Power Interrupt

I have a ESS BUSS on my electrical system with a second feed directly from the battery. If there is power on the Avionics Buss, a relay feeds power from there. If not, it flips and takes power from the direct feed. All works as advertised.

However, the relay switching time is specced at 20ms. The GNS430 copes with this. The Dynon Skyview didn't but now does with the back-up battery connected. The transponder (GTX328) trips off. It only takes a couple of seconds to reset it but with other things potentially going on, I'd rather it wasn't necessary.

Can I prevent this by using some sort of capacitor or something?

Yes, but caution is advised.

You're trying to cover that 20ms gap with a capacitor to smooth the transition, and it will certainly do that just fine with no heartburn. And when the engine is shut down and the master turned off, that capacitor will slowly leak down to zero volts and again creates no heartburn.

Now the next day you jump into the airplane and hit the master - and the battery tries to immediately handle that large load on the capacitor by feeding it some 100-150 amps for 20-30ms to charge it - and in the process it will either trip a breaker or weld the contacts on your master switch - there's your heartburn.

What you would need to do is determine the total current draw that the capacitor would need to deliver during the 20ms window, and then determine the lowest acceptable voltage to be used by the most finicky piece of equipment during that 20ms window, and choose a resistor of the correct size (both wattage and ohms) to place in series with the capacitor to limit the current by producing a voltage drop across it according to Ohms Law. That will ensure that the capacitor charges and discharges at no more than a certain maximum amp rate, and then you will need to put a breaker in your panel for about 150% of that rate, and pass the current loop on the capacitor through that breaker.

I don't think a capacitor will help

If I understand you correctly, you have a relay with coil and NO contact connected to the avionics buss. The NC contact is connected to a battery feed, and the common terminal is connected to the essential buss.

If you put a capacitor on the essential buss, it will be charged by the avionics buss through the relay. When you remove power from the avionics buss, the capacitor will discharge through the relay and avionics buss, to the rest of the electrical system until the relay hits its dropout voltage (which is probably a lot lower than any of your avionics can tolerate), then the essential buss will switch to the battery feed.

Replacing the relay with a pair of schottky rectifiers will give you the automatic switching you are looking for, but you will sacrifice about a half a volt.

My .02,
Paige

Accept the anomaly

Personally, I'd accept it like it is before I'd add more complexity to the system.

A transponder is one of the things I would consider shutting down if I was only on battery power, depending on the situation.

Even in IMC it's not that big of a deal. Transponders fail all the time.

There's my two pennies.

Don

Correct me if I'm wrong but the relay sounds like it could be a single point of failure.

rocketbob said:

Correct me if I'm wrong but the relay sounds like it could be a single point of failure.

Click to expand...

..... except that the system is wired via an AUTO/OFF/ON switch whereby I can manually switch power to the standby route in the event of the relay failing in some intermediate position Aeroelectric Connection circuits involved using a diode to prevent back flow down the standby section of the circuit. I just had a philisophical objection to having 2 sources permanently connected to the ESS BUSS.

I have a standby 20A alternator which can power the ESS BUSS indefinitely so the power drain of the transponder is not an issue. I was just trying to find a way to smooth out the voltage drop so as to keep it on line for 20ms. I seemed to remember from my school physics that a small capacitor might do that. The idea would be to run it in series from the transponder CB on the ESS BUSS to the transponder only. Everything else copes.

<Edited out series diagram - a bit more research makes me think it needs to be in parallel - Wikipedia is a wonderful thing!!>

If it comes under the "too hard" category then I shall live with having to reset the GTX327........

Paul,

Time to review capacitors again in your physics book

If you connect a capacitor in series with your transponder circuit as your diagram shows, your transponder will not work, ever. A capacitor acts as an open circuit to DC. Your transponder will not receive power even when the ESS BUSS is powered.

I think what you meant is to connect a capacitor in parallel with the transponder circuit, i.e. positive terminal on the power side and negative terminal to ground. In this configuration, it could supply power to sustain the transponder through the transition. However,

1) A "small" capacitor may or may not be sufficient. The transponder is a substantial load. You'll have to do the math to figure out the minimum capacitance needed to sustain it through the 20 ms dropout.

2) If it ends up being a sizable capacitor, you may have to contend with inrush current issues, as Greg alluded to.

3) Without further provisions, the capacitor will not only feed the transponder, but all other loads on your ESS BUSS as well. Current will travel just as easily "backwards" through the transponder's CB and back down the other CB's to their respective loads. More loads sucking down that capacitor will mean you'll need an even bigger capacitor to bridge the 20 ms dropout.

If you want to have the capacitor only sustain the transponder (issue #3), then you would need to also add a diode to prevent current from flowing from the cap back to the ESS BUSS. Something like this:

HTML:

ESS BUSS ----- CB ------|>|---------------------
                       diode      |            |
                                __|__+         |+
                            cap _____       GTX327
                                  |  -         |-
                                  |            |
GND---------------------------------------------

Note that the circuit above only addresses issue #3. Issues #1 and #2 still have to be addressed.

On a "philosophical" note , diodes, and solid state devices in general, are highly reliable. Switches and relays, and electromechanical devices in general, are notoriously unreliable. If I were you, I'd reconsider the architecture and eliminate that relay. FWIW.

-Roee

Glider pilots use a 22000 microfarad capacitor to hold up the avionics (typically a logger and nav system) while switching batteries. This is about 1" dia x 4" long.

However, Paul, are you sure you have the system wired in the best way? The whole point in Bob's architectures is to eliminate single points of failure.

Pete

I'm no EE, but when I googled schottky rectifiers, this is what came up:

"Schottky rectifiers...carries advantages such as very low forward voltage drop and switching speeds that approach zero time...ideal for output stages of switching power supplies."

Sure sounds like something that would be good to transfer power between a main bus and a standby bus.

Interested to learn what you decide to do.

-Jim

Thanks for all the advice and the remedial physics course - it's been a while!!

I take on board your comments concerning the relay being a potential single point of failure. However, as I said, I have an AUTO/OFF/ON switch where I can isolate or by-pass the relay so there are 2 ways to connect the standby power to the ESS BUSS. Furthermore, I have a switch breaker between the Avionics and ESS BUSS. Together with the avionics master and the battery master, I can isolate any buss in the event of a dead short. I accept the comments concerning the reliability of relays and maybe I have just tried to be a little too "elegant" with the design, but I have tried the system out and it certainly works as advertised. I am down-route at the moment but will post a diagram when I get home which I hope will show the system better and how it deals with various failure modes.

To the physics! Roee's diagram helped a lot - I was getting there slowly and had already realised it needed to be a parallel circuit. The need for the diode had escaped me. I believe that all that is needed now is a resistor on the positive side of the capacitor circuit to limit the charging current. BUT, won't that drop the voltage to the transponder when main power is removed? So the next question is - would the resistor have the desired effect if placed on the earth side of the capacitor?

CB
|
Diode
|
| ------ GTX327 ------ Earth
|
Capacitor
|
Resistor
|
Earth

Somehow, I don't think that works

I have solved (I think) the maths for the size of the capacitor - amazing what you can do at 3 in the morning when you are jet-lagged and 6000 miles from home!

Max draw of GTX327 is 22W for 20ms = 22x0.02 = 0.44J (say 0.5)

For a capacitor, W=1/2 C V(sq)

So 0.5=0.5xCx14(sq) so C=1/200 or 5mF

However, as the capacitor discharges, the voltage drops. Assuming we wish to maintain at least 10V or about 14x0.7 then the capacitor must be not more than half discharged (0.7(sq)). So I need a 10mF capacitor.

Which is all pretty academic as I am rapidly coming to the conclusion that it will be much simpler just to take the few seconds to switch the transponder back on! Interesting discussion, though

The only final thought I have before heading for breakfast is why does the GNS430 cope with the power transient and the GTX327 doesn't? I would have thought that any certified avionics would have dealt with a short interrupt.

paul330 said:

... I believe that all that is needed now is a resistor on the positive side of the capacitor circuit to limit the charging current. BUT, won't that drop the voltage to the transponder when main power is removed? So the next question is - would the resistor have the desired effect if placed on the earth side of the capacitor?

CB
|
Diode
|
| ------ GTX327 ------ Earth
|
Capacitor
|
Resistor
|
Earth

Somehow, I don't think that works
...

Click to expand...

Paul,

Yes, the resistor will limit the inrush current. The peak inrush current through the capacitor would equal Vbuss / R, where Vbuss is the buss voltage and R is the value of the resistor. Note that the value of the capacitor doesn't factor in to the peak inrush current. The value of the capacitor only affects the time duration for the inrush current to taper off.

Yes, the resistor will also incur a voltage drop to the transponder when drawing current from the capacitor when the buss drops out. The voltage drop across the resistor is equal to Iload * R, where Iload is the transponder load current and R is the value of the resistor.

It makes no difference on which side of the capacitor you put the resistor. They're in series, so all current flowing through one has to also flow through the other.

Since you seem to be enjoying this , have you done the math to select the value and power rating for the resistor?

Anyway, for educational purposes only.

I agree, if you do keep your existing buss architecture, don't bother adding this complexity just to keep the transponder from dropping out. Just live with having to reboot the transponder. No big deal.

Good luck,
-Roee

OK, I think this works



1000ohm-1/4W resistor

Initial charge current = 14/1000 = .014A giving .014x14 = 0.2W

On-line calculator shows initial charge time to be about 1 minute

And you are right, Roee - I am enjoying this. Fills in the time between 3am and breakfast nicely when you are in a hotel far from home.......

paul330 said:

OK, I think this works

Click to expand...

Yep, I think your circuit works. Kudos, Paul!

Two footnotes to your circuit:

1) With the diode you added, there will still be a voltage drop from the capacitor to the transponder when powered from the capacitor. But unlike with a path through a resistor, now the voltage drop will be fixed at approx 0.4V (assuming the diode is a Schottky) more-or-less irrespective of the load current.

2) Step back and look at the circuit you drew. You've independently arrived right back at the "diode OR'ed" circuit topology, i.e. two power sources (the ESS buss and the capacitor) permanently connected to a load (the transponder) through a pair of diodes. That's an often seen topology, for good reason

paul330 said:

And you are right, Roee - I am enjoying this. Fills in the time between 3am and breakfast nicely when you are in a hotel far from home.......

Click to expand...

Now, I do believe you're enjoying this mental exercise, so let's continue to improve the circuit.

One downside of your circuit is that the transponder will effectively see two diode drops (i.e. 2 x 0.4V) relative to nominal buss voltage when powered from the capacitor. That's because there's the obvious one diode drop for current flowing from the capacitor to the transponder. But there's also the one diode drop for the current flowing from the buss to the capacitor in the first place, so the capacitor will only be charged to the buss voltage minus 0.4V.

So can you think of any simple way to eliminate the second diode drop when powered from the cap?

Have fun!

roee said:

One downside of your circuit is that the transponder will effectively see two diode drops (i.e. 2 x 0.4V) relative to nominal buss voltage when powered from the capacitor. That's because there's the obvious one diode drop for current flowing from the capacitor to the transponder. But there's also the one diode drop for the current flowing from the buss to the capacitor in the first place, so the capacitor will only be charged to the buss voltage minus 0.4V.

So can you think of any simple way to eliminate the second diode drop when powered from the cap?

Have fun!

Click to expand...

Yes I can! Worked it out with my First Officer on the flight home - don't think we missed any ATC calls

Move the diode in the main power circuit to the right of the junction for the resistor ie take it out of the charge circuit. However, this will result in a voltage leakage back through the resistor during the transient - less than 0.4V though. This can be further alleviated by increasing the value to (say) 5k. This reduces the leakage to virtually nil. Charge time would be about 4 minutes.

We also worked out some solutions using transistors to switch circuits and eliminate the diodes completely.

Flame Suit On

I stand accused of firstly designing an electrical system with a single point of failure and secondly modifying one of Guru AeroElectric Bob's circuits without just cause. Here is the case for the defence:

The relay in question is NOT an automotive headlamp dipper relay. It is a high quality item with 30A rating and a quoted life of one million operations. When power is taken off the control circuit, it returns to the NO position by a cantilevered leaf spring. It's a large flat sheet of metal and simply cannot do anything else - no coil springs to break or jam. It would take a huge current to fuse the contacts closed and this could only come from a short on the ESS BUSS. In which case the ESS BUSS isolate switch breakers would trip (rated lower than Avionics Master). If there is a short on the ESS BUSS then it is lost anyway but this system protects the avionics buss. In any case, the relay can be by-passed by the AUTO/OFF/ON switch to manually restore power to the ESS BUSS in the event of some other failure mode. I should also mention that (not shown in the diagrams below) there is a 2A fuse in the control circuit for the relay. In the event of any issues there, the fuse will blow putting the relay back to the NO position and assuring ESS BUSS power.



Now with loss of Main Buss (say Master Relay fail):



In the event of some sort of internal failure of the relay:



I should further point out that I have a dual Skyview system with back-up battery on the main screen and naturally I have split my equipment between the Avionics and ESS Busses. Worst case is a short on the ESS BUSS which trips the Isolation Breaker and the fuse. This would still leave me with SL30 (NAV/COMM) and second Skyview (giving flight/engine instruments and GPS navigation) thus assuring IFR capability (Fail Operational). Main Skyview is available for 60 minutes. Worst case double failure results in total electrical loss but still leaves the main Skyview with flight/engine instruments and VFR GPS for 60 minutes (Fail Safe).

I am open to comment, especially if someone can show me a single failure which causes loss of both ESS and Avionics Busses. In any event, the parts for all this cost a few bucks although the mental gymnastics required to work it all out aren't factored into that! It would be easy to ultimately replace it with the standard always-on/diode protected circuit. Somehow (although I can't prove it) that set-up seems to me more prone to loss through a major short.

The defence rests......

paul330 said:

Yes I can! Worked it out with my First Officer on the flight home - don't think we missed any ATC calls

Move the diode in the main power circuit to the right of the junction for the resistor ie take it out of the charge circuit.

Click to expand...

Well done, Paul!

Your proposed change to the circuit is exactly what I would do to eliminate the second diode drop from the cap to the transponder.

paul330 said:

However, this will result in a voltage leakage back through the resistor during the transient - less than 0.4V though. This can be further alleviated by increasing the value to (say) 5k. This reduces the leakage to virtually nil. Charge time would be about 4 minutes.

Click to expand...

And you're correct that this does introduce a minor leakage through the resistor, but your explanation of it needs further discussion. There will indeed be a leakage through the resistor, but I'm not sure what you mean exactly by a voltage leakage. The issue is that with the resistor being upstream of the main diode now, when the buss drops out, current will leak from the capacitor through the resistor back to the buss and down to the other loads on the buss. The capacitor is no longer completely isolated to just the transponder. But note that this leakage will have no direct effect on the voltage at the transponder, other than that the current leakage through the resistor to the other buss loads will contribute to discharging the capacitor a little bit faster. But since you can tolerate a long charge time and have therefore chosen a high resistor value, the leakage current will be relatively small compared to the transponder's load current, therefore having no significant effect on the capacitor discharge time. The max leakage through the resistor is limited to Vcap / R, so at V=12V and R=1Kohm you're only talking about 12mA (and in reality will be even less, that is the short circuit worst case). And as you pointed out, the bigger the R, the smaller the leakage. True. But even at R=1Kohm, 12mA leakage is just insignificant. So we're agreed, with the values you've chosen, this leakage issue is a non-issue.

paul330 said:

We also worked out some solutions using transistors to switch circuits and eliminate the diodes completely.

Click to expand...

Yes! You can eliminate the diode drops altogether by using transistors instead. But then the circuit design and analysis will get much more involved. Extra credit if you can pull this off!

And my suggestion would actually be to do it with FETs (field effect transistors) rather than BJTs (bipolar junction transistors, or simply "transistors"). Modern power FETs have incredibly low on-resistance and therefore very high current-carrying ability. And FETs are, generally speaking, much friendlier devices to design with than BJTs. If I were to do a modern clean-sheet design for the electrical system, it would definitely use FETs for the main power switching.

P.S. You're getting good at this! Want to trade jobs for a week? What do you fly?

Is it aircrews proof?

Your topology looks very similar to a Boeing design...we actually checked the circuit during ground ops, each flight. In about 1500 hours, I saw 2 failures of the relays...which probably had 18,000 hours on them.

One potential gotcha that I see:
Let's say that you are flying along and experience an alternator failure. You are aware of the potential switching transient, and decide to play it safe and turn standby power on before you kill the master switch. When you kill the master, your relay will remain engaged, drawing power through the essential buss...your entire electrical load will flow through the standby power feed.

If there is a good load, a fuse will blow, or a CB will pop. Fuses are generally faster acting than CBs, so I wouldn't be surprised if the 25A battery fuse blows. Now you are stuck with turning your master switch on, and manually shedding loads.

If the load remains low enough to keep from blowing a fuse or CB, you will just waste power (relay holding current and other loads that you probably thought were shed when you killed the master)

So how do you prevent this? You could add a caution to the POH, or you could design that failure mode out by wiring your relay coil through a second pole in your selector switch. (what's the current rating on that switch?)

BTW, with a "diode or" circuit, there are gotcha's also...a transient essential buss short could take out both feeds.



Paige

PaigeHoffart said:

Your topology looks very similar to a Boeing design...we actually checked the circuit during ground ops, each flight. In about 1500 hours, I saw 2 failures of the relays...which probably had 18,000 hours on them.

One potential gotcha that I see:
Let's say that you are flying along and experience an alternator failure. You are aware of the potential switching transient, and decide to play it safe and turn standby power on before you kill the master switch. When you kill the master, your relay will remain engaged, drawing power through the essential buss...your entire electrical load will flow through the standby power feed.

Paige

Click to expand...

Yes, you are right. I haven't shown all the failure modes. The idea with a main ALT failure would be to leave it in AUTO - initially, you have lost nothing. Then looking at your flight time and load requirements, you start the standby generator and accept a drain on the battery or switch off the Batt Master and run on Standby Power indefinitely - yes, the POH would have to be written accordingly. The way I have the circuits configured, the main drains off the Main Buss are the external lighting which can be switched back on for landing.

Funny, I am flying the B747-400 so maybe that is where I got the idea from. A lot of aircraft I have flown have had the AUTO/OFF/ON algorithm and I think it is pretty sound.

roee said:

Yes! You can eliminate the diode drops altogether by using transistors instead. But then the circuit design and analysis will get much more involved. Extra credit if you can pull this off!

Click to expand...

So, you replace the diode on the discharge circuit with a NOT gate - controlled by a feed from the charge circuit upstream of the resistor.

I am going to be in SFO next week and there is a Radio Shack at the bottom of the street from our hotel. For a few bucks, I am going to buy the bits and build this design (just resistors and diodes) and see if it works.

And sorry, flying B747-400 out of Hong Kong and not prepared to swap....

What would happen if the negative terminal screw into the battery were to come loose and you lost that connection? Would either bus work? Would it dump charge into the airframe with a steadily decreasing potential (voltage) between the positive terminal (or alternator, or stby generator) and the airframe?

Wondering if there's not still a single point of failure in our single battery systems.

Of course you'd still have your Skyviews with backup batteries, but would anything else be operational? Or would it not matter one bit and everything would continue to work?

I believe the electrical system would still run from the alternator. It's not ideal and I believe there would be some instability since the battery normally acts as a "sink", smoothing out the power. I'm not sure what would happen if you lost the main alternator. Would you be able to start the standby with no power to the voltage regulator? Would the power spike when you closed the field fry the system before the voltage regulator kicked in?

I agree that a single battery system introduces some weakness but I guess you have to accept that properly connected terminals can't come loose. After all, we have only one crankshaft..........

Absolutely true, Paul. One crankshaft, one prop, etc. There are a lot of folks flying with electrically-dependent fuel delivery systems (no mechanical pump) who may be operating under the assumption that between the battery and perhaps even a second alternator, they are at least single-fault-tolerant for any electrical system failure that might occur. I'm trying to figure out if that's indeed true, and hoping some of the sparky experts on the forum can shed some light on exactly what would happen if our battery decided to shed its negative terminal from the battery case. Impossible? Maybe, but that very thing happened on my car battery about 5 years ago. Different technology, of course, but it does give one pause. (I can't say for certain, but I don't believe the engine was running at the time the terminal let loose.)

Great discussion all around. I love this forum!

krw5927 said:

... and hoping some of the sparky experts on the forum can shed some light on exactly what would happen if our battery decided to shed its negative terminal from the battery case.

Click to expand...

Quite simply, if one of the battery terminals becomes disconnected (doesn't matter which terminal), then it's as if the battery is no longer there. Go ahead and analyze the system with the battery removed and you'll have your answer.

roee said:

Quite simply, if one of the battery terminals becomes disconnected (doesn't matter which terminal), then it's as if the battery is no longer there. Go ahead and analyze the system with the battery removed and you'll have your answer.

Click to expand...

What I get from this is that NOTHING requiring electrical power will work. Period. You can have all the backup alternators in the world, but without somewhere for the electrons to flow TO (or FROM, don't remember that class) nothing will happen.

True?

It sounds like we are building the space shuttle here sometimes

Seriously, battery failures like terminals popping off/shorting etc. are pretty darn rare (I've never seen one). From what I've personally observed in over 35 years of aircraft maintenance is improper assembly and substandard parts/materials cause the largest majority of system failures (non "black box" failures).

Aircraft quality parts and tools cost more than radio shack parts/tools, use the former and you will have trouble free systems, use the latter and you will have problems. Never seen a catastrophic battery failure but I've seen plenty of electrical failures caused by terminals wiring and splices that were incorrectly installed/substandard parts.

RV's are pretty simple, build it to aircraft standards, use quality parts, and problems are going to be pretty rare.

PS: the electrical system will keep funtioning with the battery disconnected, some stuff may not like the high ripple but most stuff will keep working fine.

krw5927 said:

What I get from this is that NOTHING requiring electrical power will work. Period. You can have all the backup alternators in the world, but without somewhere for the electrons to flow TO (or FROM, don't remember that class) nothing will happen.

True?

Click to expand...

Not entirely true.

If the alternator is operating when the battery is removed, then the alternator might continue to operate, but its output will have high ripple and will not be well regulated. Some loads will tolerate this better than others.

If the alternator is not operating, then you will not be able to get it to start operating without power from a battery or some other power source.

Walt said:

...Aircraft quality parts and tools cost more than radio shack parts/tools, use the former and you will have trouble free systems, use the latter and you will have problems. ...

Click to expand...

I disagree.

1. High quality components can lessen the rate of occurrence of individual component failures, but they do not in themselves ensure a reliable system. System reliability is not just a function of the reliability of its individual components, but also a strong function of the fault tolerance characteristics of the system architecture as a whole. A well designed system built with less reliable components can be much more reliable than a poorly designed system built using the most reliable components money can buy.

2. Commercial grade parts are in many cases equally reliable or even more reliable than expensive "aircraft quality" parts. In some cases, "aircraft" parts are truly superior in quality in some meaningful way. But in many cases, "aircraft quality" simply means that a part was designed decades ago to some military specification, has not benefited from advancements in modern technology, and is very expensive simply due to its regulation-burdened niche market. It pays to do your homework on which "aircraft quality" parts are truly superior in a meaningful way and which are not, and where they're actually necessary and beneficial and where they're simply not.