About Loud Deaf Cards...

[ddvzlnz]

Asking a lot of questions because I'm on the verge of snagin some gear...

I want to find out which one is best in a pc (via a pcmcia - pci adapter) and which one is best in a soekris. (and supported by star-os of course).
Have any of these things been FCC certified yet?

I've read here and there that the Senao (and other high power cards) are deaf, but their published "recieve sensitivity" specs are better than Ornico's and Cisco's.

The NL-2511CD PLUS EXT2 seem to be the same card, but don't have any recieve numbers on the netgate site. I consider that to be an intolerable ommision of information.

The Microtik states -95, -93, -91, and -89dBm for it's card.

Thats not deaf.
How true are these specs?
Below is the list of relative card deafness. Sorry if you have seen it before.

Thanks
GT

I've cut and pasted the table below out of an ebay auction for some seneo cards. (the auction at http://tinyurl.com/3uhj)

Courtesy of: FreeNetworks.ORG

Card 1Mbps 2Mbps 5Mbps 11Mbps
Senao (200mW output, MMCX connectors) -95 -93 -91 -89
smartBridges airNIC (USB, SMA connector, 32mW output) -95 -90 -89 -84
Cisco 350 Series (100mW output) -94 -91 -89 -85
smartBridges airBridge Outdoor (RJ45, N Bulkhead, 100mW) -94 -88 -87 -84
smartBridges airBridge (RJ45, SMA connector, 100mW) -94 -88 -87 -84
Lucent Orinoco / Compaq WL110 (32mW) -94 -91 -87 -82
Enterasys / Cabletron (15dBm = 31.6mW) -93 -90 -87 -84
Netgear MA101 (USB 30mW) -91 -89 -87 -84
Proxim RangeLAN-DS -91 -89 -86 -83
Senao (100mW no EXT antenna) -91 -89 -86 -83
DLink DWL-650 -90 -87 -84
D-Link DWL-520 -80
Prism II Datasheet -87 -84
Cisco 340 Series -90 -88 -87 -83
Zcomax XI-300 -90 -83
Zcomax XI-325H (100mw) -92 -89 -88 -83
D-Link DWL-120 -83
Zcomax XI-325H (200mw)
Alvarion PC-DS.11b (18dBm) -89 -86 -84 -80
Tellus AP -88 -88 -88 -83
Alvarion DS.11 bridge (250mW) -93 -90 -88 -85
DLink DCF 650W/K (Compact Flash) -88 -86 -83 -80
Intel 2011 -87 -85 -84 -81
Linksys WPC11 (20mW-100mW) -80 -80 -80 -76
Global Sun GL2411-02 (20mW-100mW) -80 -80 -80 -76
Teletronics 11mbps outdoor card -80 -76
SyncbyAir -80 -80 -80 -76
Addtron AWP-100 -80 -80 -80 -76
Tranzeo TR-1000 -00 -00 -00 -84
Nokia C110/C111(35mW) -94 -83
Yakumo WLAN PCMCIA (35mW) -90 -80

[georgew]

Here is the problem with specs...

Some resellers publish the manufacturers spec, as tested during FCC testing by a independant lab using calibrated lab instruments.

Other resellers notice that the readings they get off of some software they run are lower, so they publish those.

Recently I notice a lying reseller quoted on a freenet site, then other resellers quote that freenet site, and over time the lie starts to spread, because it is good for sales.

The fact is there are not that many of the high-powered cards on the market that are FCC certified in the US. If you research the "official numbers from the manufacturer, you will get something closer to the truth.

The Senao card''s manufactur web site shows -87, -89, -91, and -93 db as the 11, 5.5, 2, and 1mbit sensitivity levels. Microtic shows the same card as -89, -91, -93, and -95... 2db is not a small difference! Also, one reseller, the owner of Netgate, told me that their 200mw cards were really only 22.5db rsather than the full 23db. The FCC testing spec agreed, stating the card was certified at around 165mw. The reason Jim doesn't publish the full specs on his site is simple... If he publishes the truthful spec, people will buy the same cards from someone else that was willing to lie. By not publishing the spec, he lets people figure it out for themselves, either by doing their own research, asking him, or assuming that they must be the same as the cards sold by the people that lie. Is it a lie to not disclose everything? Since few companies properly documents their product, everyone has some level of lying by omission.

Last time I sat through some TV commercials, at least 50% of the claims were based on less than realistic numbers. The fact is it is up to the buyer to throw the glossy literature away, and figure it our for themselves.

Keep in mind the amp in the cards will do more power if tweeked with a software setting... however this will reduce sensitivity, and most of the extra power is output as noise, so the cards become worse rather than better.

So when Demarc was taking 100mw cards and cranking them to 180mw, they were no longer FCC certified, and probably not as good as they were at 100mw. Interestingly, you have to ask, why did he tweak them to only 180mw, the card's amp is a 200mw amp. The answer is probably "the card stopped working at that output level".

As for certifying things, certification is for systems (card, wire assembly and antenna) that are sold together as a unit to end users. Strictly speaking, as long as you don't mass market your routers to end users, you don't have to have the equipment FCC certified... If you sell the units to OEM's only, or install them yourself, keeping a careful eye on keeping it legal, certification is not a requirement. If you put your own units together for mass market sales, an easy way for certification is to use antenna and radio combinations that are already certified to work together.

In my testing, the senao and zcomax cards failed to work as well as ORiNOCO cards. The Cisco cards also failed to work as well... though that testing was less extensive because it was before I started using StarOS.

So while I agree that it is unacceptable that you can't get honest and unadulterated information... this is also a normal thing. As the buyer, it is up to you to know the product better than the seller. And you can't go by manufacturewr spec, you have to try it yourself.

George

[David L. Vrablic]

I have a Non LOS shot from my house to the office.
This end has a 24 dbi GD and a mast type 500 Mw amp after 50 ft of LMR and 4 connectors.

CARD TYPE QUAL SIG NOISE

XI-300 28/92 75/192 3/192

XI-325H 92/92 148/148 12/148

Luc Silver 11/92 -74 -85

Just changed cards nothing else.

The office end stayed the same for all three. STRANGE.

[lonnie]

Actually it is not strange that the other end did not change. No matter what you fed the amplifier with it still only put out 500 mW. The thing that surprises me is that you did not hurt the amp when you fed it with the 325H which would have WAY overdriven the TX input.

The 325H does not seem to be deaf, though, from your results. It has way better receive than the other units. Now if they could just make the other units with such a good receiver (no need for more tx), we would never have to exceed any limits for the transmit power. It does not matter matter which side the gain comes on, except the regs prohibit too much tx power.

[David L. Vrablic]

You did read through what I thought was strange.
It was the receive readings.Not the xmitt.
I was very surprised of the big difference in receive signals.
The 12 db noise was right as I expected. Most likely the amp.

The TT amp specs call for an input range of from 7db min to 23 db max.
(Thats 5 to 200 mw isn't it?)
The XI-325H specs say it is 100 mw (20db) Not an HP.

I was just playing around with cards instead of plowing out my driveway.
24 inches and a blizzard, still everything worked well. I am using these HP cards on the upgrades that are in the 17 to 20 SN range. Just to have a little head room.
Thanks for the reply.

[David L. Vrablic]

When I said HP card in the last post I ment Higher Powered cards not the HP model.
Now it is time for bed 4:30 AM.

[georgew]

CARD TYPE QUAL SIG NOISE

XI-300 28/92 75/192 3/192

XI-325H 92/92 148/148 12/148

Luc Silver 11/92 -74 -85

Just changed cards nothing else.


I'm trying to interpret these numbers, and I'm having problems.... It may be too early in the morning...

But what are these numbers? The -74 and -85 I understand, but I'm having problems understanding the rest of the numbers...

George

[tony]

The rest of the numbers are raw values, and should be mentally converted into % to make more sense.

Typically, the higher the number the better, 148/148 represents 100%

Thanks.

[TheBarron]

So how do we relate those percentages/numbers to dB?

[tony]

So how do we relate those percentages/numbers to dB?

There is no method, which is why they are not displayed as such. Intersil cards are very uneven with the readings beween firmware and card revisions making it impossible to accurately calculate dBm values.

[georgew]

As far as I can tell, these numbers are all uncalibrated and are difficult to compare.

When I compared radio cards, I couldn't depend on the numbers from the cards, because they were all over the place, and didn't tell me anything I could prove. So I compared cards by putting up a AP with a 2db antenna, and with a 3db antenna and a lap-top I would test the range possible between two cards. In this case the distance I could walk and still associayte was the test criteria, and despite the numbers from the cards being nearly useless, I was able to collect some meaningful data.

In my test the senao cards were better than the zcomax cards, and the lucent cards were much better than both of these. The only calibration I did was test the cards using external measurement... how far down the street I could walk and still associate. While I can't turn that into meaningful numbers, it did comapre apples to apples.

In the test example given here, I can't tell for sure that there were any differences. The cards are clearly measuring things with different scales, so I can't line them up and tell what the numbers mean.

A percentage makes little sense as first you have to know what 100% is.. and you have to know what it is for each card, and normalize the numbers to compare them.

For example, one manufacturer might thing a s/n ration of 50 is dam good, and call it 100%, while another might think that 20db is sufficient for 100% reception, so 20% is 100%.... while a thirs manufacturer is only looking at the signal level compared to the maximum signal the receiver can see before it is overdriven.

As far as I can tell, without having the numbers well defined, and verified by an external measurement of some type, you can only "compare" the numbers from identical cards.

For all practical purposes, since the signal was so strong, and since you were using the same amp/antenna, I think what you were showing is how under identical conditions the cards measure things differently. The sensitivity of the APM is the driving factor here anyway since this is where the signal reception and initial amplification is going on. I think this just shows why you can't use the numbers from the radios for comparisons.


George

[lonnie]

The numbers I like are 0 to 100%. These are numbers that the card knows for sure. If the signal level is too low, it will be 0 or very near 0, and 100% can be counted on to be a very good signal. Another card in the same spot could show a different reading but as long as the number was acceptable to the card it is OK. Who really cares what the actual signal level is as long as the card likes what it sees and gives good results.

Bottom line to me, and some of you guys think the same way, is that results are what matter.

[lonnie]

You did read through what I thought was strange.
It was the receive readings.Not the xmitt.


This is the line I commented on
The office end stayed the same for all three. STRANGE.

The office end stayed the same on its receive because the transmit from your house did not change as the amp put out 500 mW for each and every card.

[David L. Vrablic]

Thanks for your reply,
Someday I'll learn to communicate.
Words on the right lines and commas where they are needed etc.

Looks like I am not the only one with some questions re: the "Gas Guage".Ther have been lots of questions from the Logical staff.

It gets tough when you grow up on analog.
Hit the switch and peak everything on both ends.
Now you have to start a ping to give things a "kick in the butt" and no knobs to twiddle
.
By the way, thanks for the new auto association " feature and your patience.

Thanks Dave V.

[georgew]

Bottom line to me, and some of you guys think the same way, is that results are what matter.

I agree 1000% that it is the results that count... that is exactly what I was getting at...

My whole point was that results count a lot more than numbers off a card.
My second point is that all of the numbers may have been telling the same story, the differences may have been simply how that particular chipset was made, rather than being meaningful numbers that can be compared to dissimilar numbers from a dissimilar chipset.

That is why some method of calibration is needed to make sense of the numbers. I'll work on another test, and I will calibrate it with distance.
In my past test I realize that I had 4 other wireless networks within receiving distance of my test range, so there may have been something else influencing the test. I have multiple samples of most of the popular cards, I will give it a good test for all cards with multiple passes to verify the results.

In my previous test, the zcomax cards tended to show no noise at all, which would make for terrific s/n calculations... but they were also first to [appear to] loose the signal to noise. Go figure.

George

[David L. Vrablic]

That is why I am going for the "Seat of the pants approach" when dealing with radio hardware. The thing that started this thought was the nice increase in receive readings here at my house when I compared an old XI-300 with a brand new XI-325H.

I am kind of standardizing on the XI- Prism cards for the commercial customers who use my "WILA" product .(Wireless Internet Linux Appliance) ASUS SBC etc.

The backhaul's are Lucent type cards that can use the "Long Path Opti. Feature.
I am still trying to find a good "Out to pasture" location for the dozen or so WL2400 2 M cards that came in from BSD upgrades.
They are dumb as a post but the sure do work.
I have been doing a lot of monitoring and I am getting to some conclusions regarding my deployed hardware ..
I canhonestly say that we havn't made many bad decissions thus far.

Oh well as we used to say.

"Keep your filaments hot and don't let your grid leak"

[bobbyc]

OK, I don't have much details, but I had some free time so I built a starOS box here at my house for cpe, and I tried 4 different cards all hooked up to the same 25' lmr400 going to my andrew 24dB grid.
The tower is 8 miles away and is a ydi amped lucent OEM card running on StarOS. I put my machine in infrastructure.
I put each card/pigtail into the same pci-pcmcia adapter card in my cpe machine, and SSH'd into the tower to get the incoming signal from the association list.

Lucent OEM: -69
Bromax Prism (WET11 card): -66
Demarc 200mw Prism: -62
Senao 200mw Prism: -58

All the prism cards had 1.1.0, 1.4.9 firmwares going into the test, and I had volitile firmware 1.5.6 option checked on my cpe machine. All are prism 2.5 with no internal antennas.

Clearly the senao had the highest signal output! I couldn't really test receive sensitivity.

Some other notes:
* I am going to head to one of our minicells (unamped towers) and throw the senao and demarc in to compare the incoming signals of my customers this weekend.

* I notice that the firmware, when I put the prism cards in AP mode, was like 1.2.8 when I went up the adapter config screen. Is 1.5.6 not for AP mode, or is 1.2.8 your own firmware? (I think it was 1.2.8, maybe 1.2.1)
Bob C

[bobbyc]

OK, here's my latest testing, I'm just going to give the raw numbers and then my thoughts down below:
This is ~8 miles to our YDI amped tower running a OEM antennaless agere card. Results gotten by sshing from local star-os (infrastructure) to AP tower.

Orinoco silver (fw 8.72) 19/28 -68/-66 -87/-94
Agere OEM (fw 8.72) 19/31 -68/-68 -87/-99
YDI Ruby (fw 8.42) 20/34 -67/-66 -87/-100

Senao 200mw (fw1.1.0/1.4.9) 31/NA -55/90 -86/38
Bromax (fw1.1.0/1.4.9) 20/NA -66/82 -86/44
Demarc 200mw (fw1.1.0/1.4.9) 26/NA -60/125 -86/52
Demarc 200mw (fw1.0.7/1.3.5) 28/NA -60/74 -88/14

So why the two different readings from the demarc when I changed the firmware? It is almost as if the demarc amplifies the incoming signal when using the newer firmware, or the older firmware gives bad readings and the demarc actually has very good receive sensitivity. Would the increased noise along with the increased signal would suggest that?

Check out the readings when I put the cards in AP mode, attatch to a omni, and associate to them from my orinoco client manager in the room next to the new AP.

Senao 200mw (fw 1.1.0/1.4.9) this station-52/-96 (signal/noise), test partner -66/-102
Readings from association list on AP: NA/45 36/-52 3/-96

Demarc 200mw (fw 1.1.0/1.4.9) this station -54/-97 (signal/noise), test partner -22/-93
Readings from the association list on AP: NA/47 78/-52 18/-99

Demarc 200mw (fw 1.0.7/1.3.5) Results were exactly the same as fw 1.1.0/1.4.9 when in AP mode. I would expect this since star-os loads its own firmware version when in ap mode.

So what do you think? Does the demarc really have so good of receive sensitivity that it can 'amplify' the incoming signal from my orinoco to -22 while the senao is -66? Or does the demarc misread the incoming signal level?
Thanks,
Bob C

[dkii]

I've seen this said before, I don't really think you can compare the numbers of 2 different cards, because they are all relative. The transmit signal can be compared, because you can use a reference card, in your case you used the lucent card(i believe) as a reference, however for recieved signal, I don't see any way to compare them fairly. My guess is that the new demarc firmware uses a different method for calculating the signal level, maybe on purpose to try to boost their numbers, for people like us trying to get a realistic comparison between them. The only real, solid method of testing them is when the signal is weak. Maybe hook your antenna to a lucent card, then turn it slowly until the signal is barely there, then hook it to the other cards and see what they say. If they get a link, notice the signal level, then turn the antenna more until it drops. That, IMHO would be a fair, reliable test as to the actual ability of a certain card to pull in a signal. The only factor that would be left is the transmit, the demarc 200mw would have a stronger transmitted signal than the lucent, thus may be able to work better, but it would still be a test of card for card which one will work to the furthest distance.

[bobbyc]

That is what I was thinking.... setup a test ap/client using lucent cards where the SNRs between the 2 were like 2-3dB, and then test the senao and demarc at the AP end and run a ping test. What would be a good ping packet size for such a test, to simulate 5Mb/s rate?
Bob C
As for the demarc having 'modified' firmware to exagerate the reported numbers; I loaded the 1.1.0/1.4.9 firmwares found at netgate.com (where I got my senao from) and it still reported the high receive figures. Unless there is some kind of firmware on the card that doesn't get touched when you update the station/primary I don't see that being the case.
Any more input on what I should test next?

[lonnie]

It is funny how a new firmware on the Demarc lowered the TX, just as if it was a 100 mW card with a firmware tweak in the old release. I know they said they did not do that but this sure raises the question in my mind.

[georgew]

I have not gone back to test all of my cards, but I did do some side-by-side testing with the senao 200mw cards vs the demarc/zcomax 200mw cards.

The zcomax cards clearly had more signal output, and never detected any noise. However they would loose a signal anytime they had less than 20db of s/n. The senao cards put out less power, and they reported more noise. However the senao would pass data when the s/n ratio was 2 or 3, and the readings from an orinoco client showed the senao putting out very little noise, while the zcomax cards were noisy and hard to pass data to. The client in all of the tests was a orinoco card on a 12db panel antenna at 4 miles away. In my test, I calculated that the senao had a good 10 to 12db more sensitivity than the zcomax. The zcomax had about 2db more signal, and 3 or 4 db more noise (as picked up at the client).

The bottom line of the test was that the senao cards worked better and had better sensitivity. The signals were always easy to pick up by the client, but the client's signal was lost to unmeasured noise in the zcomax card while the reported s/n ratio was still good.

The test I was performing was really aimed at cost savings. I was comparing an amped orinoco card to unamped high-power cards. The amp with the orinoco was a "cheap" 1 watt TT amp. Both 200mw cards were able to outperform the orinoco/TT combination. The orinoco/TT signal was louder, but was so noisy that it was not useful.

My test showed that a hp card was going to perform better than anything on a cheap amp. However the cheap amp did improve the performance over a bare orinoco card while running netstumbler, so I can't say that a cheap amp will make things worse in every case.

I have not tested with a "good" amp, however with costs being what they are, I can only see a need for an amp if the link I'm trying to establish is very long. With the short hops I'm normally confined to, there is nothing to be gained with an amp.

George

[bobbyc]

Cool so your findings and mine match somewhat. I haven't tested in the real world yet, just in my living room. I was blown away by the readings from the orinoco client manager when I used the demarc...
Station Partner
-54 -22 signal
-97 -93 noise
While the station readings are probably accurate, I have a hard time believing the partner signal (the receiving sensitivity of the demarc can't be that good???) But, your comparison of the senao and the demarc pretty much says it all. The senao performs much better at low SNRs than the demarc, and that's what I'm looking for.
I'm trying to get ahold of the tower company so I can put a senao in one of our unamped tower sheds for this weekend. I have a customer who has -84/-97 reading at the AP on that tower and I want to see what happens when I put the senao in the tower... just for kicks. I don't expect much of a difference except a higher power output from the tower. The senao doesn't hear any better than a lucent card, right?
Bob C

[bobbyc]

I also find it interesting that we saw the exact opposite regarding power output. Assuming the lucent AP 8 miles away is deciphering the signals correctly, the senao has a stronger power output than the demarc... about 5 dB.
Bob C

[bobbyc]

Also george, were you testing with a 200mw card from demarc or from zcomax? I know they are the same card but according to Tony the demarc has a better reference design inside... like they modify the card inside or something.
Bob C

[georgew]

The senao "200mw" card was FCC tested at 165mw (or therebouts, I don't have the test in front of me). The zcomax card was fcc tested at closer to 230mw. I /expected/ a stronger signal from the zcomax, so when I saw it I didn't spend much time looking at that... it wasn't my main goal anyway.

I tested a zcomax card purchased the day after they started shipping. I believe that dmarc does not modify the 200mw cards. I believe they were modifying the 100mw cards which were really 200mw chipset cards detuned to 100mw presumably to pass fcc tests. Dmarc was adjusting these cards to 180mw. I saw mention in a thread that there seemed to be a difference3 between the demarc and zcomax cards, but I know of no details why this might be true. I do know that zcomax has changed all of their spec sheets to show the same numbers that dmarc claims, so there is no longer a gap between the manufacturer spec and demarc's spec.

I know a former demarc partner, and talking to him I believe that other than doing the 100 to 180mw tweek, these guys are simply volume OEM resellers... they are a very small company without the capacity to do something like open cards and modify them. Perhaps they know how to make a software change, but I don't know why they would be somehow more capable than the manufacturer with regard to producing firmware. Any if they /are/doing these mods without getting a new fcc number for the cards, the mods aren't legal. The 100 to 180 tweak was not legal...

I don't know why you saw the senao putting out more power than the demarc... perhaps it had a better pigtail, or the dmarc card had a dirty or loose connector.

I have heard that N connectors can attenuate a signal up to 3db, per connector. But then I see things like lightning supressors with two N connectors that claim .1db of insertion loss, so in theory an N should be as good as .05db loss per connector. I have also seen N male connectors improperly made that would damage a female end such that the next male N connector would not make good contact. So in all of our testing, any time we see unexpected poor results we need to check our connectors to be sure one of our N's hasn't gone from .05db to 3db.

If mercury were not so toxic, I would try to mercury-wet all of my n connectors before sealing them up. The old mercury wetted relays used to be the best signal switching relays because the contacts were always "soldered" with mercury, so you always had a perfect connection. Too bad heavy metals are poisonous. But if you wanted to "solder" the pins in an n connector, a drop of mercury would do it. The solder would never be hard, but as long as it wet the surfaces, the connection would have a soldered quality.

I suspect you had a cable or connector issue when you saw your 200mw senao louder than the demarc/zcomax.


George

[dkii]

since we're on the topic of signal strength, etc... I figured I would put a TT 500mw amp and a YDI 500mw amp head to head. The TT amp is the 500mw Indoor version, connected to a lucent card with a 19" pigtail from fab-corp. The signal then goes thru a polyphaser dsx lightning arrestor, then through 20' of lmr-200, connected to 9' of lmr-240 into a 5.5dbi magnetic mount mobile antenna. A total of 3 connections about 15' away is a comtelco 7dbi "smoke detector" antenna, connected to about 10' of rg-316/u then directly soldered to a lucent card held in a usb carrier. Here is the screenshot of the signal test screen for the TT amp:

http://pub4.picturehost.co.uk/tt.gif

Just incase the link doesn't work, the orinoco cm reports a recieved signal of -49dbm, a recieved noise level of -100dbm and a corresponding SNR of 50db. The test partner reports a recieved signal of -44dbm, noise level of -86dbm, and a snr of 42db.


The YDI amp is the 500mw outdoor pole mounted model with indoor power injector. The power injector is connected to the radio with the same 19" pigtail. Then the same polyphaser, the same 20' of lmr-200. The lmr-200 is connected to the pole mounted amp. The 9' of lmr-240 is connected to the amp also via a N-Male to N-Male adapter, for a total of 6 connections. The same antenna was used on the client side. Here is the screenshot

http://pub4.picturehost.co.uk/ydi.gif

Again, just incase the link doesn't work, the orinoco cm reports a recieved signal of -43dbm, a recieved noise level of -100dbm and a corresponding SNR of 57db. The test partner reports a recieved signal of -40dbm, noise level of -90dbm, and a snr of 49db.


Conclusion:
I would say the YDI numbers are quite a bit better than TT's numbers, a whopping 8db more of usable signal(4 in signal, and 4 in noise) on the AP side, and 6db more pure signal on the client side. Even more so given the fact of 6 connections on the ydi vs. 3 on the TT. And to make it better, the YDI is cheaper and has better integral lightning protection. For both tests, both antennae were kept in the same exact position, and surrounding obstacles(that might cause reflections/multipath) were not moved. I repeated the test a total of 3 times across a 1 hour period to eliminate the possibility of occasional noise sources skewing the results. I would have never imagined the numbers would be this different.

[tony]

Also george, were you testing with a 200mw card from demarc or from zcomax? I know they are the same card but according to Tony the demarc has a better reference design inside... like they modify the card inside or something.
Bob C
From what we can tell, zcomax has no 'special' card run, and the card demarc is selling is no different that those being sold by others, such as TT.

The results of all the card testing is quite interesting.

[bobbyc]

Yeah the only difference between the demarc and the senao in my tests were the card and pigtail... I even tried pushing down the pigtail connector into the demarc and holding it there to see if the signal at the AP would change...
I talked to Tony (the demarc guy) and he's going to let me send in the card and pigtail and he's gonig to measure the power output for me.
In the meanwhile I am going to put the Senao up the tower for a couple days to see if I get any customers calling and complaing.
Bob C

[bobbyc]

Well just an update... I had the senao up at one of our unamped towers for a week and it worked great. I was even connecting to the tower from my house and I am 10 miles away and 1000' taller than the tower. The senao has higher output (obviously) but it hears so much better than the orinoco. I am assuming that my orinoco client manager is reporting the proper #'s in the test results. The only number that I know isn't being reported is the noise figure of the test partner (tower AP). It is always -101 or -102.
Also, the orinoco client manager always associates at higher rates when connected to the senao AP compared to the demarc AP. The demarc AP really reports bogus receive signal readings... they are totally inflated..
example:
the orincoo client test results reports a -74 test partner signal reading, and 11mbps when associated to a senao AP. When I have the demarc AP in, the orinoco reports a -55 test partner signal, but is associating at 1 or 2 mbps.
My hats off to the Senao and to Star-OS support for prism 2.5 AP. In the end I swapped back in the Lucent OEM card because I missed my interBSS relay!
Bob C