Fullscreen HTML5Med Res (??MB)HTML4

Review by Phil Prosser DH30 MAX Li-ion Battery Welder It is a simple idea, and it should work well. How did it go so wrong? I was asked to review one of the Li-ion battery based welders that are cropping up on internet sites of late. Having just finished the Capacitor Discharge Spot Welder (March & April 2022; siliconchip.com.au/Series/379), my reaction was: why not? This might be a cost-effective alternative. So I proceeded enthusiastically. The prices of these welders seem to reflect the capacity of the battery used, which in practice consist of one or two cells paralleled inside the welder. That translates to costs broadly in the range of $50-100. As noted in the CD Welder article, one challenge battery-based welders face is getting enough energy into the weld quickly enough. So to be fair in this review, I chose a welder at the high end, the DH30 MAX, which claimed to have a 10.6Ah battery for about $100 plus shipping. After a relatively long wait (a bit over a month), it turned up, and I must say it both looked and felt the part. The case is 150 × 28 × 80mm and has substantial heft. It is an aluminium extrusion, and it is clearly packed full of batteries and stuff. Using it Plugging the welding cables in was a delight. I wish I knew where I could buy these connectors as they are great (shown below), and I would have been tempted to try fitting some to our CD Welder. I was initially bemused at how they were insulating that connector from the front panel (it is an unclad PCB). I will get into that more shortly. Also in the pack was a length of 0.12mm “nickel” strip and a USB charging cable. I had a prototype milliohm meter sitting on my bench (to be described in an upcoming issue), and I used it to quickly determine that the leads have a resistance of 1.5mW. This is consistent with 300mm-long 10 gauge (8mm2) leads. I noted that these leads are inconveniently short, even on the first weld. Checking the maths, though, they need to be short for the welder to work. The user interface is colourful but fiddly. It took me a little while to get it to do what I wanted. With the pack fully charged, I was off to the workshop and ran a couple of test welds on flat AA cells. Three welds in, and everything went pear-shaped. After the third weld, “magic smoke” started erupting from the welder case! With some concern about the device catching fire, I moved outside. A “minor” setback With a large coffee to calm my nerves, I reassured the wife that the house would clear of the acrid smoke. This was not going to plan! The DH30 MAX welder comes with the batteries installed into an aluminium enclosure. It has a rated welding output of 4.2V at 650A. Note that the charging port is USB Type-C. 60  Silicon Chip Australia's electronics magazine siliconchip.com.au ► The DHT30 MAX welder uses a 0.91inch OLED screen. I channelled the Serviceman and took to the case with screwdrivers and pliers. Having extricated the PCBs and lithium-ion cells from the box without shorting anything dangerous, the trail of smoke and cinders was something of a dead giveaway to the fault. Ignoring the minor fact that it blew up for the moment, I will provide some comments on the construction of the unit. The cells look the part for 10Ah, weigh enough and have very wide tinned connections to the ‘power board’. The actual part numbers have all been wiped off, but without running a capacity test, I assume they are up to the task. The controller PCB has a microcontroller with the top ground off, USB Type-A and Type-C connectors, the front panel control switch and OLED and two capacitive switches that use springs from the PCB to the display panel. The construction looks OK, if not excellent. This unit can double as a USB phone charger when not welding, which is handy. The ‘power board’ connects to the two Li-ion cells, the ‘control board’ via a header, the front panel and the welding lead connectors. It has four 4N03LR8 power Mosfets rated at 30V, 240A. They are quite appropriate for this job, though I would have been tempted to use more of them. The PCB layout has footprints for six smaller devices; I would rather see them all present, given the currents involved. There are a lot of vias on the power PCB, and for the most part, both sides of the board have large copper fills carrying the current with vias connecting between them on the top and bottom layers. The left side of the PCB has VBAT running up to the output connector. The right-hand side of the PCB carries GND, the battery negative terminal. This connects to all the Mosfet source pins, with the drain on the tabs connected to the “Out-” connector on the front panel. This switching method is the same concept used in my CD Welder, but on a baby scale. Take note of those vias running right down the right-hand side of the PCB; they are connected to GND. Repairs required I found that the PCB trace for VBAT had overheated and fused. The Mosfets were OK, as was the controller. All in all, it is a credible design except for the catastrophically narrow length of track on the left-hand side trying to carry 600A or so. After scraping the charred material and solder mask away, I soldered three lengths of copper braid (solder wick) over this section. Solder wick is nice and flat and can carry an awful lot of current. While doing this, I also noticed that the Mosfets were barely soldered to the The welder ‘blew’ up on both sides of the PCB, marked with red arrows. ► ► siliconchip.com.au Australia's electronics magazine August 2022  61 board. So I soldered down the floating pins while muttering many a salty oath along the lines of “you were so close to getting this right; what were you thinking?” With the sort of conviction that you can only have when something is about to go wrong, I started reassembling the unit. Halfway through the reassembly, I had a minor conniption. The only thing stopping a dead short across those beefy batteries was the solder resist on the PCB! Remember those vias (shown directly above)? They are actually inside the slots in the extrusion! My muttering turned to the question: “Are you for real? This will burn my house down!” To fix this, I took my trusty Dremel and ground back the VBAT fill up to the solder braid I had added, giving a gap of about 0.5mm between the case extrusion and the VBAT trace. I allowed the GND side to touch the case since that would no longer be harmful. some more measurements and got the following readings: • 2.5mW from the positive battery tab to the tip of the positive probe • 1.5mW from the drain of the Mosfets to the tip of the negative probe • the Mosfet specification is 0.79mW each with VGS = 4.5V, or about 0.2mW for four in parallel • the resistance from the negative battery tab to the Mosfet sources is about 1mW This gives a total of 5.5mW or so, resulting in 650A into a short circuit. This jibes with the spec on the box. For a 200ms pulse, this would be in the region of 400J. The problem is that little of that goes into the workpiece, as that is counted as 0W in this calculation. Basically, the workpiece needs to have a resistance of at least 5.5mW between the probe tips to get even half of that energy into it, and ideally considerably more for it to take the bulk of the energy. During tests, the leads got quite warm after half a dozen welds or so, as did the tips. As you can see in the photos, while I made reasonable welds, there was significant heating around the weld spot. Is it worth it? I guess the main question is: can you use it to make good welds? A decent weld to an AA cell is shown opposite. I did that in gear 13. I found that was the minimum to get a reliable weld that would not pull off easily. But that put a lot of heat into the battery. I welded three times in succession on one battery and literally melted the plastic insulation. So you need to be very careful using this welder! So, in summary, does it work? Yes, Back to the review With that done and everything buttoned back up, it was back to the task at hand: reviewing this comedy of errors. I took a more gingerly approach, starting with a 3W resistor and testing the welder in “gear 1” through “gear 20”. These equate to power levels, which are implemented by variable pulse widths of 26ms to 300ms (see Scope 1). Using levels up to about gear 8-10 gave unreliable welds with the strip they provided. I achieved decent welds in gears 11-13. The welds were OK, but because of the 200ms weld time, things got really hot making them. To check if this made sense, I did 62  Silicon Chip Scope 1: a scope grab of the output at the “gear 11” setting. I found this gave OK welds; the pulse width is 182ms. There are a couple of short pulses at the start, which are present on all settings. Australia's electronics magazine siliconchip.com.au this device will weld after significant repairs. Is it reliable? For welding, I would give a qualified answer. It can weld, but puts a lot of heat into your workpiece. So it depends on your application. I would be very cautious using it to weld anything very sensitive, like Li-ion cells. Will it remain reliable? This device is marginal. Other similar devices could be better. The Mosfets are OK on spec; however, the manufacturing had several serious flaws. Also, an increase of a milliohm or two in battery impedance would severely impact weld quality, and that could easily happen over time or with use. I would recommend the DH30 MAX only to technically confident people willing to check it thoroughly before use, and only if you intend to undertake small/non-professional jobs. There is a world of difference between this and my CD Welder design, in terms of weld repeatability and heat in the workpiece. Granted, there is a significant price difference. I’d like to comment on how I think that the design flaws in this unit came about. My day job is in engineering in Defence, where “engineering governance” is an integral part of life. It is tedious, but it is there for good reasons! This device has all the hallmarks of a design that was originally very good, relatively simple and fit for purpose in its original embodiment. Looking at the problems I found, my guess is: • The packaging was changed and, in this process, somebody neglected to check the VBAT and GND trace clearances to the case extrusion slots. I imagine this was done by a different person than the original designer, and they didn’t even think to check. • The PCB manufacturing was cost-optimised, perhaps too much so. Six devices were reduced to four. Looking at the solder mask, it actually extends under the four Mosfet source tabs! It is hard to see the original designer finding this acceptable, but I doubt they reviewed this change. • The PCB uses lightweight copper foil. It is much cheaper to use this than heavier (eg, 2oz) copper – again, a change that I suspect occurred in manufacturing without return to design and qualification. • Someone had reworked all the Mosfets, but only fixed soldering on two of the five pins. This procedure would never get past a review or the original designer; doing it right would take a second or two extra! Also, the need for rework is indicative of a deeper manufacturing problem. All these faults could be fixed at marginal or nil cost. They might even get away with the lightweight foil with a better PCB layout. As it stands, each of these faults could lead to catastrophic failure. Therefore, I recommend that you avoid purchasing this particular unit (and be wary of other similar units) unless you will personally open it up and check that it is safe to use before powering it up. I must also admit that I have a bit of concern that one of these could go up in smoke during transportation, depending on how it is handled, given the proximity of those ‘live’ vias to the SC metal case. Improved SMD Test Tweezers Complete Kit for $35 Includes everything pictured (now comes with tips!), except the lithium button cell. ● ● ● ● ● ● Resistance measurement: 10W to 1MW Capacitance measurements: ~10pF to 150μF Diode measurements: polarity & forward voltage, up to about 3V Compact OLED display readout with variable orientation Runs from a single lithium coin cell, ~five years of standby life Can measure components in-circuit under some circumstances siliconchip.com.au SC5934: $35 + postage siliconchip.com.au/Shop/20/5934 Australia's electronics magazine August 2022  63