Whining noise from circuit board

[hpmaxim]

I designed a circuit board that primarily consists of digital logic ICs, with a few analog ICs and a 7805.  All parts are SOIC (except the 7805 which is still SMD), and automotive rated (from Digikey).  Other than the ICs, the board has ceramic caps, resistors, diodes, and one 2n3904, all surface mount.  The board is designed to be operated off a car battery (on a running car).  I am not using any electrolytics as I expect the currents involved to be very low, and the 7805 datasheet did not recommend any large capacitors and would generally prefer to avoid electrolytics for cost, reliability, and size reasons.  The board functions fine, and I did not see any significant ripple on the output of the 7805, but I can hear a high pitch faint whining noise when the board is powered.

I think of whining as something that occurs as a result of inductance, and I doubt there is any significant amount on the board (although there may be significant inductance in the wires going to the board).  The chip has digital outputs that are coming from 74HC chips.  The inputs are going into a 74HC chip but have significant amounts of protection on them (series resistor and shunt capacitor and TVS diode). Can anyone give me any suggestions here as to the origin of the whining, if its something I should worry about, or how to correct it?

[Fungus]

Ceramic capacitors can make audible noise, maybe that's the source:

https://www.google.com/search?q=ceramic+capacitor+noise

[madires]

Please see http://media.digikey.com/pdf/Data%20Sheets/Panasonic%20Capacitors%20PDFs/ECJ,ECY,ECD_Handling1.pdf section 1.8 Piezoelectricity.

[wraper]

Could be that 7805 or something else is oscillating and therefore ceramic capacitor on it's output is emitting noise.

[hpmaxim]

wraper, I dropped a scope probe on the output of the 7805 and did not notice a significant output ripple (I have 10uF 1206 on both input and output as well as .1uF on the output and decoupling each IC).  There is almost certainly significant ripple on the input.

Sounds like the noise is completely normal, not harmful and there are ways to mitigate the problem (turning the thing vertical, mounting a pair of caps above and below the board to cancel out the vibration, or using more expensive noise reducing caps) -- but all of which would cost money and are not worth it.  If the noise doesn't pose a long term reliability issue, I'm not concerned with it.  I suspect that once packaged the noise would be barely noticeable.

[wraper]

wraper, I dropped a scope probe on the output of the 7805 and did not notice a significant output ripple (I have 10uF 1206 on both input and output as well as .1uF on the output and decoupling each IC).

What do you mean by significant, what is the amplitude? There shouldn't  be almost anything at all, especially if powered from stable power supply. What is the frequency of the ripple?
Capacitors should not whine unless there is some DC/DC converter. Did you probe on AC setting to see in millivolts scale? 7805 might not like at all 10uF of very low ESR capacitance on it's output.

[wraper]

Sounds like the noise is completely normal, not harmful

...the words you will remember when 7805s will start to fail short, destroying everything what is powered from them.

[hpmaxim]

wraper, I dropped a scope probe on the output of the 7805 and did not notice a significant output ripple (I have 10uF 1206 on both input and output as well as .1uF on the output and decoupling each IC).

What do you mean by significant, what is the amplitude? There shouldn't  be almost anything at all, especially if powered from stable power supply. What is the frequency of the ripple?
Capacitors should not whine unless there is some DC/DC converter. Did you probe on AC setting to see in millivolts scale? 7805 might not like at all 10uF of very low ESR capacitance on it's output.

I don't remember, it was a few weeks ago.  I don't think I probed on AC.  I was not explicitly trying to make a measurement of the ripple.  I was having some sort of problem with the board (which I think had more to do with my inputs than the board itself) -- I eventually got everything working without any board modifications), so I started probing around, and I remember I checked the output of the 7805 and it seemed completely stable at ~5V, however I was most likely probing it in DC mode on 1 or 2V/div. 

This is the datasheet for the 7805: http://www.st.com/web/en/resource/technical/document/datasheet/CD00000546.pdf

It specifies a minimum output cap of 2.2uF, typical of 10uF, no max.  However, it says the ESR should be between .1 and 10 ohms.

The capacitor is: http://datasheet.octopart.com/GMK316AB7106KL-TR-Taiyo-Yuden-datasheet-27088958.pdf

So ESR is < .1 ohm from about 6 or 7kHz to over 100MHz.  That might be a problem.  But it might not...  The output should be nearly DC, in which case the I don't know how serious that is.  Minimum ESR seems to be around .007 ohms, but that's at around 1MHz.

There is no reason for snide remarks however, my response that it sounds normal was based on the fact that I was unaware of the fact that ceramic capacitors can make a whining sound when attached to a PCB and that there are mitigation techniques available.  You are right that the output of the linear regulator shouldn't have ripple on it, and therefore it shouldn't be making noise.  But I have the same cap on the *INPUT* and the input is definitely subject to ripple (much more so than on the output), so its quite possible that the singing capacitor is the input cap, not the output cap, and its not an issue with the 7805.

[max666]

Would the whining stop if you put your finger on it? If so that could be simple method to find the culprit.

[dom0]

You won't see a 7805 oscillate at 1 V/div, even less so at 2 V/div.

[hpmaxim]

What sort of oscillation would be required to get a 1206 10uF cap to start making noise?

[Fungus]

There is no reason for snide remarks however, my response that it sounds normal was based on the fact that I was unaware of the fact that ceramic capacitors can make a whining sound when attached to a PCB and that there are mitigation techniques available. 

You can try different brands/values of capacitor to see if it helps. Maybe even parallel it with a small electrolytic to change the characteristics of the input capacitance.

[hpmaxim]

There is no reason for snide remarks however, my response that it sounds normal was based on the fact that I was unaware of the fact that ceramic capacitors can make a whining sound when attached to a PCB and that there are mitigation techniques available. 

You can try different brands/values of capacitor to see if it helps. Maybe even parallel it with a small electrolytic to change the characteristics of the input capacitance.

If the noise is harmless, I'm not going to do anything.  If its indicative of a problem, I will try to figure out what is going on.  It sounds like it could either be normal because of ripple in the input, or a problem because of low ESR in the output cap.  I guess a low ESR cap would tend to make more noise.  I guess its conceivable shoving a 10 ohm resistor in series with the input cap could cut down on the noise too.

[rob77]

i wouldn't bet on that noise being harmless...  the capacitors are making noise because of AC content (they act like a piezo-speaker) , if there is some level of AC after a 7805 powered from a battery, then it means it's oscillating... and an oscillating 7805 tends to be harmful sometimes

[Fraser]

A noisy whining PCB is never 'normal' in production equipment. Such oscillations should always be tracked down. Oscillations can be very harmful especially if they can build in amplitude. Resonance can be quite a destructive force.

In your situation I would a) use a straw in my ear to listen to each component and assess exactly which is the source of the sound and b) apply a load to the supply to see if that stops the resonance. Some supplies do oscillate without a decent load. In such cases it is not unusual to see a small 'load' resistor across the output. If the capacitors are making the noise, I would change them out for a different make or type. I personally would also change the regulator to one from another manufacturer as some regulators like to oscillate yet others are fine.

For there to be sound, there must be oscillation somewhere. Oscillation in a power supply is bad, no two ways about it.

Aurora 

[Fungus]

If the noise is harmless, I'm not going to do anything.

A vibration in a device (which is what this is) is never 100% harmless.

Think of it as your capacitors screaming if that image helps.

[Fungus]

In your situation I would a) use a straw in my ear to listen to each component and assess exactly which is the source of the sound

Or use a BIC pen. 

[amyk]

For there to be sound, there must be oscillation somewhere. Oscillation in a power supply is bad, no two ways about it.

...unless it's a SMPS, in which case excessive oscillation is bad.

The OP mentioned digital logic - the oscillation could just be the switching transients.

[mikerj]

Aside from the possibility of the 7805 regulator oscillating, this part is not rated for automotive use and I know from experience that without a well designed input protection scheme they can fail from the high voltage transients that are often present in car electrical systems.

There are TO220 5v regulators specifically designed for automotive environments, e.g. the LM2940.

[hpmaxim]

Noise appears to still be there when I run it off a bench supply.  I was unable to isolate the sound (acoustically).  There appear to be several hot spots on the chip (primarily on a couple of the analog ICs and the 7805) however the 7805 is hot enough that I can't individually make out the 1206 caps (heat wise) although they are on the perimeter of the hot spot, so I'm guessing they aren't getting very hot.

However, I dropped a scope probe across the output cap with AC coupling, and it appears that there is a sawtooth ripple with a magnitude of 400mVpp and a period of about 70uS.  Note that the inputs were left open which I would have expected to cause excess current draw.  The sawtooth was highly non-symmetric.  My guess is that the 70us period is consistent with the noise I'm hearing.  I assume this means the ESR of the output cap is too low and its causing the 7805 to become unstable?  Is adding a 1 ohm resistor a valid solution as most caps in that size seem to have similar ESRs?

mikerj, I said it was a 7805, in fact its not, you can look at the datasheet I provided, it is essentially an automotive rated version of a 7805. 

amyk, it is conceivable its the digital logic, however, I doubt it.  The noise was there whether the inputs were stimulated or not, and the input stimulation was at frequencies much lower than the noise.  I suppose its possible its randomly oscillating when the inputs are open and when i'm hearing harmonics when they are connected.  However, the caps are decoupling, and I can't imagine the supply moves all that much as a result of the output of HC logic switching.

[Niklas]

Even if the output voltage is close to DC, there can be a small hysteresis window of a few millivolts in the regulator that causes oscillation. Add the 1 ohm resistor you mentioned and observe what happens. That is a solution that TI has for some of their switch mode regulators together with ceramic caps on the output.

[wraper]

mikerj, I said it was a 7805, in fact its not, you can look at the datasheet I provided, it is essentially an automotive rated version of a 7805. 

The only common things it has with 7805 are that it is voltage regulator and has 5V output. It is completely different IC. It is low dropout regulator, and such kind of voltage regulators overall don't do well with ceramic capacitors. Basically they cannot be used with ceramic capacitors unless stated otherwise in the datasheet. There is even written electrolytic on the schematic provided in the datasheet. Assuming that you can call every 5V vreg as 7805 you just misinformed the people that want to help you and shot in your own foot. Also, did you check the maximum input voltage? It is only 16V therefore a not a good choice for the use in the place where you put it. It will shut down if the voltage exceeds 18V, therefore some transient on the power rail (which is dirty) can cause circuit to shut down.

[hpmaxim]

mikerj, I said it was a 7805, in fact its not, you can look at the datasheet I provided, it is essentially an automotive rated version of a 7805. 

The only common things it has with 7805 are that it is voltage regulator and has 5V output. It is completely different IC. It is low dropout regulator, and such kind of voltage regulators overall don't do well with ceramic capacitors. Basically they cannot be used with ceramic capacitors unless stated otherwise in the datasheet. There is even written electrolytic on the schematic provided in the datasheet. Assuming that you can call every 5V vreg as 7805 you just misinformed the people that want to help you and shot in your own foot. Also, did you check the maximum input voltage? It is only 16V therefore a not a good choice for the use in the place where you put it. It will shut down if the voltage exceeds 18V, therefore some transient on the power rail (which is dirty) can cause circuit to shut down.

I'm not actually seeing where it says it will shut down if voltage exceeds 18V, and the absolute max voltage is 40V.  However, I do agree  it seems to indicate the max input voltage is 16V.  I honestly thought this was a 7805 equivalent.  Please understand, I don't normally do PCB design, and I had selected a 7805, and then looked for an Automotive rated equivalent, and for whatever reason I thought that's what I had found.

After re-reading the datasheet, I tend to agree that this may not be the best IC.  After looking on Digikey, I think the NCV7805 seems like a good alternative (pin/package compatible, no output cap needed, 1A max output, 35V max input, automotive qualified) -- and slightly cheaper.  Only downside is the 2V min drop, which is obviously not an issue and higher quiescent current which is a downer but not a showstopper.

[wraper]

I'm not actually seeing where it says it will shut down if voltage exceeds 18V, and the absolute max voltage is 40V.  However, I do agree  it seems to indicate the max input voltage is 16V.  I honestly thought this was a 7805 equivalent.  Please understand, I don't normally do PCB design, and I had selected a 7805, and then looked for an Automotive rated equivalent, and for whatever reason I thought that's what I had found.

[wraper]

After re-reading the datasheet, I tend to agree that this may not be the best IC.  After looking on Digikey, I think the NCV7805 seems like a good alternative (pin/package compatible, no output cap needed

Cap is kinda not required. With mentioning that <0.1uF could cause instability.