With this argument, then only one radio should be allowed on each tower. Sorry to inform you, but most people have successful installations with more than one radio per tower. Have you ever been a WISP?
No, and I refuse to become one. :-)
I suggest that you get some practical information from WISP that use more than one AP/radio per tower. There are plenty of clever WISP engineers that experiment and decide how to invest their hard earned money.
No.
I've delt with far too many of these people. They think they know what they're doing, and they don't. Quite honestly I had to deal with corporate visits from a lot of them early on at Vivato.
I don't know about the details of your example -- and without them this example is useless. If they were using 200mW senao cards, then they would have big problems because the onboard amp puts out allot of noise -- but this does not happen with the 5GHz Ath cards.
Sorry, its not a PA NF issue (and the Senao cards are no worse than many others).
That doesn't match my test and others that see the same as me.
I'm sure you're measing more noise, but the PA amplifies the noise that comes out of the transmitter (no transmitter (or receiver)) is noise-free) and adds its own.
That said, its a pretty good PA, given that its operated near its P1dB.
But it is a in-channel noise issue (due to operations on the adjacent or alternate channel, and the 23dBm tx power of the senao cards makes things worse,
Especially when the on-card PA makes allot extra noise.
You're confusing the issues. Its not card specific, and, in fact, its worse with "modern" 11g cards.
but (functionally) better than using a large antenna to generate the same EIRP, because that antenna gain will amplify the incoming signal as well, which will raise your in-channel noise power.
Of course, this is basic.
The number of people (even WISPS) who can't understand even this is phenomenal. They'll stick a bunch of 12-15dBi omnis up and then complain that other people are 'interfering'.
You sound a little arrogant -- you must be so clever to be the only one who gets it, and you have never experimented or even touched this equipment. If we only had more people in the world like you.
I used to think like you do now. No really. I was a software guy (still am), who though he understood RF becaue he passed E&M physics.
Then some very kind people (who happened to work for me) explained it to me. One of them just took a job as Intel's new CTO in their wireless division, and is world-renowned for his knowlege of all matters RF.
Vivato was a 2 year period in which I probably obtained most of a 4 year degree in RF issues.
I don't think so. It is much too complicated at the low level now. You need a PHD to start to be taken seriously.
Really?
Siavash Alamouti doesn't have a PhD, and he's taken quite seriously by the community. (He's the new CTO @ Intel's wireless division.) There are others, of course, and some at Vivato who had PhDs (who worked for Siavash, and probably will again soon) were also hella-smart.
And all of them understand what I'm talking about here. They think I understand too, and I don't have a PhD. (Must have been the simulation work I did...) For now, I'll take their word for it.
I'm not an expert my any means, but I've been down the path that your customers are facing.
I also watched a lot of money get wasted on "multi-channel" operation. By "a lot" I mean something well north of $50 million. Its all in the blog.
As I said, talk with WISP that have more than one radio per tower -- they have to compete in the real world. They are not climbing towers and installing multiple radios for fun.
I didn't say they were. I have allowed that the issues can be overcome, if you know what you're doing.
Lots of people run dual 11g cards in a single AP as well. They don't get it either. The comment about "extra capacity" is especially amusing. IALMAO.
I have only tried on non overlapping frequencies on four and more radios on 5GHz.
I think I've showed this. If you mean that you've run on four "non-overlapping" frequencies in a single (U-NII) band, then I challenge you to reproduce the results "at range". I'm confident that you can't (without a lot of additional filtering).
As before, real world experience shows the benefits of multiple radios per tower out
Wait, you didn't *try the experiment*.
Also, I have found the 11g doesn't do well with 11b. We have an 11g only setting that does help some -- in general, not pertaining to the multiradio discussion we are having here.
Actually its very nearly the same issue, especially if you're attempting multi-channel operation.
I think it is mainly a different issue in that to support 'b/g,' allot of potential highspeed data transmission time is lost supporting beacons and 'b' slot timing -- even if there is not 'b' traffic or 'b' interference for the 'b/g' AP.
Thats only an issue if you have a 'b' client associated with the AP. Otherwise, you don't need to send the beacons or support the 11b slot time.
Of course, if you *do* have an 11b radio associated, then you have to send the protection frame before any OFDM transmission, as well as back-off the slot time (SIFS).
Beacons, of course, are always sent at the lowest "basic" rate. If your BSS supports CCK traffic, then 11Mbps had better be in your "basic rate" table.
Otherwise, if you're finding that 11b doesn't interwork with 11g, then its either your implmentation, or the cards (or both). I'm pretty sure the protocol works.
There are a lot of stupid people in the world. Might as well take their money, right? I take my hat off to you for supplying what people want, but don't need. I'm sure you'll sell a lot of these.
This line of argument won't get you far in economic theory or philosophy. Again, for someone who has never touched these cards and made a setup, you can't expect to be taken seriously.
I don't need to touch your 4-way or 8-way adapter to understand the issues with running multiple miniPCI cards, especially on multiple channels.
I sometimes think like this too, but as I have aged, I find that there is no substitution for hand-on experience. Keeping an open mind is best.
Since I can show the issue with 2 miniPCI cards...
See the PCI2.2 standards and
you will see that the max power usage per PCI slot at 5V is 20W. So, 8x2=16W and consider that we have an efficient power controller (probably about 85% efficiency -- I would have to check), then we will be below the 20W PCI slot power limitation.
The issue is the 2.5W max power per slot. At 3.3VDC (the only supported voltage on miniPCI), thats 750mA. Right?
I can't find the emails that I have sent with page with page numbers on the PCI and MiniPCI specs, but you can believe me that it is 2W max. And all the cards we have (excepting what I have mentioned) claim to conform to specs. You have made statements without claim references and it is up to you to back them up (at least say that you have references for what you say -- and better, point to page number in the specs). I will do it if you won't concede this issue -- but we are all busy. You know, this is the basic research we do before designing cards and claiming compliance with standards.
I doubt you've "designed" your miniPCI radio cards, but perhaps you have. If so, you'll no doubt have the current consumption figures for them. You don't provide these on your datasheet, please quote them here.
I never thought you were making it up. But I think you've not been honest with your customers about what is possible, and what isn't.
That is a very strong statement you are making. I think you have had too much coffee or something at one time. See the response to Jack Unger for more on what I say is possible. Personal attacks are not the correct way to address technical issues.
I'm sorry that you think I'm attacking you personally. I'm not. I have offered opinions, which you have mistaken as attacks, and for this I apologise.
I accept the apology and look forward to addressing technical issues.
I'm sure that many of your customers want an 8 slot board.
I'm sure you'll sell a lot of them.
I'm sure that many of your customers won't understand the issues.
As said, our customers experiment in the real world and know what the benefits them and what doesn't. Listening to customer can give you allot more insight.
I emphatically agree, but the customer is not always correct.
Jim
John www.mikrotik.com
Jim
John www.mikrotik.com
Jim
John www.mikrotik.com
--------------------------------------
Guess all we would need to do is mount a Micro-ATX 2ghz processor board at
the top of the tower, and be done with all the extra radios now! Thank John!
Who else in the US aside from Eje carries your products please? Can you list
all sources ?
JohnnyO
>From: "John Tully" <[EMAIL PROTECTED]>
>To: <[EMAIL PROTECTED]>
--------------------------------------
John,
That is absolutely wonderful :-) but what do you suggest we use to power that puppy with? Do you have something with enough CPU horse power to push those 8 cards as I have started putting all my gear on the towers?
Thanks,
Mac
John Tully wrote:
RouterBOARD 18 -- 8 port mPCI to PCI board (Extreme)
----------------------------------------
I think it's great to see this out now... If you've got sectors and
pass-through back-hauls, you've got 6. Then you've got a couple extra ports
for redundant backhauls or extra capacity.
-------------- Mike Hammett Intelligent Computing Solutions http://www.ics-il.com ---------------------------------------- John, Can I trade in my unused 4 port version :) To think I just recently bought a 4 port and needed 6. >From: "John Tully" <[EMAIL PROTECTED]> -----------------------------------------
