Powering a Dell Optiplex Micro with 24v

[jamwaffles]

Hi all, first post so apologies if this is on the wrong board.

I've got a Dell Optiplex 3050 Micro which nominally takes a 19.5V DC power supply, however I want to install it in an industrial electrical cabinet and run it off the available 24V DC. I could rig up a DC/DC step down converter but I'd like to keep this install clean.

A long shot, but does anyone know if it's possible to run these little machines on 24V DC? Has anyone done experiments on the minimum/maximum voltage they can accept?

 

[RolloZ170]

i guess the converter Caps on the Motherboard are all 25V type ?. if 24V is highest possible voltage i see no hint for the Caps.
the best way is just buy a cheap used bare motherboard and test it, have seen some for about $35 usd.

 

[jamwaffles]

I've ordered a spare unit I might do some testing with, but thinking about this a bit more I've gone for a small DC-DC converter module I can hide in the chassis to take the 24V input down to a well behaved 19.5V.

If I do test higher voltages I'll try and remember to post an update here for future reference. For visitors to this thread in 2040, no update = no test

 

[jamwaffles]

Well, I gave it a try and hit what is in hindsight a somewhat obvious problem - the Dell chargers have a 1-wire channel that communicates the available power. I tried using a bench power supply at 19.5v but it threw up a warning about an unrecognised adapter and clamped the CPU at 800Mhz

I might revisit this in the future if I can be bothered to either sniff/hack the comm link or try a different brand of 1L machine that uses a dumber power supply.

 

[jamwaffles]

A little bit of info: the MOSFETs on the CPU power supply section are all 30V rated on my unit. The whole 19.5V from the power adapter is present on them, so it might be possible to safely power at least the CPU from 24V if one can get past the warning and downclocking in the BIOS.

 

[RolloZ170]

as i know you need a 22kOhm resistor from + to the smart pin.
worked with laptop before but not tested with optiplex desktop
edit: found out Dell made it mor complicated, using a 1-wire communication device sending power, voltage, and current.

 

[RolloZ170]

but check this:

GitHub - KivApple/dell-charger-emulator: Emulator of original Dell charger using ATTINY85

Emulator of original Dell charger using ATTINY85. Contribute to KivApple/dell-charger-emulator development by creating an account on GitHub.

github.com

 

[jamwaffles]

I happened upon that repo in my research as well, but thanks for the link! I did a bit of reading around and decided to have my own go at solving my question: GitHub - jamwaffles/optiplower: Power a Dell Optiplex 3050 Micro from a 24V industrial power supply

Well, more sidestepping it as I'm still not sure if we can run the PC at 24V directly!

 

[Moopere]

I would've thought that the 19.5v would be a nominal rating. These device PSU's, as with laptops, are meant to work across the globe with widely varying source voltages and source stability - at 24v we're nearly 20% above nominal, so probably pushing the friendship, but I'd give it a go without a lot of worry.

My main concern would be heat. Without a doubt the system will be dropping the nominal voltage down to something else internally - as we pump up the input voltage, and assuming the same current draw, we're going to be dissipating a fair bit more heat. Industrial machines are not airconditioned offices and PC's in them can suffer depending on where you are in the world and during what season. As long as you don't let too much heat build up I reckon you'd be ok.

 

[RolloZ170]

My main concern would be heat

fortunaly Dell don't use linear voltage regs, only digital with mosfet. in some cases the heat will decrease with more voltage(current rises faster)

 

[jamwaffles]

20% above is definitely pushing it I agree, but either way I've got an ATTiny85 dev board on order which I can use to just spoof the PSU validation, so I'll try that and a straight +24v to the motherboard and see what does/doesn't pop.

Designing the custom PSU PCB was just too much fun, and if my straight 24v experiment _does_ cook the Optiplex I can use my fancier solution on the other unit I have lying around

 

[RolloZ170]

i guess the converter Caps on the Motherboard are all 25V(or higher) type ?.
if 24V(regulated?) is highest possible voltage i see no hint for the Caps.

 

[jamwaffles]

All the electrolytics I can see are 2.5v, 6.3v, etc, so I'm assuming these are post-regulation. All the other caps are MLCC so no markings. 25v is the closest standard higher rating to 19.5v so I could be in luck!

 

[Pete.S.]

Hi all, first post so apologies if this is on the wrong board.

I've got a Dell Optiplex 3050 Micro which nominally takes a 19.5V DC power supply, however I want to install it in an industrial electrical cabinet and run it off the available 24V DC. I could rig up a DC/DC step down converter but I'd like to keep this install clean.

A long shot, but does anyone know if it's possible to run these little machines on 24V DC? Has anyone done experiments on the minimum/maximum voltage they can accept?

There are actally several reasons why manufacturers make fanless computers made for mounting inside electrical cabinets. Or to put it another way, several reason not to use a office computer inside an electrical cabinet.

First is obvisouly the ability to run on standard 24V power (or 12V is some cases).
Then there is the industrial temperature rating, meaning it's designed to work over a larger temperature span than office equipment. There is also an extended temperature rating that is goes both colder and warmer.
Then there is the heat dissipation / ventilation issue. Electrical cabinents usually have no circulation compared to your typical office environment. This ties into the industrial temperature range as well.
Then there is EMC rating for industrial environments. There are different standards for what the equipment needs to be able to both withstand and emit.

So I wouldn't bother putting an office PC inside an electrical cabinet. I would just get the real thing instead.

 

[jamwaffles]

Maybe I shouldn't have used the word "industrial" haha... I'm not going for capital-I Industrial here, but just to integrate a smaller PC into a new cabinet instead of having a desktop sitting on the floor next to the CNC I built in my garage . I don't mind fans or office ratings as this is just for me on an already janky setup, but I do appreciate your points and there are many good solutions out there for people doing it the right way.

I could of course just power my machine from its normal PSU, but as this is a hobby I'm going for silly but clever options over what a normal person might choose (and also space constraints)

 

[Pete.S.]

I could of course just power my machine from its normal PSU, but as this is a hobby I'm going for silly but clever options over what a normal person might choose (and also space constraints)

All is fair in war and hobby projects!

I'm surprised to see intelligence in the PSU for something likes this. If it was USB powered it would have been another matter though.
But of course it's all about vendor lock-in from Dell's point of view..

 

[RolloZ170]

I guessed it before, they just drive the PROCHOT# I/O pin

 

[jamwaffles]

There is potential to disable the throttling using this method but I have never tried it. https://www.addictivetips.com/windows-tips/disable-dell-cpu-power-throttling/

Thanks for the tip! It looks pretty easy, even on Linux. I'm actually far enough down the "spoof a legit Dell power supply" route that I got a step-down PSU design and a microcontroller to work. I'll post more updates when the boards arrive from JLCPCB.

 

[RolloZ170]

the bad of this: on VRM overheat the CPU is not throttled down and the board dies.

 

[jamwaffles]

PCBs arrived and it even works with the atrocious firmware I cobbled together from various sources. I am absolutely not shaving the "learning C" yak today . PCB and FW are here if others want to follow along: GitHub - jamwaffles/optiplower: Power a Dell Optiplex 3050 Micro from a 24V industrial power supply

I haven't stress tested the PSU yet so I don't know if it overheats or not. I'm also sidestepping the original question by stepping down to a legit 19.5v but considering the PSU is queried for voltage, it's probably best if the PC is supplied the voltage it was told it was going to get instead of just feeding in a straight 24v.

Now I have to find a place to cram this lil board in the Optiplex case...

Obligatory v1.0 bodge: