The old axiom “you get what you pay for” is often invoked in discussions of product quality. It can also inspire curiosity as to what exactly one does get when the price paid for a product is significantly less than asked by others having comparable basic specifications.
It was this spirit of curiosity that led a colleague to purchase a bench-type adjustable power supply, advertising a 0-15V, 0-2A output from a low-cost supplier for all of $17 USD. What exactly does that buy a person? Let’s find out.
Figure 1. Front panel view of the power supply to be examined.
The model in question is branded Best, part number PS-1502DD. It’s a single-output adjustable supply, with an adjustable current limit, a number of fixed voltage output settings, and one variable-output setting, in which the fine and coarse voltage adjustment outputs become active. Being a bench-type DC supply there are of course positive and negative output binding posts/banana-style terminals, as well as a third marked GND the likes of which are commonly found also in higher-cost supplies, and which typically provides a connection to the chassis/protective earth connection of a device. There’s also a pair of 3-digit, 7-segment LED displays to indicate output voltage and current. And a power button. All are rather common features, which will be examined more closely in a bit.
Edit: After completing this write-up, it was found that this same supply is available under maybe a half dozen different brand names, at price points ranging from the $17 USD range mentioned to upwards of $50 USD. Numerous resources are available demonstrating the basic functionality of the device; this article focuses more on reasons that a person might want to consider spending more on a device of this type.
Figure 2. Rear panel of supply under examination.
Scrutiny of the device’s rear panel leads to a few amusing observations. The shiny holographic sticker with the company logo stands out. At one point in time such devices were viewed as anti-counterfeit measures, the idea being that faithful reproduction of such a thing would be difficult/expensive, and thus offer an indication to observant customers that an article in question might not be authentic. Since the product at hand –is- a low-cost product already, the sticker leads a person to wonder if the producers of this item might not be concerned about the prospect of having even lower cost counterfeits of their own products undercut their revenues.
Figure 3. Close view of holographic sticker
Figure 4. Supply’s warning/cautionary sticker. Note that the referenced protective grounding conductor is non-existent on this device.
Next is the obligatory caution sticker. While the choice of line breaks seems rather avant-garde, the text of the message is disappointingly passé, the usual “don’t chop the third prong off the plug, don’t try to fix it yourself and don’t stuff the fuse holder with aluminum foil when the fuse blows” sort of boilerplate that few ever take time to read. To do so however would lead one to miss the subtle humor of this advisory, in that as we follow the power cord from the back panel to its terminus, we find…
Figure 5. The non-grounded, non-polarized europlug found on the equipment being examined.
…what appears to be a europlug which of course doesn’t have a protective ground . Very witty humor, this… The joke is made all the better by the symmetry of the plug, which cheerfully allows interchanging of line and neutral conductors. Moreover, this example appears to be labeled as having a 6-amp current rating, which is more than double the 2.5A that is typical for plugs of this apparent style. The cordage material is labeled as being 0.75mm2 in cross-section (between 18 and 19 AWG) so perhaps this labeling is more in the spirit of an Absolute Maximum rating rather than a maximum recommended operating limit sort.
Figure 6. Rear view of the supply, including cord. Users will either need an extension cord or an outlet in very close proximity.
Figure 7. Supply cable markings indicating conductor size.
Returning to the back panel, there’s a fuse holder, containing a single glass fuse in 5x20mm format, which appears to carry a 1A rating. And, perhaps most prominently, there’s a big old metal can TO-3 packaged device, labeled as being a 2N3055 transistor, atop what appears to be a mica insulator which seems to have been adhered to the device case by some means. The logo on that transistor isn’t ringing familiar, though by tradition the 2N3055 is a 15A/60V NPN type, the ON Semi and ST Micro versions of which list a 200°C maximum storage and operating junction temperature. Perhaps that temperature rating will come into play, as the supply’s chassis seems to be the only supplemental heat sinking provided to the transistor.
Figure 8. Fuse and its retaining cap.
Figure 9. The transistor used to waste energy not drawn from the output.
On the side panels, the ventilation slots stamped in the sides seem none too deep; in a number of instances, the application of paint/finish material was sufficient to cause partial or total obstruction. There are a few larger holes in the bottom that would be held off a workbench surface by some surprisingly decent screw-mounted rubber feet (the adhesive-mount ones that readily fall off cost less…) but overall the potential for convective cooling of internal components on this device does not seem particularly strong.
Figure 10. Left, right, and bottom views showing occlusion of vent slots and screw-mounted rubber feet.
Before opening the lid, the custom of the “GND” terminal on a bench supply of this type indicating a connection to chassis/earth ground was checked and validated. It’s usually green in color on this side of the pond, but this unit does have a European style plug, came from Asia and perhaps this is just one of those international culture things… The very presence of a “GND” terminal on the front panel given the fact that the equipment lacks an earth-connecting supply plug is a curious matter; it’s there on higher-end bench supplies to give the user the option of having the supply output referenced to earth ground rather than isolated from it, but since this device’s supply plug does not provide such a connection, that’s not really an option here… Perhaps the user is expected to provide their own protective earth connection via this terminal?
Figure 11. Checking for continuity between front panel GND terminal and device chassis: a connection does indeed exist.
Output connector quality
The combination banana jack/binding post output terminals also customary to lab equipment of this type were experienced for the first time in the course of this process. While the banana function works well enough, the binding post functionality appears to be cost-reduced to the point of having little or no practical value. That the internal threads are cut directly into the plastic insulator rather than a metal insert is by itself sufficient to mark these connectors as bottom-quality, but the fact that both internal and external threads both have flats cut in them (making the thread cutting operation easier/faster/cheaper) causes the area of contact between them to be near-zero when these flats are at right angles, ensuring that the already-inadequate plastic internal threads will be promptly damaged past serviceability given any serious use. The problem is compounded by excessive clearance between the internal and external threads. From a product quality standpoint, abandoning the pretense of having a binding post function and implementing a banana-only output would have probably been the better choice.
Figure 12. Closer view of output terminals, with plastic portions removed. Note the flattened regions where the external thread is removed, most clearly visible in the right/rear-most connector in the image.
Figure 13. Supply outputs with plastic portions removed. Note partial removal of both internal and external threads. This along with the softness of the plastic and a very loose-cut thread makes it difficult to bind much of anything with these binding posts.
Improper grounding and other design/construction faults
Opening the cover, we find a line-frequency transformer labeled as having two outputs, nominally 12VAC and 18.5VAC. It appears that the 12V circuit is used as a housekeeping supply for powering the display and control side of things, while the supply output is derived from the 18.5V winding. The transformer primary is fed from the AC line, with the front-panel power switch in series with one side and the rear-panel fuse holder in series with the other. There’s a few things about this that give a person cause to question the validity of the CE marks on this product…
Figure 14. Internal view of the supply.
Figure 15. Dual-secondary transformer used in the supply.
First and foremost is the non-earthing europlug in conjunction with a conductive metallic chassis. In the event of an AC-to-chassis fault, the chassis would become electrified and thus a hazard to any persons, pets, or property that might come into contact with it while also having the bad luck of making simultaneous contact with earth. A simplified schematic showing the path of fault current in a properly-implemented system with a US-style split-phase distribution transformer is shown in figure 16. In the event of an AC-to chassis fault, the protective earth conductor provides a low-resistance path for the fault current. Placement of both switch and fuse in series with the “hot” conductor that is driven to line voltage with respect to earth allows both switch and fuse to interrupt power in the case of such a fault.
Figure 16. Simplified schematic showing path of fault current in a properly-grounded device with polarized input plug.
Figure 17. Simplified schematic showing path of fault current in the supply as delivered, one of two possible configurations made possible by the non-polarized input plug. In this version, fault current flowing through an external ground connection cannot be interrupted by the fuse.
Figure 18. Simplified schematic showing path of fault current in the supply as delivered, in the second of two possible configurations made possible by the non-polarized input plug. In this version, a line fault to the chassis presents a hazard regardless of whether the switch is in the on or off position.
Second is the single-line fusing in conjunction with that symmetric, reversible AC input plug. There’s only a 50/50 chance of the fuse being connected between the “hot” line and the appliance, so should an AC-to chassis fault occur, there’s only a 50% chance that the fuse would be in a position to provide any protective effect at all. In the case where the plug is connected so as to put the fuse in series with the neutral AC conductor, the fuse opening during a fault condition might actually cause an increased hazard, since its doing so would eliminate the only “safe” current return path available. Figure 17 shows the path of current flow should an AC to chassis fault occur in the supply in question. Note that the fuse isn’t even part of the current path in this scenario.
Third, the positioning of the power switch on the opposing AC input line from the fuse creates a scenario where the switch has only a 50% chance of being able to interrupt power to the device should an AC to chassis fault occur. Since the plug offers no protective earth connection, it’s possible for such a fault to result in a condition where the device’s power switch is in the OFF position, yet the device chassis remains energized to AC line potential. This scenario is represented in figure 18; whether the power switch is in the on or off position, a fault to chassis poses a hazard until such time as the fuse opens. Which is likely to seem like a very, very long time if one should happen to become the path for the current flow required to do so…
And as an aside, the business of leaving AC line conductors flopping about and entwined among low-voltage secondary-side cabling and casually resting against the secondary-side PCB within a product of this sort seems like an unnecessary sloppiness. It’s not as if the interior is so space-constrained as to leave no other option. Is a bit of care in cable dressing and component layout in the interest of keeping the high-voltage and low-voltage conductors separate too much to ask? It would seem so.
Figure 19. Interior of supply, arrows indicate points of contact between primary-side AC line conductors and secondary-side low-voltage conductors.
A similar issue applies to the case (collector) lead to the main pass transistor on the back. Though proper care was taken with insulators to electrically isolate the transistor leads from the device chassis, the wire lead soldered to the connector tab protrudes so far as to make its contact with the chassis (or lack thereof) visually indiscernible. Careful checking with a multimeter showed such contact not to exist as supplied, but it’s within a hair’s breadth (or less) of doing so as only slight pressure with a test probe brings a sub-ohm resistance reading between collector and chassis. Should that occur in operation, there’s a good possibility of short-circuiting the AC-DC stage upstream of the regulation stage in the supply. A seriously overheated transformer is a plausible result of that, with one of the line-to chassis faults discussed earlier being among the possible outcomes of such an event.
Figure 20. Wire through tab connecting to the collector of the main pass transistor nearly makes contact with the chassis. Though it does not do so as supplied, the slightest amount of pressure applied with a test probe causes it to do so.
Excessive transistor temperature
Having established that the product does not appear to meet basic safety standards for these reasons alone, we’ll move on to another troubling aspect of its design: it’s an old-fashioned linear-regulator. Put differently, the output voltage of the transformer was selected to satisfy the maximum output voltage requirement established by the design specification, and the difference between that and the user-requested output voltage is burned using that big 2N3055 transistor on the back.
“Burned” is a deliberate word choice here.
For reference, in terms of touching a metal object a temperature of 45°C is commonly considered the threshold of pain or discomfort, 60°C a limit for maximum temperature that can be tolerated briefly, and 80°C the point above which incidental contact is gonna leave a mark, due to irreversible damage occurring in less time than is required for a typical person to register and react to the sensation of pain.
Figure 21. The test setup used to gather the data in figures 22 and 23. Supply under test on the left, electronic load indicating applied voltage and current flow center, Fluke 289 with current calibration measuring transistor case temperature on the right.
The transistor case temperature was measured as a function of output voltage and current, using a thermocouple and a calibrated Fluke 289 multimeter, and the results are shown in figure 22. With the output voltage set at 7.2V, the discomfort threshold is reached almost from the start, at an output current between 100 and 200mA. Between 300 and 400mA the safe-touch temperature is exceeded, and at output currents above 700mA ( less than half the supply’s rated output) any accidental contact with the output transistor could be expected to result in a burn injury.
Figure 22. Plot of transistor case temperature vs. output current with output voltage set to 7.2 volts.
Things go from bad to worse in this respect as the output voltage is reduced. Repeating the same stepped-current process with the output set to 1.5V, the roughly 80°C burn threshold is crossed at an output current of between 400 and 500mA, or about a quarter of the supply’s rated output. At 60% of maximum rated output current, the transistor temperature measured 141°C, and with the maximum rated 2A output being drawn, the transistor case temperature was last seen passing upward through 164°C (327°F) with much vigor. Lack of a decent camera mount influenced the decision to not see where it stopped. Because if a person is going to push a piece of ill-designed equipment to a point of anticipated catastrophic failure while working alone in an access-restricted lab after hours on a Friday night, one really oughta get it on video…
Figure 23. Plot of transistor case temperature vs. output current with output voltage set to 1.5 volts.
Beyond the thermal issue, another drawback to the old-school line-frequency transformer-rectifier-capacitor power supply is that power is only actually drawn by the system in spurts at the peaks of the AC input waveform, at a relatively low frequency of 100 or 120 Hz, depending on locality. The filter capacitor is relied upon to fill the gaps in between. The low frequency means the gaps can be pretty long, and as output current and voltage increase, there’s an escalating likelihood that it’ll start running short of charge before the next refill, resulting in line-frequency noise showing up in the output. Addressing the issue in context of this design approach means a bigger capacitor, which costs money, or a higher transformer output voltage and higher thermal losses, which also cost money, though perhaps less directly.
This supply did not cost a lot of money…
With the output set to 12V and operating from a 60Hz AC source, 120Hz ripple effects began to appear in the output at between 1.5 and 1.6A of load current. At maximum output voltage, it appeared at around 700mA, with a maximum amplitude of about 2.7V at the maximum 2A output. An oscilloscope capture showing the output voltage under this latter condition is shown in figure 24. This supply cannot deliver it’s maximum-rated output current and voltage simultaneously, due in part to this effect. Moreover, in regions using a 50 Hz mains frequency (typical where europlugs are used) the time between “gulps” of input current drawn by the supply is substantially longer than in the case illustrated, resulting in output ripple performance that would be even poorer than that shown here.
Figure 24. Line frequency ripple with output set to 15V (maximum) and 2A load current being drawn. Top (yellow) trace is AC-coupled, bottom (green) trace is DC-coupled.
High-frequency output noise
Even without this effect though, the noise content of the output is fairly disappointing for a linear-regulated supply; obtaining a nice, quiet output is one of the last and best reasons to hang the albatross of linear regulation around a design’s neck. Within the output range where the line voltage ripple effect had yet to enter, high frequency noise with a peak-to-peak amplitude of 80 to 100mV was ever-present, appearing in what looks like envelope-modulated bursts of about 12.5 MHz at a variable repetition rate in the hundreds of kHz to low MHz. The fact that the leads between the control board and the pass transistor are about 8 inches (20cm) in length probably isn’t doing any favors for the supply in this regard. The scope captures in figures 25 and 26 show this effect at two different horizontal scales.
Figure 25. AC- and DC-coupled measurements (yellow & green, respectively) of the supply output, on the 1.5V setting with 200mA output current. Horizontal time scale = 200us/div.
Figure 26. AC- & DC-coupled measurements of the supply output on 1.5V setting, 200mA output current. Horizontal scale = 2us/div.
Included cabling and leads
The included test lead accessories offer opportunity for comment also. The labeling seems to suggest a focus on phone repair applications, and while the English-language content is not the most fluent, it’s hard to fault this recognizing that things going the other direction probably don’t come out in the most fluent Mandarin either…
Figure 27. Packaging for included cabling and leads.
Figure 28. Included adapter cabling.
The multicolor lead set is an interesting concoction; the box in the middle serves as a junction point for all the red and black connectors as well as the power contacts in the USB plug. Yellow is a half-supply reference generated using two 20K resistors connecting to the red and black groups, while the green connector is connected via 10K to the black. The intended purpose of these latter two is not immediately apparent; they might perhaps have been used to implement one of the termination/identification schemes found in the USB battery charging specification, but they do not appear to comply with any of the options enumerated therein, and in any event, they have no connection to the D+ and D- contacts in the micro USB connector. It’s plausible that their intended purpose is to serve as fake thermistors, when attempting to power devices designed to require connection of a battery thermal monitor signal for operation.
Speaking of USB, the concept of providing a cable set to connect a USB device such as a mobile phone to a variable-output power supply seems rather sketchy, based on the ease with which voltages grossly in excess of the limits prescribed by the USB standard can be inadvertently applied using such a cable. It seems like an effective means of accidentally causing need for repair…
Figure 28. Internal connections of adapter lead junction box.
The included set of test probe leads of the style commonly sold with multimeters is a rather strange inclusion; the use of such as a means of supplying power to a device under test does not seem to be particularly common practice. In any event, the shrouds on these probe leads that provide insulation/touch safety in metering applications prevent their use with the banana/binding post terminals on the supply being studied, unless the plastic insulator is removed. Which is actually quite likely to occur, for reasons described earlier…
Figure 29. Meter-style test probe leads included with supply.
Figure 30. Without removing the insulator form the supply’s output terminals, the probe lead set cannot reliably make contact.
Figure 31. Removing the binding post insulator allows use of the probe leads included with the supply.
Control board and principle components
Looking more closely at the control and display circuit board itself, the electrolytic capacitors present are labeled as being the Jwco brand, indicated as having a -40 to 105°C temperature range. Cursory research suggests this is a China-based capacitor brand, the quality of which might be expected to be commensurate with that of the rest of the device.
Figure 32. Control board view showing branding of electrolytic capacitors.
The device apparently in charge of regulating output voltage is labeled as an LM723CN in a 14-pin DIP package from ST Micro. It appears that this particular item under the ST brand has been obsolete for some 3-4 years at the time of writing, making its appearance in what’s presumably a new-production product a matter of curiosity. Did the manufacturer of this supply make a last-time buy? Are they using old stock? Is new old stock being made…? It’s hard to know, though it can be said that the device does appear to perform the function expected of it given its labeling.
Figure 33. IC responsible for regulation of output voltage.
The other notable IC on the control board is labeled as a SinoWealth SH79F166AF, for which there’s conveniently an English-language datasheet available online describing the device as an 8051-compatible microcontroller, having an internal 12 MHz clock source. It appears to be in charge of driving the LED displays and possibly some other housekeeping functions such as running the overcurrent warning beeper, and seems a likely culprit for the 12-ish MHz noise found in the supply’s output.
Figure 34. Microcontroller for control of display and housekeeping functions.
In spite of all the shortcomings present in the design and implementation of this piece of equipment, it can be said that it did perform its advertised function, as delivered. Sort of. It will make up to 15 volts at the output, or up to 2 amps, but it won’t do both at the same time and won’t do either cleanly. Without additional filtering and regulation, its output noise and ripple performance are unsuitable for any sort of precision electronics work. Its lack of proper thermal management makes the main pass transistor on the back a burn hazard with a high likelihood of failure, lack of proper grounding and a polarized input plug or double pole switching and fusing make it an unnecessary shock and fire hazard, the output connectors are poor quality and non-durable, and the included cabling makes it exceedingly easy to damage the sort of equipment that the device is marketed as being suitable for helping one repair.
There are indications that at least some level of care was invested in its realization, exemplified by the screw-on rubber feet, use of adhesive agents to bind the transformer mounting screws and internal connectors against loosening, and provision of trim potentiometers for adjustment of the fixed-voltage output settings. Omitting things like this or using cheaper options could reduce production costs further.
Figure 35. Fixative agent applied to transformer mounting screw to reduce the possibility of accidental loosening.
Figure 36. Trim potentiometers on control board, allowing adjustment of output voltage at the five fixed-output settings.
Going back to the original question: is it worth it? That’s a question that prospective buyers must answer for themselves. At a price point below $20 USD it could be said to fill a need for a disposable product, or offer access to basic functionality for folks with minimal cash flow. But a durable, well-designed piece of test equipment, it is certainly not.
Edit: Since writing the original draft, versions of this supply under the same or other brand names have been found documented elsewhere online incorporating a transformer labeled as having a 24V output. Such a change would reduce line voltage ripple effects significantly, at cost of greatly worsened transistor temperature issues.