Enough filtering to car use?

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

I'm planing to use filtering as it's in a picture. What you think about it? All improvement suggestions are wellcome. Coil(L1), I just chose the biggest available in Henry's(1000µH) without any special thought about it.
Also RC-protection over the transistors wonders me. I have seen many times used RC pairs(47ohm + 100nF), so is that right?

Previously what happen without good filtering:

Regards
heguli

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

When you apply power to this circuit F1 will see a surge of current as all the caps charge up which might reduce its life.

Quote:

Also RC-protection over the transistors wonders me. I have seen many times used RC pairs(47ohm + 100nF), so is that right?

You probably do not need this, the diode should be good enough. You put in a TVS/Zener instead to provide overvoltage spike protection.

Quote:

I just chose the biggest available in Henry's(1000µH) without any special thought about it.

You should use a much smaller value inductor (if you even need it at all). 1000uH will probably cause some ringing when you connect power to your circuit.

Use a Schottky Diode for D1 if you want a lower voltage drop.

You can use larger value resistors on the base of the transistors say 1K to 4.7K

D2 is also not necessary unless you have >10uF on the output side of the regulator-- but it doesn't hurt anything to leave it there.

Hope this helps!

John

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

Wooooo ....

You didn't take half measures, Heguli ! (hoping this is a correct expression in English)
I think you will have good results with a little less filtering too.
The VDR (R3): I think it would do better when placed over C1. For C1 take a 35V version.
C2 .. C7: I think one 470 uF 25V will do fine.
For L1: 1 mH. 100 uH is fine. Mind the max. DC-current the inductor is designed for.
D2: I agree with John. It's not absolutely necessary, but I think it's a good habit. GMP. Good Manufacturing Practice.
The snubbers 47R + 100nF. When the BC547 is turned on, it has to handle the current from the 100 nF as well. And with 47R you go above BC547's limit.
A zener instead of the diode across collector-emittor will do the job for snubber and that diode. Zener 18v or 24V, 400 mW. All it has do do is to prevent that the collector-voltage rises above the BC547 spec.
And yes, as nanovate suggests: 4k7 in the base is fine.
For D1 and D2: take the ordinary 1N400x. Low-cost, simple and reliable. I don't see the need for a schottky for D1; there is plenty input-voltage left for the LM340 to regulate.

One more thing to think about: groundloops. Prevent that high currents can lift or sink ground-potential on the board in respect to chassis-ground.

I think it's very good that you put in so much effort in the supply-filter. Congratulations.
And in case you're looking for some good battery-noise/spike source: wiper-motor, interior heater fan and engine fan are the bad guys in a car. They are more disturbing than the ignition-coil.

Cheers

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

DO not forget to look at power dissipation. The LM340 in SOT-223 needs some copper area on the PCB otherwise you can put it into thermal shutdown when you try to draw too much current.

A quick calc: (13-5) * Iout = 0.8W if Iout = 100mA.

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

Quote:

C2 .. C7: I think one 470 uF 25V will do fine

It is always GMP to derate your capacitor voltages. Th 25V Nard recommended is a good value to use since it is well above your supply. Another GMP is try and place electrolytic caps away from heat sources (or hot components). Their reliability suffers if there is any localized heating on them.

Quote:

SOT-223 needs some copper area on the PCB

The datasheet has a nice section on how much you need.

Quote:

I think it's very good that you put in so much effort in the supply-filter. Congratulations.

Good job also. I think you are on your way to a working circuit. :)

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

Depending how much noise you can tolerate... I get away - i.e. no reported problems in over a hundred units - with a 47u/25v tantalum on the input and a 100n either side of a 7805, pulling a couple of hundred milliamps in a car circuit.

I do precede this with a reverse voltage protection diode - a chunky 1N4001. A tip to decrease the dissipation on the 7805 is to have two or three diodes in series - the current and dissipation is well within their capability (0.7v at your suply current) and they can dissipate energy which would otherwise be dealt with by your regulator - you're supplying a couple of volts less to the regulator.

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

Quote:
When you apply power to this circuit F1 will see a surge of current as all the caps charge up which might reduce its life.
That's ok sense device is connected to constant powerline... hopefully it has to fire up only once :D

Basically you all think that less cap's is enough, so I took off C5-C7 away.

Quote:
DO not forget to look at power dissipation. The LM340 in SOT-223 needs some copper area on the PCB otherwise you can put it into thermal shutdown when you try to draw too much current.
I checked the datasheet page 12, is it 1 sqr. inch area enough? If I need 1 ounce copper area it's going to be huge!

1 ounce = 0,02835kg
Cu = 8960 kg/m3
Cu laminate = 35µm thick copper layer

copper volume = 0,02835kg/8960 kg/m3
=0,0000031640625m3

copper volume/35µm thick copper layer
=0,090m2
=0,30m x 0,30m

Quote:
A tip to decrease the dissipation on the 7805 is to have two or three diodes in series - the current and dissipation is well within their capability (0.7v at your suply current) and they can dissipate energy which would otherwise be dealt with by your regulator - you're supplying a couple of volts less to the regulator.
Always useful way, I have done the same thing. But this time it doesn't seem to do much? Vin 13V-5V*100mA =0.8W, with one diode:
Vin 13V-0.7-5V*100mA =0.73W

Quote:
The VDR (R3): I think it would do better when placed over C1.
Done, actually I add that resistor.

Quote:
A zener instead of the diode across collector-emittor will do the job for snubber and that diode.

Quote:
You put in a TVS/Zener instead to provide overvoltage spike protection.
I'll change the RC snubber and diode to zener. Is it fast enough to kill the possible spike?

Quote:
One more thing to think about: groundloops. Prevent that high currents can lift or sink ground-potential on the board in respect to chassis-ground.
Now that issue interest me too. Sense you all think that I have too much filtering and still my last board was burned. Should I just pour a big copper area around the GND(in) pin, and all the ground tracks gather there?

Quote:
Wooooo ....

You didn't take half measures, Heguli !

Last time when I open the car doors it cost 20 euro's :lol: ...that's why I have to take some serious thinking here!

Regards
heguli

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

For the copper-area: Your calculation tells that 1 ounce copper over a square foot means it's 35 um copper, and 2 ounce --> 70 um. So your math is OK :)
Have a look over here.
https://www.avrfreaks.net/index.php?name=PNphpBB2&file=viewtopic&t=47186&highlight=
What you need to do is calculate the power in the 340, and use page 12 to see how much copper-area is needed. And it will be less than 30 * 30 cm ;)

For the groundloops: where are you mounting this device? Under the dash? or in the engine compartiment? Under the dash you would probably search for a ground-wire and use that. But what is important to think about: do not use a groundwire that could carry a lot of current. F.i. rear-window defroster, wiper-motor etc. Take a clean ground.

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

About the relay-driver: BC547 is (looking at spec's) okay for this job. Yet I'd prefer something with more balls, like BC639 f.i.
Is the use of ULN2003 an idea?

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

Quote:
Yet I'd prefer something with more balls, like BC639 f.i.
:lol: :lol: ...I'll give it some serious thoughts, just because there is a lot of more balls. Otherwise I woundn't even consider it!

...I did, what catched me was the folks who answer wrongly suffer stomach aches. I'll always answer right to you!

Could you check the grounding on the board, blue area over the upper right. Also Atmega and crystal is under groundplate ...red area in a middle.

Regards
heguli

Last Edited: Fri. Jul 13, 2007 - 03:22 PM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Quote:
I'll always answer right to you!
I see. You don't like socks or hats, do you ? :)
I tried to analyze the board, but without a schematic it's very hard ! Too hard. I'll PM you. And the Ground-potential-issue is probably not on the PCB itself. But consider this: if the ground-potential of this board is "lifted" by some other device in the car (f.i. rear-window-defroster) AND (logical AND) one of the other connections from this board is also to ground (closing the ground-loop), then turning on the defroster will create a current through this board. Pheeeuw ..... this is difficult to explain. Simular issues occur with switching PS's, audio-amplifiers etc.
In order to determine if a possible groundloop will cause problems, it is necessary to know what other in-car peripherals are connected to this board.

Btw, it looks like you are using the same 433 MHz receiver I am working with now.
And putting the SW-name in copper .... brilliant! I am used to do the opposite: In the software I describe to which assembly the SW applies.

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

Quote:

I'll change the RC snubber and diode to zener. Is it fast enough to kill the possible spike?

It will react in nanoseconds. A TVS diode (similiar to Zener) is designed for this. Take a quick look at SMAJ16A from Vishay (or others), you might like it. We use these devices for ESD protection in several of our apps.

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

I'll put the schematic also here, otherwise the thread looks silly:

Regards
heguli

Last Edited: Thu. Jul 5, 2007 - 08:31 PM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

From your burglar-alarm-board to the car-peripherals, you use relay-contacts (outputs) or optocoupler (inputs). The only groundconnection is battery-ground connection. So I think you're safe :)

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

Quote:
t will react in nanoseconds. A TVS diode (similiar to Zener) is designed for this. Take a quick look at SMAJ16A from Vishay (or others), you might like it. We use these devices for ESD protection in several of our apps.
I found from ELFA dealer this diode(CDS3C15GTA). What you think about it ...is't the same diode what are you saying? 15 volt diode can handle surge current up to 30A. And best of all, they are really cheap!
ELFA and product nr. 70-157-12.

Regards
heguli

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

CDS3C15GTA

This is bi-directional so in the case of a negative going transient you could pull the emitter above the base by Vclamp of the CDS3C15GTA. For ESD transients the duration is very small so it may not be an issue. But since you are switching an inductive load the time-constant of the transient could be much longer. I think an uni-directional diode like zener or tvs would be better.

John

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

Quote:
This is bi-directional so in the case of a negative going transient you could pull the emitter above the base by Vclamp of the CDS3C15GTA. I think an uni-directional diode like zener or tvs would be better.
You are absolutely right, didn't tought that. I'll have to stick in zener's sense my regular disti doesn't have SMAJ16A or similar available in a reasonable price. There is a transient suppressor diode(SM6T18A) manufactured by the ST Microelectonics in a horrible price.

Regards
heguli

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

Quote:

I'll have to stick in zener's

Zeners should be fine. The main difference between the two is that TVS has a larger junction area so it can handle larger peak currents. Both would be avalanche breakdown devices at the Vz you specified.

There might be TVS in other packages ... SMA/DO-214AC, SMB/DO-214AA, SMC/DO-214AC, SOD-123L and SOT-23. Also there are arrays ... NXP PESD series or Semtech or Onsemi as examples. But arrays are better suited for protecting multiple lines routed close together. But again TVS is not absolutely necessary.

PS For some reason I can not see you schematic/layout on this computer (work) so I can not look at them.

John

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

Did you tried right click over the image and show picture. Cursor might look like a magnifying glass, then left click over the picture and it's in full size. Well, it works for me and my browser is firefox.

Accidentally I order those bi-directional suppressor diodes(25 piece) instead zener's. Quite normal to me. Usually when I get my stuff from the desk I wonder how the hell order these :D ...Now I have bi-directional suppressor diode and regular fast acting diode parallel over the transistor.

Regards
heguli

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

Quote:

Usually when I get my stuff from the desk I wonder how the hell order these

I've done the same thing... more often than I'd like to admit. it is so easy to accidently add an extra digit -- smaj200A instead of smaj20A

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

Today first try with the new setup and same thing happend then previously, burned pin at atmega. PC5, which drive relay K1 wont work anymore.

Without the motor load everything works fine. If I connect all the motors it works few times, well maybe ten to twenty times and then the pin is burned.

Any ideas... I update the schematic above, it's the newest and what is now in use. I was wondering should I leave out the resistor R9 and maybe R3 as well? That way I would cut off the discharge path from the motors. Those varistors are 14VAC/18VDC.

Regards
heguli

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

Sorry to hear that !

I had a look at the updated schematics, but I am a bit puzzled: PC5 is connected to the opto-coupler, and R3 and R9 are varistors ..... What am I missing?

The current for the motors is going through the PCB-tracks. Take separate power-lines for the motors/relais-contacts.

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

Quote:
What am I missing?
You are not missing anything, I am :lol: ...PC5! Sorry, I ment PC4.
Quote:
The current for the motors is going through the PCB-tracks. Take separate power-lines for the motors/relais-contacts.
I quess you are right. I just tought I can use the same powerline then the original powerlock controlbox.

Those varistors... Why the very same pin always burn, even with the different board? Reminding that firstly relay K1 sets then relay K2. After one second K2 release and then K1 release. Why PC3-pin wont get burned sense it drives relay K2 ...because pin isn't up after cut off?
Cutting the power from the motors lowers the ground level under 0V shortly on T1 emitter? Ground track is much shorter to emitter then it's to atmega.
Positive spike doesn't kill the pin, it's the ground level. Now, how come that's possible sense K2 is released and motors don't have the powerline? ...through the varistors. Without varistors motors would be floating.

Regards
heguli

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

Does the PCB actually agree with the diagram? ie is the 1k2 resistor REALLY there as shown? Seem hard to believe that you can burn a pin with a 1K2 resistor in series.

John Samperi

Ampertronics Pty. Ltd.

www.ampertronics.com.au

* Electronic Design * Custom Products * Contract Assembly

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

Yes, the resistor is 1k2. What's the different in a board vs schematic is L1, it's now 100µH. Picture shows 1000µH.

I also checked many times shortcircuits, couldn't find any.
Removing varistors R3 and R9 makes it work. I have tried atleast hundred times and pin wount get burned. Motors seems to work reliability, I mean locks get locked and unlocked when I push the right button. Previously they worked once and after that result was what ever. Unlocking might lock the doors and vice versa.

Quote:
Cutting the power from the motors lowers the ground level under 0V shortly on T1 emitter?
I'll take this statement back, ground level must rise? ...while pin is pushing 5V... maybe :D

Regards
heguli

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

Heguli wrote:

Quote:
Removing varistors R3 and R9 makes it work.
Glad you found a solution. But I must admit it gives me an uncomfortable feeling. Because I don't understand what blew PC4 of the AVR.
The voltages in the schematic above, did you measure those with an oscilloscope ? If so ....... freaky :?

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

Quote:
But I must admit it gives me an uncomfortable feeling. Because I don't understand what blew PC4 of the AVR.
The voltages in the schematic above, did you measure those with an oscilloscope ? If so ....... freaky
no no... that picture is just my wild quess. Had to quess something sense I don't understand either what happend really. See, previous version was totally different than this one and still the same pin blow up.
Old setup on the right side, much smaller:

If you remember when I wrote about motors and how they return a bit back, you defined it as a spring effect... wont happen any more. In older version spike just didn't kill the pin, it also messed the EEPROM or during the write EEPROM it reset the chip.
Somehow I believe the varistor R9 leaked a hole lot of current and rise up the ground level on transistor emitter. Due the emitter rise also the base level rise. If base level rise above VCC while the pin(PC4) is up, it burns?
...while I write this text with my tired brains, more and more I start to be uncertain. Thanks Nard :lol:
By the way... why the pin(PC2) wont get fried sense it's also up at the same time then PC4? ...level rising is quite local?

Regards
heguli

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

Heguli wrote:

Quote:
..while I write this text with my tired brains, more and more I start to be uncertain. Thanks Nard :lol:
You're welcome :lol:

In an earlier post I wrote:

Quote:
The current for the motors is going through the PCB-tracks. Take separate power-lines for the motors/relais-contacts.
Did that help?

Nard
PS I didn't respond earlier as I missed your reply for some (unknown) reason

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

Quote:
Quote:

The current for the motors is going through the PCB-tracks. Take separate power-lines for the motors/relais-contacts.

Did that help?
Only change what I did was removing varistors R9 and R3. Still wondering why? When you are done with your RF-module project you'll be pleased with range. I can check if my car doors are closed from my balcony, distance nearly 90m... I love it!

Quote:
PS I didn't respond earlier as I missed your reply for some (unknown) reason
It happens to me also. Problem is not only with the AVR Freaks, but meny other sites loose posts too. Probably antispam works too well? ...first I tough I'm now in a "limited member" -category, only every fift post will be delivered. :lol:

Regards
heguli

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

Missing a reply .... I think it's just me causing that. Some days the list of Topic Reply Notifications is huge .... and I may overlook then one of those.

For the RF-modules: I've two decoding-programs (for different systems) finished, and am working on the third. It's fun to have an AVR responding to a ready-built system. About the range: 90 meters is very good indeed. I don't know yet how my modules will perform on such a distance.

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

Quote:

Could you check the grounding on the board

I looked but I could not see where the ground connection of the mega48 and the ground connection of the relays connect from your layout. Could your mega be floating w.r.t the relay circuit? Like JS said earlier, it is unusual that the pin fried with 1K2 ohms in series.

If it is missing then you could maybe connect the two grounds with a small resistor or ferrite bead to provide a little isolation between the higher current ground returns of the relay coils and the lower current mega48 ground.

John

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

Quote:
looked but I could not see where the ground connection of the mega48 and the ground connection of the relays connect from your layout.
Sorry John, I forgot update the layout after I change the schematic. I use 0R resistor as a jumpper, so ground should be ok... well, atleast there is a ground :)

Regards
heguli