RoadTest: Enroll to Review the Rohm Buck Converter Eval Kit BD9G500EFJ-EVK-001
Author: Gough Lui
Creation date:
Evaluation Type: Semiconductors
Did you receive all parts the manufacturer stated would be included in the package?: True
What other parts do you consider comparable to this product?: Texas Instruments TPS54560B-Q1, Vishay SiC46x MicroBUCK
What were the biggest problems encountered?: None in particular
Detailed Review:
ROHM Semiconductor BD9G500EFJ-EVK-001 Buck Converter Evaluation Kit RoadTest Review
By Gough Lui
November-December 2023
Power conversion is a common need in many products and buck converters are used whenever an efficient means is needed to convert a higher DC voltage to a lower DC voltage. The ROHM Semiconductor BD9G500EFJ-LA/BD9G500UEFJ-LA are integrated MOSFET buck converter controllers intended for industrial applications rated for a 7V to 76V input and up to 5A current. It utilises a current-mode control scheme which should provide excellent transient response capability.
Thanks to element14 and ROHM Semiconductor for selecting me as one of the RoadTesters for this buck converter evaluation kit.
The ROHM Semiconductor BD9G500(U)EFJ-LA 7V to 76V Input, 5A Integrated High-Side MOSFET, Single Buck DC/DC Converter is an integrated circuit intended for the realisation of such a converter with a current-mode control architecture targeting industrial applications. It has a suite of protections (OCP, UVLO, TSD, OVP). It uses a reference voltage of 1V with 1% accuracy, with an inbuilt MOSFET with on-resistance of 100mΩ. It covers the industrial temperature range of -40°C to +125°C operating. It uses a HTSOP-J8 leaded package that has an exposed paddle for heatsinking into a PCB.
The closest product match I could find appears to be the Texas Instruments TPS54560B-Q1 4.5V to 60V Input, 5A, Step Down DC-DC Converter with Eco-mode
. This product comes in a similar package (HSOP8) and claims to have 92mΩ high-side MOSFET, 146µA operating quiescent and 2µA shutdown current, operating frequency between 100kHz and 2.5MHz with synchronisation and a 0.8V 1% reference voltage. It is based on a current mode control architecture as well.
By comparison, the Texas Instruments chip doesn’t have the input voltage range at the upper end, but can operate down to lower voltages. The internal MOSFET is marginally better, but the shutdown current is worse. Operating quiescent on the Texas Instruments chip is superior as is the operating frequency range. The lower reference voltage leads to the ability to go lower in terms of output voltage.
Another close product match would be the Vishay SiC46x 4.5V to 60V Input 2/4/6/10A microBUCK DC/DC Converter. I previously reviewed the SiC461EVB (10A version) and found it to be a good performer. It uses a very specific PowerPAK MLP55-27L leadless package which some have claimed to be difficult to solder. In return, it is a synchronous buck converter, offering superior efficiencies peaking at 98%. It has quiescent current of 156µA not switching and 4µA at shutdown. Switching frequencies of 100kHz to 2MHz with adjustable soft-start and current limit. It is able to operate in power saving or continuous conduction mode, with a full suite of protections (OCP, OVP, SCP, UVP, OTP, UVLO). While the resistance of the internal MOSFET is not stated, the overall efficiency suggests it may be comparable or better.
Again, by comparison, the Vishay microBUCK does seem like a good choice with the higher-current capable models, wider switching frequency range and higher efficiency with synchronous buck converter architecture that has many modes. It is, however, not in as friendly of a package and uses much more current at shut-down. It also doesn’t handle as wide of a voltage range and is based on a voltage mode control architecture (which may have some disadvantages).
As a result, if one is looking for industrial applications where long product lifecycle support is important, anything above 60V input or a very low shutdown current, the ROHM BD9G500(U)EFJ-LA would be the ideal candidate.
A larger-than-expected parcel arrived for me from element14.
Surely an evaluation board would not need a box of this size! What was the reason?
The RoadTest involves more than just the evaluation board. There were three samples of the BD9G500UEFJ-LA integrated circuit as well. Of note, the evaluation kit is for the BD9G500EFJ-LA (without the “U”) which is the same chip just built on a different production line. The newer “U” part is the one which is recommended for current design.
I would say these samples are definitely over-packaged. That’s one big box for one little chip which is the size of a HTSOP-J8.
The chip itself has nice and large markings and is a “thin” SO-8 with an exposed paddle for heatsinking. A convenient and easy to use form factor, with the “leaded” nature of the package being good for reliability especially in case of shock and flexing of PCB where leadless packages tend to have less tolerance.
The evaluation board comes in a neat box, not dissimilar to the one I received for a previous ROHM Power Tree RoadTest.
A nice feature is that they label all sides, so it’s easy to identify what board is in the box no matter how it’s placed on a shelf.
Inside, the board is packed inside a black static dissipative bag with foam padding. The board’s switch seems to have poked through the bag at some point during shipping.
The board is provided with a single page leaflet. This particular board is “BOARD 39”.
The evaluation board itself is physically somewhat large compared to the size of the components. It is a green four-layer PCB with plenty of via stitching and 2oz copper on the outer two surfaces. This optimises the heatsinking capability of the PCB which, assuming the claim of 80% efficiency at 5V/5A is true, suggests the board needs to handle dissipating approximately 5W. Unusually for an evaluation board, this design seems to have an eastern influence with power input on the right and power output on the left. This is opposite of the convention I usually see in western-designed evaluation boards, although it is acknowledged that even the way books are read in different countries differs in much the same way (right-to-left versus left-to-right).
The underside houses no components although it is clear how the PCB pins for the sense lines route to individual copper traces that strategically connect close to the point of measurement for the most accurate Kelvin 4-wire measurement ability. Another nice touch is the cut-out of copper under the inductor which may serve to improve efficiency by avoiding eddy currents and reducing the noise induction into the ground plane.
The input and outputs are on these gold-plated PCB-pins which simplify making connections to heavy-gauge copper boards, however, perhaps the use of terminal blocks could also be appropriate and simpler for some types of evaluation. A three-position switch allows for toggling of the EN signal between Vcc, GND and an external input, which is a nice touch, although perhaps a bit over-the-top (e.g. a set of jumper links may have sufficed).
While the board is rated for 7-48V DC input and 5V DC output, it’s interesting to see that all capacitors on the board seem to be rated much higher than absolutely necessary – for example, this Nichicon BT 220uF 50V electrolytic is on an output that should be 5V. Likewise, a few ceramic capacitors on the board are rated for 100V DC, with one ceramic on the output rated at 25V, one ceramic in the compensation rated at 50V and the bootstrap capacitor rated at 10V. Capacitors come from Murata while resistors are from ROHM. While this can ease the prospect of board modification for other voltages, and is a nice gesture, it seems the output is still limited to 25V without dealing with C6. Custom designs may opt to use smaller and lesser-rated components depending on their operational need.
As noted earlier, this board has the “non-U” version of the chip. The markings seem somewhat less distinct. Some unpopulated positions can be seen corresponding to various options for compensation and feedback which are in the datasheet schematic but are not populated. A field of stitching vias near the chip is present for better heat distribution as the PCB acts as the heatsink.
The inductor comes from Würth Elektronik, while the rectifying Schottky is from ROHM. In all, the board is definitely a quality item, judging from the components used which are all from reputable manufacturers, and the clean manufacturing and assembly of the board.
A key parameter of any power conversion system is the efficiency as is the stability of the output. I decided to test this comprehensively by building up an “efficiency surface” plot, by stepping through iterations of current (at 5mA steps) and of input voltage (at 1V steps). Power was supplied and metered by a Rohde & Schwarz HMP4040 Programmable Power Supply, using one, two or three rails depending on the input voltage requirement, all with 4-wire connection for voltage drop compensation. Output load was provided with a B&K Precision Model 8600 DC Electronic Load.
To get to this point required attaching leads to the board which were soldered on. Then, crimp ring terminals were used to attach the output wires to the load, while the other wires were terminated in 4mm bullet sockets to be used with 4mm banana plugs (not the original design intention, but it definitely works).
Testing included ranges outside of the specification including currents above 5A and input voltages above 48V. Where measured values appeared to be sane, a value was recorded.
In this broad efficiency surface plot, we can see that the 48V input, 5A output does achieve a 79-80% efficiency which is very much spot-on with the graph provided in the datasheet. But unlike the datasheet which provides values at discrete voltages, this surface gives a bit more of an insight into how it changes as a function of load and input voltage. Of interest is that at higher voltages, the converter appeared to be unstable and generated quite a bit of voltage ripple. Also of interest is that it was possible to exceed 5A at quite a lot of voltages, with the unit tripping out in the 5.9A to 6.8A range, but the 5A output could not be sustained at 7V or 8V.
Zooming into the promised operational area, we can see its behaviour is more regular and smooth.
At low currents, the efficiency does not appear to be too bad either, with some light-load mode likely extending efficiency down into the lower currents.
The output instability is not visible in the efficiency plots but is quite clear in the output voltage plots, with a “curve” extending in the 4-6A range where the voltage rapidly takes a dive (although given contours are at 1mV intervals, this is only ~10mV or so). Voltage stability, otherwise, is excellent.
Zooming into the promised operational zone, except for the 7V in/5A out corner (top left) of the graph, everything else looks excellent.
Adding my Rohde & Schwarz MXO4 Oscilloscope into the mix with its 10:1 passive probes and a spring ground, the B&K Model 8600 generated the fastest 10%-90% step-change transients it could to challenge the ROHM buck converter.
Such current slew rates will bring many converters to their knees, but the ROHM seems to be handling it well. The peak-to-peak deviation is 666.2mV which may seem a lot, but it really isn’t given the slew rates.
Zooming in, we can see a little more detail of each “ripple” and the savage 10-20µs rise-time of the load.
Going off-load, we can see much the same detail, although the load fall time seems slightly shorter in the 8-10µs range. What is amazing is that the recovery time seems to be in the 0.5ms ball-park which is quite a bit better than some lab power supplies can manage.
Performance at 36V is similarly excellent, registering 714mV peak-to-peak deviation.
Performance at 24V is consistent, registering 731mV peak-to-peak deviation.
Performance at 12V is marginally worse at 813mV peak-to-peak deviation, but seeing that this type of transient can collapse certain converters down to 0V, this is a very impressive showing.
Looking closer, it seems both on-load and off-load performance are slightly less tight at 12V input than they were at 48V input, but this is excellent transient regulation performance for a switching converter given recovery occurs well within 1ms and that’s something that not even many laboratory-grade power supplies can boast about.
Another thing of concern relates to power quality. This is usually termed ripple and noise which encompasses periodic and random variations in voltage. In switching converters, this is somewhat load and voltage dependent as the operating points change.
With an input of 12V, I observed a no-load ripple of 9.26mV peak-to-peak which is excellent. The sawtooth appearance of the waveform suggests a low-load power saving mode is active.
At 24V, this increased to an average of 14.64mV peak-to-peak with a taller spike at each switching event. This is still a very low-level of ripple.
At 36V, this further increased to 24.96mV peak-to-peak with the spike growing taller. This is still a very good result.
At the rated 48V, this reached 35.81mV peak-to-peak with the spike again causing most of the deviation. The ripple voltage is quite small still – below 1%.
At 60V, the ripple was 47.37mV peak-to-peak, although this is outside the expected operating condition.
At 72V, the ripple reached 60.96mV peak-to-peak. While the IC appears to have an energy saving mode, at all input voltages tested, the resulting ripple is very acceptable.
Under full-load with 12V input, the ripple averaged 25.22mV peak-to-peak with a nicely sinusoidal pattern. A very good result.
At 24V, this increased to 32.06mV peak-to-peak with a slightly lop-sided sinusoidal waveform appearing.
The ripple voltage remained similar at 36V, reaching 33.79mV peak-to-peak and now having a more flattened top.
At the rated 48V, the ripple maintained a similar value, reaching 34.55mV peak-to-peak with a continued flattening of the top.
Exceeding the ratings at 60V, the converter appears to have poor behaviour resulting in a ripple with multiple frequency components reaching 327.27mV peak-to-peak.
At 72V, the pattern is less discernible but the ripple remains elevated at 334.09mV peak-to-peak. It seems that operation above the rated 48V of the kit may have unexpected consequences with regards to ripple and noise, but within the 48V rating, the performance is absolutely superb.
A key problem that a switching converter might face is an over-current condition, all the way up to a short circuit. This particular IC offers OCP and SCP which I tested.
Under a commanded load of 6.5A, the OCP can be seen to cut-out the output and automatically restart at an interval of about 21.5ms with a soft-start ramp-up until the output voltage is reached and then OCP kicks in and shuts down the output.
I did simulate a dead short using the “SHORT” command on the electronic load and saw a nearly identical result – periodic automatic restarts into the short, showing only a small voltage increase (due to the less-than-perfect short) between each hiccup. It would seem the IC is very careful about inductor current limiting and did not suffer any harm from this test.
While the IC does offer overtemperature protection, this was not invoked on their evaluation board at any time. The IC also offers overvoltage protection which features the ability to clamp a limited overvoltage condition, however, I could not think of a safe configuration that would not risk damaging the board or a power supply, so I did not attempt to provoke the OVP.
Quiescent current was measured with a Keithley 2450 SourceMeter SMU by doing a dual I-V sweep from 7V to 76V and back. With the enable signal turned off, the SMU struggled to maintain stability because the board (practically) wasn’t drawing anything that could be reliably measured using ordinary PVC banana cables.
The visual “average” of the positive and negative oscillations roughly cancel out to about 0.1µA or less, so I’d have to say that this converter is pretty much consuming nearly nothing when disabled.
When enabled but with no load connected, the output current is dependent on the input voltage. Most values ranged from 6mA down to about 1.5mA which is quite acceptable.
I assessed the thermal situation using a Topdon TC004 Handheld Thermal Camera to give an indication of operating temperature under full load for an extended period in 25°C room temperature ambient conditions.
It appears that the hot-spot temperature was getting close to 104°C which is pretty hot. The heat seems to be distributed mostly evenly through the board without any obvious impediments.
A closer look suggests the inductor was reporting the hot-spot temperature, but the Schottky diode and chip were not far behind. Nevertheless, operation up to and slightly exceeding rated load was possible and considering the extended thermal envelope of 125°C suggests there is some margin to accommodate temperatures inside a case.
However, I would also have to say that the amount of power dissipation could also be nearly halved if the controller utilised synchronous rectification with an actively driven MOSFET instead of a Schottky diode, although this does increase complexity.
In truth, upon receiving the parcel and seeing that there were some samples in there, I decided that I had to get going on a design right away, as PCB manufacturing, ordering all necessary supporting components and shipping can easily take a month. This section looks at the result of this design effort, along with the available documentation.
ROHM’s website product page offers a lot of details about the product and links to all downstream documentation:
The datasheet gives very detailed information about the capabilities of the product including graphs of performance and design considerations. Formulae are provided for calculation of parameters and component values.
However, one doesn’t have to calculate those values by hand as there is a calculation sheet available for download.
There are also web-browser-based simulations available for frequency response and thermals, provided you register for a ROHM portal/myROHM account.
It does seem one of the simulations is “under maintenance”.
An encrypted PSpice model is also available for simulation if the online simulation is insufficient.
Alongside all of this are many more generic documents and papers about design tips and tricks. The evaluation kit has its own guide as well which does double-up on some of the information.
Overall, the documentation is numerous and of good quality. Admittedly, I was a little late in discovering the simulation and calculation sheet, so I built my own version of the calculation sheet just for sanity checks.
Given the opportunity to build a buck converter for something was not something I had expected to do for this RoadTest, so I had to think a bit about what I wanted to achieve. In the end, I thought about multi-port USB chargers for older devices – as multi-port USB-A chargers become less common and USB-C PD becomes the norm, could I design a little something to convert a USB-C PD port into a multi-port USB-A charging solution, while also making a PCB that could do double-duty with non-USB-C PD input as just a bare converter?
Because of the time constraint, I took a (mostly) slapdash approach to designing my own board and deliberately decided to break away from the component values and parts that were already used in the EVK to see just how far I could push it. I also opted to keep things cheap and simple by opting only to have a 2-layer, 1-oz board but supplementing that with a small heatsink for the diode. Nevertheless, given the PCB minimum order of five, I decided to buy enough parts to make five of this unproven design just in case I broke things (and I did).
This exercise cost me a good month and about AU$250 in parts and postage. I mainly did it as an educational experiment – it would not have been worth the money otherwise. However, it should be noted that the ROHM BD9G500UEFJ-LA is not an inexpensive IC – in itself, it costs about US$5.36 right now but I suppose this is to be expected given its industrial nature.
Disclaimer: This part should not be seen as a “this is what to do to design your own switching converter”. Instead, it should be viewed as a “this is NOT what to do” as I decided to try changing things and cutting corners “because I can” just to see what happens and without the full benefit of all the tools available for simulation or modelling. I just wanted to see just how far I could push things and still have something that might (half) work. Whatever performance deficiencies you see here are likely due to my poor design choices and cost-cutting efforts. These do not represent the pinnacle of what is possible with the ROHM Semiconductor BD9G500UEFJ-LA integrated circuit.
I decided to make something quickly in KiCad that used the BD9G500UEFJ-LA as the main converter with a 22µH Bourns SRP1265CC shielded inductor. I should have used 33µH to match, but I made a mistake while ordering, so I decided to live with it. I would use an STPS8H100D TO-220-2 Schottky diode so that heat could be developed away from the board on its own little Fischer Elektronik FK 220 SA 220 heatsink, thus making my 1oz two-layer board a little less precarious (thermally-speaking). I employed an Injoinic IP2721 that I stole from a USB-C PD trigger board to do the PD negotiation, a GCT USB4110 USB-C connector, a nice Vishay SIRA80DP-T1-RE3 PowerPAK MOSFET for power pass control and Diodes Inc AP2182 USB power switches to handle over-current protection, Texas Instruments TVS0500DRVR flat clamps to provide over-voltage protection and a smattering of ferrite beads everywhere to hopefully avoid any EMI wrath.
I spent some time laying things out to try and keep as much copper as possible, throw in as many stitching vias as I thought my PCB manufacturer would allow and rounding off any sharp corners where I could. I put in some versatility to configure the D+/D- lines for picky devices and saved a bit of board space by using stacked USB-A connectors instead of individual ones. I put in a notch for cables to pass through if the units were to be “stacked” using the holes and additional terminal block connections for options in interfacing to the board. The size was just below the 10cm x 10cm “promotional” size limit. Even though each port could “theoretically” consume 1.5A under BCS (for a total of 6A) or even up to 2.4A (under some non-compliant legacy charging standards), I was hoping that the devices would back-off as the voltages fell and the USB switches would keep the most heavy current offenders at bay with the converter’s OCP as a last resort.
I thought this was all going to work when I put in the order. But unfortunately, there were a few interesting findings. Because of this, this design will not be shared as it’s a little broken and dangerous.
When the boards arrived, I hastily got out the hot air gun and started construction.
One of the boards, I decided to build “bare-bones” and just have the converter populated and terminal blocks with none of the USB-C PD or USB-A parts populated. This way, the board could be “like” the evaluation board that ROHM provided, just of my own design and configured for a little over 5.2V (mainly to compensate for cable voltage drop and give a better USB charging experience).
I populated the boards in stages, and this is a more complete construction of the board above. Two of the boards I decided to populate with the full (or almost full) USB hardware as well.
While each of the boards will do 25W, the USB-C PD input might be 65W or even 90W. So I decided to make them stackable by passing through the input voltage on the terminal blocks like so:
With the right nuts, bolts or standoffs, one could theoretically stack these converters together to build a brick. Under no load, the USB-C PD is working fine on the first board and both boards are idling happily.
But notice how only one stacked USB connector is installed and not two. Testing with two ports occupied with a power bank charging load, it seems to work just fine as it doesn’t trigger the OCP (yet), although it’s on the edge of doing so.
But when I populated all four USB-A ports and loaded them all up, strange things happened. I found the converter’s OCP would trip and hiccup, after which, the USB switches would be burned and get very toasty. Permanent damage ensued but thankfully the USB devices remained seemingly unharmed.
Getting out the oscilloscope on the non-USB-populated board under DC electronic load showed that on OCP, it seems that my board was putting out ~16V high spikes when the voltage set-point was 5V!
Those USB port switches couldn’t handle that and ended up sacrificing themselves. The downstream TVS clamps probably did their job somewhat to help with the situation and so the downstream USB devices didn’t get harmed. That was a close shave.
But I don’t know exactly why this happened. Perhaps somehow my choice of only ceramic capacitors on the secondary makes for a bad filter, or perhaps there’s an interesting situation that happens when the chip is facing not only OCP, but possibly also OTP at the same time. The undersized inductor probably is part of the reason too. Nevertheless, it is because of this that I don’t recommend you build my design – even if I matched the values, I don’t expect it to work quite so well.
Looking at my board, the chip does get hot, but not quite as hot as the ROHM EVK. This is perhaps validating the strategy of splitting the losses of the Schottky on its own heatsink to be a good choice (although at the cost of volume and expense – but it was cheaper for me to go with a heatsink than to go 4-layer and 2oz).
But no matter what I did, it seemed my design just couldn’t deliver the 5A – at best, it was closer to 4A. This is probably because of the undersized inductor and hitting the inductor current limit. I tried to game the margins a little by trying different Rt (frequency control) resistances to see how that might help.
I saw about 3.1A to 3.75A load capability across 7-30V range with the minimum Rt value (maximum frequency). Instability above 2.6A for the 20V operating point appears in the graph. At the 20V input and near maximum load operating point, I’d be expecting about 83% efficiency. The instability manifests more on the voltage stability graph, but noting the lower-spec resistors and not-truly-four-wire connectivity on this board, the results are pretty good.
With an Rt of 20kΩ, I was seeing more stable behaviour up to 3-4A. At the expected operating point, I’m seeing about 83.5% efficiency but the increased stability to 3.5A is a welcome side-effect.
At an Rt of 33kΩ, things seemed to get better. Perhaps the losses of higher frequency switching are being balanced out and thus a higher operating point efficiency of about 84.5% is seen with stability up to 4A.
If 33kΩ was better, then 51kΩ should be better-still? In fact, it seems this is the case. Efficiency is now closer to 85.2% and stability seems good to 4A as well.
Going to the other extreme for frequency, it seems things get worse – the converter now only handles up to about 3.6A at higher voltages. Efficiency has not changed much either although the voltage is stable. As a result of these experiments, I think 51kΩ seems to be a good value for this board to push its capability.
Ripple and noise under no load measured just 14.22mV peak-to-peak.
Under a 3.7A load, it didn’t break a sweat, measuring similarly at 13.74mV peak-to-peak.
Transient response looks a bit less controlled – a peak-to-peak deviation of 577.06mV.
Some ringing when going on-load is visible in the response but it’s still pretty quick. I didn’t adjust the compensation values, which may be the reason.
The situation going off-load is smoother, but the total load is quite a bit less because of the limitations of the design.
While I managed to break most of the rules to just forge ahead and build a buck converter in a slap-dash manner with sub-optimal thermals, undersized inductor, perhaps poorly tuned output filtering, the unit still worked for the most part and had similar levels of efficiency, thermal, ripple and noise and transient response performance. The one odd thing that occurred was the OCP seemed to cause the output to exceed its voltage set-point, frying my downstream USB port controllers and putting my devices at risk. Another is that the 5A rating was not possible – likely due to my (inadvertent) undersizing of the inductor and output capacitors although I am not confident that replacing the inductor would be enough. This is why this design is still considered a potentially dangerous work-in-progress.
After extensive testing of the ROHM Semiconductor BD9G500EFJ-EVK-001 Buck Converter Evaluation Kit and the BD9G500UEFJ-LA integrated circuit, my overall impression is that this is an excellent current-mode controller that features outstanding voltage ripple, transient response performance and class-leading input voltage range for an integrated-MOSFET solution, at least, within the 48V operational envelope claimed for the kit. It comes in a relatively easy-to-solder HTSOP-J8 leaded package which should withstand industrial environments better than leadless packages. It promises a long market lifetime (although this is not explicitly defined), its pricing seems reasonable considering its capabilities and a reasonable amount of design documentation is available. It also has an effective set of protections which avoid damage to the device, including OCP, OTP, OVP and UVLO. The evaluation kit itself is high quality and well-built, although laid out with power flowing from right to left.
Unfortunately, being a non-synchronous current-mode-control buck-converter, efficiency is perhaps not its strong point with a little over half the wasted power occurring in the Schottky freewheel diode. If the best efficiency is a priority, a synchronous buck converter may be a preferential option. While the frequency of the converter can be adjusted with a timing resistor, there does not appear to be any further EMI-mitigating controls (e.g. spread-spectrum mode) or acoustic noise mitigation modes (e.g. ultrasonic mode). There appears to be a light-load mode behaviour, but this is not configurable. The stability of the converter also seemed a bit problematic when exceeding 48V - perhaps there are some design changes that need to be made to the EVK to operate well at higher input voltages (despite the capacitors seemingly being within their working voltage, perhaps they had lost too much capacitance because of DC bias effects).
My deficient attempts at a cheap design ultimately backfired somewhat – the converter did not achieve full current performance likely due to a mix of an undersized inductor and potential thermal constraints. However, it did show that even my rather slap-dash approach still resulted in a workable solution that still exhibited excellent voltage ripple and transient response performance.
Top Comments
Nice review! I really like your "surface graphs", they look fantastic and they also contain a lot of informations
-
strb
-
Cancel
-
Up
+1
Down
-
-
Reply
-
More
-
Cancel
Comment-
strb
-
Cancel
-
Up
+1
Down
-
-
Reply
-
More
-
Cancel
Children