Overclock.net › Forums › Industry News › Hardware News › [AquaComputer] Aquaero 6
New Posts  All Forums:Forum Nav:

[AquaComputer] Aquaero 6 - Page 190  

post #1891 of 1963
Quote:
Originally Posted by ngzb View Post

Quote:
Originally Posted by IT Diva View Post

The silicon lottery, mostly . . . . biggrin.gif


As well as with individual parts variations and combined tolerances from spec.

Anyway we'll have a Much better idea here in a few more days wink.gif


Darlene

Anxiously awaiting for the results biggrin.gif



Okie Dokies . . . . . .

Little update, . . .

Postal service was outstanding this week, and both packages arrived today and I picked up the fans and splitter that came from from FCPU and all the electronics bits and some more PCB material that came from Digikey. thumb.gif


It took considerably longer to open 8 fan boxes than to see the problem on the scope as I plugged them up 1 by 1.


It was, as I expected to find, that adding fans with the PWM pins in parallel, causes the "off time" portion of the pulse to not go down all the way, or even close enough, to 0V. It gets progressively worse, less close to 0V, as you add each fan.

Another interesting revelation, was that the Corsair fans use a 2V pullup on their PWM line, where the standard is 5V.

That's not an issue to the A6, as it's doesn't have an internal pullup, but could be to other controllers, if there are any, and they can source as well as sink current on the PWM line. My PCB's can be configured either way.

Note that the PWM D5 needs a PWM output that can source current. . . . Which is why it doesn't work with the A6 natively, but works fine with my PCB's or mods.

For a bit of a run-down . . . .

With just the fans and splitter without one of my PCB's . .

I found, as others have, that up to 4 fans work quite well, . . . 5 fans work pretty well, with noticeable sacrificing of the lower rpm range.

Adding the 6th fan pretty much looses whatever amount of control you just had with 5.

I was getting minimum speed with near 0% pulse width at ~1800 with 6 fans, about 1200 to1300 with 5 fans, and a reasonable 900ish with 4 fans.

Adding the 7th gave near max rpm regardless of the controller setting.

With all 8 fans connected, the barely 2V PWM signal was barely getting down to 1V when it was supposed to be at 0V, and there was virtually no speed variation regardless of pulse width.


So to try something quick and easy, I grabbed the first version of the "test" PCB I had made and used for some initial testing.

Using a channel on the PCB with a small output transistor to see if it seemed stressed, and no pullup to 5V, all 8 fans are controlling perfectly.

The Corsair fans hold min rpm of about 750 from 0 to ~40% duty cycle.

From ~40% up to 100% they smoothly increase their rpms.

I set min to 40%, and max to 100% in the fans tab, and them moving the slider from 0 to 100 in the controllers tab gives a nice smooth control range.


I'll build the new PCB to take advantage of its designed in options, so that with a single jumper block on each channel, you can select a stronger pullup for dual PWM D5's, a weaker pullup for a single PWM D5, or no pullup at all for the 35X and the Corsair fans.

I'll order 2 more splitters and 7 more 2-Packs of the PWM SP120's in the morning.

That will give me 22 fans I can put on 1 channel to be sure I have a sufficiently robust output circuit.

I'll update shortly with some pics, just as soon as I take them.


Darlene

Quote:
Originally Posted by skupples View Post

Wut? & I thought some of the OCN mods were a bit over zealous...


@IT Diva Do you have access to PCB manufacturing via work or something?


The only PCB shop anywhere around here is in my kitchen.

All my boards are made by a 1 at a time, mostly manual process, once I have the design completed.
Edited by IT Diva - 3/13/14 at 5:59pm

Gear mentioned in this thread:

post #1892 of 1963
Great update, was that with the quiet or high performance editions? I am still able to control 6 of each on parallel per AQ6 header on the Swiftech splitter. The minimum rpms have slightly suffered on both though but not as bad as you described.
post #1893 of 1963
Quote:
Originally Posted by geggeg View Post

Great update, was that with the quiet or high performance editions? I am still able to control 6 of each on parallel per AQ6 header on the Swiftech splitter. The minimum rpms have slightly suffered on both though but not as bad as you described.

I got all SP, figuring they may be a little more demanding than the quiets, and for testing, stress is a good thing.

Like I said, it's pretty much a silicon lottery with whether you get decent control with 5 or 6.

Part of it is just how much the "off time" voltage stays above what the fan control electronics see as 0, what the electronics see as 0, and just how well the A6 can pull towards 0.

There's a threshold there where a little below and all is pretty well, and a fraction above and control is near nonexistent, at least for the lower half of the speed range.

I'll have some solid parts values by weekend's end, and I'm sure at this point that the little PCB can have all 4 channels repurposed for running the Corsair PWM fans.

I think I mentioned, but the 35X pump, while it works well natively with the A6, would work fine and probably more precisely, from a channel configured for the corsair fans.

Darlene


Maybe when I order more fans tomorrow, I should get some quiet editions and not all SP.
Edited by IT Diva - 3/13/14 at 6:16pm
post #1894 of 1963
You mean HPE, both editions are SP tongue.gif

Your work will be a godsend for me if I am to set up a couple of water loops in the Caselabs TX10.
post #1895 of 1963
good job thumb.gif

Darlene you have to create you own brand of eletronics stuff for modding biggrin.gif
post #1896 of 1963
Quote:
Originally Posted by geggeg View Post

You mean HPE, both editions are SP tongue.gif

Your work will be a godsend for me if I am to set up a couple of water loops in the Caselabs TX10.

Checked the boxes, I have the HPE's.

Buggers, looks like they are out of stock at FCPU, only the 2 of the QE's in stock.

Both still in stock at PPC's, but the swiftech splitter is OOS both places, buggers . . .

Guess I'll have to expand the one splitter I have with some of these:

http://www.performance-pcs.com/catalog/index.php?main_page=product_info&products_id=25413:cdf301ffeca1031dc315800f105fca00 \

Since I'll be looking to sell off the fans (with a PCB to sweeten the deal) after I use them for testing, which ones should I get, more HPE's or the QE's?


Darlene
post #1897 of 1963
I may well be one of your customers and to me the HPE makes more sense since they can always be tuned down or raised up if need be.
post #1898 of 1963
Quote:
Originally Posted by geggeg View Post

I may well be one of your customers and to me the HPE makes more sense since they can always be tuned down or raised up if need be.

I second this!
First Rig
(15 items)
 
  
First Rig
(15 items)
 
  
post #1899 of 1963
Quote:
Originally Posted by geggeg View Post

I may well be one of your customers and to me the HPE makes more sense since they can always be tuned down or raised up if need be.
Quote:
Originally Posted by MrStrat007 View Post

I second this!


+1
post #1900 of 1963
Just ordered 5 more 2-Packs of the HPE's, and some splitters to expand the swiftech one.

That will give me 18 fans to run on a channel to see that the PCB design is amply robust.


Here's the pics:

The problem that causes the A6 to not be able to control more than 5 or 6 of the Corsair PWM fans, with decreasing low speed range control from 4 on up, and no control at all at 7 or more.

This is the PWM signal at the A6 with 8 Corsair fans on a Swiftech splitter. Notice that the signal peak voltage is only 1.7V, where customarily the PWM standard calls for 5V.

More importantly, the "off time" voltage that needs to pull low to a couple tenths of a volt max, is right at 0.8V.

This is just about 100% duty cycle to the fan's electronic control, so it runs at very near full speed, regardless of how narrow the pulse is.





Test bench setup:




Closer look:







And here's the pics with the PCB installed . . . . . Easy to see that the "off time" voltage drops properly to a very near 0V level

Near max speed: 2010 rpm

]

Scope . . top is at the PCB at the A6, lower is the fans at the PCB:





And what doesn't work without the PCB . . . . Low speed control: 754 rpm




Scope:




Darlene
Edited by IT Diva - 3/13/14 at 8:54pm
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Hardware News
This thread is locked  

Gear mentioned in this thread:

Overclock.net › Forums › Industry News › Hardware News › [AquaComputer] Aquaero 6