Mini-box picoPSU PicoPSU-160-XT

The picoPSU PCU's are often used in lower power devices. I am building a low power server, that is the reason why I am evaluating a picoPSU. The picoPSU performs efficiency wise way better then a bronze 250 watt supply at below 20 watt loads as you can read in my ASRock 1U12LW-C2750 review.

Incorrect height in manual and not 1U compliant

My digital caliper measures a height of 36.11 millimeters for the 160-XT:
[img]NpafOd7D[/img]
Despite this the picoPSU-160-XT manual 1.0d writes on page 4 writes that the height is 30mm and being 1U compliant. Looks like Mini-box has not yet bothered to update dimensions, and made this manual by just copying the 150-XT manual.

150XT height: 30mm

The 160-XT contains 10 watts of horsepower more then the picoPSU-150-XT. Despite that difference this 160 watts edition is also at least 6 millimeters higher then the 150-XT. The 150-XT measures 29.88 millimeters according to the mini-box homepage:
[img=421]XerxpkOV[/img]

DC-DC power efficiency

Now I want to know how the picoPSU performs efficiency wise because the available manufacturer supplied efficiency values are quite coarse. This is what the picoPSU manufacturer (mini-box.com) states in their picoPSU-160-XT "Quick Installation Guide" Version 1.0d at page 4:
[img]ogGsiL0p[/img]

Measurement setup

Let me power up the Itech AT8512A+ electronic load, two Zes Zimmer LMG95 precision power meters to preheat. A 90% efficient Mean Well GST60A12-P1J power supply will provide 12VDC juice. I the mean time I am going to prepare an ATX extension cable by removing 5 times the 5V power pins, 4 times the 3.3V power pins and 4 times a GND pin.

ATX 24-pin extraction tool

Don't buy this ATX pin removal tool on Aliexpress or other Chinese knockoffs, it's a waste of money and resources, these Chinese ATX-Molex-Pin-Removal-Tools just don't work. The trick "How to remove to ATX power connector pins" with staples can work. Or if you have some cash on hand, I can confirm that that the Molex extraction tool mini fit jr 11-03-0044 does the pin extraction job good and lot quicker then the staples removal method.
[img]onSuzyEK[/img]

The pins are freed from the 24-pin ATX connector, bundled together with some tape and attached to the electronic load. In this picture the 3.3 volt and ground wires are attached to the load:
[img]0S5y2lyv[/img]

Output voltage measurement location

Next I do want to measure the output efficiency of the picoPSU, not any output cable losses. That is why one of the LMG95 power meters is used to measure just the output voltage, and do that directly at the picoPSU, via the yellow jumper wire (3.3VDC ATX pin = purple cable):
[img]MLn37kWR[/img]

Input voltage measurement location

However the input cable losses are included in the efficiency because you can't easy change the soldered input power cables. Voltage (and therefore power) is measured at the circular 12V DC input power jack connector.

Pre-heat PSU

After a few hours the setup is ready and the measurement equipment is warmed up. Now it is time to pre-heat the picoPSU itself by loading the picoPSU with 8.000 amps of rated current for at least 30 minutes. The maximum current rating is derived from the mini-box.com manual "Max Load" specified values at page 4. After the initialisation, the power and voltage measurements will be taken.

Measurement cycle length and number of intervals to average

The power analysers are sampling at 100 kHz, where each measurement cycle (displayed value) is 2.00 seconds and where last ... measurement intervals are shown as an average value. For 3.3VDC the average is set to 60 cycles, resulting in 120 seconds; for 5VDC averaging takes place for 120 cycles, resulting in 240 seconds of total interval displayed.

Measurement uncertainty

Measurement equipment comes with uncertainty, educate yourself what this means, I am also showing the lowest and highest efficiency values based on the uncertainty values for the measurement ranges in use at the LMG95. The only measurement uncertainty that is being neglected is the uncertainty of the electronic load. These 2 uncertainties already result in 0.1 till 0.3% points of difference in efficiency.

Eta = η = PSU efficiency

The lower case greek letter Èta (η) is also used as abbeviation/symbol for the efficiency of a power supply, defined as the output power divided by the input power.

in reality less highly efficient

	measured		spec		difference
	η max	η min	η decl.
8A@3.3V	89.4%	89.1%	93%		-4%
5A@3.3V	90.4%	90.2%	92%		-2%
3A@3.3V	89.3%	89.1%	96%		-7%
1A@3.3V	79.2%	79.0%	93%		-14%
8A@5V	88.9%	88.7%	93%		-4%
5A@5V	91.3%	91.0%	95%		-4%
3A@5V	91.7%	91.4%	96%		-4%
1A@5V	85.2%	85.0%	94%		-9%

As you can also see in the more detailed picoPSU measurement results below, for both voltage lines.

3.3V line measurement results

[img=569]7gDlYJm1[/img]

Load (A)	Load	Utrms (V)	P (W)	Efficiency	η max	η min	η decl.
8.400	105%	3.34989	31.6770	88.8%	89.1%	88.6%
8.000	100%	3.34956	30.0248	89.2%	89.4%	89.1%	93%
7.600	95%	3.34916	28.5071	89.3%	89.4%	89.1%
7.200	90%	3.34871	26.9418	89.5%	89.6%	89.3%
6.800	85%	3.34820	25.3367	89.9%	90.0%	89.8%
6.400	80%	3.34787	23.8019	90.0%	90.1%	89.9%
6.000	75%	3.34757	22.2822	90.1%	90.2%	90.0%
5.600	70%	3.34723	20.7735	90.2%	90.3%	90.1%
5.200	65%	3.34685	19.2761	90.3%	90.4%	90.2%
5.000	62.5%	3.34667	18.5323	90.3%	90.4%	90.2%	92%
4.800	60%	3.34643	17.7889	90.3%	90.4%	90.2%
4.400	55%	3.34612	16.3186	90.2%	90.4%	90.1%
4.000	50%	3.34583	14.8564	90.1%	90.2%	89.9%
3.600	45%	3.34557	13.4048	89.8%	90.0%	89.7%
3.200	40%	3.34530	11.9653	89.5%	89.6%	89.4%
3.000	37.5%	3.34515	11.2512	89.2%	89.3%	89.1%	96%
2.800	35%	3.34501	10.5376	88.9%	89.0%	88.8%
2.400	30%	3.34486	9.1199	88.0%	88.1%	87.9%
2.000	25%	3.34465	7.7115	86.7%	86.9%	86.6%
1.600	20%	3.34447	6.3103	84.8%	85.0%	84.6%
1.200	15%	3.34426	4.92051	81.6%	81.7%	81.5%
1.000	12.5%	3.34408	4.22981	79.1%	79.2%	79.0%	93%
0.800	10%	3.34396	3.54095	75.5%	75.7%	75.4%
0.400	5%	3.34379	2.15596	62.0%	62.1%	62.0%
0.000	0%	3.34368	0.82151

Actually the manufacturer specified efficiencies are never met. At 5 amps of 3.3 VDC load the declared efficiency is most close to the reality: 92% versus 90%, only a minus 2 percent difference. At 1 amps of 3.3 VDC load the spec is most far off the real life efficiency measurement: 93% versus 79%, a discrepancy of a whopping minus 14 percent points. Ituner Networks Corp. receives no points for this.

5V line measurement results

[img=577]2JU10Tmu[/img]
I do want to finish these tedious manual measurements and review writing within 2 days. That is why I have reduced the numer of different load measurment points. Sorry that my measurement setup is not (yet) automated.

Load (A)	Load	Utrms (V)	P (W)	Efficiency	η max	η min	η decl.
8.800	110%	4.94106	49.5133	87.8%	88.0%	87.6%
8.400	105%	4.94886	47.0881	88.3%	88.5%	88.1%
8.000	100%	4.95742	44.6477	88.8%	88.9%	88.7%	93%
7.200	90%	4.97292	40.0037	89.5%	89.6%	89.4%
6.400	80%	4.98719	35.3559	90.3%	90.4%	90.2%
5.600	70%	4.99974	30.8207	90.8%	91.0%	90.7%
5.000	62.5%	5.00876	27.4677	91.2%	91.3%	91.0%	95%
4.800	60%	5.01162	26.3095	91.4%	91.5%	91.3%
4.000	50%	5.02279	21.9163	91.7%	91.8%	91.6%
3.200	40%	5.03349	17.5768	91.6%	91.8%	91.5%
3.000	37.5%	5.03604	16.5028	91.5%	91.7%	91.4%	96%
2.400	30%	5.04359	13.3012	91.0%	91.1%	90.9%
1.600	20%	5.05339	9.0786	89.1%	89.2%	88.9%
1.000	12.5%	5.06051	5.94692	85.1%	85.2%	85.0%	94%
0.800	10%	5.06281	4.90561	82.6%	82.7%	82.5%
0.400	5%	5.06744	2.83319	71.5%	71.6%	71.5%
0.000	0%	5.07221	0.81608

Over here the efficiency expectations are never met. The measured 5VDC efficiency is at least 4 percent points lower then the declared efficiency at 3, 5 and 8 amps of load. 5V efficiency is 85% instead of 94% at 1 amps (-9%). No points here.

Measurement pictures

Because this Tweakers review format allows up to 30 pictures, I have included 24 random selected photos of the measured values from both 5V and 3.3V measuring setups. One example:
[img]W0N2SyKu[/img]