Hardware review: GP83 DC-DC Intelligent ATX PSU
a Big thank you to Mp3Car for giving me the opportunity to review the GP83 DC-DC Intelligent ATX PSU.
What is it? The GP83 is a 160W DC-DC converter, with a 8-mode selectable output making it suitable not for only the Carputer afficionados, but also caters for the ever-increasing ITX-community in general.
What's in the box? The packaging was done with protection in mind, and testament to Mp3Car's dedication ensuring your parcel will arrive in good nick, even if crossing the world's oceans on it's way from the good ol' USA to the Land Down Under.
Inside I found the device itself, and even though I decided to toss out all the foamy bits, I could not locate any documentation or disk containing information on the device -> WYSIWYG. Basically you'll have to scour the www to find info on this little guy - or you can use my conveniently attached pdf at the end of this review.
On my way opening the box:
Working my eager fingers through the foamy bits to get to the unit - eventually - did I say it came well packaged:
and then the other side, showing the anti-static plastic bag enclosing and protecting the device:
The contents of the bag reveals the device itself, with a 20-pin ATX cable, and your standard 4-pin auxilliary (P4) power connector. Pre-connected to the PCB is your power-led, remote on/off switch, as well as motherboard switch cable, and lastly the AMP anti-thump control cable:
Description: Looking at the backside of the PCB, I observe a relative clean track layout, with a proper coating protecting the circuitboard against oxidation/degradation. This is a common practice nowadays, but you will occasionally still find electronic circuitboards without a decent protective coating, leading to problems down the line. The solder-work is not that great though, with to big solder-joints formed, and the quality of these joints are not that great either, meaning way to much solder used to fuse components with copper vias.
On the top-side, I'm glad we have a screw-terminal connector block allowing a quick, easy connection of your input (GND, BAT+ and ACC) signals, making this a rather painless process. Here you can also see the AMP anti-thump control cable:
In the next photo, right at the lower left edge, you can make out a small white switch, and three small surface-mount LED's.
Gone are the old, traditional jumpers, and instead this small surface-mount switch, controlling, with 3 LED's indicating, in binary, the mode selected, varying from standard ATX mode up to a variety of on/off delays to suit mostly every need out there:
At the opposite end of the component-side of the PCB we have the standard 20-pin ATX-, auxilliary power, as well as aforementioned remote on/off switch, power LED and motherboard on/off control cable.
Zotac Mini-Wifi ITX board, E7500 CPU, 4gigs Ram, Intel X25-M G2 SSD.
I proceeded to test the PSU with only a single harddrive connected to load up the outputs, as some PSU's will not start up with at least a harddrive connected, even though the motherboard is connected. In this case though, it was more as a preventative measure to ensure proper function - which is not always a successful test, as the following will indicate.
Thoroughly tested the device normal startup/shutdown control via ACC line by selecting Mode1, which = 5sec on delay, 45 second off delay. (For all modes, please see page 12 of 14 in the attached pdf document, which I discovered after an elaborate search on the www)
Back to testing - all outputs behaved proper, and I selected a 12V input, to determine the regulating capabilities of this DC-DC converter. The following conditions were observed, using an industrial-standard calibrated Fluke 87 Process Meter:
Input: 12.1 Vdc
3.3V rail -> 3.38 Vdc
5V rail -> 5.12 Vdc
12V rail -> 12.09 Vdc
-12V rail -> -11.96 Vdc
These numbers are very decent, indicating a well-performing regulating circuit.
One small problem I observed here, which was indicative of what was to come, is that I could only select Mode1, thereafter the mode-selection failed to respond to any more presses of the mode-switch - I was stubbornly persisitent, however, no joy.
Anyway, after semi-successful static test, decided to connect the Zotac Test Platform1 to it - but, it was a NO GO.
The GP83 would fire up in proper sequence determined by (the stuck) Mode1 selection, but, the Zotac refused to initialize completely. The NB fan would spin up, all onboard LED's will activate, but no POST.
Indeed something got trashed in the BIOS, as I could only revive the board back to life after connecting it to normal ATX PSU, and removing CMOS battery, jumpering CMOS CLR, and then resetting to defaults after.
Please note this is a repeatable result - this GP83 doesn't go with the ZOTAC - do not try this at home, as you may end up with a corrupt BIOS, and unless you have a spare CMOS chip, you'll have a nice paperweight.
Well, not to be undermined, I decided to go onto the next step ->
Enter Test Platform2:
Commell LV-677, T7200 mobile Core-2-Duo, 2gigs Ram, Seagate Momentus 100gig, 7200rpm HD.
After the disappointment with the Zotac, I decided to repeat the test on my trusty old Commell platform - alas with similar results. The only difference is that I failed to "corrupt" the BIOS on the Commell, potentially indicative of the industrial-focussed design of this motherboard.
This is reeking of the same issue that plagued the earlier M2-ATX controller when they first came on the market -> with the exception that I only ever had issues with the M2-ATX and the LV-677, which could very well have been an issue with the LV-677 itself.
As some of you may remember, the M2-ATX was notorious for failing to start up certain motherboards, and tied with the LV-677, made for one such combo. Over here I detailed a mod to sort out "timing" issues between the M2-ATX and the Commell LV-677, and it has been running flawless since.
I have a suspicion this is one and the same problem, and one forum member has mentioned in another thread that he thought the GP83 was a M2-ATX knock-off...maybe, maybe not - the sympton on this certainly mirrors that of the M2-ATX without the capacitor mod.
1)Clean design, decent layout, and some innnovative features wrt mode-selection, and providing remote on/off switch capability.
2) Now familiar form-factor, which will allow installation in most mini-itx enclosures available to the entusiast.
3) Either headers or screw-terminals for every connection, a big plus - thank you Genius-PC (Manufacturer)
4) Decent design, with logical location of most components/connectors.
1) Not so evident, but soon someone should design this form-factor controller with 4, as opposed to 2, mounting holes - this 2-hole design is an invitation for disaster, as it doesn't mechanically anchor the board as well as it should, especially if you're the sort who will constantly fiddle with cables, especially plugging/unplugging the ATX 20-pin one.
2) Failure of the controller to work on both my ITX platforms - while the M2-ATX pretty happily performs this same chore, with 20W less to boot.
3) Absolutely no information shipping with the device.
4) Why no 24-pin ATX capability?
Verdict: With the final test outcome, I would neglect if I fail to advise against using this PSU with either the Commell LV-677 or Zotac 9300-series WiFi motherboards - I cannot speak for any other combinations, but, for now, until this issue is resolved, I will certainly avoid this one. It would be good to pen down your experience with the GP83, in this thread, if you own the same, or other brands of motherboards.
Armed with this information, we can report back to the manufacturer, and, should we prove an issue exist with this PSU, they may decided to investigate and fix the potential issue(s). This is pure conjecture on my part though, as one should not draw any conclusions from a single test-sample, as we're all well aware.
I know a few will point the obvious, stating the possibility my systems have power requirements exceeding the capabilities of the GP83 - well, my little 150W Pico-PSU, as well as 140W M2-ATX, happily starts both systems, time and time again, without fail.
On paper, the GP83 has pretty impressive specs, allowing a wide input range of between 8-28Vdc, with no less than 90% output efficiency across the input voltage range - this at maximum load!
Ripple Voltage will also not exceed 300mV p-p at 28V DC input, and maximum 15A current. Impressive, for a converter in this price-range.
Maximum current output on both the 3.3V, 5V and 12V rails is 8A, with peak loads of up to 12A's are allowed, with the provisor it is for a period shorter than 60 seconds.
Should you consider using this PSU, with an input approaching +18Vdc, the manufacturer's recommendation is to use positive airflow to cool SMD's, as efficiency, reliability and longevity all tapers rather sharply past the 50 deg C mark
For further specifications, please refer to the attached GP83.pdf file. It should also answer any other questions you may have.
As a result of the outcome of this review, I will write a report and return this unit to Mp3Car, who should use this opportunity to send it back to the supplier to investigate and rectify the potential issue. Pending this outcome, I will update this thread in the (hopefully) near future.
Thank you for reading, hope you will find something useful to take from his :)
Phillip aka mrbean.