed, fixed, retested, and reboxed (thousands of units).

We did the best we could is all I can say. But sadly there are hundreds of units out there that "may" have the issue which is why this issues is now covered for the life of the switch. If the units are outside the USA we work with the customer to help offset the cost of return shipping.

[sakita]

One of the many things I really like about Netonix is that you share details that other companies never do. It seems to me to be especially so now that UBNT, for example, is publicly traded that certain things don't get talked about for fear of hurting the stock price (or maybe just damaging their groomed image). There are certainly multiple examples (some come easily to my mind anyway) where it was mostly silence when issues came up. And when it came to addressing certain issues they just bailed (and I've got the hardware to go along with some of that). It is such a striking difference that you are so honest in a world that runs more on the political model. Personally, I prefer full on honesty. Maybe some people just can't handle the truth (insert movie sound clip here).

So for the new features, we don't use QinQ or MSTP at present. IGMP snooping is something we'll be test / using (we recently updated our core routing gear which will help us move forward with multicast). Our MODBUS equipment usually needs a straight up serial port for programming or it has an Ethernet port for MODBUS. What I could use, however, is TCP to serial. This would help us eliminate port servers at locations where we just need one serial port for older equipment that requires it.

[sirhc]

All of our NEWER switches coming later this year possibly with *cough* SFP+ but they will also have DUAL serial interfaces, one dedicated to MODBUS and remote serial devices and one for console functions.

[IntL-Daniel]

3) Defective CAP (C51) on 250 watt power supplies - Covered under warranty
This affected 1 batch of power supplies used in the WS-12-250-AC and WS-10-250-AC. The symptoms are if the power supply is not loaded with at least 50-60 watts the unit will reboot after several hours and keep rebooting quicker and quicker until eventually it gets stuck in a reboot loop. If you let it cool down it operates again as it did before for several hours. We have extended this issue to be covered to the life of the switch so if you don't realize you have the issue for 3 years it is still covered under warranty.

Could you please confirm / exclude possibility of this issue for our WS-12-250-AC - purchase date 11/2015, board revision D, MAC ec:13:b2:61:3d:24. We can see very similar behaviour as repored above. Thanks.

[Julian]

Daniel -

I can exclude that unit, unless there's RMA history on that unit, which I haven't seen. If you're having issues, we will gladly repair the unit if you open an RMA.

[IntL-Daniel]

Thanks...but could it be possible to just say first, that my unit could be affected by reported known hw issue of CAP (C51) or not? Thanks.

[Julian]

Thanks...but could it be possible to just say first, that my unit could be affected by reported known hw issue of CAP (C51) or not? Thanks.

I can exclude that unit

Your unit is not within the affected range, I apologize if that was unclear