Isolated power supply for automotive use

Go To Last Post
20 posts / 0 new
Author
Message
#1
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Hello,

I am currently developing a datalogger for car.

Here is my I/O configuration :

    * 8 DI * 4 Hi-speed counter (fmax = 100kHz)
    * 4 AI
    * CAN connection

I will have 2 uc :

    * Cortex M3 : main uc * Attiny 2313 for hi-speed counters

I want to have a good isolation (galvanic isolation?) of the I/O and of the power supply.

I intend to use optocouplers for I/O (K847 for DI and HCNR 2000 for AI) and a CAN transceiver (MAX13041).
>> This components provide enough isolation for car use?
Here is the schematic of my optocoupler. Could you check it?

I would like to use DC/DC isolated converter for generating 3.3V and 12V.
>>I need some help to choose the right DC/DC converter, unless there is a better way to handle my problem?

>>Should I take care of others problems linked to car design?
>>Have you some resource/links for automotive design?

Thanks for your comments and advice.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Take a look at Texas TPS5430.

In order to have a step down isolated dc/dc converter then for sure you have to use a transformer and not a simple coil.

Also there is the 34063al and there is a newer and faster ic like this faster and better.

Maybe you will find something there.

Michael

Michael.

User of:
IAR Embedded Workbench C/C++ Compiler
Altium Designer

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Michael.

User of:
IAR Embedded Workbench C/C++ Compiler
Altium Designer

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Why do you want to galvanically isolate? If you hope this will avoid noise coupling problems, then you're probably about to learn a lesson regarding fast transients! Most galvanic isolation techniques work from DC to low frequencies and require careful attention to implementation and design to avoid coupling the noise that you might want to avoid.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I don't think galvanic isolation will give you any advantage except for added complexity and even more noise sources.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Have you some examples of power supply use in car system?

If galvanic isolation is complex to design, is there an other design easier and good enough?

I don't have experience with such design (I take a look on google and some IC manufacturers (LT, Maxim, TI, Vischlay, ...)).

Thank you for your advices.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

National Semiconductor have regulators specifically for automotive use.

As for the other signals, do some research on ECU designs on the interweb and these should go a long way to solving your design problems.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Just a normal regulator, and you add a bit of bulletproofing to the inputs and outputs, with current limiting resistors, voltage clamps like zeners or tranzorbs and bit of filtering etc.

I have a fair number of AVRs in my car, and did not use optocouplers or galvanic isolation and they just keep on working :)

For digital inputs I usually use a series current limiting resistor, a pull up or pull down resistor and a tranzorb and a little cap.

If you wanna know how car manufacturers do it, look here ;)

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

And as additional advice to what Jayjay and Kartman already mentioned: pay a lot of attention to your groundpath !

Nard

A GIF is worth a thousend words   They are called Rosa, Sylvia, Tessa and Tina, You can find them https://www.linuxmint.com/

Dragon broken ? http://aplomb.nl/TechStuff/Dragon/Dragon.html for how-to-fix tips

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

@Jayjay1974: You are saying that you simply use a 7505 or something equivalent to do your voltage regulation? Don't you have problems with charge dump during starting/halting the car?

Have you some PCB/schematics examples?
Is my schematic above good enough to isolate my inputs?

In any case, if you have more documentation about power supply, it is welcome.

I found this document to do power supply for car. Is it an overkill? http://www.maxim-ic.com/appnotes...

Thank you.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Never had any problems with my AVRs during starting or shutdown, they don't notice it and a few of them are always on. I always include a 15V tranzorb and a series diode which should take care of all that. I've never tried disconnecting my battery with a running engine, or disconnecting the alternator or jump starting with 24V :D

I think the circuitry in the maxim appnote is overkill for most applications. I haven't seen any such complex circuits in the few car modules I've looked at. I'm fairly sure it's meant for safety circuits like airbag ECUs which simply should work at that single critical moment and need to be extremely immune, after all, you don't want the airbag to suddenly deploy without reason :)

Cars are extremely cost sensitive, manufacturers are not going to include all those components for each and every module (some cars up to 100 modules :?).

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

My ECU widgets have nothing more than a series diode for polarity protection, then a massively overspecced 7805 (1A rated vs the 20mA or so I'm using) with the usual capacitors on either side.

Never had a failure in a vehicle, and only one failure on the bench: a circuit board manufacturing error had shorted the output to ground.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I look around the tranzorb, but I haven't found a good documentation on it.
Could give me some details? Is it to remove the transient?
What is the input voltage range (12-48V is available?)

Thank you for your comments.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Resurrecting this thread :)
I also need to design a gadget for automotive use. I've been searching around and found several solutions, from the most complex to the most simple. One of the simple solutions would be to have a TVS and a poly-switch. However, for several reasons, it isn't practical for me to have a poly-switch. Therefore, I devised the solution below and would be glad to know your opinions.

I would use the TVS "alone", but the datasheet mentions that its failure mode is short-circuit, and I can't sleep very well at night knowing that. So I added a "small" fuse on the ground path. The circuit this will feed will draw little current compared to the 1A the 78 can handle. Also, for the TVS to handle it all alone, I would probably need some big tracks to handle the current. Note that the event that "scares me" is a load dump, which can last several hundred milliseconds according to info we can find around (noone seems to agree on the maximum voltage on a load dump, though, but 60V appears some times). The thing concerning me the most is of course security, I don't want to have a fire starter nor a component shrapnel launcher :).
I chose to put the fuse on the ground path because I'm going to have the (same) +12V coming in from other places too...

Thanks

Attachment(s): 

Embedded Dreams
One day, knowledge will replace money.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I guess it is unclear to me why the fuse is in the ground line. This would not be the traditional approach.

Is there ANY other ground path for the circuit? Perhaps through other load devices? If so, a blown fuse will not protect the circuit.

JC

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I'm also perplexed with the idea of the fuse in the ground line. Rather than use a transzorb, I'd be using a MOV as they're more robust. These tend to explode if the energy isn't limited so a series resistor or a poly switch would be a reasonable solution - or simply use a regulator specced for automotive use like a lm2940 (I think).

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Given that most engine sensors reference either to engine ground or ECU ground (connected but indirectly) I'd want to see that fuse in the positive line, too.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

There's no other ground path, but there's another 12V path. At some point, the 12V coming from the battery is split in 2 and those 2 will be used by the gadget. That's why I put the fuse on the ground, so that the entire circuit is removed from the supply in case of problems. The other 12V input is protected with series resistors and diodes to the regulated 5V. The ground fuse was an attempt to use only 1 fuse+TVS, otherwise I would have to use the same set for the other 12V inputs...

The MOVs I can get only clamp at very high voltages (95V and up) so that's not an option for me. I also didn't like the fact that MOVs explode on overload. Even using an automotive regulator, that wouldn't be enough because I still have the other 12V inputs.

Quote:
(...) most engine sensors reference either to engine ground or ECU ground (connected but indirectly) (...)

Humm... I guess I'll have to use fuse/TVS set for each 12V input, then. It's the most practical solution for me...

Embedded Dreams
One day, knowledge will replace money.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

If there's enough energy to make the MOV explode, then a TVS diode will do much the same thing, or the pcb goes up in smoke. There's no problem in getting MOVs in the required voltage range - try Farnell or RS. In fact, the Delco ECUs I've played with use a MOV - TVS diodes don't seem to be too popular in the automotive stuff I've seen for power protection. I've learnt the hard way that by adding protection devices, you need to protect your circuit in case the get activated - on low current circuits, you can simply use a resistor to limit the current, on a power circuit, this might introduce too much heat and/or voltage drop. One a board I designed recently, I had a MOV across the circuit feeding a motor to clamp inductive spikes. Worked well, until someone wired the power transformers for 110VAC and applied 240VAC - result was blown up MOVS and cooked pcb tracks. Next circuit mod is to add a polyswitch in series so the board will survive until the transformers go up in smoke or the circuit breaker trips.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

One of the problems I had was that I had no specs for the transients we can get in a car. But I finally found some, in the brochure of an automotive transient surge simulator equipment:
http://www.noiseken.com/english/...
Let me summarize below for future reference. These are worst case values. The "Duration" is the time to go from 10% of peak, up and back down to 10% of peak again; the signal is not constant at the Peak for the entire duration (except for pulse type 4), take a look at the graphs in the document.

 PulseType    Peak   Duration  RaiseTime  SourceZ  Repetition
    1        -720V     2ms       500ns     10ohm     500ms
    2a       +300V     500us     500ns      2ohm     100ms
    3a       -350V     195ns     3.5ns     50ohm     10us
    3b       +350V     195ns     3.5ns     50ohm     10us
    2b       +66.5V    4.8s      500us     ~0ohm      -
    4        +66.5V    100s      900us     ~0ohm      -
    5a(12V)  +100V     400ms     5ms       0.5ohm    30s
    5a(24V)  +200V     400ms     5ms       1ohm      30s
    5b(12V)  +40V      400ms     5ms       0.5ohm    30s
    5b(24V)  +40V      400ms     5ms       1ohm      30s

They say the equipment supports 12, 24 and 42V car systems so, some of the peaks above may not happen in typical 12V car systems (66.5V for 100s looks a bit too harsh for 12V systems...). But we can see pulses peaking 100V in a 400ms time span; from the pulse shape I can say it is above 50% peak for around 50% of the time, so it's 200ms above 50V. Seems harsh enough to me...
I attached pulse 2a waveform from the document as an example.

Ok... my problem with Farnell/RS is that I already have 2 component sources for this project and adding another one will only increase its cost due to transportation costs, that's why I was trying to use components I can get on one of my sources: TVS... zeners... but not MOVs for voltages below 95V.

Some people suggest circuits that simply disconnect the power. This could be an option, but my problem is that I need more than one... and want obviously keep my costs down (already too up).

There's something I don't understand in using a resistor to limit the current, or maybe I'm not seeing it in the right context. Let's say we have a 10 ohm resistor and a 20V clamping device. Let's say surge voltage is 70V during 100ms. Assuming that the circuit being fed takes up a negligible amount of current, the current on the resistor would have to be (70 - 20) / 10 = 5A. So at what power should my resistor be rated? For constant power, it would have to be rated at more than 5A * 50V = 250W (obviously impractical), but since it must be able to handle it for only 100ms, could it be rated for 250 * 10% = 25W only (still high, tough)? And it's already taking 2.5W at 0.5A to the load under normal work. It seems to me that this resistor method isn't quite practical in the auto environment (long and high transient); maybe that's why you Kartman say you've seen only MOVs...

Ok... assuming the higher/longer transients are rare events (the load dump that happens with an intermittent connection to a low battery), what do you thing about designing the clamping_device + low_value_resistor for the more frequent transients (lower over-voltage) and let a fuse burn above that threshold?

And for the negative spikes up to around -720V, is it enough to have a series diode with the positive power (appropriately chosen), or is this too simplistic?

So many questions... thanks for your patience guys.

Edit: I found a datasheet of an automotive semiconductor component that was tested for the pulse types described above and has specific values for the voltages, durations, and so on (they say an external TVS is needed for 40V if load dump higher than this (pulse 5b)):
http://www.st.com/stonline/produ...
See below the interesting table from the document.
space

Attachment(s): 

Embedded Dreams
One day, knowledge will replace money.