filtering very dirty DC

So for this project, it's going on our racecar. Brief electrical synopsis:

2 high current solenoids.
High power ignition system
high power charging system
Lots of relays
and other electronic-unfriendly stuff.

The charging system is dual 16 volt batteries in parallel (16 volts at 1200 amps) with a pair of 16 volt 120 amp alternators, also in parallel (16 volts @ 240 amp peak charge).
The MSD systems inflicts a minor AC voltage into the system we've found. MSD says it's normal due to the high power it puts out.
The alternators are direct connect one-wire units, so no regulation or anything.
The high power solenoids and several many relays can induce voltage spikes on the power lines (alot of them are un-dioded, though some are).

I have a pair of 3A 5V voltage regulators that will be supplying power to the project. #378R05-C1-B21AD

So I know I need capacitors to keep the regulators happy. I also know I need a capacitor on the input lines for filtering.

My main question is, aside from capacitor filtering, do I need to add diodes anywhere? I was thinking to put a diode on the power and ground lines, so current can only flow in on the pos cable, and out on the neg cable. This would block out the voltage spikes from the solenoids, as well as any AC voltage off the MSD box, correct?

The solenoids should all have flyback diodes if they don't already. The relays are also inductors, so they need flybacks also. Make sure the diodes are current rated/not already burnt out. If you don't have these, adding them will clean things up a great deal. If you already have them, stick around, I'm sure someone will have better ideas soon..

I think only 7 relays on the car have flybacks....out of 18 or so. Most of the relays have been added on with quick connects to get signaled by buttons and timers and have less than half an amp @ 16V on the coils.

The shift solenoid pulls 8 amps @ 16V

The transbrake solenoid pulls 5 amps @ 16V

The transbrake I could add the flyback at the timer terminals, not sure where I'd add one for the shifter, as it's powered off a relay.

There's also the alternator solenoids and the starter solenoid, none of which can get flyback'd (120 amps at 16 volts each to charge and >600 amps @ 16 volts to start) plus the starter and alternators themselves.

Also, the 2 solenoids, plus the shift relay all have to activate/deactivate quickly (.01 of a second can lose a race) so flybacking them is not the most practical in the world. Also, this device will be going between multiple cars, and not all are built the same. So flybacks are to be expected and suppressed. All other racing electronics suppress it, else they wouldn't work at all.

Not an expert on racing but here we go. They don't need to be rated for the current the solenoid is pulling. They are not wired in series with the solenoid. These flyback diodes are wired parallel to the inductive load and in an OPPOSITE direction to short the high-voltage spikes induced when current stops flowing through the solenoid.

These currents are not big but the voltage could get pretty high (hundreds or thousands of volts), if not shorted. As long as the flyback diode is connected to the solenoid, it shorts all negative spikes from the coil and prevents the high voltage buildup.

Using a Schottky diode would give a signifiant speed increase as well. Regular PN diodes switch in the hundreds of nanoseconds, Schottky diodes switch virtually instantaneously (0 rebound time.)

As for suppressing voltage spikes other ways, I can't think of another way that won't be either much slower and also less efficient.

BTW, why are you worried about noisy DC? Is it interfering with interoperation between components?

I haven't finished the project (which is going to be used in technical inspection across multiple cars in a league), but I don't know how "dirty DC" will effect the voltage regulators or arduino, especially if some of the AC voltage off the MSD hits it. Our data logger shows over 23 volts at times, on a 16 volt system. I chose a regulator with a 35 volt maximum input (regulating down to 5 volts), one regulator (for 56 LED's) has a 5.8 C/W heatsink on it, the other (39 LED's + arduino) has a 7.8 C/W heatsink on it. I'm building this for the extremes.

Get a scope and see how dirty it actually is. Do an FFT (if your scope can do those) and see where the noise components are and design a low pass filter to eat it: Low-pass filter - Wikipedia

Filter it on the in and the out for double plus good filtering.

Put a 1kOhm resistor with a 4700 uC capacitor and it'll filter anything with more than 0.033 Hz. It will take a few seconds to "power up" since rise time is 10.8 s. Meaning you hook it to your batteries, it'll take 10 s for the 1st cap to hit 90% of 16V. As soon as the voltage hits the cutout voltage of the regulator the second cap will start 'charging'. It'll also give you a few seconds if for some reason power cuts out.

darkscout:
Get a scope and see how dirty it actually is. Do an FFT (if your scope can do those) and see where the noise components are and design a low pass filter to eat it: Low-pass filter - Wikipedia

Filter it on the in and the out for double plus good filtering.

http://www.pronine.ca/rccir.htm

Put a 1kOhm resistor with a 4700 uC capacitor and it'll filter anything with more than 0.033 Hz. It will take a few seconds to "power up" since rise time is 10.8 s. Meaning you hook it to your batteries, it'll take 10 s for the 1st cap to hit 90% of 16V. As soon as the voltage hits the cutout voltage of the regulator the second cap will start 'charging'. It'll also give you a few seconds if for some reason power cuts out.

I don't have a scope, but this will be going in different cars (that aren't ours) as it's for tech inspection. So relying on clean power isn't the best thing in the world. I just have to be able to filter it. That's why I thought just having a diode on the pos wire and a diode on neg wire would be sufficient?

If the relays and solenoids are electronically switched, they MUST have protective diodes backwards across the coils to protect the switching circuit. If you are worried about signal noise and that induced 30Vac, put .1uF and 30-50uF caps in paralell across your power buss (positive feeds to ground) in as many points as you can, particularly a set close to the arduino power feed. That will take care of switching transients and the induced ac. The closer that you can get the .1uF caps to the switching sources, the better.

magnethead794:
I don't have a scope, but this will be going in different cars (that aren't ours) as it's for tech inspection. So relying on clean power isn't the best thing in the world. I just have to be able to filter it. That's why I thought just having a diode on the pos wire and a diode on neg wire would be sufficient?

Diodes aren't filters. You can turn them into rectifiers if your voltage ever goes negative (you'd have bigger issues at this point).I'd say get some big caps and filter the DC. The bigger the better they'll eat a ton of junk.

Big caps will take care of the slow stuff, but you need small ones to take care of the fast spikes too.

123Splat:
Big caps will take care of the slow stuff, but you need small ones to take care of the fast spikes too.

Big caps will take care of the slow stuff and the 'fast' stuff. "fast stuff" is just a high frequency spike.

RC circuits are 1st order systems. With a cut off frequency of 0.3 Hz at 0.3 Hz you're at 1/2 power. (-3 dB) and it just rolls off at 20 dB/decade from there. -23 dB at 3 Hz. -43 dB at 30 Hz. I'd say that'd more than take care of "fast spikes".

I was going to use the diodes to clean out the AC signal that's present in the system. Then capacitors will clean up the DC enough for the voltage regulators not to fluctuate?

If you have an AC component of 10V-pp and a DC offset of say 20V. You're going to get a sine wave that goes 15-25V. A diode won't clean that up since voltage never reverses. Rectifier - Wikipedia

If they're introducing an AC component enough to bring your total voltage negative, that's not good.

darkscout:
If you have an AC component of 10V-pp and a DC offset of say 20V. You're going to get a sine wave that goes 15-25V. A diode won't clean that up since voltage never reverses. Rectifier - Wikipedia

If they're introducing an AC component enough to bring your total voltage negative, that's not good.

I thought that by nature, AC voltage takes the negative lead pos and positive lead neg? Or is that what you mean by negative voltage?

magnethead794:
I thought that by nature, AC voltage takes the negative lead pos and positive lead neg? Or is that what you mean by negative voltage?

Say you have a perfect 20V DC voltage. Then you start getting leakage from the ignition system. Say that's 2sin(ft). So now your signal is

20+2sin(ft).

Say you're doing this in a work shop with some overhead florescent lights, you're going to get some noise from the line voltage in the walls at .1sin(2pi60t), then there is noise from random other sources. Eventually you'll get a "full" fourier decomposition of the signal.

20+2sin(ft)+.1sin(2pi60t)+.01sin(1000t)+....

The NET signal is still positive so the diodes won't do anything other than make sure the signal flows in one direction. This is where the lowpass filter comes in.

(f can easily be figured out from firing frequency)

Is there a reason the Arduino can not run on it's own batteries? Depending what you are doing with it you can run a long time on AA's or even some Lipo batteries. Batteries would mostly isolate the Arduino from all the problems listed.

I have used a lcd module, gps, and Arduino Uno on a 9volt for more than 2 hours + testing. AA batteries would last longer.

darkscout:
Caps in parallel are just going to reduce capacitance by (R1*R2)/(R1+R2).

Not exactly.

Why are you using R?

See: Capacitors in Series and in Parallel

Here, we have made use of the fact that the voltage V is common to all three capacitors. Thus, the rule is:

The equivalent capacitance of two capacitors connected in parallel is the sum of the individual capacitances.

But also: http://arduino.cc/forum/index.php/topic,82030.0.html

cyclegadget:
Is there a reason the Arduino can not run on it's own batteries? Depending what you are doing with it you can run a long time on AA's or even some Lipo batteries. Batteries would mostly isolate the Arduino from all the problems listed.

I have used a lcd module, gps, and Arduino Uno on a 9volt for more than 2 hours + testing. AA batteries would last longer.

IMO for this purpose, batteries would just complicate things and drive cost up where it doesn't need to be. Right now I have about $73 into something that can be bought for $450. And the cheaper the better.

I'll just put a 10uF capacitor across the input leads and hopefully that will filter things out. The regulators are rated 35 volts max and put out 5V at 3 amps, I have one with a 5.8 C/W heatsink and the other with a 7.5C/W heatsink. So they should be able to take the full voltage that our logger is reading.

justjed:
Not exactly.

Why are you using R?

New baby sleep deprivation.

You are correct:
Parallel:
Resistors: 1/R=1/R1+1/R2
Caps: C=C1+C2

Series:
Resistors: R=R1+R2
Caps: 1/C=1/C1+1/C2