Seasonic Power Angel: Could it be Hacked ?

[ToasterIQ2000]

I recently bought a Seasonic Power Angel, a nifty little device to measure power consumption.

I openened it up and I see that there are two circuit boards in it: one houses the AC in / out plugs; the other seems to be dedicated to the LCD display and buttons.

There are six wires connecting the two boards. From my familiarity with electronics I could resolder the wires to separate the butttons and LCD from the larger board and mount that half in a sigle drive bay faceplate -- mostly an informative and amusing bling-bling mod for a pc.

I'm curious though: what exactly is happening on those six wires? Do the buttons on the LCD panel signal the power monitoring board at all, telling it what to report? What sort of signal(s?) does the main board send to the display panel? 0-5 volt or 4-20ma process signals? RS485 protocol? Something uniqe designed into the power angel?

Those questions are over my head: all I can do is speculate. But I am curious if anyone on the SPCR forums would look at this and know... It occurs to me that somene far more familiar with the tech might investigate and come up with a way to make the meter report Amps, Volts, Power Factor, or Watts over a serial or USB port in the same manner that better uninterruptable power supplies can report things. Or extract info off those six wires and turn some part it into something that can be read via a fan tachometer header...

It is beyond my keen. Just tickles my imagination. I think it would be neat...

[Brians256]

Throw some pictures up! The six wires may be (from your description) the to/from for each of the three plug lines: hot, ground and neutral. That is, if the "ac board" has no electronics on it.

[ToasterIQ2000]

The stock unit comes with a short ribbon cable between the two halves.
I soldered in test leads with jumpers so that I can break connections and see what happens, or attatch a multimiter.
This is one of those devices that contain no user servicable parts and I am way over my head.
I suspect that:

the wall socket voltage is sent to the LCD as a voltage scaled linearly in a range from around 1 to 1.3 volts AC between pins 4 and 5, as numbered on the larger board. 110v=1.023v, 110v=1.132v, 120v=1.22v,130v=1.33v.

Either the watts or Volt Amps are sent to the LCD as millivolts AC between pins 3 and 5. 10w=8.5mv, 20w=13.6mv, 30w=19.3mv, 40w=25.1mv, 50w=31mv.

Aside from plugging in and unplugging various AC DC transformers, I have no way of creating a load with variable power factor. And I haven't measured anything on the 6 wires that correlates to the different power factor 'ed loads that I have tried.

With J1 off the Hz reading appears doubled ... or meaningless.

Pulling J6 by itself powers down the LCD and resets it's clock.

Pulling J5 by itself blanks the LCD display, but the display is restored with J5 without resetting the Power Angel's clock.

[bigben2k]

For those trying to follow this, here's a pic and product description:
http://www.seasonicusa.com/products.php?lineId=8

The first (big) board regulates power to the ICs. The LM2902, a quad op amp, may be setup to read the voltage and amperage, once it passed a resistor and the capacitor. The output would be a simple 20 ma current driver.

The second circuit board would be the display control.

The six lines you tapped into would include; ground, power to the display and control ICs, the signal feed for voltage and amps. I'm not sure about the other two lines, but the device only needs the timer on the control IC to calculate kW/hr.

I'm not sure how it's reading the PF (Power Factor).

[Brians256]

It looks highly integrated enough that it would be difficult to do any serious hacking. If you know enough to reverse engineer it, it would be easier to build your own power monitor. The unlabeled chip on the left looks like a std microcontroller with the part number rubbed off to prevent easy reverse engineering (although it doesn't work that well to prevent anyone who is serious).

Good luck. It looks to take more time to reverse engineer than I would be willing to invest.

[ToasterIQ2000]

BigiBen's quick take is precicely what I was hoping for here, thankyou.

My interest was not a reverse-engineering of the whole thing. I was quite pleased that my skill and tools let me find the 'watts consumed' signal. I measure it as 4-8 ma AC, with voltage increasing with power draw: around 1.2 volts perhaps at the Power Angel's maximum rated 15 load.

My main curiosity at the moment is: How complex a cirquit would it take to convert that milliamp 0-1.2v AC signal into pulses that could be read by any old fan-tach header? A fan tach type output where a pulses - per - revolution type signal could be taken as total watts out of the wall socket could be read by any number of fan management utilites, would not require any new logging software, and would be informative in the work done / ( power consumed == waste heat out) sense.

[Brians256]

1) Rectify AC voltage into DC with capacitors (depending upon input impedance of next phase)
2) Turn DC voltage into frequency with something like the LM231 voltage to frequency chip.

It should be quite easy if you do electronics well (which I think you can).

[bigben2k]

No problem.


Like I wrote, the op amp output is a 0-20 ma output, but you don't want to add another resistor in line there, to have your own voltage drop, because it'll throw off the original circuit. So with the existing voltage reading, you can tap a VCO (voltage controlled oscillator) to generate a tach signal. You might have to amplify the signal (voltage) to the vco, probably using another amp; this would also nicely isolate the VCO from the original circuit, and minimize any effect of tapping into the signal.

[Brians256]

Hmm... you wanted the org circuit to work... Good point, Ben.

If so, put a voltage follower before the capacitors and LM231 to show a high impedance interface to the original circuit.

I'm not sure that a VCO would work well on an AC signal, Ben, which is why I suggested rectifying it first.

[bigben2k]

Hum yeah, forgot about that (signal line being AC). Yep, a rectifier, after the high impendence tap, that'll work.

[ToasterIQ2000]

Well, I thumbed through some of my old electronics texts and looked over a LM231 *.pdf spec and decided that this route is too far afield of what I am familiar with.

On the broader topic of how to automate data logging of watts consumed by a PC I did find a couple of other products.

I did net run out and buy a CR5110 from http://www.crmagnetics.com/newprod/transducersg.asp , though that looked perhaps the best off the shelf solution. It would require a freee com port or serial to usb adapter and additinal logging software.

There are quite a few products out there under a search for 'watt transducer' or 'current transducer'. There is one other product that I wanted to mention here, but my link to it is somewhere else on this multi-boot.

I did buy a WattNode WNA-1P-240-P and a 5 amp torroidal coil from http://www.ccontrolsys.com/products/overview.html

Sigh. I think I overloaded it's pulsed output generator with a used pot on the pull-up resistor that is a little jumpy at the low end; One should get a more ocnsistant output and higher sensitivity on the low end than I am getting, I suspect.

Here is just a 'snapshot', ie result of first flush of excitement, from putting the wattnode on my Crystalfontz 633 along side a flowmeter and temp probes. & my curiosity is satiated for now.