Discussion found cheap but good working alternatives of custom regulators

Discussion in 'Wii' started by xsping, Apr 3, 2019.

  1. DeoNaught .

    Joined:
    Dec 10, 2016
    Messages:
    556
    Likes Received:
    586
    Location:
    Constant Fear
    Portables:
    sin^2(x) + cos^2(x) +e^(i*pi)
    Have you ever heard of a time bomb?
    Just because it works now, does not mean it will work later.

    You didn't even attach any load to them. There was someone actually who made a portable with them, he "finished" the portable with the regulators, but never shows it turning on.... because it wouldn't.

    but, you seem to be dead set on proving everyone here wrong, hey, it's not like they are studying in those fields and know what they are talking about.... oh wait

    even if it's cheaper, it is still not worth it, because it is a HUGE safety hazard, those things warm up really fast, and can melt plastic, which can melt wire insulation, which can short wires together, starting a fire, and injuring someone, or killing someone at worse, if you are that reckless. You aren't just endangering yourself, but others around you. Buying from china, they probably have like five versions, and you miraculously got one that kind of works, great, but for how long?

    Please throw them away, or return them, you can pay either $1 regulators, and $1000+ in fire repairs, or $35 regulators, and you'll be happy and safe knowing you decided to not cut corners, and you have a good working portable.
     
  2. Trimesh .

    Joined:
    Jul 9, 2019
    Messages:
    22
    Likes Received:
    32
    I'll be honest, I was fully expecting it to exhibit horrible behavior and/or blow up like all those low-cost LM2596 based switchers available from various sources in China. But it didn't, and actually seemed very well behaved.

    And yes, I also did short circuit and step load testing (easy, since that programmable load has a function that does exactly that) - the main reason I was interested in the low voltage load test is that it's a very effective way of blowing up switchers because the duty cycle is low and the dI/dt is high - but it worked fine.

    Sure, I wouldn't want to use it in anger without more characterization (especially since I can't find a datasheet for the chip), but at this point I'm certainly not willing to claim it's unconditionally unusable either.

    Guess what I do for a living?
     
    BocuD, Noah, jefflongo and 3 others like this.
  3. Shank Certified Wiitard Staff Member . .

    Joined:
    Jan 31, 2016
    Messages:
    1,065
    Likes Received:
    1,990
    Location:
    Texas
    Portables:
    3
    @Trimesh You certainly seem like you know what you are talking about. I think the problem here is your information conflicts with a lot of old knowledge we have cemented in our brain from prior experience. In the early days of Wii Portablizing, we tested a lot of voltage regulators, and all the cheap Chinese ones did not work well at all. On BitBuilt we try to give helpful advice, and shut down advice that could misguide uneducated users, which explains the backlash to your recommendation. Our current stance is that generic Chinese regulators are problematic, unreliable, and can destroy consoles, and we recommend using tested ones like the PTH08080, because that is what our experience has been. We have had many horror stories on the forums of melted chips from poor regulators.

    Personally, I have never used those regulators specifically, but this is the first positive reception I've heard of them. It's possible some of our old information is incorrect and outdated, but unfortunately the burden of proof to prove otherwise lies on you. You seem to have the knowledge and equipment to do in-depth tests. I appreciate the knowledge you have provided so far, and would personally would love to see more information and thorough tests on these boys for applications in portables. It would certainly be exciting to have a new off-the-shelf low cost regulator option we can recommend for the first time in... who knows how long.

    Thank you for taking the time to do the tests so far. Looking forward to seeing more.
     
    Stitches, Noah, Gman and 4 others like this.
  4. jefflongo Broke BitBuilt Staff Member . .

    Joined:
    Dec 15, 2016
    Messages:
    766
    Likes Received:
    1,029
    Location:
    Ventura, CA
    Portables:
    1
    You can cut the condescending nonsense off. Trimesh clearly has experience and is providing useful information. He also said he was testing them under a 2A load and monitoring temps. Having a post count title on the forums doesn't give you the privilege to shut down any conversation that disagrees with the status quo.

    That said, I believe it's worth spending a few extra bucks on something that's reputable and reliable. It's great that you're performing these types of tests on other power management solutions. Many of us have been exploring using more efficient regulators that come in a smaller package, albeit this isn't aimed towards decreasing cost. There certainly is a niche for people that are looking for something cheap, however in the long run, building a portable is expensive so the efforts in making your portable more robust, efficient, and easier to maintain makes it worth it to head in the direction that we're currently going IMO.
     
    StonedEdge, Aurelio, cheese and 5 others like this.
  5. Stitches 2 and a Half Dollarydoos Staff Member . .

    Joined:
    Feb 5, 2017
    Messages:
    1,238
    Likes Received:
    1,130
    Location:
    Banana Bender Land, Australia
    Portables:
    6
    Well, that's new. Maybe some chinesium regs are less shit than they used to be, but I won't trust them until we have results from several batches across multiple vendors.

    Edit: Since you have the equipment, would you be able to run a test of those cheapo regs inside a small plastic container with a small hole in one side? This would give us a better indication of thermal performance under actual portable conditions (tucked away in a cramped corner with poor ventilation and covered in tape and wires). The open air results seem possibly promising, but shit gets fukkin' baked in portables (see the recent WaveRider and PiiWii Pocket meltdowns).
     
    Last edited: Jul 25, 2019
    Aurelio and cheese like this.
  6. Trimesh .

    Joined:
    Jul 9, 2019
    Messages:
    22
    Likes Received:
    32
    OK, I'll do some more testing - probably at the weekend. I'll try to get some comparison data. To add some context, I've dug out a few other regulators, too:
    regulators.jpg

    The top one is one of those terrible LM2596 regulators - it's mostly included for comedic purposes, since I fully expect it to grenade on at anything like full load. The next (from the left) is our mystery regulator using an unknown IC from some unknown vendor, the next is a PTH08080 (bought from Avnet, so presumably real) this uses a TI TPS54350 and should be properly designed since it's from the same vendor as the chip - the last is another little switcher based on the Monolithic Power MP1584 - this is a high frequency (1.5MHz) chip, which may explain the small size of the inductor. Or it might just be underspecified...
     
    Notjohnny, Aurelio, Stitches and 3 others like this.
  7. Stitches 2 and a Half Dollarydoos Staff Member . .

    Joined:
    Feb 5, 2017
    Messages:
    1,238
    Likes Received:
    1,130
    Location:
    Banana Bender Land, Australia
    Portables:
    6
    marvelous.png
     
    Pacal and Aurelio like this.
  8. xsping .

    Joined:
    Dec 9, 2018
    Messages:
    75
    Likes Received:
    20
    the left one is what i used in my first portable,it works well until now,but the quality of these regs vary from the chip quality they used. Now when powermii pcb becomes opensource, i turn to make it , and i found the inducer quality is very important to avoid overheating
     
    GingerOfOz likes this.
  9. BocuD .

    Joined:
    Jul 21, 2016
    Messages:
    443
    Likes Received:
    478
    Location:
    The Netherlands
    Portables:
    2
    This all looks good to me, especially the efficiency's of those earlier tests sound neat - but what about voltage ripple? From the limited knowledge and experience that I do have in the "field" (lol) I'd expect the cheap regs to underperform here. I see you have a scope you can test with, so I'd be interested in the results when it comes to that.
     
  10. Trimesh .

    Joined:
    Jul 9, 2019
    Messages:
    22
    Likes Received:
    32
    OK, the results are in, and it took a lot less time than I expected, because two of them were so bad no extensive testing was done.

    The chosen testing conditions were 1.0V output and 2.0A load, since this is an operating point that will make the regulators work really hard.

    DUT 1: LM2596 based regulator.

    Well, I expected it to suck, and it didn't disappoint me - I had to test this at 1.2V because that's the lowest specified output voltage. The output voltage was highly unstable and had about 200mV p-p noise on it. DC efficiency was 49%. I terminated the test early when the IC hit 86.4c after about 5 minutes of operation. OK, it didn't actually blow up, but I have a feeling it would have if left longer.

    DUT 2: MP1584 regulator

    I expected this to be a decent performer, but it turned out to be nearly as bad as the first one - at least it could be trimmed down to 1.0V, but on anything more than about 500mA load it was constantly shutting down and restarting. When I trimmed the voltage up to the point where it worked normally (about 2.4V) then the DC efficiency was 61% - I didn't bother testing the thermals since I was pretty sure they were going to be bad given the low efficiency. (*1)

    DUT 3: Mystery Taobao regulator

    This one at least gets credit for actually working - this time I tested it at 1.0V, and the performance was somewhat degraded from the results I got earlier at a higher output voltage. This time it hit 66.3c (on the chip) and the measured DC efficiency was 78%. Noise is about 60mV p-p - neither input or output filter caps made much difference to this, so it's presumably inherent regulator noise.

    DUT 4: PTH08080

    As expected, this was the best performing device - this was connected with a 100uF input filter cap as per the datasheet recommendations, but no output filter cap. One interesting result was that this was the only part where the magnetics got hotter than the chip did - the chip hit 42.2c and the magnetics 45.6c. Measured DC efficiency was 93% - noise was about 25 mV p-p.

    Conclusions:
    The obvious favorite here is the PTH - the Taobao regulator did work, but it is significantly less efficient and runs noticeably hotter than the PTH did and is hence not possible to recommend in this sort of battery powered application.

    *1: This board had parts fitted with values that were quite different from the ones in the datasheet. It's possible it would work better if they were changed, but that rather negates the purpose of buying a pre-assembled regulator module.
     
    Stitches, Shank, GingerOfOz and 5 others like this.
  11. Gman RTFDS Staff Member . . . . .

    Joined:
    Feb 25, 2016
    Messages:
    881
    Likes Received:
    1,941
    Not surprised here, but very interesting nonetheless. Thank you for sharing the results.
     
    Stitches likes this.
  12. Trimesh .

    Joined:
    Jul 9, 2019
    Messages:
    22
    Likes Received:
    32
    Yeah, I'm not terribly surprised either, to be honest - I always expected the PTH to work better simply because it's got a more efficient architecture.
     
  13. Stitches 2 and a Half Dollarydoos Staff Member . .

    Joined:
    Feb 5, 2017
    Messages:
    1,238
    Likes Received:
    1,130
    Location:
    Banana Bender Land, Australia
    Portables:
    6
    That's more testing than I expected, cheers for the effort and data my dude! RIP LM2596, not sure how they still exist.
     
    Aurelio likes this.
  14. VanBrunning .

    Joined:
    Jun 22, 2019
    Messages:
    2
    Likes Received:
    0
    How much proof do you need I'm wondering. You can drive those cheepos pretty hard. If you wire 2 of them in parallel how I mentioned before, you have effectively 0 chance of over driving them. Aside from that, their efficiency improves when they run cooler.
    Really the only risk is cooling the regs if you slightly undervolt, because undervolting the CPU or GPU either stable or simply not enough to power it. Over volting is where you get into dangerous territory.
    Usually the LCD driver board has some tolerance level for over voltage built into the board, the same can be said for GC controllers if you are using an original controller, basically anything on the 3.3v rail can be undervolted.
    You can't kill a wii by not providing enough voltage for it to turn on.
    A 001 model wii (the 45W one that shouldn't be used) can and has run off of 6 of these. That was before I had little thermisters to test for heat, and they stayed warm, but no more than a 30min old cup of coffee.
    If your handheld went off like a hand grenade, my guess is you had 0 airflow, only one of those, and didn't check it before hooking it up. Measure twice cut once.
    I have no issue with "Because it just works" but some people are in here for the DIY experience and don't mind tinkering to save $20 or more. Not to mention they should have a multi-meter and a work bench power supply/load.
     
  15. BocuD .

    Joined:
    Jul 21, 2016
    Messages:
    443
    Likes Received:
    478
    Location:
    The Netherlands
    Portables:
    2
    Oh boy..
    First, the thing you mentioned about putting two of them in parallel, is kinda stupid; Using a resistor to "isolate" the regs from each other? Smart I guess, but it really doesn't work like that. You are also forgetting the fact that you will cause a very significant voltage drop this way and you are essentially just wasting power. That voltage drop means there is wasted power, directly related to the amount of current drawn by the sink (in this case the wii). Say we calculate the losses for the 1V line. We can roughly do that by considering how much it draws. That would be about 1.5A. (so about 1.5W, again these are all rough guesstimates) We could then represent the components attached to the 1V line on the wii as a 2/3Ω resistor. The entire circuit would look something like this:
    upload_2019-7-30_0-28-45.png
    See the problem? Those resistors there each will result in about a 0,06W loss. and a 0.08V drop.. Thats not insignificant. Considering the output voltage will drop anyways given the load and their rated current, (which is usually inaccurate with these Chinese regs) you would need to set them at around a 15% or so higher voltage then the voltage you are trying to get to the wii in the end. Anyways, as @Trimesh found, the highest efficiency you can expect from the best reg he tested is 78%.. Not that bad, but catastrophic in a battery powered device. given the fact that there's an 8% loss from these resistors by putting two regs in parallel, and there's a 78% efficiency overall, the total efficiency drops quite dramatically to just 72% (best case scenario, this gets worse the more current you draw).
    Then, there's the part about "undervolting the CPU/GPU isn't bad". This may be true to some extent, but not entirely. I don't think I can give you a good opinion on this however as I lack the knowledge or experience to properly say anything about this. What I do know however, is that having this much noise on the output is bad. Quite bad. This noise on the power line isn't just going to make the system unstable though, it radiates everywhere. You will notice that on the audio and video output unless you shield everything very well.
    Third..
    wat
    The reason we don't use these 6 layer Wiis is because they use a lot of power. Inefficient regs combined with a system thats power hungry in the first place isn't going to get you a lot of battery life.. Other then that, a lot of power draw means a lot of heat output. 80% efficiency on regs with a system drawing 12W (not unreasonable for a 6 layer wii) means you waste 3W of power with those regs, and thats not even considering the resistor thing you had going on. 3W of heat is a lot of heat to just passively radiate. For reference, its more then what the GPU on a 4 layer Wii draws. I find it hard to believe these regs not getting this hot given the facts lol. But even then:
    You are saving 20$ with this? My b, the PTH regs cost 6$ a piece if you buy them from the right place - thats 18$ to power a full system. Even if these regs of yours are 2$, and you bought 6, well, you saved 6$.. And got a barely working inefficient setup. The reasons we stick with for example the PTH regs is not jxust because they work great and we were too lazy to find alternatives, its because they work well, they are well priced, they are efficient, and they are reliable. I think thats where the difference is most clear. I wouldn't trust these regs in a portable I would make, especially considering everything else that can go wrong in a system as complex as it. If I have a PTH in there, and something breaks, I can expect it was my fault because, and I would be able to fix it. I won't have to fully disassemble everything to then figure out the reg blew on me. We take pride in our work, and we would like it to keep working. Using reliable parts makes that a lot easier. It really isn't like these PTHs are the only way. Look at PowerMii, the Wii PMS, or even some of my own things. They are all using different regs. Some examples, like the Wii PMS, are even more economical for most people (still) given the features it has and the size of it. I agree with the part about the DIY experience, but there's a difference between going for the "diy experience" and plain ignoring research other people have done already and doing something else. While this may work for you right now, if you are serious about making a portable and making it any good, please get better regs.. Then I'm sure we'd all love to see it.
    Sorry for the rant btw, I just got triggered and started typing..
     

    Attached Files:

    Pacal, StonedEdge, jefflongo and 2 others like this.
  16. StonedEdge .

    Joined:
    Nov 16, 2018
    Messages:
    35
    Likes Received:
    89
    Location:
    Japan, Tokyo
    Portables:
    One G-Wii
    upload_2019-7-30_10-1-33.jpeg
     
    VanBrunning and Stitches like this.

Share This Page

Loading...