With your logic, even my phone charger will fail with overheating eventually.
You have to consider thermo heat transfer by air, which is around 100W/m^2 K.
My 8"X10" is around 0.2m x 0.22m or 0.04 m^2. So, it will allow 4W/K for the front and perhaps 3W/K for the back. It is stable at 7W/K * 30K or around 200W. So, a 99% inverter will work.
Possible they control the rotor current, takes a lot less power
Cheers
Klaus
Well, 100W/m^2 K might be pushing it. perhaps 50 with good convection air flow.
Thermal conductivity of: Still Air:0.024 Glass:1 Convection Air:50 Aluminum:200 Gold:300 Copper:400 Diamond:1000
Surprisingly, Diamond is the best. So, I am making my box with Diamond studded Gold Box. May be for $100K.
Den tirsdag den 29. juli 2014 01.18.16 UTC+2 skrev snipped-for-privacy@gmail.com:
this is even better
-Lasse
Diamond-studded really isn't the way to go. You can vapour-deposit sheet diamond and get both the mechanical strength and the high thermal conductivity in a purpose-deposited box.
It might be cheaper than studding the box with gem stones, but probably not all that cheap, even if you were prepared to tolerate graphite-rich "black" diamond which can be laid down relatively fast.
-- Bill Sloman, Sydney
But how does sheet diamond look like? If it looks like glass, not much "wow" factor added. I might not be able to make the "coolest" box, but i can make the most expensive box.
Sheet diamond looks like glass - black glass if you go for the cheaper graphite-rich vapour-deposited diamond - and won't have much "wow" factor with the kind of people who'd be impressed by a diamond-studded gold box.
Since it will dissipate more heat, it will have a much larger "wow" factor with people who might buy it as an inverter, rather than a piece of interior decoration.
-- Bill Sloman, Sydney
here.
I know of one other, pumping aquaduct water up hill over 3000 feet (calculate the head on that in psi) for Lost Angeles. The pipes are about
12 feet dia and there are at least 4 of them. That'sa more than a fire hose.
W.
e.
ulate the head on that in psi) for Lost Angeles. The pipes are about 12 f eet dia and there are at least 4 of them. That'sa more than a fire hose.
You may be talking about a succession of hills totalling 3000 feet, but Los Angles isn't high enough to need a 3000 foot head.
Getting water over one hill - up one side and down the other - can't rely o n the syphon effect if there's more than 30 feet of head invovled, so you r eally do have to pressurise the water enough to get it up the highest hill, but once you've done that, what you lose on the subsequent up-slopes you g et back from the down slopes.
You do have to supply extra head to make up for friction losses, but there you can trade off against larger diameter pipes carrying the same volume of water more slowly.
Pumping water through - full - pipes doesn't waste energy in wave generatio n, so you only lose energy as velocity squared.
-- Bill Sloman, Sydney
That's correct. You replied to the version of my calcs that were edited and somewhat truncated by James Arthur. Here's the original. Note the word "adiabatic".
(...) At 95% efficiency and 2Kw of output, that allows about 100 watts of dissipation. 1 Joule = 1 watt * 1 sec 56,000 Joules / 100 watts = 560 seconds The spec says nothing about environmental conditions during the test except that the temperature will be somewhere between 15C and 30C. Assuming a somewhat adiabatic (insulated) system and still air, the inverter will run for: 560 sec / 60 sec/min = 9.3 mins before going out of spec and overheating. Even if one makes the heat sink out of diamonds, with a specific heat of 0.519 Joules/gram, it will only run for 12.5 mins before overheating. In other words, without secondary cooling or science fiction efficiency, this inverter is not going to run for 100 hrs without going overtemp.
In effect, that's the worst case, where no heat is radiated, convected, or conducted to the surroundings. I would not expect the heat to build up continuously towards melt down, but rather the floor, walls, ceiling, and HVAC system will move much of the heat away. How much will be determined by the mounting configuration and the immediate environment. For example, Google claims that the room temperature will remain between 15C and 30C. The allowable case temperature rise is 35C. If Google maintains the room at 15C, I can dissipate roughly *TWICE* the power (watts) as I could if they maintained the room at 30C. Similarly, if they maintained the room temperature by HVAC air flow in a small room, the module would be much cooler than if they did it with still air, in a much larger room (i.e. auditorium).
Things get ugly if we make some assumptions based on typical solar inverter installation. If the purpose of the tiny package is to cram it into a tight space, the air flow will be zilch and all cooling will be by radiation and conduction. If it's to be buried inside a NEMA box to make the authorities and fire marshall happy, the only cooling will be by conduction. If it is intended to be mounted on the 1" standoffs, forget the conduction. Ad absurdum. I would be happy to estimate the final temperature after 100 hrs if Google would kindly offer some environmental and mounting clues.
Ummm... Still air? Natural convective flow? Forced convective? Volumetric flow rate (m^3/sec)? Over what parts of the case? Ambient air temp? Estimated thermal resistance of the case and heat sinks (C/W) based on material and surface area? Copper, Aluminum, or diamond? Radiation losses? Ambient air temp?
That's 20w dissipation. Color me skeptical as I don't think it can be done. I also don't see how you obtained those numbers, but I would be more interested in what the package will do at 95% efficiency (100 watts dissipation), which I think is a more realistic number. Why Google wants such high efficiency, when they don't even bother dealing with the rather high 10 ohms series resistance on the input which will burn 198 watts somewhere at full load, will probably remain unanswered.
Also, nobody has yet addressed my question as to how such a miniature inverter is going to help: "Making them smaller would enable more solar powered homes, more efficient distributed electrical grids, and could help bring electricity to the most remote parts of the planet."
-- Jeff Liebermann jeffl@cruzio.com 150 Felker St #D http://www.LearnByDestroying.com Santa Cruz CA 95060 http://802.11junk.com Skype: JeffLiebermann AE6KS 831-336-2558
That's why we have to have a balance between the expensive nonsense that many vendors would /like/ their customers to buy, and something that is home-made and only one person understands. Thus we use the power supply arrangements that are in common use - just because I cannot influence this situation, does not stop me from wishing for something better.
Backups are certainly essential - RAID is about uptime, not backup.
The difference is that if a hardware RAID card fails, you lose access to all your disks. Unless you can get exactly the same card, with the same firmware and features or restrictions, you cannot access any of the disks. It is all gone. With software RAID, all you need is another computer with enough SATA/SAS ports (or USB disk bays, in an emergency).
Did you read my previous paragraph? I know you are trying to be helpful, but I feel I am repeating myself in every post. It would not help to have a couple of 30V switches - that just means that the 220V AC to 30V DC conversion is done in another box. It would not avoid the low voltage DC to mains AC to low voltage DC waste, it would only help on a small fraction of the power in the server room, and it would not help with using the mainstream and vendor-independent equipment that I need to use.
It is certainly a solution for some use-cases - but it is not a solution that would suit my needs. I don't mean to be rude, because I know you are trying to be helpful here. But you are missing the point entirely - perhaps I have not made it clearly enough.
I am not expecting a solution here - my posts are more of a rant than a request for help. The "solution" is a vendor independent standard for low-voltage DC supplies for off-the-shelf equipment in low and medium range server equipment - UPS's, rack-based servers and storage, and networking equipment. Such a standard does not exist - thus there is no solution available to me, and I simply have to live with the waste of extra conversions. I have been trying to outline why such a DC standard would be a useful thing for everyone involved - vendors as much as users.
On a sunny day (Tue, 29 Jul 2014 13:07:11 +0200) it happened David Brown wrote in :
Na ja, if that is your thing,..,.
Somebody once wrote here 'designing is making new things from things you have'. Philosophy is moaning about things you do not understand or are to lazy to figure out or whatever.
I do not know your 'agenda', but as you presented your self as 'IT manager', and yes that already made me suspicious, then from your job perspective there is no need for a universal DC standard, just use AC. There would be if there was no AC (you were on mars for example, or the moon).
From an electrickal engineering POV it would indeed be nice to omit the AC step, mainly because of rectification in computah supplies causing large peak currents and thus I^2.R losses, but modern computah supplies have power factor correction and what not.
So, once you get the money and NASA on your hand and have moved to moon or mars, or Sahara desert for all that matters[1], I think the laptop idea works great, I have tested it, on solar too. There are not many aliens there with intent anyways to hit your site, if they are unfriendly they may hit your site with missiles, also in that case a few laptops as backup would work great.
Its all about money, just create impossible specs (I take the F35 as example), then have some clueless idiots design something that meets those specs on paper, then if everybody sags it is a disaster, silence them by demonstrating it works (for a few seconds), then keep 'fixing' the design for the next 10 years, add expert after expert, well... Then you be the boss of many, at that point your pension is high enough to fly Virgin Galactic (once ever), and your worries are over,. The project will then be canceled and everybody will be happy.
No
[1[ Sorry my inglis I alien
Then my posts on this topic are more philosophical than designer by that categorisation - though I do understand the situation. But I am not currently trying to change my server room setup (it works fine as it is)
- I am just ranting about how I would like it to be.
Certainly there is no /need/ for a DC standard here, and server rooms like mine work perfectly well on AC. But they would work /better/ if there were an appropriate DC standard.
the wires are short (except for the lines coming into the room and the
worse with a low-voltage DC system, assuming the same wire thickness - but I expect they would still be negligible.
The UPS takes the AC down to low-voltage DC (for the batteries, smoothing, protection, etc.) and then up to high voltage AC, then the power supplies take it from high voltage AC down to low voltage DC again. By cutting out AC between the UPS and the servers/network equipment, we could cut out two large conversion steps that each waste something like 10% of the power (as well as having a cost and taking space).
Certainly laptops are great if you have a limited power supply.
Your English is not a problem - I've seen native speakers write much worse.
Yes, depends on what number we pull out of the air. But 20W (99%) to 40W (
98%) is reasonable.
The current industrial average is 97%. So, it might not be too far off. I t will not be easy, but not impossible. Having 99% inverter does not mean capturing 99% of the energy. It's just pushing most of the energy loss out side the box, into connecting wire, block diode and panel itself.
0.05 ohm gallium nitride transistor might make it possible.Not according to Google. 97% is already the norm.
On a sunny day (Tue, 29 Jul 2014 16:03:38 +0200) it happened David Brown wrote in :
I am just joking about. On the more serious side: You claim 'DC would work better' There are many issues with DC, starting with the switches, if your servers have those. Switching DC is not that easy. Second I have been involved with powering a big computer room, and the I^2.R was a big problem although it has been many years. The rectifiers in the supplies charge the caps at the peak of the sine wave, at that point the current peaked so much that the I^2.R losses caused the cables to fry. I dunno how much you know about electronix but we will see.
Anyways it occurred to me that _last time I looked(tm)_ most of them computah supplies were 115/230 V, where in case of 115 the voltage was doubled to some DC, say 2 * 115 * sqrt(2) = 325 V or so, and in case 230 V it was just 230 * sqrt(2) = also 325 V. That was before the power factor jive an all sort of tricks. So maybe it is still that way, and your universal DC voltage IS already present just feed the computahs after the rectifiers, with 325 V DC[1]. Just about 27 12V car batteries in series, just tap the cars in the parking lot for free energy.
[1] This needs some holes drilled, and other work, you need to ask that a technical department be created, make sure you convince the management that you know nothing about technical stuff, best chance to become head of both departments.This message will self-destruct after reading.
In a three phase system with a balanced load, the _fundamental_ currents from all phases will cancel out in the neutral conductor. The neutral conductor is supposed to carry only the phase unbalance current and hence in some cables, the neutral conductor is smaller than the phase conductor.
However, in old television and computer power supplies without PFC, the third harmonic current is _summed_ in the neural conductor. If the neutral conductor is weaker than the phase conductors, there is a real risk that the neutral conductor is fried, potentially causing fire or at least loose neutral/ground.
In old computer classrooms with 20-40 tabletop PCs (without PFC) in the same room, supplying electricity to all machines caused all kind of problems, in addition to those loss issues, also causing problems with GFIs.
95% is the minimum spec. As near as I can determine, nobody is going to get extra points for being the most efficient. Therefore, my guess(tm) is that the contestants are going to scrape the edges of the important specs as closely as possible. Even worse, Google didn't bother specify the load range that this inverter is expected to operate. Is that >95% at full load, half load, over a 20% to 100% range (assuming it turns off with no load), or whatever?
It might be easier to obtain 99% efficiency than to dissipate 100 watts in such a small package, especially with no secondary cooling.
Looks like the range is about 97 to 98.50% (click on the "show 100 entries" pull down). Note that these are "peak efficiency" which is the highest efficiency that an inverter can provide. That's cheating. The California Energy Commission uses "weighted efficiency" which is an average over the operating region. That drops the efficiency down to about 96%, which what I would guess(tm) that Google has specified: Click on any of the reports below for efficiency versus load graphs:
I would be interested in how Google handles that almost 200 watts being dissipated in the 10 ohm input resistance, which presumably is dissipated in the solar panel, power source, or input wiring. It's not part of the efficiency calculation until someone decides to calculate the overall system efficiency.
Average or peak?
-- Jeff Liebermann jeffl@cruzio.com 150 Felker St #D http://www.LearnByDestroying.com Santa Cruz CA 95060 http://802.11junk.com Skype: JeffLiebermann AE6KS 831-336-2558
LOL- typical geeks nitpicking parts technology.
Who cares about the low end efficiency? They are not measuring efficiency. They are measuring temperature.
ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.