Regulator Rectifier (longish and techie)

[chanik]

As many of you know, I am critical of the electrical system Rotax/FD have devised. So for my final tweek to the Rotax electrical system, I decided to upgrade the crappy Ducati regulator/rectifier (RR). This site does a pretty good assessment of it but with just 75% efficiency and terrible regulation accuracy, I wanted a better one. http://contrails.fre...c_ducati_en.php

So I picked up an ESR510 from www.Electrosport.com Now, as an EE, I am constantly frustrated by the high noise to signal ratio WRT technical information available from vendors. The ESR salesperson couldn’t provide me with anything but did assure me that it was a very modern design compared to the Duc/Energia unit commonly used on the ROTAX 912, since the ESR510 used MOSFETs, etc. and was designed to be much more efficient. But no graphs or tech data. MOSFET is a nice, high-tech sounding term but in practice it means about as much as ‘computer designed’ or ‘scrubbing bubbles’. Fortunately, I have all the equipment to emulate a charging system. I used a 5KW varactor to emulate the stator output for various RPMs and an HP6060a 300W DC load to serve as a battery in various states of discharge along with an actual Shorai LFX18L1 battery to test compatibility with a real load. http://www.johnea.ne...1026_125941.jpg In order to calculate power input, I monitor both voltage and current and use waveform math to find dynamic power which I can then average, again using the oscilloscope to do the math. http://www.johnea.ne...r/scopemath.jpg Power out is more simply just average current times the constant battery voltage. From there I just ratio the two powers to yield efficiency.

 

And so? At 268Watts AC in, I get 204W DC out. 317W in gave 17.3Amps mean at 13.7VDC (237W) which is 75%, just as poor as with the Duc. (Step down switcher regulators can readily be made ~93% efficient so this is pretty disappointing). You can see in the thermal image where the lost power is going. http://www.johnea.ne...r/rrthermal.jpg The RR got to a toasty 80C after 12 minutes. One can expect about 1.2C/W thermal coefficient from die to case with big TO-220 packages bolted to the case heatsink spread across 4 devices which means the silicon is at least 105C. Silicon devices never last run at these elevated temperatures which is reflected in all the complaints about blown RRs (both the Duc and ESR if you peruse the various BMW and Ducati motorcycle forums).

The only advantage the ESR really has is that is regulates voltage out very well: 14.2V regardless of temperature and current loading. However that problem is relatively easy to fix on the Duc with the diode trick to the control pin described in my earlier thread. Plus I would have to re-make the electrical connector to use the ESR510. But the 510 also has one huge flaw. It switches flowing AC current. SCRs cannot do this and that’s a good thing. Slamming off a 20A current is bad for three reasons. It generates large voltage transients meaning more electrical noise and more stress on especially the thin varnish insulation in the stator windings. There are indeed some complaints on the motorcycle forums about shorted stators after switching to aftermarket RRs. The 3^rd problem is the large vibrational stress imposed on the stator coils, analogous to water hammer. You can hear what it does to my 50lb varactor once the RR starts chopping current as the voltage hits 14.2V. Startled me the first time. http://www.youtube.c...bed/FAwFNo4lVbU

In conclusion, swapping out the Duc for the Electrosport is not a good idea. On a motorcycle, it is just a bad idea. On a ROTAX in an airplane, it is a dangerous idea.

Postmortem: There are a few other RRs out there that might work, like the Euromotosport Enduralast and the Shindengen FH020AA. But they also do not have any tech data I could find and don’t have a lamp output. The RR51 is available from a different Electrosport in England which seems to be the precursor of the ESR510 and is rumored to be the design which was copied in China to make the ESR version. I’m skeptical about all of these. There are two that might be viable candidates: this one looks like it is made in small volumes but intriguingly was adopted by Remos http://www.schicke-e...nic.de/dgr3.htm And this one which boasts an impressive 47A rating: http://www.silent-he...m/Duc_R_Duc.htm and is drop in compatible with the FD wiring harness. I’m a bit jaded at this point but if someone wants to get one of these, I’d be willing to test it out for them. Otherwise I suppose I’ll stick with the old Duc.

One up note, the Shorai LFX18 battery did great through all of this. Kept less than 5deg C rise even at 20Amp sustained charge. The LiFe is a bit quirky though. It is basically fully charged with anything over 13.7V input, rapidly rising in voltage after that. It merely needs to be kept under 14.8V which is an easy target for any RR. On the discharge side, it stays very flat from full, 100%SOC which is ~13.3V loaded, down to 12.8V then drops off a cliff below that. I only find a 0.4V bounce back at a 5Amp load so to stay above their recommended 12.8V cutoff, one should not let the loaded voltage drop below 12.4V. There is only about 10% capacity left below 12.8V anyway. If overdrawn, perhaps because the RR died on a long trip, then the ideal treatment is to use the balancer/charger Shorai supplies. Else, use a trickle charger to slowly recover the cells. It is during deep discharge that the cells tend to severely unbalance. After the battery climbs back to 13V, you can charge normally, but it would be worth testing their capacity in that instance. This is really something everyone should do if they have any doubts about their batteries or charging voltages: Fully charged, pull the alternator and time the voltage to 12.6V with a LiFeP04 battery or 11V with a PC310 (both result in ~10% SOC remaining). Or do it on the ground if you have an AC charger handy. Both batteries are about 80Whr at 5Amp draw so you should have 1+hr of time until the battery is depleted. My CTSW with Dynons draws about 70W without lights. When I got my plane it only ran about 15minutes to full discharge of the PC310 mainly due to under-voltage charging. For once, I agree with Rotax's choice

[Jim Meade]

Interesting and useful. Thank you. I can't say I'm educated enough to understand it all, but I'm confident I got the gist.

[chanik]

The gist is that we're all still stuck with the inefficient Ducati and we should all be very leery about the supposed better aftermarket options out there. SIlent Hektik seems it might be a nice microprocessor based unit but it is also very small volume production with no technical data so I don't trust it either

[Jim]

Kurt, as an electrical engineer currently designing switching power converters, I found your data very interesting. I first got involved with this sort of stuff when I was in high school and I designed a poor battery eliminator for my Honda S90 motorcycle.

 

Here's a link to schematics and analysis of many different types of motorcycle rectifiers and ignition modules. Hope you find it interesting

 

http://www.motelek.n...chema/spannung/

 

BTW, I'm wondering about your varactor.. USA usage for varactor is a diode whose capacitance changes with applied voltage. I don't think that's what you're talking about.

 

EDIT: Watched your video. We call it a Variac.

[FredG]

Kurt, thanks for sharing your rigorous investigations.

[chanik]

Kurt, as an electrical engineer currently designing switching power converters, I found your data very interesting. I first got involved with this sort of stuff when I was in high school and I designed a poor battery eliminator for my Honda S90 motorcycle.

 

Here's a link to schematics and analysis of many different types of motorcycle rectifiers and ignition modules. Hope you find it interesting

 

http://www.motelek.n...chema/spannung/

 

BTW, I'm wondering about your varactor.. USA usage for varactor is a diode whose capacitance changes with applied voltage. I don't think that's what you're talking about.

 

 

That's my Lexdysia showing through. It should be Variac. And if you look closely at the oscope, you will see that the RR is not switching OFF the current during conduction. Still worse: the ERS510 is regulating by shunting current. The voltage collapses and the AC current shoots up, saturating my current probe. I had to review my data to even notice this since it is really a bad design. By shunting all the current, you end up with saturated current flowing through the stator coils. All the available power becomes heat (300W at 6K RPM) and ends up in the stator windings and ALL THE TIME even if the electrical system is drawing little.

[Jim]

If I were to design a regulator from scratch, I'd start with an active bridge rectifier made from power mosfets, a filter capacitor, and a switching regulator. And I'd probably duplicate it for reliability. Done right, it should be about 90% efficient overall and never get much more than warm to touch. And it would have a datalink to a display or my Dynon.

[chanik]

Sure, but it is not so easy. You need to hold off 200Vpeak AC voltage, Open circuit from the coils and get the timing just right with a synchronous MOSFET bridge so as to be low-loss but crucially not get shoot-though. Two of the FETs must be high-side driven which is also tricky and all must be rated for over 50Amps if you want high efficiency, like a set of 4, IPP110N20N3 at $4ea. It needs a Microcontroller to orchestrate all this well. Adding the caps inside would be great but costly. You could barely fit 10mF of X7R MLCC caps in the case and the 400, 22uF cap array would set you back $30 just in parts. Sort of $35 COGS or $70 with the big internal caps so would have to sell for $175 or $350 with the caps retail.

[mocfly]

Seems like our regulator has given up the ghost. Details to follow after troubleshooting in the am.

[chanik]

Coincidentally, I think mine is about to. It is wandering in regulation voltage over a +-0.5V range. Shorai is getting me a Silent Hektik unit to vet. If that one checks out, then I would prefer to swap it out than spring for another Ducati. In any case, I will be posting efficiency, regulation and thermal data once I have it in hand

[Anticept]

Why are the ducati voltage regulators dying so much? Heat?

 

Anyways, I too was thinking about some kind of new regulator, but since you are already on the case, keep us informed!

[chanik]

Heat.

Notwithstanding many datasheet claims of operating junction temperatures over 100C, I have NEVER seen a reliable circuit that runs hot. Case temperature of 60-70C means die junction temperatures over 100C

[Roger Lee]

Not all electronics die from heat. Everyone wants to blame heat and for some it may be true, but there are just failures and incorrect wiring and electrical setups that kill them too, but you can't tell someone it was their fault because it's always the MFG's fault. Just ask anyone. If you put a heat strip on the reg/rec and it doesn't go over 175F then it wasn't heat.

[chanik]

Not all electronics die from heat. Everyone wants to blame heat and for some it may be true, but there are just failures and incorrect wiring and electrical setups that kill them too, but you can't tell someone it was their fault because it's always the MFG's fault. Just ask anyone. If you put a heat strip on the reg/rec and it doesn't go over 175F then it wasn't heat.

Not so. You're in my field of expertise now. 175F is 80C which would yield a die temperature for the Silicon of quad devices in the bridge of at least 105C operating at 160W delivered (80W dissipated). The life of such a device can be measured in hundred of hours at most. As for the empirical angle, if you scan the BMW and Ducati motorcycle forums you will see that there too, the Ducati's have a long history of short life. The technical model derives from the Arrhenius equation for Silicon http://www.siliconfareast.com/activation-energy.htm

There are many mechanisms for failure but they net out to an effective activation energy of 2-3eV if you feel like running the math. The short version is that every 10C rise in temperature will cut life in half. So if the RR SCRs and diode devices would last 10,00hrs at 60C junction temperature, then 5000hrs at 70C, 2500hrs at 80C, 1250hrs at 90C and 625 at 100C and just 300hrs at 110C. Once one device starts to fail, the increased internal resistance will raise its die temperature even higher and send it quickly to its demise

[mocfly]

You think Flight Design had a reason for moving it? Of course they did.

 

Kurt,

Check this one also?

 

http://www.lightaircraftassociation.co.uk/engineering/Standard%20Mods/SM12970%20Rotax%2091X%20Schicke%20GR6%20Regulator.pdf

[Roger Lee]

Temp strips in our planes usually never see 140F. After 32 years using the Ducatic reg/rec they have held up, but I will agree there have been failures, but that anything mechanical part has failures. I have a 10 month ol TV that went bad in 4 months. Doesn't mean they are all bad. If the failures were over abundant like everything else Rotax would have changed them like they did the AC fuel pumps after 6-7 years and other parts over the many years. You guys are forgetting to take in the thousands of engines with the Ducati reg/rec and the millions of hours with them. So the failure rate is very small. Just because you read about a few failures on line you never read anything about all the ones doing well in the field and some over 30 years old. Remember people post about bad things, but rarely ever post how well it works. I haven't seen anyone come on and say how good the oil pump works, but let one leak and they are all bad.

[Tom Baker]

Temp strips in our planes usually never see 140F. After 32 years using the Ducatic reg/rec they have held up, but I will agree there have been failures, but that anything mechanical part has failures. I have a 10 month ol TV that went bad in 4 months. Doesn't mean they are all bad. If the failures were over abundant like everything else Rotax would have changed them like they did the AC fuel pumps after 6-7 years and other parts over the many years. You guys are forgetting to take in the thousands of engines with the Ducati reg/rec and the millions of hours with them. So the failure rate is very small. Just because you read about a few failures on line you never read anything about all the ones doing well in the field and some over 30 years old. Remember people post about bad things, but rarely ever post how well it works. I haven't seen anyone come on and say how good the oil pump works, but let one leak and they are all bad.

 

I'm just getting ready to pull and replace seals in a leaking oil pump. I think they are all bad.

[Anticept]

We've had to replace 2 regulators at about 1500 hrs on both of our airplanes.

 

The mechanic in delaware county airport is constantly replacing the volt regulators too, sometimes seeing a person bring their plane back more than once for that same problem.

 

I also have a little bit of experience with EE, but not on Chanik's level. The ducati regulators are garbage. I figured it was heat that was causing the failures, but I wanted his opinion. He is right though, the junctions will be SIGNIFICANTLY hotter than what you will read on the case. On a basic math level, applying the inverse square law for the heat would give you an idea of how intense it can get at the die level.

[mocfly]

Point to ponder: the entire assembly is potted in epoxy and there are only 2 components that are actually ATTATCHED to the heat sink surround.

[chanik]

All 4 components must bolt to the case or they would fry at 20Amps, but two of them need to be isolated (with thin mica sheets usually) The Schicke unit http://www.lightaircraftassociation.co.uk/engineering/Standard%20Mods/SM12970%20Rotax%2091X%20Schicke%20GR6%20Regulator.pdf is standard equipment on Remos so that one is worth evaluating. Like I said, if someone wants to get it I'll measure it up. Admittedly, noone has done statistical sampling but even here, the number of people mentioning RR failures is quite high. In the motorcycle world, failures are very common because people tend to mount these low and behind the engine, right next to where the exhaust come out and sweeps back. So they are exposed to enormous radiative heating and often fail in 10's of hours.

The point is that Semiconductor thermal failures are not like stalls, they are not abrupt so below some temperature you are fine. Well done COOL electronics can easily have an MTBF of over 100,000hrs. Think of how long an alarm clock lasts. If the Ducati were even 10,000hrs MTBF you would almost never hear of a failure.

[mocfly]

Seems like all is well.

I pulled out the plug, cleaned all the connectors, checked the wiring per the guide and resistance was good, AC VOLTAGE after reinstalling the plug. At engine start AC voltage at the plug measured 18v and decreased to 14.4-14.1 volts over the next 1/2 hour. Prior to shut down when turning the avionics off the AC voltage rose to 15.1-15.3

 

[Roger Lee]

The reg/rec plug is hard to push in and usually takes two hands to do a good job. Then look at the plug and see if it is straight and fully embedded into the receptacle. I have seen some that look in, but aren't. They are partially out and making poor contact.

After hose change is an important time to double check. The two yellow AC wires have to be removed and many do not reset the tab that locks the connectors back in plcae. so when the push the plug in it pushes those two wires outward so so they make poor to no contact and start arcing. This will cause damage.

[chanik]

I finally received the Silent Hektik RR and did some testing. It is much superior to both the Ducati and the ESR510. First off it dissipates about 1/3rd the power, operating at a calculated 89% efficiency at 250W in versus 75% for the Ducati and the ESR510. It only got to 50C on a 10 minute high current charge. It charges to a precise 14.0V which is a good semi-float charge for either the LifePO4 or the Odyssey battery. Like the Ducati, it opens the stator coils when not charging. It is a solid unit made out of machined Aluminum and is very conveniently drop in compatible with the Ducati. It will be installing in my E-LSA promptly. Very nice performing unit.

[Runtoeat]

Kurt, thanks for passing on this info. If someone installs this regulator in their CT, please ask for a "blanket" LOA from FD!

[Roger Lee]

Hi Dick,

 

FD would never approve this even if it was better because it would go against Rotax and they would never want to take on the liability of that and it's still be experimental. It would take thousands of hours worth of actual running test to prove its worth and reliability. Even though it tested okay on a bench doesn't put Kurt out of the woods yet for the short or long term use.

There may be hidden factors in play that none of us know. I have no issue with people trying different things, just remember how expensive things are. Right now if Kurt lost a lifter in his plane Rotax would help replace the engine. Once this is installed any help from Rotax on a part or engine failure regardless of what happens is out the window. He is in uncharted territory.

For better or worse it has a ways to go for proofing.