The 2014 El-Cheapo Power Supply Roundup

Honli ATX 680

First Look

Honli is a brand that seems to have appeared only fairly recently on the Australian market. Most of their units are priced very low for their implied wattage, with this 680W unit being available for only $29 from a few eBay sellers. Let’s see how they manage to sell them at such a low price.

Again, we have a more powerful 5V rail than 12V. The power supply is plain grey, but it has a gold coloured wire fan grille.

 

Test Results

Test 1 (77.19W Load)

Rail	Load	Voltage	Ripple
12V	2.48A	12.58V	37.4mV
5V	5.08A	5.08V	10.2mV
3.3V	4.91A	3.34V	11.0mV
−12V	0.1A	−12.04V	45.6mV
5VSB	0.51A	5.07V	19.2mV
AC Power	100.6W
Efficiency	76.88%
Power Factor	0.63
Intake Temp	20°C
Exhaust Temp	22°C

.
Test 2 (104.93W Load)

Rail	Load	Voltage	Ripple
12V	4.8A	12.22V	45.6mV
5V	5.09A	5.09V	11.6mV
3.3V	4.94A	3.36V	12.0mV
−12V	0.1A	−11.81V	53.4mV
5VSB	0.51A	5.1V	22.2mV
AC Power	134.6W
Efficiency	77.96%
Power Factor	0.64
Intake Temp	22°C
Exhaust Temp	24°C

.

Test 3 (157.84W Load)

Rail	Load	Voltage	Ripple
12V	9.32A	11.98V	68.6mV
5V	5.09A	5.09V	14.0mV
3.3V	4.93A	3.35V	15.8mV
−12V	0.1A	−11.96V	72.4mV
5VSB	0.51A	5.08V	30.8mV
AC Power	200.3W
Efficiency	78.80%
Power Factor	0.60
Intake Temp	22°C
Exhaust Temp	26°C

.

Test 4 (204.48W Load)

Rail	Load	Voltage	Ripple
12V	9.48A	12.19V	87.8mV
5V	9.9A	4.95V	21.2mV
3.3V	10.09A	3.33V	20.2mV
−12V	0.1A	−12.36V	90.0mV
5VSB	1.0A	5.02V	45.6mV
AC Power	261.1W
Efficiency	78.32%
Power Factor	0.59
Intake Temp	23°C
Exhaust Temp	28°C

.
Test 5 (253.49W Load)

Rail	Load	Voltage	Ripple
12V	13.81A	11.96V	105.2mV
5V	9.88A	4.94V	25.2mV
3.3V	10.03A	3.31V	23.2mV
−12V	0.1A	−12.56V	106.0mV
5VSB	1.0A	5.0V	51.8mV
AC Power	331.2W
Efficiency	76.54%
Power Factor	0.6
Intake Temp	24°C
Exhaust Temp	31°C

.
Test 6 (301.85W Load)

Rail	Load	Voltage	Ripple
12V	18.18A	11.71V	126.4mV
5V	9.92A	4.96V	30.2mV
3.3V	10.06A	3.32V	26.8mV
−12V	0.11A	−12.88V	119.6mV
5VSB	1.0A	4.99V	56.2mV
AC Power	408.5W
Efficiency	73.89%
Power Factor	0.6
Intake Temp	25°C
Exhaust Temp	35°C

.
Erm, yeah. This explains why it’s so cheap. The results aren’t that great at all. The 12V rail started at 12.58V in Test 1. That gives us 0.58V (4.83%) worst case regulation. Bear in mind that 12.6V (5%) is the upper limit in ATX specifications, so this is very close to being out of spec. The voltage dropped to 11.71V in Test 6, giving us a total drop of 0.87V, or 7.25%, which is a poor result. The 5V rail had maximum and minimum values of 5.08V and 4.94V, giving us 0.08V (1.6%) worst-case regulation and 0.14V (2.8%) drop. This is a much better result than we saw on the 12V rail, but there is still some room for improvement. The 3.3V rail fared the best, with maximum and minimum values of 3.36V and 3.31V respectively. This works out to be 0.06V (1.82%) worst-case regulation and 0.05V (1.52%) variation. This is the only rail which I would consider to have good voltage regulation. The average result for the 3 rails is 2.75% regulation and 3.86% variation, which is not a very good result.

The efficiency peaked at 78.8% in Test 3. That’s somewhat better than our previous contender, but still not great. The power factor results are interesting. They are barely any better than what we usually see on products with no Power Factor Correction (PFC). This unit, however, does (appear to) have Passive PFC, as there is a large coil mounted to the back of the case. I have my suspicions that it’s a fake one where the wires just go in and straight back out. We’ll find out for sure when I take the unit apart. The delta between the exhaust and intake temperatures was 10°C during test 6 – which is actually slightly better than the Tsunami – as it had the same delta at only 250W load. The most I could pull from this power supply was 300W. When I tried pulling 350W, the fuse blew (probably as a result of a failed switching transistor).

Rail	Test 5 (253.49W)	Test 6 (301.85W)
12V
5V
3.3V
−12V
5VSB

.
Like on the Tsunami, the ripple suppression was poor. The 5V_SB and 12V rails both had more ripple than the maximum allowable limits with the power supply under 300W load. The 5V and 3.3V rails fared better, but the ripple was still above half the maximum limit, which is more than what I would really like to see.

Disassembly

The input filtering consists of two X capacitors, two common-mode chokes and two Y capacitors, which is just enough components. Instead of a bridge rectifier, there are four individual 2A diodes. This is tremendously undersized for a 680W unit. The primary side capacitors are also very undersized, at only 330µF, and are supplied by Zhila – a small obscure company. The switching transistors are Huashan KSH13009s, which are rated at 12A. This unit uses a two-transistor 5V_SB circuit; the main switching transistor is a JD Semi BU3150, and the smaller drive transistor is a C1815. Again the critical capacitor is from a no-name brand – Zhila in this case.

The 12V rectifier is an MUR2020CT 20A fast recovery rectifier. I don’t recognise the manufacturer’s logo, but its specs are probably similar to the International Rectifier part. The other two rails use PanJIT SB2040CT Schottky rectifiers, which are also rated at 20A. These parts are all insufficient, since the label said that all 3 rails were capable of more than 20A – especially the 5V rail, which claimed over double what this rectifier is capable of. The secondary capacitors are mostly from Zhila, except the two on the 5V_SB output which are branded H.Q, and one 3.3V output capacitor which is Japanese from Nippon Chemi-con. Unfortunately, it is a KZG series, one of NCC’s few unreliable series.

I was pretty keen to disassemble the PFC coil after the rather low Power Factor results, and my suspicion was correct – it is fake. The wire just goes in and comes straight back out. What is disturbing is that there is no insulation where the wire is joined. Fortunately, it is covered by a few layers of tape while installed in the “coil”. If it was to come loose, however, it could be a disaster, as it is live with 230V, and there would be nothing stopping it from touching the case or another low voltage component – possibly even causing it to dump 230V into one of the outputs! Another problem is that it is installed on the back grille behind the secondary heat sink. This reduces the airflow around it and the cooling it gets. I criticised the Cooler Power GX850 for a similar thing last year.

And here’s a new feature this year. For units that blow up, I’ll include some close up pictures of the carnage. Note the blown fuse and the melted washers around the switching transistors. I’ve seen these same KSH13009 transistors blow themselves to pieces on other units when overloaded.

The fan is branded HXS. It was quiet up to about 200W load. After that, it started to get more noticeable. It was never disturbingly loud, though. Unfortunately, it has a sleeve bearing without much lubricant. The heatsinks are small. They are both fairly thin, and don’t have a great deal of surface to air contact. This unit actually bears some internal similarities (mainly with respect to the transformer markings and the manufacturers of the parts used) to the Power Logic Magnum Pro 315. I wasn’t able to identify the OEM for that one, but this unit has some markings under the primary heatsink suggesting that the OEM is WangLinZhong, a company I have never heard of before.

Specifications and Conclusions

Real Wattage	250W
OEM	WangLinZhong
PFC	Fake
Price	$29 AUD
ATX Connector type	20+4 pin
Worst-case voltage regulation (12v, 5v, 3.3v)	4.83%, 1.6%, 1.8%
Worst-case ripple (12v, 5v, 3.3v)	126.4mV, 30.2mV, 26.8mV
Worst-case efficiency	73.89%
Input filtering	Adequate
CPU Connector	ATX/EPS12V (4+4 pin)
PCIe Connectors	1x 6 pin
Molex (Peripheral) Connectors	2
FDD Power connectors	1
SATA Power connectors	4

.

Pros: None

Cons: Can’t deliver half of labelled rating (−3), Low quality capacitors (−2), Poor ripple suppression (−2), Poor voltage regulation (−2), Fake PFC coil reduces airflow (−1), Uninsulated wire in fake PFC coil, Low quality fan, Inefficient, 5V heavy

Score: 0/10

< Previous Page | Next Page >

Pages: 1 2 3 4 5 6 7 8 9 10 11