SDG

JLCPCB 4 Layer PCBs for $2: jlcpcb.com/
Miniware MHP30: bit.ly/386xgic
MHP30 on AliExpress: bit.ly/3mpmkkH
Support the Channel with Patreon: / sdgelectronics
A big thank you to all of my Patreon supporters!
-------------------------------------------------------------------------------------------------
0:00 Review from last video
3:06 Audio Testing
4:54 1st Culprit
5:47 2nd Culprit
6:30 Testing a fix
8:45 What is this component for?
11:39 Summary
-------------------------------------------------------------------------------------------------
Soldering Tools and Equipment:
Solder Paste: bit.ly/2E1P5jR
Solder Flux: bit.ly/2MQn7we
Quick 857DW+ Hot Air Station: bit.ly/2OWDvf4
T12 Soldering Station: www.banggood.com/custlink/3KK...
AC Version: www.banggood.com/custlink/3GK...
T12 Iron Tips: www.banggood.com/custlink/3DG...
Daniu Tip Cleaner: bit.ly/31VHwWM
KSGER T12-BCM3 Tip: bit.ly/2xgYXCQ
Test Equipment:
Mustool MDS8207 Multimeter: bit.ly/2kKHr7A
FY6900 Generator: bit.ly/2yZI40n
Differential Probe: bit.ly/31xNTOF
Microscope Equipment:
Eakins Barlow Lenses: bit.ly/3aBY5tR
Eakins Black Microscope: bit.ly/2VEcYaW
Eakins Camera Microscope: bit.ly/2x86hnT
Camera Adaptor for Microscopes: bit.ly/2KBZQMQ
Amscope Microscope Head (US): amzn.to/2YmZWNx
Amscope Trinocular Microscope (US): amzn.to/2YdBE8k
Amscope Trinocular Microscope (UK): amzn.to/2V8w9Gh
Swift SS41-WF20 Binocular Microscope US: amzn.to/2VLWEpb
Swift SS41-WF20 Binocular Microscope UK: amzn.to/2GYGYFV
Visit my website: www.sdgelectronics.co.uk/
#jlcpcb #mlcc #noise #ElectronicsCreators

[@dreamcat4]

Wow this is a fantastic video! For some reasonm whenever you get into aspects of electronics design that include a mechanical, or phsics element. The challenge of trying to solve around these other important factors just teaches me so much more than something that was purely electronics based. Ie. why we design products in real world in specific ways, and not just only to BoM cost alone. What these other important and critical choices are all about. Another first for me is actually seeing somebody demonstrate whine coming from a capacitor, rather than a magnetic coil. Which of course we have known about for years people saying this stuff... but not actually knowing how common this is, and for which specific ccapacitor types, or the principle for how the capacitor generates motion when it's job is about switching an electric field rather than a magnetic one etc. etc. BTW steve sorry i totally forgot to mention previously the great technique for finding the location for narrowing down to individual component. It is quite simple actually (again, not my idea but gleaned from another guy).. what you can do is simply make a temporary listening cone, by coiling a sheet of paper or thin plastic into your hand. Like a reverse gramaphone, or reverse speaker cone. Then slowly move the narrow end of the cone across your noisy board. Listening with your ear in the big end of the cone. (or a microphone). The amplitude of the noise will be loudest once you hover the cone over the area of pcb generating the most sound. Noting that it might be the pcb board itself vibrating due to the action of soldered other cmponent nearby or on the other side of the pcb if double sided board. Also try to think of the relative local and global stiffness or Q factors at play due to the entire ensemble of the whole pcb. Vs the subset of bad actors who are working as drivers. Since different sized shaped or stiffness (thickness) pcb will also play a factor. Finally if there are multiple component driving the same shared pcb (but at each different frequencies), then the compined effect might indeed be to create a difference or 'beat' frequency, that is a lot lower and then falling into the audible range. Which otherwise individually you would not expect to hear any of them! Hope it helps.

[@sdgelectronics]

The PCB was definitely playing an important role here. The first suspect capacitor I think was positioned in such a way that it probably saw the worst of the vibrations/acoustic wave, hence sounded much louder than the capacitor at fault. I think it shows up on the microphone a lot better, but the output capacitor had much more high frequency content (hence the very harsh tone). Thanks for the comprehensive reply.

[@sincerelyyours7538]

@@sdgelectronics Very interesting diagnosis and solution. I like the reverse megaphone idea as expressed by Dreamcat 4, but I fear that sometimes the frequency emitted may be outside the range of human hearing. Get to a certain age like me and you may not be able to hear anything over 8 kHz or so. In that case, could combining the megaphone with the mic help localize these sounds?

[@_ATHONOR]

Nice Video - when I was Uni I had a similar issue with a Power Supply for an audio project, so I swapped my ceramic caps for poly film capacitors and worked a charm! You can keep the low ESR element by using a film without the piezoelectric effect, but looking on Farnell can be quite costly when not in bulk. Would like to see you try this again with Film Caps (1210 drop in replacement available!)

[@wi_zeus6798]

Thank you so much for this mini-series! I have an issue with audible noise coming from a Nixie PSU. I always suspected the inductor and my initially bad layout. After improving the layout (and maybe critically) changing the type of ceramic cap on the output, the noise was significantly reduced. Initially I had just a few very large ceramic caps, now a combination of smaller ceramics (both physically and electrically) plus one electrolytic. I noticed the noise would go down if I increased the capacitance on the output. Maybe the ripple current in the inductor is lower but now after seeing this I realized, that the ripple on the ceramic caps also is lower. I'll definitely investigate that! Very interesting. Normally I'd just put my finger on all components to just feel for heat etc when testing but on a HV PSU that's not such a great idea. If I had done that I may have noticed that the caps were the culprit?

[@sdgelectronics]

Maybe, sometimes changing the capacitor will affect the noise from the inductor as a result of different ESR. There's many interlinked factors!

[@akhurash]

Nice video. Would not have thought it was from the capacitor.

[@sanches2]

I made a probe at work to find the source of such issues. It's made of a piezo buzzer plate and a sewing needle sodered on it. A small bidir. Suppressor is put in parallel to avoid hv pulses damaging my scope input. You touch caps and inductors with the needle with the fft on and you look for the cleanest tallest peak location :) works like a charm.

[@erikisberg3886]

Great idea, may copy that! I have used piezo discs, the smallest ones from RS, as sensors before. I think You can get them down to 10mm dia. Perhaps use something non conducting instead of a needle.

[@rwils6333]

Fantastic for exploring this further. Glad it took so quick

[@JeromeDemers]

what a great timing. I have the same issue with my DC/DC I design. Well one proto makes sound but not my other proto. I was always wondering where the sound came from. I thought it was my inductor. I will definitively touch the components and take measurements. Thank you very much!!

[@vizkoze]

Nice update. Thanks for sharing this knowledge

[@ZeedijkMike]

Well you just keep on giving. A very interesting series. Learned a lot. A very Mary Christmas to you and your loved ones.

[@dreamcat4]

Thanks again! Specifically for explaining this piezo effect is occuring inside MLCCs. It turns out that googling the right keywords will turn up a PDF document called 'capacitor singing FAQ' by TDK. And this is a really good and valuable source of information. To learn more about how to understand and deal with this problem in electronics. Really useful, cannot thank you enough. Because all my previous research was into coil whine. And of course... then the search results are not usually going to mention capacitors.

[@rrowan327]

Thanks Steve, I really enjoyed that video and learn from it. Cheers Rick

[@wthornton7346]

Superb demonstration. Thanks.

[@peerappel2012]

Thanks! This is a very interesting and useful video!

[@19janiboy96]

super interesting stuff!

[@rudigerhoppe5708]

Nice video. Great you followed up on that.

[@markrichards5630]

I was just having another look at this (I think we started experimenting with both the 3409 and the XHP70's at full brightness at the same time - your insight has given me more than a leg up so, thank you) and I realized as you said, "the data sheet circuit layout is pretty much the same as the one I have here", that it's not, and in a significant way. The data sheet does not use a cap in that position that is causing the noise on your board. No cap, no noise! (C4 that is: whether that allows the ripple to get to the LED or not I"m just about to find out).

[@neillerm3234]

Learned a lot from this, thanx so much!

[@coldfinger459sub0]

I just got to see and experience what we studied in electronics glass about noisy ceramic capacitors. 👍

[@sdgelectronics]

Nice to see a real life example of something you've read/learned about. I'll certainly be more wary in the future.

[@deangreenhough3479]

Learned a lot, thank you 👍😁

[@SidneyCritic]

The debugging vids are always interesting. I wonder if the old car trick would work on electronics. To trace where a noise is coming from on an engine you rest a large screwdriver on parts when the engine is running, and then rest you ear on the screwdriver handle end. Basically the metal screwdriver transmits the vibration directly to your ear. You didn't need it because your mic test worked well, but it would've been interesting if there was a metal rod from the part to the mic.

[@pcelectronicsaudiorepairan4164]

Hi Steve, AS I mentioned in another comment, I used to teach, and part of that involved using a SHURE wireless cardioid mic. Even when the mic capsule was removed from the transmitter section, it (the transmitter) would produce audio by tapping or yelling at it closely, due to those pesky capacitors. I hope this clears up the mystery for your viewers.

[@Mr.Leeroy]

Great coverage! After you play a couple of days with formulas in Excel you can even approximate output ripple from ESR, Fsw, delta L current and it reminds calculation of smoothing capacitance for sine wave rectifier. Shielded SMD inductors are potted most of the time and are very monolithic, they are rarely making any sound. Maybe in DC-DC circuits of GPUs where there are dozens of amps switching at any decent load, supplying billions of low voltage transistors in the range of 200-400W per BGA, but even there you are more likely to hear ceramics.

[@antigen4]

fascinating - have a metal lathe with a variable frequency drive board making EXACTLY this sound and it's LOUD .. i always thought it was due to a coil but maybe it's a cap! briliant. thanks

[@arenaengineering8070]

I confirm. And even when applied especially in SMPS, in addition to the microphone effect, due to the piezoelectric effect of ceramic capacitors, you can get parasitic interference in the form of amplitude modulation of the output power, which cannot be eliminated. So it is more correct to use a parallel connection of an electrolytic capacitor and a ceramic one.

[@piconano]

This is why I don't see any thumbs down on your channel!

[@sdgelectronics]

There's always a few, normally people that dislike sponsored videos. Thanks :)

[@tinygriffy]

@@sdgelectronics The dosage makes the poison....

[@FlyingShotsman]

Thanks for the follow-up video, Steve. It was neat to see how that played out on the PCB. Perhaps aluminum polymer caps are the answer. Still very low ESR, but without the aging issues inherent in electrolytics. I just wish they were cheaper...

[@sdgelectronics]

I do have some polymer ones from the original GPSDO refurbishment project. I think it'll end up fine with a small electrolytic though.

[@EngineeringVignettes]

There is some legitimacy to still using through-hole technology, less mechanical coupling to the PCB. The MLCC was using the PCB as a sounding board. I'm guessing that, long term, the solder joints would end up fatiguing due to the mechanical movement over time as well (I think Dave Jones did a video on that)... Good alternatives may be to use thin film or maybe even poly. I don't think through-hole (round) ceramic caps can be easily found anymore... Season's Greeting and cheers, - Eddy

[@PetreRodan]

nice catch! but I think you properly named this phenomenon as piezoelectric, since microphonic relates to generating voltage from surface deformation/vibration.

[@sdgelectronics]

Yes I think you're probably right

[@SchmurtzAlex]

There are so much devices which making this noise (my PC motherboard, my internet box, my led bulbs, my garage door motor) ! Everybody says that it comes from coil ! I will investigate again now I know that it can comes from capacitors. So replace capacitors is a way to do... 🤔 A video about all the ways to resolve it could great (replace capacitors, different solder technics, epoxy ? ...).

[@thekaduu]

Nice one! I though when you poked the capacitors, the noise was coupling to the soft antistatic mat more, thus dampening the noise... What you heard and what we heard was not the same too. The first capacitor you replaced may have been fine...

[@sdgelectronics]

I swapped the input caps back and there's no noise, so I think it just happened to be a particular spot on the PCB that resonated well.

[@George-gz5zm]

Hi Steve, thanks for taking it further and doing some analysis. Should have known it would be the cap. I think it would still be interesting content to wind your own inductor in some sort of oscillator or filter project. Hows the gps oscillator project coming on?

[@sdgelectronics]

I plan for an AC-DC SMPSU project, so there will be some coil winding soon!

[@alffriman3826]

Capacitor is the mechanical sound source but practically noiseless if not using a PCB as membrane. Make a tombstone and almost all noise is gone. If very critical to make circuit noiseless, add few mm solder wick as additional sound-isolation on both sides of capacitor. Solder wick is better then a thin wire as it have low RF-losses.

[@invendelirium]

I once had noise coming from a snubber network in a small AC-to-DC flyback converter i made as a part of a bigger board. Tombstoning was exactly how i got rid of the noise. Tombstoning is not a very production-friendly solution, though. I also sometimes tombstone my ceramic caps for reliability (often after finding out that my circuit is not working because the caps cracked while i was tossing the board around while assembling it).

[@Sheevlord]

I didn't expect the capacitor to be the culprit. I guess I was so used to linear PSU transformers buzzing that I automatically assumed that any inductor or transformer would be the "singing" component. I wonder if covering the capacitor in epoxy or other compound would quiet it down. Some throgh-hole ceramic caps look like they are dipped in resin or something similar. Something that feels plastic-y. Maybe those are less noisy than the normal SMD caps?

[@MarcoPolo-zv1ft]

capacitor layers vibrate in the frequency of our hearing range

[@antigen4]

i guess it's converting the extra ripple into mechanical movement (vibration) - either at the frequency of the ripple or at a sympathetic frequency

[@SuperBrainAK]

What about putting a tantalum capacitor instead of the MLCC?

[@rwbishop]

Related... look up 'Magnetostriction' sometime.

[@piconano]

You're welcome! [insert smilie face here] I wonder if instead of using one cap, what if you use 4 ceramic caps in parallel, would the problem would go away completely? I think it will. The ripple currents are too high for such small size. By reducing the ESR even farther with paralleling, they'll have higher ripple current handling capacity. Great followup. The mystery was killed and buried in this video.

[@Mr.Leeroy]

No it won't. You simply won't be able to solder all caps close enough. Parasitic inductance comes into play hard. Every additional cap illustrates the meaning of diminishing returns vibrantly, even if you solder them on top of each other. Every cap would have a different equivalent impedance proportional to distance. >=10mm is like useless. Low ESR ECaps are the only thing better, but bring requirement of maintenance compared to ceramics.

[@piconano]

@@Mr.Leeroy OR, do what I do! you solder them onto 2 parallel copper wires and make a vertical stack, which can be soldered on the the same pads, vertically. Just like an electrolytic cap. Works like a charm. Go try it before shooting people down.

[@Mr.Leeroy]

@@piconano Does not matter which direction you are stretching your current path. First caps sees the most current as is has lowest impedace, every other proportionally less. And I tried it, seen with my own eyes on the oscilloscope how ripple is affected. On low switching frequencies current takes path of least resistance, on high - of least inductance. ECaps are different because plate surfaces are much larger. And there is a clear correlation in max rated ripple current of ECap to its physical size.

[@Mr.Leeroy]

@@piconano If you saw difference, that was purely by adding more capacitance and ESR/ESL was not limiting in your case.

[@piconano]

@@Mr.Leeroy good for you. bye.

[@cmuller1441]

Maybe you should use the analog dimming method. See datasheet example 9.2.2 Analog Dimming Application for 4 LEDs

[@sdgelectronics]

I wish I had not been so impatient now, it's too late and I'm stuck with the PWM dimming.

[@cmuller1441]

@@sdgelectronics just send the pwm with an rc filter to the analog control...

[@tinygriffy]

Hey Steve ! Did you take a look at the layout guidelines (chapter 11) of your lm3409 datasheet ?? Vin seems to be a much wider track (main current path), also it states that the Vout cap (one of your main culprits here) should be placed as close as possible to the leds ! Wonder if that would make a difference. ... I still think you don't really need the buck regulator when you drive the whole shebang with an esp anyways. It's like having an extremely good pwm controller (the esp in that case) to control a very limited one (the lm3409). I am really very curious wether it would work or not feeding the mosfets gate directly with a signal from the esp. Anyhow, I love the debugging videos, they are most valuable. Thanks for sharing !

[@sdgelectronics]

The new PCB is much improved on the layout - video in the next few days. If it was a PIC or some other microcontroller, I'd offload the converter functionality, but I've had many occurrences of the ESP32 rebooting. During reboot the pins aren't in a guaranteed state so there's a risk of damage to the LED.

[@tinygriffy]

​@@sdgelectronics Hm, I thought the esp had a couple pins that are definitely low on boot, with a pulldown on the gate I have never had any incidents... anyhow, I am eager to see the new boards.. Thanks !

[@sdgelectronics]

I'm sure it's possible, however a switch mode controller should really have all (or at least the highest) the attention of the firmware if you are the implement it on a MCU. I don't have that control on the ESP32 whilst running FreeRTOS.

[@aleksandarperovic2520]

I have same noise on my monitor samsung syncmaster 740. When its OFF make same noise, and when its ON nose disappear Can you help me ?

[@drthik1]

So the pcb was acting as an amplifier and resonating that capacitor. I wonder if you could rotate the capacitor 90° in relation to the board so its layers aren't in the same plane? New sub btw

[@sdgelectronics]

Yes, I think the PCB was acting like a sounding board for the capacitor. This explains why I heard some strange harmonics when sweeping the frequency range.

[@Gooberslot]

I never would have suspected the capacitors. Are there any other types of surface mount capacitors you could use?

[@SeanBZA]

Almost all ceramic capacitors are piezoelectric, so you can use tantalum, electrolytic or SMD film in place of them, or get ones with better dielectric that is not a piezo, but then you are limited in value for the size.

[@sdgelectronics]

As with most things, there's a tradeoff and no perfect capacitor. They all have their own non-ideal behaviour.

[@Mr.Leeroy]

2:57 like that you are introducing more parasitics

[@sdgelectronics]

True, I did a 2nd take and missed out my original thought - that the extra capacitance slowed down the gate waveform slightly.

[@gogs991]

you should change diode

[@FixDaily]

Can you test it out with a tantalum capacitor and tell me if the problem remains the same? Keep up the great work

[@sdgelectronics]

No problems with tantalum capacitors, though they have their own issues...

[@FixDaily]

@@sdgelectronics You mean by this getting into flames?

[@sdgelectronics]

@@FixDaily yes towards their end of life

[@antigen4]

@@sdgelectronics CONTRARY to popular opinion tantalums don't really have 'aging' issues -if you use a tantalum properly there is no reason why it shouldn't last 100 years but most people have no idea HOW to use them. they will most assuredly 'die' however if you expose them to voltage spikes at or near their rated capacity - even for microseconds. it's important to use them only downstream of well regulated and buffered supplies

[@erikisberg3886]

I learned to hate tantalum caps. They work like a charm in the lab. But it is the single component that has caused me more grief than anything else in noisy industrial environments. I try to use dry electrolytics if possible.

[@JeromeDemers]

where is C4 in the schematic @4:30 ? It looks it's after L1 and connected to Vo. So is it Coff? That is "off time programming" so not related. Did you added that yourself? (EDIT) ok it's "Co" page 28 in datasheet. They recommend 2.2-µF X7R 10% 50 V murata GRM43ER71H225MA01L in that examples.

[@Omgbroccoli]

You should try using NP0/C0G class ceramic cap. Unlike x5/x7, NP0 caps are not prone to microphonics.

[@brendandsilva4498]

All ceramic caps are prone to microphonics. In my opinion, polyester or polysterene caps should be used.

[@guillep2k]

They don't come in such large values

[@Omgbroccoli]

@@guillep2k yep, my bad

[@BoB4jjjjs]

Avoid SMCs that's what I learned. Hate then anyway, but things are going that way!

[@antigen4]

stick an electrolytic on it!

[@_Jkobson]

can u get name of caps?

[@sdgelectronics]

Pete and Dave

[@sdgelectronics]

Samsung X7R 10uF 50V CL31B106KBHNNNE

[@_Jkobson]

​@@sdgelectronics i look at datasheet samsung, it's regular mlcc x7r caps, next time use Film Chip Capacitors if u need low height or small size, or use multiply small caps to same capacitance, it solve this problem without disadvantages of electrolytics caps

[@tinygriffy]

@@sdgelectronics 🤣 because Pete is shiet, and Dave misbehave

[@y_x2]

It is piezoelectric effect not microphonic effect.

[@hoggif]

If you have sound effects, it is probably due to pwm chopping due to charge up of the inductor, there are quite a lot of audiable harmonica from the inductor during charge-up. It is not all high frequency as shown in your previous video. Typical inductor.. I think lower esr caps could compensate for that though.

[@harinerdnook]

Plz make a video on basics of electronics