How to stabilize 3.3V rail of stand-alone PSU?

[logan893]

I'm looking for ideas for how to stabilize an almost unloaded 3.3V rail of a stand-alone PSU.

Current setup has me using a stand-alone PSU to power additional HDDs which my server's main PSU are not equipped to handle. Also, the Fujitsu server PSU supplies only 12V to the motherboard, and the motherboard only provides 5V in addition to 12V for HDDs. There is no 3.3V available as far as I know.

I have two Intel DC S3700 1.8" drives, which run on 3.3V only. I have passive adapters for these, as I was previously not able to find active ones with powerful enough (2+ Amps) converters.

The drives work great in another 6 year old machine, which does supply 3.3V from the PSU, and the PSU also drives a motherboard which seem to keep the 3.3V rail above idle. SSDs work great in this system. I'm looking to move the SSDs to the server.

Here's where the problem starts to appear. The server itself does not provide 3.3V, so I'm relegated to use the stand-alone PSU. The stand-alone PSU does not have a motherboard connected, and thus its 3.3V rail is mostly unused. Drives start up, and work mostly fine when idle. Using dd to write a mere 1-2 MB of data at a time works fine. Increasing this further, to 3 or more MB, will cause the drive to tax the otherwise idle 3.3V rail too much too quickly, and the voltage drops. My not-so-fancy multimeter shows a drop from the regular 3.3V down to just below 3.1V. The voltage drop causes a power loss and reset of the drive. (Specs state minimum 3.13V; 3.3V +/- 5%.)

Any ideas for how to stabilize the 3.3V rail of a stand-alone PSU? What's a good item to connect to the 3.3V line to provide some base load?

If I cannot stabilize the PSU 3.3V rail, I also have some 3.3V step-down converters rated at 3A which I may try to solder in-line with SATA power adapters.

I also found a micro SATA to SATA adapter with a 4A converter, however a unit price of $25 (according to their ebay listing) is steep.
www.circuitassembly.com/product_view.html?partno=H709259

Edit:
The stand-alone PSU is a Corsair CX430. Nothing fancy.
CX430 — 80 PLUS® Bronze Certified Power Supply

 

[ttabbal]

If it's just that you need a load for the PSU regulation to work properly, the best option is a power resistor. Maybe about 1 ohm 10W between the 3.3V and GND. That would pull 3.3A, which isn't that much as PSUs go. That's 9.9W, so a 10W resistor is a bit on the edge, maybe bump up to a 20W. Make sure to heatsink it, it's going to get hot. I like the metal cased units, pretty cheap on ebay or a local surplus electronics place. I then mount them to the metal case of the PSU, in the fan airflow if possible.

Note, I've only done this on the 5V line for a ATX->bench PSU conversion. That might be what you need as well, not positive.

 

[logan893]

I like the idea of a single large resistor. I don't have any high current low resistance resistors on hand, unfortunately. I could attempt a proof-of-concept using multiple higher resistance ones, but seeing as all I have are tiny 0.25W ones, I'd have to use 40 or more at for example 68 Ohm. That's just not practical.

I'll have to order some heavy duty ones via ebay to test the concept.

Perhaps I'll figure out some other way to test it while I wait.

 

[logan893]

Unfortunately the dummy load idea has not been very successful.

I've tried to hook up both 1 Ohm and 2.2 Ohm resistors (capable of handling 50 and 25 W power dissipation respectively) to the 3.3V line, and the PSU does not like this one bit. It goes into some power-down protection mode, and needs to be powered off for a while to recover.

If I simultaneously hook up one 2.2 Ohm resistor each to 3.3 V as well as 5 V, it runs for a little while longer, before ultimately powering down automatically.

Voltage measured is also on the low end with this load.

Perhaps I'll string two 2.2 Ohm resistors in series on the 3.3 V line, and also add a capacitor, to see where that gets me.

 

[ttabbal]

Try connecting the resistors between 5V and GND, leave 3.3V out to start with. The 5V line seems to be what most PC PSUs want to have a load on. I don't think capacitors will help here, try putting a few resistors in series, you might be pulling too much power and triggering the PSUs short circuit protection mode. More resistors in series will increase the total resistance, lowering current.

ATX to Lab Bench Power Supply Conversion - All

That link has some info, and in the comments people talk about different things they did to get their PSUs to work. Perhaps one of those will help get yours working better.

 

[logan893]

Thank you for the suggestions.

I did a few attempts with smaller resistors, and capacitors, on both 5 V and 3.3 V, with just as poor result as all previous.

Then finally I strung all three large resistors I have together in series (2x 2.2 Ohm (25 W) and 1x 1 Ohm (50 W)), to create roughly 2.5 A dummy load on the 12 V rail, and this seem to do the trick! 3.3 V rail now seem stable. I'll try to put the drives through some varying load to verify the PSU reliability.

At first I didn't quite understand how this load would behave much different than having two HDDs connected, as I had previously. I thought the two HDDs should be able to add a comparable load on 12 V rail as these three resistors. Further reading of the spec sheets for Seagate NAS 4TB drives however reveal that while they consume 2 A from 12 V during spin-up, they consume only 0.283 A each during operation.

This should serve as a temporary measure to verify the solution. I may procure a single large resistor capable of dissipating 50 W or even 100 W, at somewhere between 4.7 and 10 Ohm.