Power Supply HDPLEX 250W alternative - What would you like to see?

[QinX]

Qinx,

I have a Canada customer using Voodoo 350W with HDPLEX 250W DC-ATX(latest version). Here is what he reported.

Quote" I am currently running it with a Voodoo 350W power supply and my PC only has a 4670k (no GPU). However, the unit keeps shutting itself down."

Then he measured the DC output from the Voodoo adapter.
Quote" I measured the voltage, and there would be massive dips to 12-14V when power consumption suddenly increases."

Thanks for the response!
I've tested with several AC adapters and my own mini PCB. All had similar problems. So I do believe that his case was one of a DOA brick.

As it stands right now I've tested with the following configurations
HDPlex 250W + HP Voodoo 350W = Not stable
HDPlex 250W + Meanwell RSP-1000-24 @19V = Stable
HDPlex 250W + Meanwell ERP-350-24 = 99.9% Stable ( I have seen a crash here and there but much much less frequent)
QinX Passthrough PCB + 12V 192W AC Brick = No Stable
QinX Passthrough PCB + Meanwell RSP-1000-24 @ 12V = Stable

Ofcourse the 1000W PSU is overkill but shows nothing is wrong with the DC-DC board itself.

my old generation design and dimension

Please do tell what future goodness you have in store for us

 

[michaelmitchell]

I am a little bit confused as well. It seems the project is basically to have Pico-BOX copy my old generation design and dimension, label it at 300W and most likely to sell it at 65-75USD on Aliexpress or ebay.

Nothing wrong with more competition.
I would suggest Frank from Pico-BOX to use Germany WE inductor which should have better response curve at high A application.

Well I guess the idea is to pretty much have a drop in alternative for the HDPLEX with minor changes which a few small cases already support in its current dimensions, I don't know much about this kind of thing but I imagine there aren't many configurations of components to achieve that and the component choice looks similar to the previous two revisions of the X3 apart from inductors which might have just been for prototype convenience.

The current price for the X3-ATX-300 is $54.00 and I imagine this revision will be about the same.

Only Frank could tell us more, but I don't expect he checks these forums much so I will send him an email for some feedback .

Thanks for the response!
I've tested with several AC adapters and my own mini PCB. All had similar problems. So I do believe that his case was one of a DOA brick.

As it stands right now I've tested with the following configurations
HDPlex 250W + HP Voodoo 350W = Not stable
HDPlex 250W + Meanwell RSP-1000-24 @19V = Stable
HDPlex 250W + Meanwell ERP-350-24 = 99.9% Stable ( I have seen a crash here and there but much much less frequent)
QinX Passthrough PCB + 12V 192W AC Brick = No Stable
QinX Passthrough PCB + Meanwell RSP-1000-24 @ 12V = Stable

Ofcourse the 1000W PSU is overkill but shows nothing is wrong with the DC-DC board itself.

Please do tell what future goodness you have in store for us

Josh used a Dell 330W in his recent R9 Nano tests and it seemed ok. I use one also, but do not have an R9 Nano to test with.

 

[michaelmitchell]

Here is some feedback from Frank about component choice.

Dear Michael,

As for the inductor choice, the criteria is to reduce the cost while maintaining the similar efficiency as that of WE inductor.
- We will change the appearance of inductor, so that looks different from WE. Our inductor supplier will do that.
- We can make some comparison test between PICO-BOX and HDPLEX, in terms of efficiency.

Another difference between PICO-BOX and HDPLEX is the ATX controlling circuit. HDPLEX uses TPS3510 from TI while PICO-BOX adopts a STM8S MCU from ST. This is the main feature of PICO-BOX series PSU. The MCU based controlling circuit provides the following advantages:
- In the past we faced with the problem that some motherboards cannot work with the ATX PSU that uses TPS3510. However, we can change the firmware of STM8S to make those motherboards work well with PICO-BOX PSU.
- MCU based circuit can monitor more signals (12V, 5V, 5VSB, 3V3).
- In the future we will extend more functions based on MCU, such as we are planning to develop software and apps so that customers can configure and diagnose the PSU by themselves.

 

[MarcParis]

I fear that inductor choice from Pico box will be only cost saving vs WE Hd plex whereas we will need additionnal function to power R9 Nano and hungry GPU..

 

[michaelmitchell]

I fear that inductor choice from Pico box will be only cost saving vs WE Hd plex whereas we will need additionnal function to power R9 Nano and hungry GPU..

All indications point to the power bricks at this time, I don't know that any innovation in the DC-ATX boards will help if the power brick is tripping. @hdplex post about his customer switching to Dell 330W is very interesting, it suggests that perhaps the Firebird 350W may not be the best choice even though its stated 20W more capacity to attempt powering the R9 Nano if it has some issues in dropping voltage under load increases, the Dell 330W has run as high as 420W by Dan before.

I very much doubt inductor choice will effect the ability to power an R9 Nano otherwise HDPLEX 250W would already handle it just fine.

@QinX Can you get a Dell 330W adapter? I would llike to see the same tests run with both HP 350W and Dell 330W run on HDPLEX and X3-ATX-300 when I send it to you.

 

[MarcParis]

In fact I've tried Dell 240W unsuccessfully also...really, right now, I'm not that motivated to try Dell 330W right now

 

[iFreilicht]

In fact I've tried Dell 240W unsuccessfully also...really, right now, I'm not that motivated to try Dell 330W right now

That makes little sense. Just the 175W of the Nano + 91W of the CPU are 267W, which is more than the 240W the Dell PSU is rated for, and as QinX said, the Nano and many other GPUs can have a peak load much higher than that.
All evidence we have right now points towards the Dell 330W stably running the Nano, nobody has yet built a system with that combination that ran badly. You should probably give it a shot.

 

[MarcParis]

That makes little sense. Just the 175W of the Nano + 91W of the CPU are 267W, which is more than the 240W the Dell PSU is rated for, and as QinX said, the Nano and many other GPUs can have a peak load much higher than that.
All evidence we have right now points towards the Dell 330W stably running the Nano, nobody has yet built a system with that combination that ran badly. You should probably give it a shot.

Well, when I've tested Dell 240W on R9 Nano, I meant to power it only R9 Nano..just connecting pci express 8 pins port to HD plex+Dell 240W.

Please also note that it was several times reminded that R9 Nano has power peak till 500W (apparently mainly through the pci epxress port) to manage...currently both voodoo 350 (Qinx and my feed recent feedback too) AND dell 330W (Josh feedback) seems uncapable to manage it

 

[EdZ]

R9 Nano has power peak till 500W (apparently mainly through the pci epxress port)

The PCIe port will supply a maximum of 75W. The other 100W has to come from the PCIe power connectors, and any peak loading will be coming from those connectors.

 

[QinX]

I believe currently I'm the only one that has actually ran the R9 Nano on a single supply. All other tests have had the PCIe slot power as well.

@michaelmitchell I will see if I can source a Dell 330W unit, that would give us the widest range of tests.

 

[MarcParis]

I believe currently I'm the only one that has actually ran the R9 Nano on a single supply. All other tests have had the PCIe slot power as well.

@michaelmitchell I will see if I can source a Dell 330W unit, that would give us the widest range of tests.

Thanks Qinx, what is your revision on HD plex, as yours is not coil whining ?

 

[QinX]

Looking at the original photo I can see it says Revision 1.2 on the board. So that one is already outdated again.
The whining version is on the top and revision 1.2 is on the bottom.
Perhaps @hdplex can explain what has changed and why the coil whine has returned. Maybe he is also willing to send me the latest revision, perhaps that also reduces the crashing with the R9 Nano. Then I can make a comparison based on the latest products of but HDPlex and Pico-Box

 

[MarcParis]

My HD plex is rev 1.4 and it's coil whining loudly powering R9 Nano..
I guess it's like R9 Nano I need to have luck to pick up the good one..

 

[hdplex]

My HD plex is rev 1.4 and it's coil whining loudly powering R9 Nano..
I guess it's like R9 Nano I need to have luck to pick up the good one..

So your HDPLEX 250W is whining powering R9. Is it stable?
Qinx, you are using HDPLEX 250W to power R9 as well? Is it whining?

 

[QinX]

My HDPLEX 250W doesn't have any whining, but isn't fully stable with the R9 Nano.

HDPLEX 250W Rev 1.2 + HP Voodoo 350W = Not stable
HDPLEX 250W Rev 1.2 + Meanwell RSP-1000-24 @19V = Stable
HDPLEX 250W Rev 1.2 + Meanwell ERP-350-24 = 99.9% Stable ( I have seen a crash here and there but much much less frequent)

 

[MarcParis]

Well unfortunately it's not stable..and it's coil whining. But I'm not sure it's related to hd plex directly. I've tested with dell 240w and voodoo 350w. The latter is a bit more stable but not much. Last option is dell 330w power brick but we still don't if it could handle current peak.
Do you think there is any solution possible (except going to atx/sfx power supply)?

 

[hdplex]

Perhaps @hdplex can explain what has changed and why the coil whine has returned.

Yes, my engineer discovered the early coilwhine version has "self-exciting"happening during peak load. Changed the inductor and adjust some parameters to solve this. The early black inductor was made in Taiwan. As you could see Pico BOX uses the same black inductor as my early coilwhine version on their sample.
The none whine version V 1.3 you have uses the grey color inductor which is from Japan.
The latest version V 1.4 uses German WE inductor。
QinX: you don't have coil whine using V 1.3 with R9 Nano,correct?
I tested the latest version using GTX970. No coil whine.
But if QinX V1.3 does not have coil whine, why would MarcParis V 1.4 has coil whining?
Maybe R9 Nano is a different beast?
Hmmm...Maybe I should purchase a R9 Nano and test myself with my upcoming internal 250W AC-DC adapter.

 

[MarcParis]

In my case, I'm only powering r9 nano using hdplex (not the full system).
I didn't try to power full system directly using hdplex as I should be above 250w on average.

Maybe I can try just to know if there is still coil whine or not.

Ps : how big is your internal ac-dc 250w?

 

[MarcParis]

Well I guess, It's better to show than writing..
(SORRY in advance for bad quality image..but sound is good..)
1st video : with all system powered by HD plex 250W+voodoo 350W...not stable even on Heaven benchmark. Average power consumption on AC wall meter : 250W

2nd video : only R9 Nano powered by HDD plex 250W+voodoo 350W. Stable on heaven benchmark, but not on more demanding games. Average power consumption on AC wall meter : 192W

I hope this could help..

 

[QinX]

Yes I've got the V1.2 and I assume the V1.3 uses the same inductors.
It doesn't have any coil whine, aside from the R9 Nano itself producing coilwhine.