Boost converter help.

[yngndrw]

Hi,

I watercool my computer but my pump has just died (Again) so I'm looking for a new type of pump which will be more reliable. I've got a lot of suggestions to use a particular pump, however it's 24V and I only have 12V inside my computer.

I'd like to make a 12V to 18V-24V converter. (Adjustable between 18V and 24V, however I expect it to mainly stay on 18V.) The pump draws 2.5A at 24V, which is 60Watts.

I have found this on google: http://www.epc.com.cn/meeting/resource/2005power/download/linear.pdf
However it is a buck-boost converter. I only require a boost converter, so I'd like to simplify it.

So in summery, I need a boost converter to convert 12V to 18V, with the option to adjust the output upto 24V. (Although it will be setup mainly for 18V) The converter must be able to supply 2.5A at 24V, which is 60Watts continuously, but must also be able to cope with the startup current of a motor.

The circuit must be very reliable and must NOT add any noise onto the input power rail.

Here is a link to the pump: **broken link removed**

Any IC used must be available in a DIP package, because I'm on a laptop so I havn't got any software or librarys from Eagle. I'll need to build the circuit on stripboard while I design a PCB for it.

Any chance of a circuit being designed for me, seems as I can't do it on here. (Laptops, can't live with them and can't live without them !)

A list of part numbers available in the UK would get you cookies. :x

I have an oscilloscope and adjustable powersupply so I can test the finished product with no problem. I also have a Peak Atlas LCR meter so I can make my own inductors if need be. (But please tell me what guage of wire to use and the expected number of turns, becuase I have no idea about them.)

Thank you !
-Andrew.

 

[Sceadwian]

You should be able to use a boost inverter to make a positive voltage negative. Then you can use the -12 and +12 lines as a +24 line

 

[Papabravo]

Well I think you're going to have a problem. That 60 Watts on the output at 80% efficiency will require 75 Watts on the input. At 12 volts that's 6.5 Amps. You won't find an IC with an internal switch that will handle that kind of current. You will need an external switch. Next thing to figure out is if the 12V supply inside the computer will handle a load transient caused by the the motor startup. SMPS are notoriously lousy at load transients. Probably the most challenging aspect of the project is finding an inductor that will survive twice the peak input current (2 * 6.5 Amps = 13 Amps) without saturating or demagnitizing the core material. It needs to be wound with wire of sufficient guage to prevent heating as the 13 Amps builds up when the switch turns off.

If you have never built an SMPS before and have no access to any SMPS CAD tools your chances of designing one and selecting components without a great deal of pain and agony are rapidly approaching a very low order of probability.

That said, if I were you I'd find the badest 100W 18-24V adjustable linear supply I could find and fagedaboudit.

 

[yngndrw]

Would that not be less efficent ? Also I require the output to be variable, down to 18V, so it wouldn't be a simple inversion.

For them reasons, I think that I may aswell just use a standard boost converter, especially for 60Watts.

Edit: Papabavo, you replied while I did.
I hear what you're saying. I'm interested in learning about SMPSs but I thought I'd have a bit more time to do so. I thought that maybe someone would have a go at designing one for me to get me on my way, but I never realised that a 60Watt or so PSU would cause any problems.

My computer's PSU can provide atleast 35A on the 12V rail, although I'm not exactly sure what "Load Transients" are unfortunatly. Can you explain this term and how I can find out if my supply could support this ?

After a quick google, I found an inductor which will cope with 13A. On the datasheet, there's a decent range, doing upto 30A. Assuming that the required inductance is in that range, that problem is solved. They don't look too big either which is always a plus. Link:
**broken link removed**
Suitably rated inductors in terms of saturation current are of values: 0.33uH, 0.68uH and 0.78uH.

I have a book on SMPSs but no experiance and very little time. I need the computer for Uni and the like. Also with the lack of access to any files on my computer, I don't have the usual designing applications I'm used to. (Not that I have any specific SMPS design programs - Any suggestions for any ?)

Regarding yor last comment, do you mean a 230V -> 18/24V PSU ? The problem with that is firstly size and secondly the heat output. I also wanted the unit to be self contained, with no power cable coming out of the back. It also means that the pump will automatically turn on with the computer.

My bench supply will only supply upto 15V unfortunatly.

 

[Sceadwian]

Why liquid cool anyways? Modern peltiers and a few extra fans with variable speed controllers are just as good, quieter and more energy efficient.

 

[yngndrw]

Why liquid cool anyways? Modern peltiers and a few extra fans with variable speed controllers are just as good, quieter and more energy efficient.

I disagree.

Liquid cooling is SILENT if done right, my PC is. (Was)

Using a large radiator means that you can use much larger fans at a lower speed, making it quieter and cooler.

Peltiers are VERY inefficent and require a lot of cooling. You cannot run a decent peltier on an overclocked processor without watercooling, even with watercooling on a peltier, there's a hell of a lot of heat to remove.

For a modern overclocked CPU, a watercooled 320Watt peltier will NOT be suitable.

Anyway a little update about the specs of the pump. The manufactorer states a 10A startup current, in it's worst case scenario.

 

[Super_voip]

use a simple semi-regulated power supply made from a tapped transformer(voltage selection) to a bridge rectifier to a smoothing cap, dead easy and quite good enough for a motor

 

[yngndrw]

The problem with that is that I'd need to bring in a mains feed. The reason why I wanted to make a DC-DC converter was for size reasons and also to neaten it. By bringing in a separate mains feed, I risk forgetting to switch on the pump or the fuse blowing in the pump's power supply, leaving my pump off and my CPU fried.

I'm going to go through my SMPS book and do as much as I can on paper, I'll need suggestions on:
* Reducing the noise created on the input line, A pair of mutually wound inductors followed by a large capacitor ?
* What controller IC should I use ? Should I just make a simple controller out of a saw-tooth oscillator feeding into an op-amp or use an IC ? What IC would you suggest ?

I know it's normally a big project to make a SMPS, however the semi-fixed output voltage and the lack of an AC input section should make this simplier. It will also be simplier as it's just a Boost converter - No custom transformers to wind.

Thanks.

 

[Super_voip]

A 10 Amp peak, voltage adjustable switchmode is hardly the easiest solution.
Me, I love switchmodes, I think people got tired of me recommending them for their projects, as for the safety, feed the computer power supply from the pump side (after the fuse) no power for the pump = no power for computer.
Remember that the computer power supply is switchmode and at turn on will have to provide an addition 10 Amps to get the pump running.

 

[yngndrw]

You have a point.

Regarding it being adjustable - I suppose it would be fine as a fixed PSU setup for 18V, however the voltage feedback circuit would have a potentiometer to allow slight over-voltaging. As long as the components are rated high enough, I don't think that it would cause a problem ? (Other than efficiency.)

Regarding feeding the computer from the pump's feed: Good point, but I really don't want the mess of feeding mains cables in and out of the back. How do you recon the sizes would compair of a linear regulator Vs a DC-DC converter ?

Also heat output is an issue - If the unregulated DC supply was 30V (To allow headroom for over-voltaging to 24V), then the regulator would have dissapate 30Watts of heat. If I need to add a fan I want it to be very slow to reduce the noise to nil. That would require a reasonable sized heatsink - I don't think I'd have enough space.

Regarding the computer's powersupply. Very true, however I already have a 12V pump running off it, which is rated at 18Watts continuous. (1.58A @ 12V) Based off this new pump's startup current, let's say it requires 6A at startup. A lot of people run two of my old pumps. That's 12A of startup current required based off my previous guess.

On that note, the pumps that I'm running now have a delayed start. They also have a soft-start, however it's bugged and doesn't actually work. :c

A simple RC timer could hold the converter in a standby state for 1-2 seconds, then the converter would start up. By putting a capacitor across the Voltage Feedback line and Ground, the regulator will have a slower rise which should help things out, no ?

Thanks, I'll get doing some calculations from this book.

Edit: Just did some calculations from my book.
The expected peak inductor currents worry me:
Boost-mode converter
Startup Peak Current: (24V, 10A load) 110A
Running Peak Current: (24V, 10A load) 27.5A

Half-Bridge converter
Startup Peak Current: (24V, 10A load) 56A
Running Peak Current: (24V, 10A load) 14A

Looks like I'm going to be winding a transformer ! Fun.

Edit 2:
The book shows an "XMFR" on the base of the top half-bridge transistor, what is an XMFR ?
From what I can tell (Google), it's an inductor of some sorts, but what's it for ?

Edit 3:
Okay, I think you're right about using an AC-DC converter.
After more calculations, the input requirements are: (Assuming 70% eff)
At startup: 28.6A (343Watts !!)
While running: 7A (86Watts)

I'm using a 580Watt PSU, which can supply 420Watts on the 12V rail. It says on the side 12V @ 35A, which would leave 6.4A on the 12V rail for my computer at start-up. (76.8Watts) My processor alone takes more than that.

I'll have a look around for a pre-build SMPS, if not I'll have to use a linear PSU I guess.

On the other hand - This is all just startup current, I wonder if I could get away with it ......

Thanks everyone.

 

[Super_voip]

The recommended power supply, from the link, is a 150Watt switchmode and while it says the pump will run from 18V it's not hard to generate a higher voltage. Transformers work on reflected impedence arrangement so if, at startup, the caps need all the charge they can suck out of a power supply, the inrush current will be high (slo-blo fuses). A tapped transformer should be able to supply 18v from a 14V tap and 20+ from a 18V tap, the size of the cap determines smoothing and for a motor, it doesn't need to be too smooth. If you do use an external radiator just bolt it to the top of the power supply enclosure, that way you won't forget it, maybe move the pump to the exterior and locate it near the radiator. If all the plumbing is internal use a locking connector for the power, if it slips out.....

 

[yngndrw]

So you mean that the inrush current through my current PSU if I was to build this DC-DC converter would be too high ?

Also you're suggesting that I use an unregulated power supply for the pump and just use different transformer taps to change the voltage ?

All my cooling is internal, but I have a bit of space above my existing PSU along with a couple of spare fan holes, so I could mount a new PSU to a fan hole and use the other hole for the mains cable.

I think that the pumps have some startup electronics in them, so I think I'll need to look at regulating the supply regardless. This however would prevent me from using transformer tappings to select the voltage.

Maybe I could use a transformer followed by a small home-made buck converter. That would be simple enough because it's only step down - There must be loads of ICs available which can do the job, each with a datasheet with the circuit pre-designed.

Just need to find an IC if I use that route - Preblem is choosing one.

 

[Hero999]

Why liquid cool anyways? Modern peltiers and a few extra fans with variable speed controllers are just as good, quieter and more energy efficient.

I disagree, water cooling is much more efficient than a peltier.