Server converted power supplys.

Do you enjoy making your own equipment or antennas? Discuss construction and design in here.
Post Reply
User avatar
sureshot
Veteran
Veteran
Posts: 2680
Joined: 24 Jan 2012, 21:26
Call Sign: 26TM413
Location: South East Coast UK.

Re: Server converted power supplys.

Post by sureshot » 15 Mar 2018, 22:26

This is only a small offering, literally. I had to give it a few days of use to see if it behaved reasonably. As most come here looking for "BIG POWER" this might not be for everyone. Although it does cover a multitude of sealed smps units.

The one below was surplus from some old thin clients I had. I started by measuring the voltage whilst the unit was intact. Then the case was split by scoring round the case edges multiple times with a sharp stanley knife. On opening it up I had a quick scoot around the pcb, make sure all capacitors looked ok.

As the wires to the lead had age hardened I decided to replace them with twisted pairs, total wire capacity could carry 8 Amps. As this is a 12 Volt 4.16 Amp power supply, I thought that adiquate. There is no space for terminal or binding posts in this, so I had a rethink and used good quality speaker terminal plate with spring back terminals. I've seen these on retail units stating up to 5 Amps current. If you get good quality ones you could go to 10 - 12 Amps easily with these terminal plates.

Mounting the terminal plate was easy, but I did use nylon machine screws, there are two internal heatsinks, whilst they might be at a low potential I was not taking that for granted. ( in times gone by I've come across heatsinks in atx units that where at mains potential) Put a piece of card glued over the soldered terminal. A test whilst the unit was in two halves, all looked good so it was reassembled. Standard araldite epoxy worked great, allowing over night drying. Tape or a small clamp does the job nicely.

As I don't have a scope at the moment, it's a case of taste it and see. No problems powering an 80 channel cb and an old tagra 25 watt linear amplifier.
Some overs later and all seems fine, current drain was just over 3.2 Amps with that load combination, voltage drop 0.2 Volts under load. So it won't win any prizes in terms of power, but for what it is I'm pleased with it. This process could be adopted to many sealed smps type units, this is what I had laying around.

I'm tempted to look for an oscilloscope, but eBay has only junk, I've got my eyes on a 40mhz hand held unit, might invest in one. It would be interesting to see ripple noise for a known load current, with any smps. Next offing will be back to big server units. Hope some can see potential in this modest offering. :)
You do not have the required permissions to view the files attached to this post.
CB call. Shipwreck.

User avatar
Auldgeek
Veteran
Veteran
Posts: 2858
Joined: 05 May 2014, 09:18
Location: North Lanarkshire, Scotland

Re: Server converted power supplys.

Post by Auldgeek » 16 Mar 2018, 12:17

Some interesting posts Sureshot. :thumbup:
I have 2 large boxes stuffed full of power supplies of various types, most are rated at 12 volts, a few are somewhere between 5 & 9 volts.

While I don't see me having the need for something like this, I may look out some of the larger 12 volt bricks and see what they're like.
Auldgeek (Previously Kilimax) - Drew
Amateur Callsign GM0BRJ
Tango Mike Callsign 108TM144

I've changed radios that many times, I've forgotten what I have :think:

User avatar
MrWeetabix
Admin
Admin
Posts: 14514
Joined: 26 Aug 2008, 13:34
Call Sign: 26TM175
Location: Tynemouth

Re: Server converted power supplys.

Post by MrWeetabix » 16 Mar 2018, 12:21

Cracking little project.... be nice to re-purpose some older PSUs. That one from the thin-clients would be ideal for someone who was travelling. Much easier to pack than a 3/5 standard CB PSU or similar

:thumbup:
26TM175. OP: Craig. QTH: Tynemouth. Also 163TM175 when mobile in Wales 8)

PO Box: 62 Town: Albrighton City: Wolverhampton Zip: WV7 3WR

TM1 is now on Facebook http://tinyurl.com/TM1onFB || New Tango Mike QSL cards https://tinyurl.com/TM1-NewQSL

User avatar
sureshot
Veteran
Veteran
Posts: 2680
Joined: 24 Jan 2012, 21:26
Call Sign: 26TM413
Location: South East Coast UK.

Re: Server converted power supplys.

Post by sureshot » 16 Mar 2018, 13:16

Thanks for the comments guys, yes there are lots of sealed psu units that could be viable for use. Couple of pictures for size scale. The 3 - 5 Amp is a standby power supply. But as for the little hp smps, you could put it in your pocket its that small.
If your needs are modest, this type of psu might do you fine. :)
You do not have the required permissions to view the files attached to this post.
CB call. Shipwreck.

User avatar
sureshot
Veteran
Veteran
Posts: 2680
Joined: 24 Jan 2012, 21:26
Call Sign: 26TM413
Location: South East Coast UK.

Re: Server converted power supplys.

Post by sureshot » 12 May 2018, 22:00

Give this a mention, but only the one image.
Bet somewhere in the country, world for that matter. Is currently pulling there hair out with a noisy power supply fan. This is for the Xbox 360 power supply grey brick 175 watt version (different technique for the other two models of grey brick psu ) I've not attempted them yet, as heatsink runs the length of the unit in the 203 watt psu.
This one's fan was sounding like a cement mixer, quick explanation.
Unpowered ! Remove the lid after taking out the four selftaping screws. Snip the fan lead as close to the old fan lead as possible. Find a 40mm axial fan with as low as a power rating as you can, I suggest sunnon fans.
Next remove all the plastic ducting in the top half of the case from the inside, including the old fan.

You will see a rectangular recess in the case rear where the old fan was located. Centre that recess drilling a pilot hole, use a 38mm wood spade flat bit. It must be sharp, drill slow half through one side so you break the surface with the 38mm bit. Now come in from the other side until your through the case. Slow and steady so as not to crack the case. Expect noise and mess :D
Now line up your fan drilling 4 x 3.2mm holes for the new fan and guard. 20mm M3 machine screws work well with washers and nuts. Once fitted dry fit the case lid, making sure no screws touch any internal metal parts, heatsink etc. That bit is very important to check.
Now strip back enough of the internal fan lead and bare the two wires, + - Shorten your fans flying leads, so you've got enough to solder both leads. Push heatshrink tube on before solder leads.
Solder the leads, shrink the tube over the spliced leads. Route the cable such that nothing is fouled, and is neatly tucked away from the fan blades. Use hot glue if you like. Refit the case halves and power on to test. Quiet fan years more service happy days.

At some point I will tackle the 12 Amp and 16 Amp grey psu bricks, as there fans are just as likely to be worn out over the years. :)
This modification is for the 175 Watt 14Amp unit.
You do not have the required permissions to view the files attached to this post.
CB call. Shipwreck.

User avatar
sureshot
Veteran
Veteran
Posts: 2680
Joined: 24 Jan 2012, 21:26
Call Sign: 26TM413
Location: South East Coast UK.

Re: Server converted power supplys.

Post by sureshot » 30 May 2018, 22:38

Been looking at something a bit different. I don't have big amplifiers like RM KL 500 or Zetargi high power amps. But wanting to test these different server power supply, whilst using the amplifier I have. But with the psu loaded down quite a bit, really just to see the effect of ripple on the psu output whilst a modest amplifier is connected. And loading it down. I don't have the means or area to put together a precision load device. But at the same time this is very high current, where precision resolution isn't really necessary.

What I've got is a replacement for halogen lighting as a load piece of test gear, be it homebrew. I hope to find a modest oscilloscope sooner or later, and this load device will be interesting to see the output waveforms under load. And also while driving a modest amplifier. The idea being it could then in theory drive large amplifiers in place of the load device. I would rather prefer a psu failure if any, connected to a load and modest amplifier, as this will shut down before it goes South if that was to ever happen.

To the point now...
It's a pair of 100 watt heatsinks with three high power resistors on the interface. I couldn't think how to get super close power ratings, so it's 220 watts to 300 watts and from 12 Volts to 14 Volts input. Two fans for cooling to dump that heat, and another fan pulling it out of the case. I did want to go to 500 watts loading, but the case and metal structures where a bit beyond me at the moment. It's sourceing heatsinks and cases large enough to house a 500 watt load. So 220 - 300 watt it is. I'm using relays for longevity, although not really necessary. The host power supply is put in to the 30 Amp front posts, this is idle until one of the six switches powers one of the resistors. And so on up to all six if needed, or as little as one resistors for loading. It can be used equally on linear power supply as well. It is just resistive load, as a combo of inductive and resistive would need much more space.

The host psu under test powers the connected amplifier and any added load resistors. The panel meters will show any voltage drop under load, at also status of power / current being drawn from the host power supply. Six toggle switches and corresponding relays control each load resistor and status led. Two fan lights for each fan when there powered on, using thermal switches to control these. No fans needed at light loads, and two thermal switches via another power relay to cut the input and load in the event of a fan out or fault. And finally a crowbar to cut the power if the voltage climbs to high. Unlikely in modern server power supply, but best to be on the safe side to protect any connected equipment. I hope to get in real time use, some conversations with locals, and off shore contacts if there are any. As any strange behaviour from the power supply would certainly show if it was unhappy with the load it's under. If the units I use as hosts work well under increasing load, less any horrible noises coming across in TX and RX, I would consider that power supply a successful one.
Just the one image at the moment, it's early days. Hopefully more to follow soon.
Any thoughts always welcome. :)
You do not have the required permissions to view the files attached to this post.
CB call. Shipwreck.

User avatar
sureshot
Veteran
Veteran
Posts: 2680
Joined: 24 Jan 2012, 21:26
Call Sign: 26TM413
Location: South East Coast UK.

Re: Server converted power supplys.

Post by sureshot » 31 May 2018, 22:59

A little bit more progress, these heatsinks are rated at about 100 watts a piece. Not less the material missing for holes, it all counts. At 12 Volts it works out at a maximum load of 221.5 watts, at 14 Volts it's a maximum load of 301.5 watts. So at the nominal 13.80 Volts it's a load of 292.9 watts. It's pushing it a bit at 300 watts end of the scale, but with a 100 watt cpu temperatures are around 45°C with around 65 - 75 % fan speed. These resistors are rated at 150°C. With a fan at 100% things should be around 85 - 100° C I'm hoping :D might be less 8) If needs be I can over drive the fans a little with a small boost converter. I did um and arr about using these heatsinks, as I noticed the bulk of the aluminium is not uniform across the whole heatsink. It kind of has a convex profile where it meets the cooling fins.

Trying to space the power resistors as best as possible to get even heat distribution, had I gone in the other direction the two outer resistors on each heatsink would have run hotter. For a cpu it's of no conciqence, but I've had to stagger the resistor layout for what looks like best even thermal dissipation.
That's it for the moment, not to grim with out a drill press. Although you really do have to concentrate to avoid wandering or elongated holes, neither luckily.
More to follow when I get to the front and rear case panels. Then finally the wiring up. I've wanted to use a digital meter for a while now, but thrashing it out on another forum I was informed they tend to drift somewhat with time. So analouge meters it is this time.
You do not have the required permissions to view the files attached to this post.
CB call. Shipwreck.

User avatar
sureshot
Veteran
Veteran
Posts: 2680
Joined: 24 Jan 2012, 21:26
Call Sign: 26TM413
Location: South East Coast UK.

Re: Server converted power supplys.

Post by sureshot » 03 Jun 2018, 22:20

Some more progress...
It's a passive unit really for load testing, I realised I missed out the relay bank, and two bus bars in the form of 30 Amps terminal blocks. after I'd taken the pictures. It's really quite a simple construction. Had an rarer moment when I slipped whilst deburing the holes, slight scratch I know it's there. Annoying, might blow some spray over it yet. Or liven it up with some colour.
These boxes are off eBay, the shells are tough rugged plastic abs type stuff. But the front and rear panels are more a polyethylene or polypropylene type material. The drill grabs, even when using progressive drill bits. Still a crowbar circuit and fan board to put together.
Bit of wiring up soon... :)
You do not have the required permissions to view the files attached to this post.
CB call. Shipwreck.

User avatar
sureshot
Veteran
Veteran
Posts: 2680
Joined: 24 Jan 2012, 21:26
Call Sign: 26TM413
Location: South East Coast UK.

Re: Server converted power supplys.

Post by sureshot » 12 Feb 2019, 23:56

Been a while since I've got in to anything hobby related. A few unfinished projects to get round to...
Recently found an interest in something I haven't tried before, well, part of it is new to me.
Just looked at the date at the post top of this page, wow almost a year ago now. That little Hipro power supply is still working fine today. I only have the one 80 channel radio connected up, but it's always on 24/7 mostly on the cept 40 channels.

So I've got an interest in buck and boost converters, it's really a taste it and see thing at this stage. These devices are very cheap and offer plenty of output current. They are switching devices, and might need additional filtering. Haven't got that far yet. The primary part of this modular duo is pc power plug pack and laptop power supply. There cheap as well, and components and build quality is very good in branded units.

You might see where I'm going with this..
A single unit is an option, with no need for any interface connections. But I'm leaning towards a two module idea, where the pc plug pack power supply is up stream from the unit suppling the radio equipment.
The two new test subjects are both hp units, the largest is 19.50 Volts rated at 7.1 Amps 135 watts. The smaller unit is 18.50 Volts 4.9 Amps at 90 watts.

I did pick up a selection of converters, but for now going with dropping voltage, so buck converter as opposed to boost converter. I'm still missing an invaluable piece of kit, that being an oscilloscope.
As these are switching devices I'm yet to try these on RF equipment, I really hope I can come up with something. Most converters are less than £10 and even cheaper if you don't mind waiting for the slow boat from China.

Most have voltage adjustment and current limiting should you want to implement that. Almost completed two hp power supply units, opting for the speaker style spring clip connections. There is very little room for anything else, unless you keep the trailing lead that comes with these units. DC sockets are hard to find as they are proprietary to hp systems. You could always chop off the hp plug and add a 5.5 x 2.5 mm plug or similar. And like wise for a chassis socket in the converter box powering the equipment.

Once I've put the second part of this duo together, I will give it a try, hopefully I won't need additional filtering components, but will cross that bridge where I come to it, if its necessary.
Anyway, below are a few pictures of the testing of a 8 Amp buck converter off Amazon, eBay also have a few different types as well. The images are testing resistive load, and that's a 50 watt halogen lamp. The converters heatsink barley warms up with just shy of a 4 Amp load. Once in an enclosure, its feasable to at a voltage and current meter. Also the number of outputs you want. Air cooling will be needed at higher current levels, but a descreate 40mm axial fan an thermal switch is all that's needed.
Be back when I've tested, and have a completed an offering. :)
You do not have the required permissions to view the files attached to this post.
CB call. Shipwreck.

sec1223
Top Poster
Top Poster
Posts: 1687
Joined: 28 May 2010, 09:28
Location: west yorkshire

Re: Server converted power supplys.

Post by sec1223 » 19 Feb 2019, 23:31

iv ordered a few more 800's. just cant beat them !
cant stand IGNORANT b******ds

User avatar
sureshot
Veteran
Veteran
Posts: 2680
Joined: 24 Jan 2012, 21:26
Call Sign: 26TM413
Location: South East Coast UK.

Re: Server converted power supplys.

Post by sureshot » 22 Feb 2019, 20:26

sec1223 wrote:
19 Feb 2019, 23:31
iv ordered a few more 800's. just cant beat them !
Yes there not bad at all. There's a seller on eBay, not me btw, he has job lots of hp 460 watt and hp 750 watt server units. Same unit as the hp dps 1200. £89.00 for 10x the 460 watt units, just over £100 for 10x the 750 watt units.

I've picked up a few hp power bricks now, just an occasional tinker.
Yet to try a buck converter on a transceiver, there's a buck converter that's rated at 300 watts. That's way above the power bricks capacitys, the hp 4.9 Amp units are fairly easy to open. Just patience and scoring with a sharp Stanley knife, a little challenging running the output wires. They mustn't foul any metal internally, and heatshrink the terminals is an absolute must.
The speaker terminals I've been using are really heavy gauge in wire size that they except.

I did buy another few from a different source, and there garbage. Just flimsy junk. Annoying, but I should have realised why they were so cheap. Once you get above hp 4.9 Amp units that are plastic welded, the 135 watt 180 watt and 320 watt hp units are fasted with self tapping screws under the small rubber feet. I know it's an extra step, dropping a power bricks voltage, I just thought id try it. And 13.80 Volts to 14.40 Volts is easily achievable. I do have a couple of buck boost converters, but they where for putting together a lab bench power supply.

Got to get round to putting the converter box together...
I did concider a single unit solution with the hp power brick and converter box as one unit, it would only require some M3 machine bolts and nuts for a seamless construction.
I would say with any power brick that's plastic welded, be sure to key the two plastic half's with sandpaper. Also go for standard epoxy, the slower drying time makes for an incredibly strong bond. I'm looking at different interface connection ideas for wire quick discontents.

On the server units, I have spotted another viable unit worth trying. That rated at nearly 1000 watts, I won't pull the rabbit out of the hat just yet. Just need to be sure its works first.
CB call. Shipwreck.

User avatar
sureshot
Veteran
Veteran
Posts: 2680
Joined: 24 Jan 2012, 21:26
Call Sign: 26TM413
Location: South East Coast UK.

Re: Server converted power supplys.

Post by sureshot » 22 Feb 2019, 23:55

Just a couple of pictures of these power bricks. Have to shrink the images, but some numbers should be identifiable.
On no account connect these directly to any radio or related equipment, these units put out between 18.50 - 19.50 Volts. The output must be dropped to a safe level for radio equipment use. You can do that with a buck converter, or even a linear regulator with pass transistors. Both will work well. If you opt for dropping the output voltage with a linear circuit, the final output would be very clean in terms of electrical noise.
And if you do any testing with the unit open, exercise extreme caution. The primary side of these switch mode power supplies can hold a charge for some time after removing the power input.
And some primary heatsinks can also be at mains potential. If your in any doubt, go and buy your power supply from a retailer.
This is just an exploration in to smps bricks and DC buck converters, or linear dropping circuits.

Smallest to biggest, with my Sons chewed ruler for scale.
19.50 Volts 4.9 Amps 90 watts

19.50 Volts 7.1 Amps 135 watts

19.50 Volts 9.2 Amps 180 watts

19.50 Volts 11.8 Amps 230 watts

There are variants that have 18.50 Volts with similar current outputs, I think it depends on the revision of the hardware, and what it was intended to power.
I did grab a few of the Dell power bricks, the unit that powers the usff optiplex 745 and upwards. Those are 12 Volts straight out of the unit, just wanted to revisit those with some new ideas. Some time in the near future..
You do not have the required permissions to view the files attached to this post.
CB call. Shipwreck.

User avatar
sureshot
Veteran
Veteran
Posts: 2680
Joined: 24 Jan 2012, 21:26
Call Sign: 26TM413
Location: South East Coast UK.

Re: Server converted power supplys.

Post by sureshot » 28 Feb 2019, 00:16

Had a bit of a tinker tonight, this was with the Dell 12 Volt 18 Amp power brick.
Dell DA-2 AC DC 12V 18A Power Supply PSU .
Having had one of these in the past, I thought there was only one revision of this power supply. After a lot of searching, and various sites suggesting wiring configurations to get these working, I've come to a conclusion.

As far as I know now, there are two revisions of the same power supply. But the wiring to get them working correctly is different, in the image below are the two variations of the same unit. The one with the copper shield is the older of the two. To power that one up, both red and blue wires go with 0 Volts or ground. The blue wire is the remote to turn it on, the red wire sence voltage drop and compensates for any line drops. If you don't connect this it won't power on, and power a load. Articles I've read say the red wire is not needed, but I believe it is. So the white wires are + Volts, the black are - Volts, both red and blue and the bare ground wire are bundled together in the copper shield variant.

The newer aluminium shielded unit has all white wires + Volts together, and add just the red wire to them. - Volts black wires have blue remote wire and the bare ground together. If you leave the red wire out of the + Volts side, the unit still works (aluminium variant only) but then there is no compensate for voltage line drops. I ran both through a buck converter, just to see if they would work through one. I plan to go with a boost converter with these power supply, there by getting 13.80 Volts or up to 14.40 Volts. You could happily leave them at 12 Volts, they would still power radio equipment. Eventually I will be back with a unit that is hp and dell, one buck converted the other boost converted. I'm still undecided whether to remove the cable, incorporating a small project box to the psu case, so it's a single stacked unit.
At the moment XLR connectors look the best option for cabled units.

Anyway the two units are wired differently, I expect in the case of powering the Dell PC, the computers hardware sorts out the connection configurations.
There is another 150 watt 12 Volt Dell power brick, although harder to find. It was for the earlier usff PC's.
It would probably be straight forward to work out the wiring configuration for that unit.

The easiest way to tell the two Dell supplys apart is the power indicator lens, or light pipe. In the copper variant, the lens is just plain defused. In the aluminium variant the power indicator lens is also defused, but has a cross pattern across its surface. Just on a final for now, there might be other revisions of this power supply, but as yet I've only found the two.
You do not have the required permissions to view the files attached to this post.
CB call. Shipwreck.

Post Reply