Rotax and the 22,000uF smoothing capacitor

[RFguy]

Now in its own forum

suggested DWG attached

 

r1.pdf

[skippydiesel]

More questions:

 

• Why do you have a fuse before and after the 22,000uF capacitor?
• Cant remember ever seeing a fuse on the + starting cable (battery to stater solenoid) of older/simpler cars - it would need to be a very high capacity - sugestions?
• So do I go  to JayCar and ask for a "TVS  diode - type 5KP16A or 5KP17A""  and use it to bridge between the positive + & negative - terminals on my 22,000uF capacitor ????
• Same as above regarding the "10 OHM 5W resistor in line between the battery & the 22,000uF capacitor ?

After I have done all this, what will I have done for my aircraft?

[RFguy]

19 minutes ago, skippydiesel said:

More questions:

 

• Why do you have a fuse before and after the 22,000uF capacitor?
• Cant remember ever seeing a fuse on the + starting cable (battery to stater solenoid) of older/simpler cars - it would need to be a very high capacity - sugestions?
• So do I go  to JayCar and ask for a "TVS  diode - type 5KP16A or 5KP17A""  and use it to bridge between the positive + & negative - terminals on my 22,000uF capacitor ????
• Same as above regarding the "10 OHM 5W resistor in line between the battery & the 22,000uF capacitor ?

After I have done all this, what will I have done for my aircraft?

"Why do you have a fuse before and after the 22,000uF capacitor?"

GE : because it is a component that may fail and cause a fire or at least alot of smoke.

 

"Cant remember ever seeing a fuse on the + starting cable"

GE : yeah this is more common nowdays. the fuse needs to be rated at the sum of all currents likely to be encountered.  so for a 150A starter, a 200A fuse is suitable. Some mfr split the master and starter fuses and solenoids so the cable that goes off to the master is not so huge (maybe 60A?)

 

"So do I go  to JayCar and ask fo"

Jaycar wont have those big ones . element14 might. I usualy do, they're a couple of bucks.

That will provide spike protection, and blow a fuse if there is an extended high voltage scenario.

The capacitor will not hold very fast transients down. The TVS will.

The best thing here is a "dissipative  clamp" but that is a few electrical parts and not readily available.

 

"Same as above regarding the "10 OHM 5W "

GE Yes , Jaycar have those. the only thing is- if the regulator draws current and there is no battery isolator, the resistor will permit battery power to go to the regulator when nothing is happening and flatten the battery. So, knowledge of the quiescent current is required.  The big issue is without the soft start, that the giant cap looks like a huge short circuit and is likely to damage the switch contacts when connected. 1000 amps will flow, briefly..... 

--- if you are not having fault problems with that alternator isolator switch (contacts welding) , then this 10 ohm resistor is probably not warranted. Using a wire size that is just big enough for the job  would also be another way to attack this- there would be sufficient resistance to ensure the current does not get too high.  But length and all sorts of things will come into it. 

 

The capacitor is I think bad news. It creates a number of problems that need to be solved.  I will draw another doc. 

 

"After I have done all this, what will I have done for my aircraft?"

GE : it will be non standard. if you are happy at the moment, then I guess dont touch it. 

 

 

 

Edited December 27, 2020 by RFguy

[RFguy]

Just to make things clear- I do NOT like the setup using a 22,000uF capacitor !

 

This is a bandaid for a design problem

 

Its use creates many other problems.

 

 

 

[skippydiesel]

I often make/do Hansen that have no discernible performance improvement but may assist with component longevity and to sum it all up I get that warm fuzzy feeling of having "done good" - this might be one of those.

[aro]

The diagram is incomplete... we need to see the C regulator terminal and switching to understand the design.

 

Rotax call for a 25A slow blowing fuse to protect the regulator. Since that circuit needs to take the entire regulator output anyway, the fuse between the regulator and capacitor seems redundant. The fuse on the battery side protects that circuit. I'm not convinced that a 25A fuse would provide much protection for the capacitor.

 

My understanding of the capacitor function is:

a) the PM alternator output is not very smooth because it comes from a single phase, and the capacitor helps there. Sure, a better alternator might be nice, it might also be heavier, bulkier, more expensive.

b) if the battery and load is disconnected, it slows the voltage rise enough that the regulator should be able to respond and cut the output without damage to the regulator & charging circuit.

[RFguy]

yeah

I'll update the dwg. 25A slow blow is surprising - isnt it a 20A continuous output ?

I gather it is 25A to cope with flat batteries. Pretty hard to blow a slow blow fuse without 2x the current.

And of course there are many types of fuse curves..... If the cap goes full short, we need to stop the wiring catching fire, hence the fuse. I would accept those that a fault condition of say 50% of the fuse rating could leave the wiring intact but the capacitor would be smoking like a tear gas can !

 

I understand the alternator is also a PM (permanent magnet) type, so the only thing that the regulator can do is adjust how often the load is connected. An advanced version of this would have a thing called phase controlled rectifier, but in this case I dunno whats inside it. If regulator components are rated for the max open circuit voltage, that's probably OK but the cores in the stator will not like it....

 

Your understanding of the reason for the capacitor I think is reasonable: The capacitor  yes does smooth the output out of the (I expect) full wave bridge rectifier. The battery should be doing this job, but the connections to the battery dont permit the battery to do that job .  and it also controls the rate of rise of the output, slightly... the regulator loop should be fast enough though. pretty crappy regulator otherwise.

 

I don't doubt the capacitor is required  because of the way it is wired up.  Either way, I think it is an extremely low performance solution for a engine hi-tech company.... 

 

However- most people dont have any problem with the rotax charging setup  so I guess form that point of view, it is satisfactory.....

 

 

 

 

 

 

 

 

[Kyle Communications]

The Rotax generator is a dirty simple thing. The std regulator is also a dirty and poorly designed and built unit. Typical of most motorbike stuff. The reg we have redone and running is far better it is also capable of very good current limiting and voltage regulation. We have a unit testing now and it has gone through some "fine tuning". I am currently doing the final layout of the pcb. It will be in 2 parts. You can have a very good regulator only and with another unit that bolts on underneath you can then have current limit and voltage limiting which will allow you to use a Lithium battery safely. It will of course charge other chemistry batteries perfectly fine.

This is a early version the regulator only.

 

[RFguy]

Hi mark, thanks for chiming in...., are those scope pics of the factory regulator or yours going in and out on a cycle by cycle basis ? looks like a suitable method, cycle by cycle basis.

That waveform- looks like not too much problem with stator core saturation when unloaded ? That can lead to the cores being biased up one way. Depends on the core material used

I am going to do  a simple buck-boost PFC style regulator-

I will be able to get more out of the alternator this way, and probably achieve better alternator stator temps due to a more favourable powerfactor than a straight rectifier-cap.  I'd done buck-boost PFCs up to 5kW, so a 250W one should be straightforward...

-glen

 

[aro]

14 hours ago, RFguy said:

25A slow blow is surprising - isnt it a 20A continuous output ?

The installation manual shows 20A at 5000 rpm, and lists the maximum output as 22A. Also, as you point out the capacitor can draw a much higher current from the battery for an instant when it is initially connected.

 

You obviously want to be sure that the fuse won't blow unless there is a real problem, otherwise it becomes a reliability issue in itself.

[Kyle Communications]

Thats one of ours....the std generator is max 18 amps but it gets pretty hot. The big issue is the actual ducati regulator..its a POS. They use a std bridge rectifier in it and the amount of heat often creates bad a cooked solder joints on the main current components. Ours uses high current single package diodes and SCR's. Thats from a year ago those waveforms..its even better now. We decided to limit the current to a base of 10 amps so not to cook Litium batteries if the internal BMS craps itself in the battery and also by limiting the voltage you should never have any issue with the Lithium. There is a plug selector to allow higher currents up to 14 amps for charging. We decided that was the limit for the generator. Ours has far far better heatsinking as well. It also has the same hole mounting specs as the original

 

[RFguy]

Hi Mark. yeah sounds reasonable. A std bridge like a PB40  will be about 40 watts dissipation at 20 amps !

What's the maximum AC frequency on a rotax setup ?

 

 

 

 

[RFguy]

16 minutes ago, aro said:

The installation manual shows 20A at 5000 rpm, and lists the maximum output as 22A. Also, as you point out the capacitor can draw a much higher current from the battery for an instant when it is initially connected.

 

You obviously want to be sure that the fuse won't blow unless there is a real problem, otherwise it becomes a reliability issue in itself.

Whether the inrush would blow or weaken a 25A fast blow fuse would depend alot of the capacitor, the wiring  , and the fuse curve... many many things.... Hence I like to see the soft-start- the resistor connected across the switch so that the capacitor charges up , even if just for very short period.  if the regulator draws no current in that direction, that will achieve the desired outcome.  Mark might have something to add here on quiescent of the Ducati regs ?

Edited December 27, 2020 by RFguy

[aro]

4 minutes ago, RFguy said:

Hence I like to see the soft-start- the resistor connected across the switch so that the capacitor charges up , even if just for very short period.

How likely is that to be a real problem? I have never heard of it in practice.

 

That switch needs to be rated to open a 20A+ DC circuit - is the capacitor going to draw enough energy to damage the switch when it closes?

[RFguy]

OK, let's do some MATH  since things should be ENGINEERED and not relied on "never failed so far"

Capacitor: Something not too exotic : 22,000uF  25V 85degC (needs to be 105C !)  but nevertheless a good quality capacitor

A Kemet cap, ALS30 series :  Max ripple current 100/10k = 10 amps. 100Hz ESR = 17 milli ohms

1.5 meter total of 14 AWG (in + and -)  = 10 milliohms

20A fast blow fuse and holder  :  8 milli ohms

Battery connectors : 5 milliohms.

Switch , 25 A toggle :  15 milliohms . a good one. 

total = 63 milli ohms

voltage = 12.5V. time constant = 1.4mS. Let's assume it's all over in 1mS.

peak current = 198 amps.

Fuse I^2T limit = ~ 500A^2/sec. get sa bit rubbery here because the exact pulse shape is not yet known, however : 

Pulse I^2t = 39.2. limit is 500. need to be 30% of I2t rating for 1000 hits . or I^2t fuse > 100.  and it is 500, so the fuse will be fine. 

As for the switch, an tungsten lamp inrush is fairly similar, switches are derated usualy around a factor of 3x to 4x for a DC supply.

Applying something similar for available datasheet, we want a switch that is rated at approx 100 amps to withstand 1000 cycles.

hence, use of a soft start for the capacitor.....

 

-glen

 

Edited December 28, 2020 by RFguy

[skippydiesel]

WOW!!!!!!! - I dont think I will need to be reminded NOT to engage either RFguy or Kyle in any conversation that may involve electronics - just wait for the install instructions, when all has been sorted out.

[aro]

1 hour ago, RFguy said:

OK, let's do some MATH  since things should be ENGINEERED and not relied on "never failed so far"

 

Sure, but when there are a lot of assumptions it's also worth verifying that the engineering calculations successfully predict what you see in the field.

 

If a tungsten lamp inrush is similar, should we have a soft start for a light switch?

 

It also depends on the wiring - I know in my aircraft the capacitor remains charged when the master is switched off so there won't normally be a large current when it is switched on. I can't remember the details of the switching of the charging system - I will have to double check what is switched. 

[RFguy]

" should we have a soft start for a light switch?"

yes, derating is normal. It depends a bit on the service

for DC service, derating for tungsten frmo the specified resistive load  is ~ 4 x

for AC service, 50 Hz, derating is far less because of the AC waveform. Sure you can hit it at the top of the cycle but that is unlikely to happen all the time. 

 

ARO , yes do have a look how it is wired. I'd expect some leakage from the cap and the regulator, so if the cap is still fairly well charged after a couple of days, I'd say it is still connected (somehow) to the battery...

-glen

[Kenlsa]

Reading this topic confirms that I am the long lost fourth brother of the 3 stooges.

Ken

[aro]

I have been re-reading my info on switches. Switches can have different ratings for resistive loads vs. tungsten loads however the switches that are readily available to us typically seem to list one value. Which value is it? Are we asking for problems using e.g. a 15A switch for a 12V 100W light?

 

Re wiring:

- I know the capacitor is not connected to the battery with the master off - even though it remains charged, it can be discharged.

- I know the battery is disconnected from the bus and from the regulator with the master off.

- I know the regulator is disconnected from the bus with the master off

- I know the regulator C terminal is disconnected from the battery with the master off

- I think the capacitor remains connected to the regulator B terminal when the master is off

- The capacitor does not discharge, so it isn't connected to the C terminal with the master off.

 

The C terminal is the one I am not sure about. The requirements are difficult to satisfy:

- needs to read the battery voltage within 0.2V

- needs to be disconnected from the battery at shutdown

- must not be disconnected from the B terminal while the engine is running or regulator damage can occur

 

The last one is the one that might be a problem for me. I think the C terminal is connected to the bus which means it would be disconnected if I shut off the master. I don't think previous versions of the installation manual were as explicit. I could connect it to the B terminal, but then the voltage would be less accurate relative to the battery because it would be influenced by the electrical load on the alternator.

 

Ideally perhaps I should run both B and C wires back to the battery, and another separate wire forward to the bus. Then they all need some sort of fuse at the battery end...

[RFguy]

ARO, what schematic are you working from ? anything I can look at? any wire sizes given to permit clues ?

 

I have seen many schematic variations for the bog std rotax, and also many many variations from LSA to LSA. Jab have changed things around a few times, also.

and...

 

"Are we asking for problems using e.g. a 15A switch for a 12V 100W light?"  yes. if tungsten.... maybe after many cycles it might be unreliable.... it depends alot of the contact material.

 

-glen

 

Edited December 28, 2020 by RFguy

[Kyle Communications]

Glenn the Rotax has 8 poles and at 5800 rpm its about 464 hz out of the generator. You can see at around 5000 rpm on the first cro pic the freq is 416hz

 

 

[RFguy]

ahh. could use one of those fancy 400 Hz full wave rectifier MOSFET helpers from LT.

I've built my own discrete units for the brain stimulator (1200V, 6000 amps (pulse)) .. but for something small, the LThelper  is a  good solution to squeeze a little more efficiency out of it.

[Kyle Communications]

I am using high speed hi current TO-247 package diodes

The bridges they use almost unsolder the tags from the pcbs..we have pulled a heap apart and repaired them. other issues as well with heat too...the bridges they use are std and good for 50 hz but not for 500hz

 

[RFguy]

as usual diodes are the toughest thing in electronics...they desolder themselves to save themselves...:-) . ya know I had some silicon carbide diodes that the plastic encapsulation melted off.. and they continued to work.

Edited December 28, 2020 by RFguy