Needing to filter Capacitive Discharge feedback

[magnethead494]

I'm trying to put some LED's on our racecar. They're 7 segment LED panels where each segment is 5 LED's with a 220 ohm resistor on the end.

The car runs on 12 volts with engine off and 14 with engine on.

I have power (V_in) coming off a distribution block, ground also goes to a distribution block.

Vin then goes to a 512-KA378R12CTU 12 volt regulator, comes out, and splits off to a screw terminal and to a 511-LD29150DT50R 5 volt regulator, which sends 5 volts to an atmel 328P TQFP package. I have 47uF tantalum, 10 uF tantalum, 1 uF, and 0.33uF capacitors to clean up the input/output voltages.

The board works as designed (querying binary coded decimal switches and sending a duplicate readout of the LED panel to an indicator display) with th engine is off (water and fuel pumps don't affect anything), however, once the engine is running, the display goes flat blank, EXCEPT some tiny, tiny pulses (1/500 sec maybe?) every so often. Shut the engine off, goes back to normal. I put my multimeter in ammeter mode, and it recorded 0.1 amps between the Vin lead and a power source. The board pulls 0.4 amps with the ignition off.

I put my multimeter on frequency settings, and I get zero hz with nothing on or with all the pumps on. Once the engine is running, I get a 1.2 Khz wave in the system (presumably sinusoidal but without an oscilloscope I don't know...).

I was powering the displays from a distribution block very close to the ignition box, so I made a cable going directly to the battery posts (4 ft longer cable). This made things significantly better, now only the 3rd digit (the last on the daisy chain of power) flickers on/off every second or so.

I'm guessing I need to make some sort of filter circuit, but I'm not really sure?

 

[NascentOxygen]

I'd guess that one of the regulator ICs is reacting badly. Try more electrolytics across their input terminals. Try 4700uF and 47uF and 0.47 uF all in parallel, 35 volt rating. Electrolytics don't like heat, so locate in a ventilated spot or they may not last long.

 

[magnethead494]

I'd guess that one of the regulator ICs is reacting badly. Try more electrolytics across their input terminals. Try 4700uF and 47uF and 0.47 uF all in parallel, 35 volt rating. Electrolytics don't like heat, so locate in a ventilated spot or they may not last long.

Nascent,

Given the decoupling caps on the power supply board and on the controller, I have no decouplers on the LED boards themselves.

The 5 volt mcu and LED indicator don't flicker, and they run as a subset of the 12 volt circuit. This leads me to believe that the power supply board is working properly, and that I need to focus my attention on the LED panels.

I don't know if the CD ignition box or the alternator or both are causing the 1200 hz wave (and I don't have an o'scope), but as best as I can tell, that's the only thing that changes when the engine is running.

Not shown in these pictures is the 47uF tantalum that was added in parallel to the 0.33uF cap.

The 1.0uF decouples the voltage before it gets to either regulator

The 47uF tantalum decouples the output of the 12 volt regulator, while the 0.33uF decouples the input of the 5 volt regulator (they are in parallel)

The 10uF tantalum decouples the output of the 5 volt regulator. That's really the only element that the 5 volt line sees (besides the .1uF's on the controller) that the 12 volt line doesn't- maybe that's all I need on the other side?

On the controller (not shown), the 328P has 2 different 0.1uF decoupling caps on it, on VCC and AVCC.

http://www.mouser.com/ProductDetail/Lelon/REA0R1M2ABK-0511P/?qs=sGAEpiMZZMtZ1n0r9vR22Z8b58wL7Nwv4bv%2fa21%252beXo%3d
http://www.mouser.com/ProductDetail/Lelon/REA010M2ABK-0511P/?qs=sGAEpiMZZMtZ1n0r9vR22YgcI%2fJYgoutoRXBBckjC2Y%3d
http://www.mouser.com/ProductDetail/EPCOS/B41827A6106M000/?qs=sGAEpiMZZMtZ1n0r9vR22b0Wmj03gb3uE3xfdGLFOIs%3d
http://www.mouser.com/ProductDetail/Nichicon/UVR1H470MED1TA/?qs=sGAEpiMZZMtZ1n0r9vR22afkwEr%252bqyzvsXww%2fp7SF0M%3d

http://www.mouser.com/ProductDetail/AVX/TAP104K035SRW/?qs=sGAEpiMZZMtZ1n0r9vR22SKv3nCGDP6cmjwZVaWgu7k%3d
http://www.mouser.com/ProductDetail/AVX/TAP105K035CCS/?qs=sGAEpiMZZMtZ1n0r9vR22X84dCiTW0OjwZB4uN8T5Aw%3d
http://www.mouser.com/ProductDetail/AVX/TAP106K035SCS/?qs=sGAEpiMZZMtZ1n0r9vR22Xu88swJg3nY7eBNkT2YtbA%3d

 

[jim hardy]

I don't know if the CD ignition box or the alternator

would it be worth disconnecting alternator belt (or field wire) and running for a few seconds?

 

[sophiecentaur]

Ferrite chokes do a good job of tidying up waveforms, too. This applies to the signal lines as well as the power lines to your electronics.

 

[magnethead494]

going to stack the electrolytic capacitors on the connections and put a choke on the power wire. the .1uF has a 3milliAmp ripple current spec so I'm debating to put it on there, 1uF has a 22mAmp, 10 is 50 something, 47 is well over 100.

 

[sophiecentaur]

Did you think of buying a suitable filter unit? You can get them as bulkhead fittings instead of a simple feed-thru and the cost may not be all that high for the 'right thing'.

 

[jim hardy]

Most car radios have a filter in power line to keep out "alternator whine"
and most junkshops have junk car radios.

It's just a simple LC filter, probably the inductor from a speaker crossover would do.

We're pullin for you !

 

[magnethead494]

I'm 2 blocks from mouser headquarters, I can get anything at the will-call office within an hour of ordering.

So here's what I came up with today. Notices i said our car...I should have said a friend's car. Our car already has one.

Using regular 18 gauge doesn't work at all. I have to use my 6x22gauge wire bundle. It's not shielded, but I think the 4 unused wires are acting as a shield.

The ignition box gets power from the barrier strip, which is 4 feet of 14 gauge from the battery. The charging posts on the back of the car go directly to the battery via 4 gauge cable.

single 18 or 6x22 don't work at the barrier strip
single 18 or 6x22 don't work directly at the battery terminals
single 18 doesn't work at the charging posts, 6x22 does.

I stacked 2 47uF, 2 10uF, 2 1.0uF, and 2 0.1uF all together on the input line, slowed down the flashing just a bit, maybe from a couple thous of a second to maybe a hundredth or so per second (.01 on .99 off).

What we did for testing, was relay the board off the charging posts, switching with the ignition switch. I still get 1200 hz at the input terminals of the power supply board, but the LED's stay steady.

Something to think on- the ferrites are making a HIGH pitched hum/whine- around 15kHz or so. If I unsnap them, it goes away. I have one on the power lead for the power supply and one on the power lead for the ignition box.

We're going to do a quick test rig on his second car tomorrow, so see if it's roadster or if it's something I need to make a design compensation for. He has 4 cars and bought a display for each one. Just makes me really nervous not being able to start the other car I've already put one on.

on his car:

https://fbcdn-sphotos-a.akamaihd.net/hphotos-ak-snc7/417024_10150856967159606_771454605_12582084_408841564_n.jpg

on another car (couldn't start it, may or may not work?!):

https://fbcdn-sphotos-a.akamaihd.net/hphotos-ak-ash4/417181_10150849242164606_771454605_12558482_690369463_n.jpg

on our car (works perfectly fine):

https://fbcdn-sphotos-a.akamaihd.net/hphotos-ak-snc7/431409_10150829834529606_771454605_12489056_227043297_n.jpg

 

[magnethead494]

So we hooked up his second car today, and it works just fine, we tested it with the LED boards sitting on the ignition box with the car running. I didn't get a frequency measurement though. The roadster has an older msd 6-AL box in it, the others have a slightly (maybe 1 year) newer 7AL2 box. So it's something with that specific ignition box.

 

[magnethead494]

Also, on the roadster with the issue, I put a piece of sheet metal between the MSD and the power supply, and that didn't help any. So I don't think it's something coming as magnetic waves, but since I'm getting the same 1200hz signal at the charging posts with it working, I don't think it's anything in the electrical signal, either.

I'd wonder if it's a ground loop or something of the sort, but it doesn't even work when tied directly to the battery.

Based on a calculator I found:

4 gauge spaced 50mm apart:
9pF/m
1.3uH/m DC
1.2uH/m HF
355 ohm HF

x 2/3 meter = 6pF, .8uH DC, .72uH HF, 250 ohm HF22 gauge spaced 1mm apart:
24pF/m
.57uH/m DC
.46uH/m HF
141 ohm HF

x2 meters= 48pF, 1.14 uH DC, 0.92 uH HF, 282 ohm HF

18 gauge 1.5mm apart:

26pF/m
.53uH/m DC
.43uH/m HF
129 ohm HF

x 2/3 meter = 18pF, 36uH DC, 28 uH HF, 76 ohm HF

x 2 meter = 52pF, 1.06 uH DC, .86 uH HF, 258 ohm HF

The only working combination is 2 meters of the 22 gauge with 2/3 meter of the 4 gauge..

6pF, .8uH DC, .72uH HF, 250 ohm HF
48pF, 1.14 uH DC, 0.92 uH HF, 282 ohm HF

check my math:

5.3333 pF, 1.94 uH DC, 1.64 uH HF, 532 ohm HF

 

[NascentOxygen]

Something to think on- the ferrites are making a HIGH pitched hum/whine- around 15kHz or so. If I unsnap them, it goes away.

That shows they are working! It also confirms that there is high frequency current in that cable. Did you just clamp the ferrite cores onto the cable? The use of ferrites I've seen had the cables looped through the cores 3 or 4 times, to increase the inductance. Try that.

Ferrite is magnetostrictive, it expands and contracts slightly in time-varying magnetic fields.

 

[magnethead494]

That shows they are working! It also confirms that there is high frequency current in that cable. Did you just clamp the ferrite cores onto the cable? The use of ferrites I've seen had the cables looped through the cores 3 or 4 times, to increase the inductance. Try that.

Ferrite is magnetostrictive, it expands and contracts slightly in time-varying magnetic fields.

The multiconductor I had just going through it once since the jacket OD matched the ferrite ID. The 22 gauge I had looped 3 times.

 

[jim hardy]

So it's something with that specific ignition box.

Something to think on- the ferrites are making a HIGH pitched hum/whine- around 15kHz or so. If I unsnap them, it goes away. I have one on the power lead for the power supply and one on the power lead for the ignition box.

That tells me they BOTH are passing HF current.
Which suggests that the ignition box takes huge gulps of current at that frequency
and it is easier for Mr Ignition (who's a sort of brute force guy) to get them from your LED's filter caps than out of Mr Battery.
That's because Mr Battery's wires are spaced wide so have considerable inductance.
That makes it easier for the HF current to come right out of Mr LED's filter cap, it has lower impedance
so feeds backward into what looks to it like an external fault.
Proverbial candy from a baby?I had a similar situation once on a huge 130 volt battery bus. It was modulated by 3 inverters(~65 amps each) making over 100 volt peaks at 120 hz. They wouldn't run if not in synchronization.

I would bet a beer that if you open Mr Ignition you'll find his electrolytic filter cap on input has developed extreme ESR which let's him modulate your battery voltage to point Mr LED can't run on it.
Tack a huge electrolytic right across Mr Ignition's terminals, and if both ferrites quiet
down you have established probable cause to open him up.

old jim

 

[jim hardy]

I think you're an electronics guy
you will appreciate these capacitors.

http://www.vishay.com/docs/28336/120atc.pdf

 

[magnethead494]

So I went back to that dragster's shop so we could start it. Did the same thing as the roadster- blinked (has a Mallory ignition box, completely different). So I stacked a 1000 uF and a 470uF in parallel at the LED board itself, and it took the frequency down to 20 Hz, and slowed down the blinking rate to about 3 tenths of a second on and 7 tenths off.

So you're saying I need to tack inductors and make an LC filter?

On this one, I'd say there's probably 14 feet of the 22 gauge multiconductor. And the box mounts right above the coil and distributor, probably doesn't help but I have to make it work.

On this dragster:

no caps- 230-330 Hz flunctuating
58uF of stacked caps- 86-98 Hz flunctuating
1470 uF of caps- 20Hz fairly steady

22 gauge spaced 1mm apart:
24pF/m
.57uH/m DC
.46uH/m HF
141 ohm HF

x4.25 meters= 102pF, 2.42 uH DC, 1.955 uH HF, 599.25 ohm HF

but I don't know if the waveform is triangular, square, or sinusoidal since I don't have a scope.

 

[magnethead494]

reading this: http://www.allaboutcircuits.com/vol_2/chpt_8/2.html

and this: http://linuxcar.sone.jp/reg.en.html

and this: http://www.play-hookey.com/ac_theory/ps_filters.html

I am borrowing a spare of the offending units, an MSD 6AL, to bench test with. It is rated at 1 amp draw per 1,000 RPM. I will hand-turn a spare distributor to fire off the ignition, running off a small 12V gelcel, to replicate the issue. I can then use a drill to gain speed. The issue shows up at a measly 500 RPM, which is only 250 RPM of the distributor, or about 4 rotations per second.

Things I see on the second to last image: 150uH inductor, 20V 600W Zener (is that really necessary?), 1000uF decoupling cap before regulator, 2200 decoupling cap after regulator.

seeing as 1470uF brought it down to 20Hz, The next step of 2000+uF and an inductor is all I need?

----------------

http://www.mouser.com/ProductDetail/Fastron/HBCC-101J-02/?qs=sGAEpiMZZMueR%252blcXtRMwTF%2fyJ2D16pKJ2Kbgoo7HPU%3d
http://www.mouser.com/ProductDetail/Bourns/5800-151-RC/?qs=sGAEpiMZZMueR%252blcXtRMwXbDwFg%252bO9QWnKibw9JhM6M%3d
http://www.mouser.com/ProductDetail/Bourns/5800-221-RC/?qs=sGAEpiMZZMueR%252blcXtRMweTvlE%2fCoLxyD6jdE9zdVC8%3d

http://www.mouser.com/ProductDetail/Vishay-BC-Components/MAL202117102E3/?qs=sGAEpiMZZMtZ1n0r9vR22ciSY6ngmHpL2xflyuHcHrs%3d
http://www.mouser.com/ProductDetail/Vishay-BC-Components/MAL202117152E3/?qs=sGAEpiMZZMtZ1n0r9vR22ciSY6ngmHpLrF%252bfbmNQIFU%3d
http://www.mouser.com/ProductDetail/Panasonic-Electronic-Components/EEU-TP1V202/?qs=sGAEpiMZZMtZ1n0r9vR22RLWfCw%2fHjMd4a0byORSsPc%3d
http://www.mouser.com/ProductDetail/United-Chemi-Con/ELXZ350ELL392MM40S/?qs=sGAEpiMZZMtZ1n0r9vR22fqp084l1X3esgjl60Q7HGo%3d

 

[magnethead494]

playing with http://www.calculatoredge.com/electronics/ch pi low pass.htm

for a 10 hz .05 ohm 3db 3 component setting, says 566uH inductor and 1F cap.

I must be missing something...

 

[magnethead494]

So I have an MSD going to a gelcell battery, ect.

the 12 volt LED's flicker as expected when I spin the distributor, but the whites don't...asterick.

If I turn the booster down on the white LED's, then they'll start flickering. I know the spot I have marked is 17.5 volts.

I have the filter configured as a pi filter, 3900 uF input, 470uH inductance, 3500uF output.

on a related note: This would make a fun halloween project with 40,000 volts floating around from the coil firing... FYI that's what is causing the flicker. The MSD is eating all the juice from the battery and screwing with the supply for the LED's.



https://fbcdn-sphotos-a.akamaihd.net/hphotos-ak-ash3/548085_10150875048389606_771454605_12643308_1791448242_n.jpg

 

[magnethead494]

So the guy with the roadster went from an electric fuel pump to a belt drive fuel pump, and now the LED boards are displaying random segments, like the shift registers are getting some sort or crap signal. But I don't know if it's coming from the controller or from the power line. The shift registers are just getting the signal at 5 volts and using the signal to switch the 12 volt transistor loads, more or less.

 

[jim hardy]

have you tried filtering at MSD, post 14?

 

[magnethead494]

It did not help. I used the 3900uF.

 

[jim hardy]

It did not help. I used the 3900uF.

Ohhhh, bummer. Sorry about that.

Do you think it's picking up radio-like interference from the sparkplug wires?
Automobiles made of metal have a ground-strap to the hood to separate them from radio antenna,
and i notice 1950's Corvettes have antenna on rear fender far from igmition..

http://americanclassicars.com/wp-content/uploads/2010/10/1956ChevroletCorvette.jpg

http://americanclassicars.com/wp-content/uploads/2010/10/1956ChevroletCorvette.jpg

 

[magnethead494]

Ohhhh, bummer. Sorry about that.

Do you think it's picking up radio-like interference from the sparkplug wires?
Automobiles made of metal have a ground-strap to the hood to separate them from radio antenna,
and i notice 1950's Corvettes have antenna on rear fender far from igmition..

http://americanclassicars.com/wp-content/uploads/2010/10/1956ChevroletCorvette.jpg

http://americanclassicars.com/wp-content/uploads/2010/10/1956ChevroletCorvette.jpg

no, that issue was related to solid core wires, which nobody uses anymore due to high resistance and EMI.

I'm taking the whole ignition setup to the university next week and putting it in an EE lab with a professor's assistance. Going to scope it and try some simple things. I found a 12->12 @ 1/2 convertor/isolator, but they're $15 each...ouch.

Could I replace the base 12 volt regulator with it though? (16-12)*.470 = 1.88 dissipated watts. (12*.470) = 5.64 drawn watts.

http://www.mouser.com/ProductDetail/Cincon/EC4A02H/?qs=sGAEpiMZZMtwaiKVUtQsNVEV%2fwrjnskQvJKQLG%252b%2fma8%3d

 

[magnethead494]

didn't try that regulator/convertor/isolator, however, did try a $5 27,000uF capacitor from Mouser, on an O-scope.

the green line is without the 27,000 uF cap, the red is with it (slightly less amplitude than shown, unknown wavelength difference)

https://fbcdn-sphotos-a.akamaihd.net/hphotos-ak-ash3/546200_10150904649719606_771454605_12753056_1003185896_n.jpg