IDE/ATA specs

[NinjaKitten]

I'm building a box with a set of six 500GB IDE hard drives, each connected by an IDE to USB cable to a USB2.0 hub. I've purchased a cheap 100W computer PSU to power the drives.

There are two things I want to do that I need help with:

1) I want to have an "activity" light for each of the drives and, according to Wikipedia's entry on Parallel ATA, pin 39 is the activity pin. However, on HowStuffWorks the pin is listed as DASP (Drive Active / Slave Present). I'm assuming that because each drive is only connecting to a single USB convertor cable (i.e. there's no slave drive) there should be no problem using this to drive the activity LED, right? What would be the best way to drive the LED from that signal given that the drive is powered by an ATX-style (+5V, 0, 0, +12V) molex connector?

2) I want the drives to power on when the +5V line of the USB cable (from the PC) goes high. The power supply can be turned on by shorting pin 14 (PS_ON) of the ATX header to a ground pin. Would it be possible to solder the coil of a 5V relay over the +5V and GND pins of the USB cable, then connect the switch over PS_ON and a ground on the ATX header? Would I need to put a capacitor over the switch to reduce jitter?

Cheers,
NK

[dangus]

1) Use some sort of buffer chip (or discrete transistors) to drive the LEDs. Like a ULN2003 kind of thing, or an HCMOS hex inverter. Presumably you can solder a wire to pin 39 on the USB adapter. For that matter, some USB adapters might even have activity LEDs on them.

2) The relay could work, but a transistor should work too. NPN transistor, base to the USB 5V through 1k, and 10k from the base to ground. Emitter to ground, collector to the ATX on line. Probably.

Personally, I'd rather put the drives in the computer and use a Promise IDE controller which adds 4 IDE ports. But then I don't completely trust this newfangled USB yet.

I presume you'll be adding some generous cooling for these drives; I'd use two or three salvaged power supply fans blowing incoming air directly over the drives, maybe running on reduced voltage (7 to 9 volts) to quiet them down. Passive just doesn't cut it, unless maybe you're taking advantage of special power-saving features. Just leaving a 3.5" drive plugged into a USB adapter on the desk, the case temp hits 50 C or so after running for an hour. Not good for longevity.

[NinjaKitten]

For part one, do you mean put pin 39 of the IDE adapter to an inverter (NOT gate) and then drive the LED through the output? I thought it was active high? Although in that case would I just use two inverters in series?

Regarding the transistor idea, could I still put a 5uF (or possibly 10uF) cap over the LED without it causing issues? I don't want the LED to be flickering uber-fast.

The issue is that my PC already has 6x 1TB SATA hard drives in it, which uses up all available slots. There's no space to put the drives anywhere. Eventually I'm going to remove the USB controllers and have a cheap PC build set up as a server with this unit attached. A cheap socket AM2 board with Athlon X2 + cooler and 1GB of DDR2 comes to £100 right now, without a case. Add 3 or 4 two port IDE controllers for £13 each and you're looking at £150 overall. That's without drives. My current build is costing me around £50 in total, again without drives.

The case I'm making is going to use a PWM control. I have a pair of thermistors going to analog inputs on a PIC, with digital outputs driving the fan PWM pins. It's a simple linear relationship between input level and output frequency. The drives sit on their sides to allow better convection cooling, and there's a pair of fans at the top and bottom of the unit for when convection isn't enough. I've attached a basic idea of my design for airflow. The view is from the side; left = back of the unit, right = front.