DC-DC converters with galvanic (optical) insulation

I have to write this DC-DC stuff before I forget, and in case someone else might be tempted to buy the Aviogard from MGL. Updated Mar 10 2018.

MGL Aviogard

First the Aviogard from MGL Avionics. It consists of a Murata PS UQQ-12/8-Q12 at it's core (at least the unit I purchased). This is a €140 or less industrial off the shelves device that MGL repackage and sell for €600+.



The rest of the Aviogard consists of the casing and terminals, two ferrites, two TVS diodes (transorbs), two capacitors and two SMD stuff that I'm not sure of what is (too small to see, resistors I think ?) It's actually a nice package though, and the Aviogard will offer protection from over voltage surge, and will filter noise, even without the UQQ DC-DC converter. It's a nice terminal block with protection and filtering all by itself, but without the UQQ, that package is worth maybe €20-30? in components. What's wrong with it?

  • The UQQ is not installed according to the specs of the UQQ. The voltage is set 10% too high, and the "sense" terminals are not connected to the outputs. The UQQ specifications specifically warns about these two things things. Ut uses terms as "MUST NOT" in fat capital letters, although the manual does not state what exactly will go wrong.
  • It cannot possibly charge a lead acid battery as per Aviogard documentation. I tried several times, and if the battery isn't already fully charged (meaning fresh from the charger in float charge), it will enter into "hiccup mode". Even with a fully charged battery it would enter into hiccup mode 2 out of 3 times or something. This "hiccup mode" is a self preservation mechanism of the UQQ, and as I understand, a standard safety mechanism for regulated converters. If the load isn't manually removed, it will stay in hiccup mode indefinitely. Upping the voltage to 10% above specs will only make it worse.
According to the manual, the UQQ tries to hold the voltage constant. Since U=RI, the only way to do that when R is the load, is to increase (or decrease) I. This is called "voltage mode" in the manual. When the current reaches a certain point (probably around 8 A? which is the nominal current, the manual doesn't specify exactly), it enters into "power mode". In power mode, the power is held constant by reducing V and increasing I. This continues until the current reaches 9.5 A and is defined by a drop in output voltage by 2%. If the current is increased even more, it enters into hiccup mode.

So far so good, but what happens when a battery is put on the load side? A Lead acid battery should be charged according to the figure below, and where the max current should be between 10-30% of the Ah of the battery.


With an empty battery, the current must be limited, or it will just surge unrestricted into the battery. Only when entering the "topping charge" in the figure, when the battery starts to become saturated, will it prevent amperes from surging. In float charge, very little current is entering, at about 2.25 V per cell (13.5-13.6 V for a 12 V battery). A battery charger clamps the current by reducing the voltage.

A battery that is 50% charged has approximately 12 V open circuit voltage, E=12 V. The internal resistance, Ri, is about 0.05 ohm. If it's charged by a terminal voltage, U=13.8 V (output from the Aviogard), then the current will be: I = (U-E)/Ri = (13.8 - 12)/0.05 = 36 A. 36 A = hiccup mode for sure.

A 100% charged battery is 12.7 V open circuit voltage: I = (13.8-12.7)/0.05 = 22 A, at least initially, but will be reduced fast, because it is 100% saturated. A battery is also a capacitor. But, if the UQQ is equally fast as the capacitance effect in the battery, which it obviously has to be, it will enter into hiccup mode nonetheless. 22 A is still far into hiccup land.

A problem number two is the voltage drop of 2% that determines if the converter shall enter into hiccup mode. A 2% drop from 13.8 V is about 13.5 V. A fully charged battery is only 12.7 V. The second it is connected it will cause a dip in voltage below 13.5 V, and this can cause hiccup mode all by itself. It then shuts down while the battery is connected, tries to fire up, but the battery is then already fallen back to 12.7 V, if it even momentarily should have increased a bit. 

When it goes into hiccup mode with the battery connected. It heats up enormously. Way too hot to touch. I measured the voltage over the battery, and it went slowly but surely down. The last time I tried, it went all the way down to 7 V. Then I switched it off. This can only mean there is some short circuit inside the UQQ. The open circuit voltage was about 12-12.5 (don't remember exactly, could also be fully charged). Using the battery equations, this means a current of approximately 100 A was continuously drawn from the battery. Of course this causes heating, LOTS of heating. Glad I tested it on the table before mounting it in an airplane. That's what toasted the PCB, and eventually toasted the UQQ. I have fixed the PCB now, and I have measured the output of the UQQ. It's stuck in short circuit - fried to death :-)

The UQQ (MGL Aviogard) is simply not a battery charger, end of story. It does not work. Not in theory, not in practice. MGL have stopped responding to mails when I told them what happened to my unit. It is not all together impossible that I got a broken device from the start, but highly unlikely. Without the backup battery it worked exactly as it should.


The Mascot 8660

This is a very different beast. It's purpose made for vehicles, and is used all over the word in ships, trains, dumpers and whatnot. The 8660 is just one model of a ton of other models with different voltages and capacities, different specific functions. The 8660 comes in two versions, 12-12 and 24-12.


What's special with this one, is that is does not supply constant voltage. Instead the voltage drops linearly when the current increases.


For instance, if the input voltage is roughly 13.8 V, the output voltage at zero load is around 14.5 V. Then at max load (10 A) the voltage has dropped to 11 V. Now let's hook up a fully charged battery with E = 12.7 V at the output lines. It will immediately supply about 5.5 A at 12 V, but since the battery is saturated, the current will drop to just above zero, while the voltage increases to 14.5-ish. What about the 50% charged battery of 12.0 V? It will immediately drop to 7.5 A, and end up at approximately 6.5 A, 12.3 V. There it will charge the battery, and the current goes down while the voltage increases.

In essence: It will work. Probably not the best treating of the backup battery, but it will cause no harm as far as I can see. It is 1/4 of the price of an Aviogard.

I received an answer from Mascot. Using a DC-DC converter to charge a battery is no good idea according to them. There is always the possibility of the battery being completely flat or close to it (ruined essentially). This will overload the DC-DC converter. He said it would be better to use the Mascot 2544 charger as DC-DC converter, than to use a DC-DC converter as a charger. I got a small battery however, only 6 Ah, and with the Mascot 8660 things do indeed work, tested on my table.


Traco Power TEP 150 WI

Yet another very different beast. Looks to be a blend of the Mascot 8660 and the UQQ.


Haven't really looked too much into it, but it looks like an industrial version of some of their other power supply stuff, thus more of an accurate constant voltage unit rather than an "always supply something useful" unit like the Mascot. Maybe it is not meant for vehicles at all for all I know, the specs mentions only telecom and stuff like that. But it does have a very interesting characteristic, because it is designed to have a battery hooked up to it on the load side. It is just as much a battery charger as a DC-DC converter. Essentially exactly what I want, except the max current is very high (14 A).


The 12V output version will keep constant voltage (12V or whatever chosen) up to about 14 A, the  CV region on the picture. From that point the current is held constant by decreasing the voltage, exactly like a battery charger. Then, and not until the voltage drops below 4.3 V, will enter into hiccup mode. Anyway, this will with 100% certainty work exactly as I want. The backup battery can be less than empty, and it will work. It is essentially what the MGL Aviogard should be, but isn't, not even close. And this is 1/3 of the price of an MGL Aviogard.

The problem with this is the large current. 14 A is OK for a 47 to 140 Ah battery, not a 6 Ah battery. But still, I believe it will work just fine even though the charging is not done in a way that is good for the battery. Batteries can always be changed easily, and does not cost much.


Alfatronix DDi 12-12 072   

This device is also specialized for vehicles and boats, trains. It's one of many other devices doing some variation around the theme of DC-DC conversion, similar to all the Mascot devices.


Seems like nice devices, and the 12-12 V versions come in 3, 6, 9 and 14 A versions. The 072 is the 6 A version. Their official site in the UK does not have any real specs however. Mainly only sales mumbo jumbo, nothing that explains how it works, and more precisely what it can be used for. More importantly; what it cannot be used for. Maybe it's just like the UQQ (Aviogard) - useless and essentially broken when hooking up a battery on the load side? It's very odd not to supply real and detailed specs on industrial devices like this IMO.


Others

I had no idea, but there are literally thousands upon thousands of different isolated DC-DC converters out there. All shapes and sizes, voltages and currents. Must be one of the most used slightly complex electrical component out there. I guess there are several others that could be used, but I have no knowledge of them.


Conclusion

The perfect system would be a nice light weight and compact little thing that supplies clean (isolated) 13.2 -ish V at araound 10A max. Hooked up to this it should be a 4-6 Ah LiFePO4 or similar light weight battery. This should be run by a purpose built electronics that enters backup battery mode seamlessly whenever the main supply goes under 9-10 V, is short circuit'ed, or there is some fault in the DC-DC converter itself. In normal operation the converter should just maintain the backup battery in optimal condition as well as "clean" the power from the main system. This device should have a CAN interface to the EFIS for monitoring. I have found no such device, but I'm sure they do exist somewhere. The closest I could find is the IBBS from TCW. It would probably do the job, but looks very fiddly and odd with Dsub connectors instead of terminals, and backup only for a few devices, not all in one go. It's not clear if it has galvanic isolation either, no mention of it. Surely infinitely better than the Aviogard though.

Ayway, I have found something that works. The Mascot 8660 DC-DC converter and the Mascot 2544 DC-DC charger. I have tried it on my kitchen table, and it even though I can hook up an empty 6 Ah battery directly to the 8660 with no adverse effects (no heating, no puffing and clicking noises), I will also use the charger to maintain the backup battery in optimal condition. I couldn't find any radio noise either, and both the 8660 and the charger are protected from over voltage by TVS diodes. They are meant to installed in vehicles (boats, trains, cars, trucks, tractors etc) and are robust industrial devices. A bit bulky and not optimal connectors, but it will do the job 100%. So I am happy. Installation is next.

No comments: