I've just had to return my first (and last) RB333 to my distributor as although the board powered up every time it failed to boot three times out of four, regardless of whether I was using PoE or jack - tho' in truth I was unhappy with the design which piled three radio cards on top of each other (Cross-talk? Heat?), and its minimum power requirement of 12v which is a problem when you're using batteries on a remote site.

I gather it wasn't the first 333 he's had returned with the same problem and he's no-longer supplying them. I have a need for three cards in this situation but the 433 won't be available in this neck of the woods for at least another three weeks and the RB600 won't fit in any enclosure I use.

My distributor has no 532s left but fortunately has a couple of daughterboards for them left and I do have a 532 I can pull from another location so with that and a daughterboard should be able to get the site running in a couple of days.

I've also had problems with the 133 that I never had with the 112. In fact I had a superbly functioning, trouble-free network built on the 112 and the 532, and the 532As with their 400MHz CPUs, 128MB hdd and extension memory slots I found to be rugged and reliable and quite fast enough to handle anything I was ever likely to throw at them.

I miss them. R.I.P. 532 - victim of change for the sake of marketing.

HI,

why only 2 cards in 433? : /

3 cards work fine without issues, even XRs and other high power ones.

I agree with you.. I tried it.. but.. RB433 have lots of problems with firerwall in ROS v3.9. I can't enter my RB433 with LAN or WLAN, and with console it makes me a lot of timed out errors.. It says that I have to make supout file and send to support.. but I cant download it in telnet.. hahaha.

Now I try v3.5.. let's see how it works : P


Martín

3 cards work fine without issues, even XRs and other high power ones.

Is this on a testbench in a nice air-conditioned workshop in a Latvian winter or in something like a nice white airtight DCE Die Cast Aluminum Enclosure from Pacific Wireless mounted on the sunny side of a pole in midsummer anywhere between 45'N and 45'S latitude?

I was considering three Mikrotik R52Hs in these slots and the middle one would have had the heatsink of the lower one around 1mm from its underside and its own heatsink 1mm from the underside of the card above with one of the four sides completely blocked and two others obstructed by the clips. Seems to me that the 'issues' that kind of thing might cause could take a while to develop.

Even putting aside heat and cross-talk 'issues' it's simply bad and thoughtless design - changing or replacing a wireless card is fiddly enough when you're hanging in a harness 10m up a mast. When you have to take out two other cards before you can even get to the one you want - and either fiddle with those flimsy u.fl. connectors or leave the cards hanging from them - you kind'a wish you had the joker who thought it looked a great idea on paper up there with you.

Sorry Normis but even if you can get away with it I personally don't like it enough not to ever buy a 333 or 433, and I'm going to leave it to others to find out how well cards double-parked in the 600 work over time.

But the 532 with a 502 daughterboard? Give the man a coconut.

Combination of XR+R52+XR works good and there are no heat or RF issues. If you are working in a hot environment, you can install fans inside the box, there are Fan headers on the board.

there are no heat or RF issues. If you are working in a hot environment, you can install fans inside the box, there are Fan headers on the board.

If there are no heat issues, why are there fan headers?

In any case installing fans inside the box is a great idea when you're sitting around a table somewhere with the project groupies drawing circuit diagrams on the backs of envelopes. I use these things outside in the real world where they have to be protected from moisture and dust and insects so external ventilation is not something to be encouraged while the box I use is 250 x 180 x 50mm so stirring the air inside it isn't going to achieve much - especially as there isn't going to be much air actually passing over the cards anyway given the way they are enclosed on three sides.

Moreover two of my current sites and the third in planning are having to be remotely powered, and while you can talk casually about installing fans, having to cater for them when planning solar panels and battery capacity is no small thing, both as part of the design headache and the additional expense.

Although the voltage monitor on the 333 is a great idea I hope you'll stick with, as far as I can tell there is no temperature monitor let alone temperature-triggered switch on the board for the fans, so you either envisaged them being scheduled by a script(?) whether they are needed or not or turned on an off manually when someone thought they might be needed.

Anyway, I'd argue that better design would avoid the need for fans in the first place but what does my opinion count for? After all, I'm only a customer.

dunno.. I just don't know.... but...

We have like ± 40 RB333's installed all over the place...

Summer tops at +45C
Winter - down to about 4C (only early mornings)

We have NEVER had ONE issue with ANY of the 333's - except... drop the power .000001 -- volt below 12Volts and you're looking for trouble. We standardised ALL of our RB333 Base Stations on batteries using "nice" battery chargers - and like I say - NO issues....

Standard practice is when ANY RB arrives we do a /system reset-conf and load the latest WORKING ROS (3.9)

We only replace the RB333s as and when we require a RB600 to take up the load - and then move them down the line as "edge BSS"

Most - if not all have 3 x cards installed in mixtures of R52/H/XR5 - and as long as the power is good they work really fine.

Tanker

.
.
.
ahhh... forgot to mention -

We only use the Power Input Socket off the batteries and not the PoE

T

why have Fan connectors? Strange question. What if you want to put it into the cener of Death Valley? I'm saying that the cards don't cause much extra heat, and the Fans are there if you have too much heat combined from all factors.

.......

or....

If you suffer from sweaty palms or sweating brow... plugin a fan or 2 to keep you cool while making off that troublesome LMR..... hehehehehe
(sorry)

What if you want to put it into the cener of Death Valley? I'm saying that the cards don't cause much extra heat, and the Fans are there if you have too much heat combined from all factors.

OK. No doubt it's better to have them and not need them than vice versa.

But I'll bet you don't have a mains supply of power in the center of Death Valley, and while presumably solar isn't a problem there you now have to factor into the panel and battery sizes the need to power fans and think about when you want them to run and not run and probably pay for two so's there's a backup and get an enclosure that can cater for them and I'll bet you'd still worry about what might happen if both fans seized up, as they will eventually being mechanical devices.

Whereas I'll bet you could put a 532 in the center of Death Valley in a standard enclosure and forget about it, and it would go forever.

Only you can't 'cos they don't make 'em any more.

And even if heat isn't a problem the arrangement of these cards still makes handling them in the field an unnecessary pain. Every time you touch one you risk damaging the connectors or the card itself, yet you've no choice but to interfere with one if not two perfectly good cards to get to the second or third. With the 532 and 502 daughterboard you could get to any one of four cards without touching another one.

I suggest it's the little differences - the little bit of thought and care in design that can make life so much easier for the guys at the sharp end - that make a big difference when it comes to deciding who's product to buy. In that regard Mikrotik had been doing a damn good job but to my mind that's beginning to slip a bit.

R.I.P 532.

I think the 532 is going the route of the old cisco 2500s. A fantastic workhorse platform that is getting replaced by new rookies that many of us view as untested.

Seriously, great work during the lifecycle of the 532. It has had the type of reliable history that you guys should shoot for for all of your future platforms.

when the RB532 was the actual product, everyone was crying for RB230 and said that RB532 was bad and unreliable That's how product evolution works.

when the RB532 was the actual product, everyone was crying for RB230 and said that RB532 was bad and unreliable That's how product evolution works.

*I* certainly wasn't crying for the 230. I thought the 532 was great.

Of course *that* might be because I only came in when any problems with the 532 - the problems making it "bad and unreliable" - had been sorted out.

Because *that* is how product evolution, any evolution, *should* work - you take a product, iron out the bugs, listen to what people say about it and use that input to make it even better. Getting to the stage of having evolved a great product like that and then scrapping it to start again with another set of bugs and unreliabilities because of a perceived need to always be 'bringing in the new" is certainly not "intelligent design"!

As an RF engineer, why would you LOAD up a single router with more than 2 Radios to begin with.
You can take the stance of power consumption, but we all know the real consumption is with the cards themselves and not in fact the routers.
Real performance is realized when one, loads up as few cards as possible onto as many routers as possible, all in individual enclosures.

Truth = 532 was at end of life.

Just because 2 - 3 people do not like the evolutionary path of a product, does not indicate that it was the wrong road taken.

Just my point of view.

As an RF engineer, why would you LOAD up a single router with more than 2 Radios to begin with.
You can take the stance of power consumption, but we all know the real consumption is with the cards themselves and not in fact the routers.
Real performance is realized when one, loads up as few cards as possible onto as many routers as possible, all in individual enclosures.

Truth = 532 was at end of life.

Just because 2 - 3 people do not like the evolutionary path of a product, does not indicate that it was the wrong road taken.

Just my point of view.

I agree that more routers with fewer cards is to be preferred and hence would rather use two 532s each with two cards over one 600 with four. However the "evolutionary path" taken by MT seems to be to load more and more cards onto one router. Is the next in the 600 series going to have two Triple-shots offering six cards?

So I can't agree the 532 was at the end of its life. It's just my point of view but I would still be buying 532s if they were available and happily letting others solve the teething problems with the 333, the 433 and the 600.

Well said!

I do so love grumpy Friday nights.

MT would do well to understand that their customers are not all idiots.

We have beaten up on Wisp-router big-time this week about the product changeover stragegy that MT used this spring. Hopefully they have enough muscle with MT to pass it along. I did hear that we are one of a chorus of customers who are not happy.

Guys, its build season in the Northern Hemisphere!!!! So lets make sure none of our customers can get product! And the product they can get is brand new, buggy to one degree or another, (like all new products from any manufacturer) and is only available in very limited quantities.

MT seems to feel that if you don't go to MUMs you don't care.

Without wishing to offend anyone, I learned a very long time ago that the people who are easy to reach simply don't matter. The people who make the real decisions aren't generally the people that have time to go to tradeshows.

Better, faster, cheaper is great. Really cool and to be congratulated!! MT is generally speaking doing a heck of a job making that happen.

Unfortunately, as mentioned above, there are some serious concerns about RF management on some of the new multi-card Routerboards, but I have a much larger issue with the broader strategy on product lifetimes and end of life plans.

Please, MT -- there is a better way to manage your product release strategy, and we really hope you are going to take some of this constructive criticism on board so we all have a better experience next time.

Its a shame we don't hear from John Tully much on the forum any more. John was always great about listening to customer concerns. Seems like senior management today is either disengaged, hears but doesn't listen, or listens but doesn't react. Unfortunate...

George

http://s279.photobucket.com/albums/kk128/Tanker_jbl/

Mountain = 2000M (±6600ft)
Mast (we're ½ way up at 100M (330ft) and this is before the "clean-up and tidy-up and close-up".

This is a BSS with 4 x RB333's and all of them have a combo of XR's and RB52H's

It services ± 160 clients and runs 3 x Backhauls

It's been in service for 9 months now.

We feed 12 Volts up the mast from a charger and battery system.

No fans & no heat or RF issues.

T

dunno.. I just don't know.... but...

We have like ± 40 RB333's installed all over the place...

Summer tops at +45C
Winter - down to about 4C (only early mornings)

We have NEVER had ONE issue with ANY of the 333's - except... drop the power .000001 -- volt below 12Volts and you're looking for trouble. We standardised ALL of our RB333 Base Stations on batteries using "nice" battery chargers - and like I say - NO issues....

Standard practice is when ANY RB arrives we do a /system reset-conf and load the latest WORKING ROS (3.9)

We only replace the RB333s as and when we require a RB600 to take up the load - and then move them down the line as "edge BSS"

Most - if not all have 3 x cards installed in mixtures of R52/H/XR5 - and as long as the power is good they work really fine.

Tanker

hi Tanker,

You really gave me back some confidence. I have 8 rb333's running as important AP's and all have three R52H cards in them.

Ater reading this tread (and some other with sort a same atmosphere) I started to feel pretty worried about my setup with the rb333's. I mean, I made the setup of my new network during the winter here in inland Spain and we are just waiting for the hot summer months with the backing sun to come. Secondly, I have up to now only a couple of test clients hooked up to the AP's but the promotion to get many is starting this summer too. I hope cross talk or extra heating of the cards is not going to be an issue.
Third; I also had ´mucho problemos´ with previous rosv3 versions and was very reluctant now to use the latest update.

You gave me back hope that I might not see problems out of any of the three problem areas I just mentioned.
On some of the rb333 I have the middle slot used for 2,4Ghz and the two others for 5Ghz and space the freq's as much as possible and suitable. So far not seen any crossover yet. Signal strengths are good with very good s/n ratios. Some of units are back hauls and traffic, although not very heavy yet, has seen no problems so far.

So, although I can understand the general feeling among some users the design could be better (and actually EOL the rb333 after such short live proves them right. MT must have had reasons to EOL these boards that fast) some real live tests (which you obviously are an example of) show it ain't all that bad as some suggest.

Thanks for that.

Rudy

P.S.
I have a whole series of rb433 coming in so should have spare units available for any failing systems.
Up to now hope that they won't have the rb333 problems if they had to start with.....

[quote=... drop the power .000001 -- volt below 12Volts and you're looking for trouble. We standardised ALL of our RB333 Base Stations on batteries using "nice" battery chargers - and like I say - NO issues....

Tanker[/quote]


Just one Question: How do you manage this? If the units run on battery power only, that is, the charging current is gone, then the voltage drops below 12V rather fast. How do you get around this?
I mean, a battery is only considered to be in an un-charging process only if the voltage is only just dropping below 12V?

And by the way, I tested a rb333 and it stayed working untill the voltage dropped to somewhat near 11,5Volts?

rgds.
Rudy

WR

I have a habit of over-stating things...

A full battery (set "off" value) will top off at 14.1- 14.2 volts - depending on the charger (some top off at 13.8 volts)

The set "on" value is normally around 12.8 volts. some chargers will "shunt-charge" ie: switch on and off until it "thinks" the battery will stabilise at 14.x volts. Batteries being what they are will always try and establish a 12 volt stable state (hense the huge resistance). A battery going "flat" is a battery voltage anything below 12 volts - being a state of discharge.

OK - so the power fails...

The battery's ability to maintain anything close to 12 volts (depending on the amperage load) will actually determine the quality and capability of the battery. We only use Deep Cycle Solar type Maintenance free batteries - because a water battery in this sort of environment will steadily boil off the water unless you have a circular cell type water battery with a higher SG than normal - to sustain this sort of usage.

But - yes - we have experianced (this particular BSS) a 32 hour uptime after power outage - and **if** we catch the system before the power comes back on the terminal voltage is normally around 10.4 volts. Once the power comes on the charger MUST have the capability to pick up the battery load draw as well as starting the BSS.

Maintenance Free or Low Maintenance batteries are the only answer (more expensive) but in a critical and pivotal BSS as this it warrants the extra cost for this type of battery.

I think what I meant was make sure you have a battery which can "hold" the load for as long as possible at a voltage of around 11 volts for as long as possible.

The RB333's do run down to about 10 volts - but the amperage for 4 x RB333's as well as all of them with a combo of XR5's/52H/52's does pull the voltage down kinda rapidly - but then it's up to the battery to hold until the power comes back on.
Remember - as the voltage drops the amps climb - which doesn't add to the "the good cause of clients NOT calling" - hehehehe.....


Another thing - we "cycle" the batteries every 6-8 months - ie:- change them out for "shop checked refurbished" batteries - whether they are troublesome or nor - it's just part of our maintenance schedule. we bring them in and they go through a proper discharge-charge cycle on the bench - ready for the next visit.

All the best

Tanker

Tanker:

Thanks for your explanations. It's nice to know that the setup I planned in all respects is done by somebody else before, with success!
So it gives me confidence in staying on track with my plans.

Rudy

Anybody who says the RB333 has no issues with crosstalk between the radio boards must be on some strong drugs.... Being in radio engineering, and not just wifi slap the gear up and go, I can personally tell you how flawed the design of the 333 is. For those who say heat is not an issue, it is an issue under sustained high throughput. We tested a RB333 with 3 XR radios, 2 XR5's and an XR2. With an infared thermometer, we measured the heat at 2m and after 1 min of throughput, heat was not an issue, at 10 min of constant throughput, the cards were 10C over room temp, at 1 hr of max troughput, almost 40C over room temp... Based on the operating range of the XR cards, 80C is the limit without the extended temp version, thus, at 40C temp inside an enclosure (Easily doable in most areas), that's going to destroy the cards in nothing flat.

Moving on to the RF issues, as has already been covered, the pigtails and MMCX or UFL connectors have serious loss. Put a few of them on top of each other and you will destroy your RX sensitivity very quickly in your radio cards. Deaf radios don't make anybody happy. While I am happy with the performance of the RB333 and can't gripe about that, I have to say that I feel it was a bad move to use the tripple shot connector. We only use the 333 and 433 in single radio applications due to the above highlighted issues. More seperation between cards would be a nice consideration for your engineers to take into account when designing high performance boards.

It's nice to get expert confirmation of something that seemed simple common sense.

I might use the 433 with only two cards, especially where one is b/g and the other is a, but never with three - which means that in situations where I need three I now require two boards.

Unlike the position with the late, lamented 532 where with the 502 motherboard I could use three or even four well ventilated, easily accessible and widely-spaced cards in the one enclosure.

R.I.P 532. I miss you.

Anybody who says the RB333 has no issues with crosstalk between the radio boards must be on some strong drugs....

expunge:

I am sober, no drugs and don't even smoke. Let's do a test.....

I have a rb333 with three MT-RH52 cards on default power setting in a production environment. Standard MT pigtails (and the boxes are MT too). All three are ap bridge mode and serving in backhaul links for AP's in a routed network. (Not a lot of traffic though, but some). All power settings on default.

wlan1 (bottom one in my case) sends on 5600Ghz, wlan2 sends on 5280 and wlan3 (top card) sends on 5500Ghz.
All in the default 5ghz band (no narrow or ´super´ channel band)

My wlan2 which is on a relatively short range (1km) with 25dBi antenna has the worst signal. The other two have both same antenna's but 7km links and better signals (I know, wlan2 needs some lining out.)

I am not having all the radio technicians skills so lets presume I am a dummy here....
How can you now tell if I have problems with cross talk?

As test I just shut-down both wlan1 and wlan3 radios to see if this has any influence on the card in the middle.
I see nothing changing.... Signal strength levels stay the same, s/n shows no difference and CCQ levels are the same. (Its not a static picture, the levels jumps a digit, but no more then that and that is normal.)

I fully agree on your statement that the situation of the cards could have been better. Things can always be better, otherwise we all still would have been working with mail pigeons.
But if I cannot see any changes in any reading of a a card that is either working on its own in a rb333 board or with two other cards in the slots beside it, how real is then the cross talk issue?

Don't mistake me, I am not arguing your opinion re design of the board. My worry here in hot spain is the temperature. For that reason only I am considering now to replace these units with separate 433 base units with less radio's in them.
But if you make a bold statements like you opened with I would like you to show us how cross talk then effects the cards.

I don't see no negative result in cross talk in the cards and MT is not that stupid that they would put something together while it would definitely give major problems.

So please some more consideration in your wordings and a show case to backup your statement.


Rudy

MT is not that stupid that they would put something together while it would definitely give major problems.

Combination of XR+R52+XR works good and there are no heat or RF issues.

We tested a RB333 with 3 XR radios, 2 XR5's and an XR2. With an infared thermometer, we measured the heat at 2m and after 1 min of throughput, heat was not an issue, at 10 min of constant throughput, the cards were 10C over room temp, at 1 hr of max troughput, almost 40C over room temp...

I'm sure MT would not deliberately put something together that would cause major problems, but you have to wonder how thorough their testing is. Even if it causes no issues with some combinations of cards when it's sitting in the open air on a bench, that's hardly a realistic test.

Unless Normis can support his claim with a full description of the test conditions and results, expunge's results must carry more weight.

But even putting aside the heat and crosstalk issues these Triple-shots, while they might be manageable on a bench in a lab, are a real nuisance to deal with in an enclosure up a tower, ie the kind of situation a customer has to deal with all the time.

Good day

I have added a pic...

http://s279.photobucket.com/albums/kk12 ... lbum_album


This is a 532 loaded with 4 x XR5 cards...
It has been in the field for 20 months mow.
We have visited it twice
It carries 1 x Backhaul and... 3 x Sectors 5.8GHz and 80 clients ...

This thread has been badly hi-jacked from 532's to the RB333 - but getting back to RB532
IF and I say IF the RB33 is such a **terrible** cock-up on MT's part design-wise" - please explain this RB532 scenario from a heat perspective?????????

We have them loaded with the RB564 daughterboards as well..!!!!!!! That's 6 wireless cards.. and they run pivotal BSS for us...

OK - so I have NOT had 100 years of RF experiance - but I come from an Engineering background where we were taught - and we proved that heat loss will eventually equal heat gain... so ... 1 x XR5 temp will equal 5 x XR temp when they are all warmed up and working. If this heat is greater than the other components in the enclosure it will be the max heat at which thet TOTAL enclosure will operate.
The only other temps to be considered will the ambient (room or outside) temps - which by all accounts - should be subtracted from the max heat given off by the XR5's - with a variance for the enclosure insultaltion, also known as cooling effect (little as it may be).

Just my .02 cents

T

If this heat is greater than the other components in the enclosure it will be the max heat at which thet TOTAL enclosure will operate.
T

Nice pics.

If you're saying what I think you're saying then yes, on paper x cards will eventually heat up the interior of an enclosure to the same degree however they are arranged. Almost everything a wireless network engineer does is uphill against the Laws of Physics. The best we can do is mitigate them to the extent we can. On the 532 with or without a daughterboard the cards are at least placed as far apart from one another as is physically possible, so while they could eventually cook themselves they are at least deferring that moment to the greatest possible extent rather than applying direct heat to each other as three cards in a Triple-shot and the paired cards in a 600, are.

I'll see if we can get a bi-metal temp probe into one of the BSS when we next visit it... and record ACTUAL temps inside the enclosure.

I think that the readings are going to be lower than we all expect - but let's wait and see... the closest one to me is ± 60Km - so I'll schedule a maintenance visit and see if we can get this recorder fitted.

The trifecta slots haven't given us any hassles in either the RB333's or RB600's.

I'll keep you posted as we get the recordings back.

T

1 x XR5 temp will equal 5 x XR temp when they are all warmed up and working.

Every physical engineer can tell you that what you state here is not the case. 1x a certain heat source is not the same as 5x the same heat source. If that would be the case one radiator in my house would be enough to keep my house warm....
5 XR's will produce 5 times the amount of heat (measured in calories, joules or ultimately Watts).
Thus 5 cards will in the end heat the inner temp of the enclosure more then only one.
Off course, the hotter the enclosure get the more heat loss towards the outside environment, and the lower the outside environment temperature the more heat loss from within the box.
Also, if your enclosure is receiving direct sun light the sun's infra-red radiation will heat up the enclosure to higher temps then the actual outside air temp. would suggest. It is just a greenhouse made of plastic/alu/steel.

On the other hand, wind will cool down a enclosure faster if the air temp is lower then the enclosure. Off course it will heat up the enclosure if the air temp is higher then the enclosure.

It's all about balancing heat production against heat loss.

So, factors rising the radio temps are:
- heat produced by the radios themselves
- plus the heat received by the sun
- the outside air temp. if this is higher then the inside temp.
- the thermal insulation value of the enclosure.

On the other side of the balance we have the heat loss factors:
- heat loss to the outside air temp. if this is lower then the inside temp.
- the thermal insulation value of the enclosure.

Temperature=energy is always looking for a balance. That is a physical law nobody can deny. Thus the more heat you produce with the radios the higher the temperature will rise. But the higher the temperature of the enclosure rises above the surrounding air mass the more heat loss will take place, until the heat loss balances out with the heat production.

In this respect I think it well possible that if a enclosure is placed in the sun, in a hot surrounding environment, like in a desert, no wind and you have one or more radio's working hard, the inside temp can reaches temps that easy outrun the max. working temps of the radio's.
Theoretically there is no max temp the enclosure would reach. In real life their will be a maximum, probably the temp the radios will die with. (If they are dead the temp probably drops since no more heat is produced by them.)

Food for thought....

It's all about balancing heat production against heat loss.

Agreed. Unless you're going to fit vents and/or fans with all the resulting problems of dust, insects, internal condensation etc, there isn't much you can do about heat build-up inside the enclosure(*) but there are things you can do to help the equipment inside cope with it - ie locate it at the bottom of the enclosure rather than the top.

In this regard if you place a 333 or 433 so that the cards are at the bottom of the enclosure the solid base of the Triple-shot is vertically above them, forming an air trap so that the air the cards are heating has to escape through the partially obstructed sides. It is only fully open to the bottom. Hence logic - I haven't tested it - suggests that the insides of the outer cards cards, and the entire middle one, will be in an oven of their own making. And all the ethernet ports, and the power supply, will be at the top of the enclosure where the ambient temperature is going to be highest.

Place the card the other way up and although the cards are now open at the top for maximum air-flow over them, they are going to be in the hottest part of the enclosure to start with.

I can't get the 333, nor the 433, in the enclosures I use sideways.

So whatever their merits as regards speed etc. I can't avoid the conclusion that the 333/433 are not as intelligently designed as the 532 and they are certainly harder to work with in the field, so

R.I.P. 532. I miss you.

(*) Presumably having white-coloured enclosures rather than dark ones, and ones of a plasticised metal rather than plain metal will reflect rather than absorb heat from the sun. Perhaps I should look at adding cooling fins to the tops of the most vulnerable ones!

I agree on the design, like said before, it can always be better.
But when I come to think about it, a simple sun shield, or fitting the enclosure in a shaded place, or at least the north side (on the northern hemisphere) of the mast would already be of a great help.

I think the full broad sunlight shining on the box for hours will probably the biggest heat source for the inside of the enclosure unless there is a good cooling wind.

It would be interesting is someone could do some test with a temp. probe in different boxes, with different cards with different radio's and different routerboards.
Actually MT should perform such a test but I'm afraid some of the voluntary testers of MT stuff (that makes someone of us!) has to do it.

Maybe this summer, given some time I will do some test myself. I've got 20 standard MT boxes (hard plastic) coming and 20 rb433's. So must have some available for some tests.

Tanker:
Can you tell me what enclosure you used in this photo: http://s279.photobucket.com/albums/kk12 ... LEANUP.jpg

Does anyone know where I can order this enclosure?

It looks great.