Mastero MK-4.5 250w 12v DC-DC ATX PSU + SDC - Page 4

darkhalo · 09-06-2004, 07:43 PM

Eternalightwith : what are you using to print your circuits? and where did you get all of the parts?
Mastero: how long till the kit is out?

Best regards,
Eric

Quattro · 09-06-2004, 07:49 PM

I'm also waiting for the kit, I hope it not too long before its out.

Mastero · 09-08-2004, 04:08 PM

By the 20th of this month the Kits will be for sale

Mastero

Quattro · 09-08-2004, 06:27 PM

Mastero, will the kit thats coming out have autostart up included or are you still working on that? If your still working on it when do you expect to have it finished?

Mastero · 09-16-2004, 06:00 AM

The Mk4.5 shall be with autostart up feature.

MK-4.5 almost ready. Just calculating the prices. if anyone wants one let me know A.S.A.P

Mastero

Quattro · 09-16-2004, 10:38 AM

I'm interested. PM sent.

MrPerfectionest · 09-16-2004, 12:08 PM

Is the output of the LM2587T still set at ~24V?

Mastero · 09-16-2004, 12:19 PM

Quote: Originally Posted by MrPerfectionest

Is the output of the LM2587T still set at ~24V?

yes why ... ?

Mastero

MrPerfectionest · 09-16-2004, 03:53 PM

There is no way your 12V section will live (without liquid cooling) to the specifications you have outlined (8V-30V and 10A output).

For starters there may be a built in limitation with the 2587T that will cause it to fail how you have designed it. If we assume no losses between the Boost to Buck stage then the 2587 must deliver 2.5A@24V to get the necessary 60W for 12V@5A. When you enter 8V for Vin and 24V@2.5A for Vout into national.com's online webbench tool it says the 2587 won't work because the switch current is greater than 1.31A. Even at 14V Vin it still says the switch current must be less than 2.36A. At the specified 8V Vin the best you can hope for and not violate this 'parameter' is about 2.7A@12V (per controller). I have read over the data sheet several times and I can't figure out where this calculation is coming from. I don't know if it is a thermal one or some other one. If it is a thermal limit then you can do some engineering and figure out if it will be ok because the controller is only going to spend a small amount of time at 8V (during crank). Perhaps a message to tech support will answer this (I am also going to play around with the simple switcher software to see if I can figure it out).

That isn't the biggest problem though. The real problem comes with stepping down the 24V to 12V. If someone were to try and run this at 10A (5A per regulator pair) then you will be asking this poor little LM1084 regulator to dissipate 60W of heat. If we only consider the junction to case thermal resistance of 2.7 C/W then the temp rise will be 162 C. If you want to run it like this then you need to affix it to an infinitely large heatsink liquid cooled to -12C or colder.

Why have you chosen 24V to step the device up to? Ideally you should step up to just above the dropout voltage for your regulator. With a boost circuit the regulator (the 2587) will stop regulating when Vin + Vdiode > Vout and go into a 'short circuit' state (current passes through the inductor and diode to the output) which under the nominal condition (car running, battery at ~14V) should be fine. The only problem is what happens when the voltage suddenly dips below Vout (car cranking)? I am not sure if the regulator is quick enough to begin regulating (maybe the data sheet will tell us). It would be an interesting experiment to try this and see what happens (probably need a 'scope to see it). If it isn't fast enough then you certainly can't spec the supply up to 30V Vin. If you can't test this then you should set your output from your 2587 to around 15V IMHO, and lower the input voltage specification.

MatrixPC · 09-16-2004, 05:06 PM

I think the design has 2 independent +12V@5A sections.
And I have no problem with heat/cranking/or any problem for the past 8+ months.
My 3.5 has only 1 instead of 2 sections and I had test it to run stable @ 4.5A (2 CD/DVD ROM 2 72K RPM WD HD).

rando · 09-16-2004, 05:17 PM

@Matrix, the MK3.5 12V section uses a switching regulator to generate its 12v (technically it uses two regulators, buck to 5V then boost to 12V). As such, it doesn't have the issue that Mr.P. is pointing out. The newer design uses a switcher to boost to ~24V and then a linear regulator to drop back to 12V.

I'm not certain but I would guess that this was done to save board space. There just isn't much room to throw in the extra inductors (of the size that are being used) in that layout -- so another set of switchers just won't fit. Tapping the 5V rail, like the 3.5 did, isn't a great solution either since it robs power from the 5V output to feed the 12V.

I've noticed that other PSU designs get away with much smaller (physically) inductors and very small (if any) heat sinks. Instead of fully integrated switchers, they employ fast switches and switchmode controllers. I'm not sure how/why this design avoids the need for such big components. SMT components help cut down the size a bit too but ultimately, it's the inductors that appear to be driving the size of the last 3 generations of sproggy designs and subsequently its design.

MrPerfectionest · 09-16-2004, 05:57 PM

Quote:

My 3.5 has only 1 instead of 2 sections and I had test it to run stable @ 4.5A (2 CD/DVD ROM 2 72K RPM WD HD).

So you are saying that everytime you are in your car both DVD drives and both hard drives are running at full power all the time? I don't even think the OS could task switch fast enough to make 2 hard drives and 2 DVDs run at full blast at the same time.

In reality the average power drawn by all that might be near 2A all the time depending on what you are doing, which the circuit can handle with adequtae heat sinking (that you have).

MrPerfectionest · 09-16-2004, 06:09 PM

Quote:

I've noticed that other PSU designs get away with much smaller (physically) inductors and very small (if any) heat sinks. Instead of fully integrated switchers, they employ fast switches and switchmode controllers. I'm not sure how/why this design avoids the need for such big components.

The 2587 isn't really ideal for use anymore because it switches at a rather slow 100khz.

Newer controllers (I would have used an LM3488 for example) switch up to 1MHZ!!!

This same design using the LM3488 probably could have gotten away with 5uH inductors or smaller.

MatrixPC · 09-16-2004, 07:26 PM

Quote: Originally Posted by MrPerfectionest

So you are saying that everytime you are in your car both DVD drives and both hard drives are running at full power all the time? I don't even think the OS could task switch fast enough to make 2 hard drives and 2 DVDs run at full blast at the same time.

In reality the average power drawn by all that might be near 2A all the time depending on what you are doing, which the circuit can handle with adequtae heat sinking (that you have).

The 3.5 I have is the custom make which is almost the same as the 4.0 that Mastero redesign due to some issue with the 3.3V and 5.0V rail base on National Chips that some people have. So basically the 12V section is the same as Mastero 4.0 and 4.5.

This is how I do the test.
1 DVD ROM and 1 DVD+-RW by Maddog Multimedia both rated at 2A on 12V.
2 WD HD (40G and 60G) 7.2K RPM 0.24A on 12V.
Each IDE channel has a HD and a DVD
Copy 1 550 MB ISO image file from one HD to the other while 1 DVD playing music CD and the other play DVD movie. All running fine for 15 min while engine off, start car, and run for another 10 min. No reboot, no problem and the setup is in the Supra.
Of course, I only have one DVD ROM and a HD as part of the setup.

rando · 09-16-2004, 08:40 PM

Quote: Originally Posted by MatrixPC

The 3.5 I have is the custom make which is almost the same as the 4.0 that Mastero redesign due to some issue with the 3.3V and 5.0V rail base on National Chips that some people have. So basically the 12V section is the same as Mastero 4.0 and 4.5.

Ah gotcha. Your 3.5 changes design so frequently that I can't keep track any more.

You might consider giving a new name in the future, to avoid confusion espcially since the 3.5 and 4.x supplies have so little in common.

As for your power measurements, I'm a bit skeptical that you're using 4A. Can't you just put a meter on your 12V rail and check it out? Even at 2A you would be dissipating ~24W in the LM1084, wouldn't you?