Posted: Wed Feb 28, 2007 12:40 Post subject: The whole WHR-HP-G54 power issue
Hi
I've read a lot of the posts on the amplifier and power settings for the HP Buffalo router. I've also read the Buffalo documentation which doesn't really help as it gives the same range and power specs for the HP as for the standard WHR-G54S. I'm not in any way an expert but I would like to use these units on a regular basis, I'm currently using Linksys WRT54GL's and WHR-G54s's, and therefore have an interest.
It seems that the whole controversy rests around the correct power setting in the GUI for the internal amp and what affect, if any, changing this has. I have a theory but I stress it's only a theory and I really would appreciate any informed input. Brainslayer doesn't seem to have got involved in any of these threads and I suspect that's because, the same as the rest of us, he probably doesn't have access to real Buffalo issued information.
It is interesting as a side issue that the whole of this only really came up because Linksys started playing games with the alternative firmware market last year. Otherwise I suspect we would all still be using WRT's and Buffalo would never have got a look in. It would probably be in Buffalo's commercial interests to throw some light on this issue if any company employees subscribe to this forum.
Anyway my theory is as follows and I'm quite happy to stand corrected but only by informed comment:
Most external amps apply to both transmitted and received signals therefore boosting the receive sensitivity as well as the transmitted power. Th HP has an onboard amp that boosts the transmitted signal to around 90mw in the factory firmware from an input of 10mw out of the radio chip. I think from previous posts it also has a separate receive amp which will not be affected by any power setting changes in factory firmware or dd-wrt. The reason I guess the transmit power at around 90mw [19db] is that the EU limit is 100mw and Buffalo being a corporate entity are not going to mass market a product that exceeds EU limits. They claim 60% improvement over normal routers which have a power output around 32mw [15db] and that stands up with that level of boost to transmit and receive.
The big question is what happens when you flash the router with dd-wrt. In my case sp2 special. Any input from the code writers would be useful here but I think that the dd-wrt firmware recognises the HP from the board flags and automatically sets the radio chip at 10mw output whatever value is actually written in the power setting. You therefore have a router outputting around 90mw as it was supposed to do. Amps don't really multiply signal they take a specified input and output a clean signal at whatever the amp is set to. Anything else just causes noise and blown amps.
I suspect the reason everyone then gets trouble is that they manually change the power settings in wireless/advanced settings. Once you've done that even if you go back to the default 28mw you would overwrite the original dd-wrt firmware default and overpower the amp. I'm basing this theory on the concept that dd-wrt firmware is generic for these routers and cannot control the amp in the HP only the radio chip output so the figure in the GUI has to be coming from the radio before the amp! The most important point here, if I'm correct, is that the only value you can manually change the GUI to successfully is 10mw!
The amp cannot put out any more than Buffalo have set it to however much you output from the radio. All you will do is increase noise and stress on the unit by cranking up the power.
If I'm right about all this what it means is that there is no point paying extra for an HP to use with dd-wrt! The WHR-G54S is about £5 cheaper in the UK and runs quite happily at 100mw which is all you're going to get out of an HP whatever you do. Anything above 100mw is pretty pointless without a corresponding increase in receive sensitivty as the average wireless card / usb only outputs 15db and can't "talk back" to the router anyway. The 5db difference betwen the router and the receiver in the link budget is usually compensated by a higher receive sensitivity at the router end.
Any constructive comment on the above is welcome / scriv
I am also curious about this issue, and also, what differences there are between the versions of DD-WRT on the exact meaning of changing the power setting. I am also curious about receive sensitivity of DD-WRT compatible routers - which is most sensitive, and would the Buffalo WHR-G54S (or any other Buffalo products) have the same receive sensitivity as the WHR-HP-G54.
The receive sensitivities are all given in the company product manuals for the appropriate router and there is nothing in dd-wrt that would change that spec. There was a previous poster on another thread who had identified a separate receive amp on the board of the WHR-HP-G54.
Posted: Wed Feb 28, 2007 13:53 Post subject: Re: The whole WHR-HP-G54 power issue
scriv wrote:
Th HP has an onboard amp that boosts the transmitted signal to around 90mw in the factory firmware from an input of 10mw out of the radio chip. I think from previous posts it also has a separate receive amp which will not be affected by any power setting changes in factory firmware or dd-wrt. The reason I guess the transmit power at around 90mw [19db] is that the EU limit is 100mw and Buffalo being a corporate entity are not going to mass market a product that exceeds EU limits. They claim 60% improvement over normal routers which have a power output around 32mw [15db] and that stands up with that level of boost to transmit and receive.
However, using SP1 with 10mW setting, I pump out more than 200mW of signal. That corresponds with the specs of the amp (~251mW@802.11g and ~500mW@802.11b).
How did you come up with the 90mW value?
Uppon sale, the total radiated power dBi+dBm (device output and antenna) has to be 30dB or less for EU and 36dB or less for US. Or so I think
My experience at one hotspot is that there were several distant clients who were more consistantly satisfied when I went from the Linksys WRT 2.2 to the Buffalo HP...all other things being equal. Don't have any numbers, but it aligns with what others have said about the receive sensitivity being better.
As far as TX, I have played with this a little, but really would like to have a better metering system than just what iStumbler and KisMAC give. But, I can say that there is small shifts in power at 10 meters as one increases the TX in the advanced page. It isn't as extraordinary as increasing the power on the WRT.
I have used the type of amplifiers that you speak of, where the power output is set for a given level...and in fact, they are better driven by lower input levels. The HP unit seems to give a small increase, where those units wouldn't give any increase at all. But I haven't experimented enough to issue a firm report.
I concur that Buffalo would not go over 100mw budget in the EU. But I also doubt if they would go under it (including the 2dB antenna.)
The receive sensitivities are all given in the company product manuals for the appropriate router and there is nothing in dd-wrt that would change that spec. There was a previous poster on another thread who had identified a separate receive amp on the board of the WHR-HP-G54.
Actually, I find the receive sensitivities are usually left out of the manufacturer info. For example, can you find and point out the exact locations of this info for Buffalo and/or linksys routers?
Also, would one be able to tell if a Buffalo WHR-HP-G54 has a blown internal amp from (mis)use of DD-WRT transmit power setting by testing in the factory firmware that 100% power is stonger than 25% and that if not, the amp is blown? I had turned my transmit up briefly when I had DD-WRT trying to find the point of maximum file transfer speeds across my house and want to make sure my router was not damaged. Thanks.
Thanks for constructive input, with a bit of luck we'll end up getting to the bottom of this issue.
u3gyxap EU is 100mw [20db] at the antenna. As I actually try to earn a living from this stuff I can't comment on how closely that's adhered to
I suspect from what you say that buffalo restrict the amp output in the original firmware and that dd-wrt resets the amp somehow to its default spec and the units are therefore banging out 251mw
This would explain the problem with the external amp on http://www.dd-wrt.com/phpBB2/viewtopic.php?p=64859#64859 which was my original reason for starting this post
I think we're getting there / scriv
Actually, I find the receive sensitivities are usually left out of the manufacturer info. For example, can you find and point out the exact locations of this info for Buffalo and/or linksys routers?
They're given in the cd product manuals that come with the routers.
Actually, I find the receive sensitivities are usually left out of the manufacturer info. For example, can you find and point out the exact locations of this info for Buffalo and/or linksys routers?
They're given in the cd product manuals that come with the routers.
Yep, and you can even download them from the website (look under 'available downloads', then 'user manual'.
u3gyxap from your post on the other thread and this one you seem to have a good knowledge of the amp spec and possibly the board design?
If you've got a copy of the amp spec or a link to it could you please post it? Thanks
Do you know whether the amp is purely for transmit or for receive as well or whether there is definitely another amp for the receive?
I think we're getting pretty close to a definitive answer here I just need to fill in some holes. Can anyone else please keep to on topic posts ie the specific characteristics of the HP amp/power setup under dd=wrt not divert the thread into information that is easily available elsewhere!
Thanks a lot / scriv
Thanks for constructive input, with a bit of luck we'll end up getting to the bottom of this issue.
u3gyxap EU is 100mw [20db] at the antenna. As I actually try to earn a living from this stuff I can't comment on how closely that's adhered to
I suspect from what you say that buffalo restrict the amp output in the original firmware and that dd-wrt resets the amp somehow to its default spec and the units are therefore banging out 251mw
This would explain the problem with the external amp on http://www.dd-wrt.com/phpBB2/viewtopic.php?p=64859#64859 which was my original reason for starting this post
I think we're getting there / scriv
Yes, it is 20dB for the EU.
It is fairly easy to check if Buffalo are setting the power output based on regulatory domains. Just load US firmware on the HP, then reset it to defaults, and see if the signal changes
I may buy one additional HP unit to play with it sometime next week.
The integrated amp, which is the reason these to be called HP (High Power) is a Tx amp. It is Anadigics AWL6153 (http://www.anadigics.com/products/addrefs/Datasheet/AWL6153_Rev_1.3.pdf).
If you take a look at the board you will see, that only the Tx line is connected to it.
The units also have a separate Rx amp, which is the same low noise Rx amp used in the WHR-G54S.
OK u3gyxap I think we're there Definitive answers I think as follows [please correct if I'm wrong]:
1. HP units leave the factory with a Buffalo driver that controls the radio and tx amp to give an output within the regulatory limits for the area in which it is sold. That driver is not open source so can't be used in DD-WRT.
2. DD-WRT [any version] can only control the radio output so that the amp reverts to default spec 10db [10mw] input around 250mw [24db] output.
3. The only real difference between sp1 and sp2 is that 2 recognises the amp and 1 doesn't so you have to set the 10mw manually in sp1.
4. The rx amp is irrelevant to all this and boosts the performance of the g54s as well as the HP. I wondered why the g54s was better in point to multipoint than the Linksys.
5. Boosting the power output of the radio to the amp past 10mw just creates noise for no gain and probably shortens the life of the tx amp.
Basically all the above means to me is that the HP might have a longer life due to reduced power output from the radio and a cleaner signal than the g54s has when set at 100mw. In point to multipoint 150mw is probably wasted but that doesn't matter as long as the signal is clean. It just means I'll have customers who can "see" the signal but not logon at extreme range.
Like I say think we've got there so for anyone else watching this thread apart from u3gyxap please only post if you strongly disagree with the conclusions for good technical reasons.
Thanks everyone / scriv
1. Buffalo's driver does not control the Tx amp itself. It controls the output to the amp, very much like DD-WRT's driver does. The difference is, that DD-WRT has fixed output (you set the power the web interface), and Buffalo's firmware changes power interactively, based on the readings of signal and noise, and _possibly_ on the regulatory domain set.
3 & 4. The amps are not PnP, so they are not recogniseable. The firmware does NOT recognize amps. The firmware knows predefined set of boards and knows the hardware of every board. And it knows how to run on each and every one of them.
Based on the readings of the boardflags, it understands what board it is working on, and if it is in the predefined set, it makes the necessary adjustments (that is why support for new devices has to be added).
Default installation of SP1 is controlling the output to the Tx amp exactly the same way as SP2. But, by the time SP1 came out, nobody knew that HP had the same Rx amp as the non-HP. So that was not being adjusted (it was working fine on the non-HP). What happens to be the best way, is changing the boardflags, so the firmware believes it is working on a non-HP unit, and the low-noise Rx amp is being fully utilized. Then, since the firmware doesn't know it is actually working on a HP, you have to limit the Tx power to 10mW, so the Tx amp is not being overdrivven.
The Tx amp has input limit of 10mW and that is clearly stated in the cpecs.
But let's not forget that the HP may have significantly higher Tx power SOLELY because of the higher (5VDC) power supply. The data sheet for the amp confirms this difference, assuming the higher voltage for the HP actually appears at the Vcc pin of the amp. I suppose it's possible that the Broadcom chip could perform better at 5VDC too.
Hell, maybe the routers are identical, and only the power supplies are different.
Default installation of SP1 is controlling the output to the Tx amp exactly the same way as SP2. But, by the time SP1 came out, nobody knew that HP had the same Rx amp as the non-HP. So that was not being adjusted (it was working fine on the non-HP). What happens to be the best way, is changing the boardflags, so the firmware believes it is working on a non-HP unit, and the low-noise Rx amp is being fully utilized. Then, since the firmware doesn't know it is actually working on a HP, you have to limit the Tx power to 10mW, so the Tx amp is not being overdrivven.
The Tx amp has input limit of 10mW and that is clearly stated in the cpecs.
Sorry I don't quite follow...
Are you saying that it's best to change the board flags and run sp1 or change the boardflags and run sp2.