Why do my DC converters keep dying?

[harrisonpatm]

I have a 24s LFP battery on my e-motorcycle. I use a 12vdc converter to power accessories (headlights, signal, horn, phone charger). I started with a model similar to this; not exact model, just this style. I got one with a 10 or 15A rated output, cant remember which, I just knew that I only needed about 5-8 amps of DC for headlights and whatnot, so I picked one overrated for my use. After about 6 months of use, it had suddenly failed, not sure the cause. The failure symptom that I found with my multimeter was that the wires for the 72v input were shorted, read as 0 ohms resistance. Luckily I had fused the input with a 2A fast blow glass fuse, so the battery itself didn't short, and even if I didn't fuse the input, my BMS is rated against short circuit (not something I want to test).

My guess at the time was because that style of converter is just a board in an aluminum case, not weather or vibration proof. No big deal, I replaced it with one of these, the 10amp version (exactly model i used this time). Went with that style because it's all potted, so I figured I'd be good against weather and wear and tear this time. Cut to yesterday, and it failed again, except this time I was paying attention and I think I found the cause:

I had just finished charging my battery to 100%, recently. I pulled out of the driveway, went down a few houses, and let my regen stop me at the stopsign. I felt regen for a half second, then cut out. When I checked the log, sure enough, the BMS recorded an overvoltage protection. That first burst of regen on a mostly full cell had briefly pushed it into overvoltage territory, so it had disabled charging to the battery. Since the battery wasn't allowing charging, and the controller was still on, it attempted to "charge" the only thing directly connected to the battery, which was the converter. So the overcurrent fried the 72v input, and on this second failure, I checked the same symptom, which is that the input wires read as a dead short.

I've tweaked my BMS settings a bit to hopefully prevent this from happening again, and I could also lower my charging voltage by a bit as well, giving my battery more headroom for that first regen current. However, what i'd like to know is, why did this happen twice, even though the 72v input is fused? This is specifically why I used a fast-blow 2A glass fuse, rather than something like one of those inline auto fuses, because I was trying to protect my converter from sudden overcurrent. Except now it seems like overcurrent is getting past the fuse, to the DC converter, frying it, causing an input short, which then blows the 2A fuse. How can I change this so that the fuse blows before the converter?

Or am I wrong about the reason why the converter has failed twice?

 

[calab]

The generic/cheap/ebay-style dcdc's can be optimistic in wattage ratings. Like rc lipo c-ratings at hk, like sellers wattage ratings for hub motors who will state any wattage # to get a sale, like light lumen ratings, non-genuine (flashlight battery and lesser quality) lithium-ion mAhr ratings (like 10,000mAh 18650's), like distance range ratings for ebikes.

It might be a different story if you bought a good, half decent dcdc like from Grintech, or a unit from Meanwell from a reputable seller like Mouser.com because then you'd know for sure its genuine.

 

[harrisonpatm]

The generic/cheap/ebay-style dcdc's can be optimistic in wattage ratings. Like rc lipo c-ratings at hk, like sellers wattage ratings for hub motors who will state any wattage # to get a sale, like light lumen ratings, non-genuine (flashlight battery and lesser quality) lithium-ion mAhr ratings (like 10,000mAh 18650's), like distance range ratings for ebikes.

It might be a different story if you bought a good, half decent dcdc like from Grintech, or a unit from Meanwell from a reputable seller like Mouser.com because then you'd know for sure its genuine.

Agreed, but that's not the question (I am well under the wattage ratings).

 

[E-HP]

Except now it seems like overcurrent is getting past the fuse, to the DC converter, frying it, causing an input short, which then blows the 2A fuse. How can I change this so that the fuse blows before the converter?

Or am I wrong about the reason why the converter has failed twice?

Unless the laws of physics have changed, the converter won't fail due to overcurrent, but on overvoltage; which is why the fuse isn't going to blow. I'd go with one with a higher input voltage rating. I have this one that can take 120VDC in

https://www.amazon.com/Pro-Chaser-regulator-Scooters-Bicycles/dp/B07B6B6HQZ

 

[harrisonpatm]

Unless the laws of physics have changed, the converter won't fail due to overcurrent, but on overvoltage; which is why the fuse isn't going to blow. I'd go with one with a higher input voltage rating. I have this one that can take 120VDC in

https://www.amazon.com/Pro-Chaser-regulator-Scooters-Bicycles/dp/B07B6B6HQZ

Hey, thank you, that actually makes sense. If the controller tried to use regen to "charge" the converter when the battery isn't available, then it could cause a voltage spike because the battery isn't there as a load to bring it back down.

Thanks, it's obvious when you mention it, but it didn't occur to me previously. Guess that's what a forum is for!

 

[amberwolf]

That's what I would expect to be the problem--the voltage spike issue from having no battery connection destroys controllers too.

There doesn't even have to be a regen mode available--anything that causes the motor to generate a voltage higher than the battery voltage would go to can do this, like spinning faster than normal fastest speed coasting down a hill, etc.

One of the reasons we have so many problems with ebike/etc stuff is that almost all of it is being run *at* it's max capability. Good engineering would leave much much higher margins, but that would also cost quite a bit more, so almost nothing you can get for these purposes does that. Instead, you have to go pick parts that are for much higher current or voltage or power applications than you are ever going to use them for, and pretend that capability doesn't exist, so you don't upgrade yourself into a failure.


Same thing for batteries--designing the pack to have much more capability than you will ever need*** and thus not using them to their full capacity, such as having a charger set to only charge up to say, 90% of full voltage, means there is room for regen or other input to them even at full charge so they cannot go overvoltage, and can thus remain a load for sinking voltage spikes, etc.

Then if the BMS is set to stop at 100% so that it protects against literal overcharge, while not stopping normal sinking events, will help protect against damage to the other stuff like controllers and DC-DCs, etc.


*** having say, 25-50% more capability than required, both for capacity and current/etc, means that you still have enough for dealing with unexpected detours, headwinds, cold weather, other adverse conditions, and pack aging over time.

 

[amberwolf]

For fusing, remember they don't blow *at* their rating. That is the current they are essentially guaranteed to *not* blow at or below.

Above that current, then they will blow eventually, but each fuse (or circuit breaker) will have a chart from the manufacturer to show how long it takes to blow (or trip) at any specific current.

If there's no such chart available, I recommend not using that fuse, as you can't predict what it will actually do under a specific set of conditions unless you test some to destruction to be sure they'll actually work.


So, for instance, if you need a fuse to blow immediately *at 2A*, it will also eventually blow at lower currents, and probably won't handle even 1A for very long.

There are "fast blow" fuses, delay fuses, etc.; picking a fuse for a specific application can take some time and research until you have learned enough about different ones to have a good idea of where to start. (no, I'm not an expert at it yet either)

 

[Eastwood]

yeah, I’ve done this once before, pressed my throttle regen when the battery was fully charged coming out of my driveway. Thankfully nothing blew but made the controller a little warm.

 

[harrisonpatm]

So, is there a way to safeguard against it? For either the controller or converter? Since fuses won't do it...

 

[Comrade]

So, is there a way to safeguard against it?

Have you determined that overvoltage is actual cause? I don't see evidence of that in this thread. Suspicions. Yes. But not evidence. Why not ride around for a little bit with a voltmeter that can detects and store voltage peaks?

 

[harrisonpatm]

Have you determined that overvoltage is actual cause? I don't see evidence of that in this thread. Suspicions. Yes. But not evidence. Why not ride around for a little bit with a voltmeter that can detects and store voltage peaks?

You're right, we don't know for sure. Unfortunately, I don't have the desire to repeat the conditions and potentially blow another converter, costing me 15 bucks, for the third time. I think we have a good idea that overvoltage causes it, because as @E-HP pointed out, my fuse is protecting against overcurrent, so eliminating that as a cause... Plus we have @amberwolf noting that overvoltage is the C.O.D. for many controllers and converters in the past, and I trust his judgement

 

[Chalo]

My guess at the time was because that style of converter is just a board in an aluminum case, not weather or vibration proof. No big deal, I replaced it with one of these, the 10amp version (exactly model i used this time). Went with that style because it's all potted, so I figured I'd be good against weather and wear and tear this time. Cut to yesterday, and it failed again,

Those fully potted, cast enclosure Chinese DC-DC converters are notoriously unreliable at loads more than half of their nominal rating. If I wanted to run one at a steady 5 amps, I'd get a 20A model. For 8A, I'd get a 30A model.

Regen is a can of worms that makes a reliable bike unreliable. This is yet another example of that principle at work. Check your axle nuts to be sure you're not loosening them.

 

[Eastwood]

So, is there a way to safeguard against it? For either the controller or converter? Since fuses won't do it...

My remedy is to just go full throttle for a short period when the battery is fully charged to bring down the voltage so I don’t have to worry about accidentally pressing the regen.

But yeah, if your controller is adjustable, you should be able to set the High voltage cut off, that should prevent from blowing another converter.

 

[harrisonpatm]

Those fully potted, cast enclosure Chinese DC-DC converters are notoriously unreliable at loads more than half of their nominal rating. If I wanted to run one at a steady 5 amps, I'd get a 20A model. For 8A, I'd get a 30A model.

5-8 amps is my headlights, high and low, flashing signal, brake light, and blaring horn, all simultaneous. I've actually measured that during the build phase last year. Hence me getting a 10A option. Constant draw is more like 2-3 amps, and only at night when I have the headlights on for my 5 minute commute at 4am every day. During day rides, no headlights, 1-2 amp output. I also run 2 similar units at 50-80% load in my house's solar system, 24/7, no failures for the past year so far. So I still am betting on overvoltage from regen being the issue, not overcurrent. Especially since the last failure happened immediately following a recorded overvoltage event. Occam's razor and all.

Regen is a can of worms that makes a reliable bike unreliable.

Can't disagree with that at all. I like it a lot in principle, in practice it's tricky at best.

Check your axle nuts to be sure you're not loosening them.

No issues on my setup. Zero wiggling for the past year.

 

[harrisonpatm]

But yeah, if your controller is adjustable, you should be able to set the High voltage cut off, that should prevent from blowing another converter.

This is a good plan, I can't remember the exact voltage I set mine at, I'll have to check and it should be another safeguard.

 

[amberwolf]

So, is there a way to safeguard against it? For either the controller or converter? Since fuses won't do it...

Assuming it is overvoltage (which makes sense given the stated failure conditions, without a failure analysis of the electronics involved), then there are brute-force things like TVS diodes (transient voltage suppression) and other types of fast-reacting voltage limiting that could be installed on things the same way they are on surge protector power strips, etc., and often are installed inside various wall-powered power electronics' input stages.

Zener diodes, if any are available in the right voltage, could also be used, but these require a dropping resistor which will then limit current into the device and waste power.


The best way to safeguard against it is to use that "good engineering" principle of significant safety margin, and use things that are capable of much more than you need them to do. If there is a chance that a 100v spike in voltage could occur in a system, use a 200v part and it is virtually certain to not be vulnerable to any spikes the system could generate under expected conditions. If there is a chance that a 100A current spike could occur, use a 200A part. Etc. The margins don't *have* to be that high, but if you want a system that is "proof" against events caused by these things, that's one "easy" (but expensive) way to do it.

Using known-quality well-engineered parts and modules in the system, that are known to be made of genuine parts with significant margin already built into them, is another way, but also not usually cheap.

It's too-often so much cheaper to use the cheap stuff and just replace it when it fails, than to spend enough money (and research, testing, and design time) to build something unbreakable.

 

[amberwolf]

This is a good plan, I can't remember the exact voltage I set mine at, I'll have to check and it should be another safeguard.

Remember this will only prevent intentional regen-generated voltage events (assuming the controller is fast enough to shutdown regen in an event).

It won't stop anything not generated on-purpose by the controller itself.

 

[Comrade]

I don't have the desire to repeat the conditions and potentially blow another converter, costing me 15 bucks, for the third time.

Disconnect the converter for testing. That's easy enough. Put the multimeter on and find a hill to test regen on. You don't know what's going on. Maybe you need a 200V tolerant converter.

 

[E-HP]

Disconnect the converter for testing. That's easy enough. Put the multimeter on and find a hill to test regen on. You don't know what's going on. Maybe you need a 200V tolerant converter.

The BMS reported overvoltage when it occurred.