Help diagnosing wind fluttering/rustling noise in D-87/U87i DIY build

[dchang0]

In Mic 1, I removed the switches and switch wires completely, and the intermittent rushes of rustling noises seem to have gone away. They did happen a few times early on after the switches+wires had been removed, so I could not confidently say that the switches+wires had been at fault.

But I've been listening to Mic 1 for an hour or so, and there is no light wind noise and no intermittent rustling so far.

UPDATE: 2hrs so far, and no noise. I tempted fate and opened up Mic 1 and pushed C4 slightly tilted to make room for my soldering iron to go in and re-install the Pad wires, and then I tested it without doing anything. The light wind noise is back. Is it because of the physical tilting or is it because of the cleaning I did on the lead that goes to the isolation pin?

UPDATE 2: It looks like the light wind noise comes back if Mic 1 is detached to the preamp for a while then plugged back in, and then the light wind noise goes away after enough time has passed while still attached. Not sure why this is.

 

[dchang0]

Confirmed: cleaning the boards with a Q-tip and isopropyl alcohol introduces light noise for a while, after which the noise goes away after several minutes have passed.

However, in Mic 2, the light wind noise never seems to fully go away, regardless of how much I clean the board and let it dry after cleaning. I'm guessing the capsule is the one I initially thought to be noisy, since the noise floor is quite a bit higher than the other mic.

Only way to find out for sure is to swap the capsules (again). The problem with the switches or switch wires obscured the proper diagnosis of the capsules earlier.

UPDATE: Never mind on swapping the capsules. What I did was reflow only the two 60M (68M) resistors, R1, and C2 on Mic 2, and the noise floor dropped to the same as on Mic 1. I do not know why this is--the joints looked fine to me.

At this point, both mics are super-quiet in Cardioid mode. If I were to forego the switches, these are more than good enough for studio recording use.

Next, wire in a jumper for Fig 8 on both and test for a longer period of time. What I'll do is use the same straight wires that I would use with the pattern switch, just solder the ends together. This will allow me to differentially diagnose whether the switch or the switch wires or both are responsible for the louder intermittent and continuous noises.

 

[dchang0]

Adding the switches back in Mic 1 seems to have worked well. It does not have continuous light wind noise and only rarely has the rush of rustling noise. I reflowed the solder joints and that seems to have improved things.

Mic 2 still has the continuous light wind noise in Fig 8 and Omni (more noise in Fig 8 ). Reflowing the switch contacts makes no difference.

UPDATE: It seems that the best way to reduce the noise is to keep reflowing and cleaning the contacts for the pattern switch in whatever mode makes noise. It will result in the widest change in volume of the wind noise. I basically kept doing it until I got the lowest noise I could get, then gave up. There is still some wind noise in Mic 2 Fig 8 mode, and it is certainly louder than Mic 1 Fig 8 mode (by about 10dB), but nothing I do seems to get it better than that. Cardioid modes in both mics have the same level of noise. Omni is roughly the same.

UPDATE 2:

I went back to a jumper on the pattern selector and noticed that the noise on Fig 8 still exists. This probably means that something is wrong with C2 or R4 (only present on the Fig 8 part of the circuit), but reflowing/cleaning those did nothing.

It may be time to replace the polystyrene caps to remove them from suspicion.

 

[dchang0]

I put Mic 1 away (it has intermittent but rare rushing rustling noise) and tore apart Mic 2.

I completely detached the switch board and the capsule and pulled the boards out and gave them as aggressive a cleaning as possible (basically, everywhere but around the Styroflex caps), and then put the boards back in without a capsule attached (and without test caps). It took a little more Q-tip swabbing of the exposed leads that would be touched the most often on both boards, but I finally got the boards whisper quiet again in Cardioid mode with no capsule attached.

I used acetone instead of isopropyl alcohol. If you do this, be very careful with acetone, as it will dissolve the polystyrene caps even faster and more aggressively than isopropyl alcohol!

Cleaning the attachment points for the capsule AND the A, AA, B, BB, C, CC contacts AND the exposed leads of the high-resistance resistors seemed to make the difference.

So this looks like a matter of cleaning/contaminants on the components. Looking back, it would have been wise to buy the no-clean solder (I am using a high grade, leaded solder meant for SMT/SMD work, but it is not no-clean, because the residues are expected to be bathed off later). However, I probably still will not use no-clean solder, because it is harder to clean off the residues IF one chooses to (see here: http://www.assemblymag.com/articles/84286-why-clean-no-clean)

It would also have been wise to wear disposable gloves all the time, even while handling the boards, but I think the biggest difference would be in using no-clean solder and leaving the no-clean residue on the board, rather than trying to clean the boards without enough solvent and just pushing the solder residues around.

I'm back at a headless but noiseless build. Time to add back in the other stuff. I have removed the switch board completely from the mic body and will give it the scrubbing of its life.

 

[dchang0]

I added just a jumper to activate Fig 8 mode without any capsule nor any test capacitors added.

There was a light burst of wind noise at the beginning, but after waiting a bit, it remained super-quiet.

Next, add back the front and back sides of the capsule.

 

[dchang0]

Crap. Added back the capsule in Fig 8 mode, and it brought with it loud rustling noise.

Then, I remembered that it might simply be a matter of cleaning, and I used acetone on a Q-tip to clean the solder joints for the capsule, and voila, the noise went away. The capsule is very quiet (same as the other mic) right now, in Fig 8 mode.

So, now to add back the switches. First, the pattern switch, which seems to have been the only challenging one so far.

 

[dchang0]

Adding back the pattern switch went well. The mic is still quiet. I had prepped the switch board beforehand with lots of cleaning and also cleaned the contacts on the main board before and after soldering. No noise.

Interestingly, the previous issue of the back capsule being slightly quieter than the front in Fig 8 and Omni modes went away. It may also be a consequence of dirty contacts. I will later clean Mic 1's capsule solder points (not the capsule!) using acetone--it may even solve the intermittent rush of rustling problem.

MIC 2 IS SOLVED! (NOT REALLY--see later in the thread.)

Tempting fate, I will open up Mic 1 and clean the capsule solder pads on the PCB with acetone to see if the volume of the back diaphragm increases to match the front and the intermittent rush of rustling goes away. Sadly, it takes a long time to hear the intermittent rustling, so what I have done is set up the mics and run them through loudspeakers and just go about my day.

 

[dchang0]

Mic 1 isn't quite as good as I thought. Mic 2 is now actually quieter!

I'll probably have to do a tear-down and clean of Mic 1 also.

UPDATE:  Cleaning Mic 1 with acetone on a Q-tip also reduced the noise to slightly above Mic 2. I did a partial tear-down but did not go as far as with Mic 2.

There are still intermittent rushes of rustling in both mics after a long time has passed. Probably charge building up in some residue over time.

I'll probably still have to tear down Mic 1, as the rustling is more frequent in it than in Mic 2, which is now the stalwart of quiet.

If I had to do this all over again, what I'd do is buy silver mica or polypropylene film caps instead of polystyrene, solder the boards including the switch board and wire them together, then immerse them in an ultrasonic flux remover bath to get them absolutely clean.

Then, I'd attach the transformer, XLR and capsule and spot-clean the solder contacts for the capsule.

 

[ricardo]

UPDATE:  Cleaning Mic 1 with acetone on a Q-tip ....

You used ACETONE on a PCB with polystyrene caps????

If I had to do this all over again, what I'd do is buy silver mica or polypropylene film caps instead of polystyrene, solder the boards including the switch board and wire them together, then immerse them in an ultrasonic flux remover bath to get them absolutely clean.

Use NPO/CGO ceramics instead.  Best caps you can buy if the size/value/voltage fits.

This cleaning is standard for mikes.  Clean all the wires to the caps as well.

 

[dchang0]

You used ACETONE on a PCB with polystyrene caps????
Use NPO/CGO ceramics instead.  Best caps you can buy if the size/value/voltage fits.

This cleaning is standard for mikes.  Clean all the wires to the caps as well.

Yep, I like to live dangerously, LOL.

I was VERY careful about the acetone application using a trimmed-down Q-tip. Basically, I cleaned just up to the polystyrene caps without touching them. That means that under the polystyrene caps, all the original soldering residue is still there (it never got cleaned out).

I've been looking into the proper flux cleaner, and it depends heavily on the type of flux used. There are some that are extremely damaging to polystyrene and some that are not.

I have two fluxes: whatever is in the non-no-clean solder I used (probably rosin) and whatever is in the non-corrosive flux I used (active ingredient: zinc chloride). It turns out that isopropyl alcohol is pretty much worthless against rosin--all one will succeed in doing is softening the rosin and making it easier to push around the surface of the board, creating conductive film.

Acetone did work well.

I went even more dangerously and used lacquer thinner on a Q-tip per this discussion:

http://www.eevblog.com/forum/suggestions/myth-busting-solder-flux-cleaning-with-isopropanol-alcohol/

My mics now do not have the rare intermittent pops, so the mics are both fixed as far as I know. I'm testing them one at a time for hours on end, so we'll know tomorrow.

Thanks for the tip on the NP0/C0G capacitors. If I get even a single pop or rustle out of these mics again, I'll order a set of those to replace the polystyrene caps and go borrow my buddy's ultrasonic flux remover bath and solve this once and for all.

 

[dchang0]

This whole ordeal does make me wonder: why did I not have the same problems with my two D47fet builds, even though I assembled them at the same time with the same techniques (same solder, same iron)?

This is true also for Matt55, who built a flawless D47fet and also had noise issues with the D87.

It may be that the isolated/floating capsule diaphragm attachment point in the D47fet is the reason why noise is not an issue. The backplate connection is very far away too.

Is it possible that the layout of the D87 could be improved to reduce the chance of introducing noise? Maybe something as simple as moving the capsule connections an additional 0.1in apart would result in 90% fewer noise complaints even while using non-no-clean solder....

 

[dchang0]

Mic 2 has gone for a few hours with only one pop, and it was very quiet.

Mic 1 sits for a while and then has light rustling. It looks like I'll have to tear Mic 1 completely apart, clean, and reflow/resolder it after all.

 

[dchang0]

I found that acetone beats lacquer thinner in terms of cleanliness. Acetone leaves no residue at all, as far as I can see if shining a bright light on the PCB surface and looking at the reflection. Lacquer thinner does leave a visible residue using this test, and isopropyl alcohol leaves an even more obvious one.

I'm now slowly working backwards on Mic 1. I removed the switch board entirely and cleaned all the switch solder pads on the hi-Z PCB. There was no noise in Cardioid mode (no jumper, capsule still installed). I added back the jumper for Fig 8 and noise returned, so I cleaned the components again (except the polystyrene caps of course). Now it is in Fig 8 with capsule and without noise. I will leave it like this and do the loudspeaker test for a few hours to make sure it is noise-free.

When I am soldering, I am still using non-no-clean solder. What I do is add a tiny bit of clean solder, then dip the 25W Hakko DASH soldering iron's entire tip (shape I = super-fine) in non-corrosive flux. I let the flux burn off for a few seconds, then wipe the tip on a wet sponge, which leaves the tip clean and slightly dull. There should be no flux or rosin on the tip at this point. Then, I very carefully heat the pad with the iron almost vertical so that no fumes can deposit on the PCB around the pad. I insert the wire, heat it more till the bead is smooth (but still dull, due to the lack of flux). If the bead cools with a smooth surface, I leave it alone. If it does not, I dip the soldering iron tip 1mm into the flux and then use it to reflow the bead, which should turn shiny even with the tiny amount of flux. I then clean around the solder pad with acetone on a Q-tip.

Seems to work well enough; at least it did for Mic 2.

 

[Matt55]

Ha, I saw my name mentioned!

I built 2 x aml ez 1073's, a 47Fet, an elam 251E + power supply with tube point to point no problems. All worked perfect first time. No noise at all.

I did change to a different solder for the 251E and I prefer it, its a no clean, that actually flows and solders well. I can did out the name if anyone is interested!

My DIY 87 still doesn't work. I am about to open it back up and have another go with it! Thanks for the info, first thing im going to do is remove the switches.

 

[dchang0]

Thanks, Matt55--Please let us know how your D87 fix goes, especially if you solve it.

And yes, please share the make and model of the no-clean solder you like.

Congrats on the successful builds! At this point, given how much trouble I've had with the D87s, I am loathe to build any more mics.

 

[dchang0]

Morning update:

So close yet so far! I set up Mic 1 with a loudspeaker next to my head overnight. This morning, after maybe six hours, I was awakened by a loud storm-like sound.

It seems the mic will behave well for hours on end, then get into some sort of nasty oscillation that goes on for a few minutes. Right now, no switches or switch wires are attached. There is a jumper wire installed to force Fig 8. The capsule is installed and working.

So I still have to clean something out. I am pretty much forced now to disconnect the capsule, either to test the boards headless OR to take the boards completely out of the mic body for a rigorous scrubbing like Mic 2 got. Diagnostically it's better to go headless and do the listen test for another several hours to rule out the capsule.

UPDATE: mic is now headless (no capsule, no testing caps), and now for the listening test to see if any rustling occurs at any point during the next several hours. If nothing happens, then boards are probably clean enough, and I will clean the capsule wires and reattach them with minimal solder. When I was solving the noise issue on Mic 2, I cleaned the capsule wires before soldering them in and used tweezers to avoid putting any fingerprint oils on them.

If the noise comes back, then the capsule is probably slightly dirty inside or out, and it will be RMAed.

I did notice that the capsule has a little flap of mylar sticking out on the outside of the ring. Would this be enough to cause noise? I figure no, since there is no backplate behind this portion of mylar, but if it is out there like an antenna, then that would be reason enough to RMA the capsule.

 

[dchang0]

I wonder if it might be wiser to attach the capsule diaphragm wires directly to the nearest components, rather than to the solder pads on the hi-Z PCB, mimicking the isolated/floating connection point on the D47fet. Maybe this will completely prevent any stray capacitance/conductivity across the board surface...

The FD would apparently go directly to the -10dB switch pad on the switch board, cutting out lots of middlemen solder points and maybe 4in of wire, round trip, and the RD would apparently go directly to one leg of R3, cutting out one solder points.

Thoughts or comments on this idea are welcome.

 

[dchang0]

I'm also preparing to use the nuclear option in case I just can't find the source of the once-per-6-hour storm noise: replace all the polystyrene caps and use the ultrasonic flux remover bath.

To anyone considering this approach, here are the parts I have sourced but not yet tried. Quantities and prices are for one D87 build.

U87 Option 1 (mostly polypropylene film)

10pF 500V Silver Mica 5% CDE CD15CD100JO3F Mouser 598-CD15CD100JO3F 1x$1.79
470pF 1000V Polypropylene 2.5% WIMA FKP2-470/100/2.5T Mouser 505-FKP2-4701002.5T 2x$0.26
560pF 63V Polypropylene 1% Kemet PFR5561F63J11L4BULK Mouser 80-PFR5561F63J11L4 1x$0.72
220pF 100V Polypropylene 2.5% WIMA FKP2D002201D00HSSD Mouser 505-FKP2220/100/2.5 1x$0.38

U87 Option 2 (all NP0/C0G)

10pF 100V MLCC C0G 5% TDK FG28C0G2A100DNT06 Mouser 810-FG28C0G2A100DNT6 1 1x$1.79
470pF 100V MLCC C0G 1% Kemet C317C471F1G5TA Mouser 80-C317C471F1G 2x$1.06
560pF 1000V MLCC C0G 1% Kemet C315C561G1G5TA Mouser 80-C315C561G1G 1x$0.81
220pF 100V MLCC C0G 1% Kemet C317C221F2G5TA Mouser 80-C317C221F2G 1x$1.30

If you want to buy test capacitors to replace the capsule, these are close to 72pF and within the range of real capsules.

82pf 200V MLCC C0G 1% Kemet C052G820F2G5CR Mouser 80-CCR05CG820FR 2x$6.31 (Pretty expensive but small enough to fit inside the SYT-5 body.)
68pf 500V Silver Mica 1% CDE CD15ED680FO3F Mouser 598-CD15ED680FO3F 2x$2.65  (These are 11mm long and may be too large to fit inside the SYT-5 body!!)

I have NOT tried any of these yet, so I cannot vouch for them!

 

[dchang0]

After running Mic 1 headless for a couple of hours and getting no noise during the listening test, I cleaned the capsule wires, removed excess solder from the pads with a Hakko 808, cleaned the solder pads with acetone, and reconnected the capsule with a bare minimum of solder and no solder flux. Will do the listening test for another few hours, this time for longer, to see if the once-per-6-hours storm sound comes back. If it does not, then I am pretty much home free, since I expect the switch board to be good (removing it did not remove the noise, so it is probably clean but could be bad too).

If any of you use a Hakko 808, please be careful with it. I accidentally melted a corner of one of the polystyrene caps with it because the tip is so fat. It did not seem to do any damage beyond the cosmetic.

 

[dchang0]

Curious as to why Mic 2 had the very rare occasional single pop, I plugged it in for the first time in many hours, and it immediately had lots of tiny, quiet pops, like static electricity discharging or like a bundle of tiny firecrackers going off in the distance. I guess it wasn't fixed after all, so I started to work backwards on it too.

What I found was that there was two huge blobs of solder on the underside of the hi-Z board at the pattern solder pads that were staying with the board even when the switch wires were removed. After removing the blobs, I am now working my way back up (currently, both mics have just the capsule, no switches, and they are silent during the listening test). It is possible the blobs were not the cause of the pops.

UPDATE:

I added back one switch at a time and did a listen test in between each change. Both mics are now completely assembled and quiet. I did have the light crackling noise when adding back the Low Cut switch on both mics. What I did was reflow all the solder joints for the Low Cut with absolutely no flux, and I noticed that I burned off something (fumes came off the solder during reflow). My guess is that if there is anything to burn off, whatever it is can cause noise. Earlier, when I did any reflowing, I used flux to wet the solder and make it super-shiny, but this was almost certainly why I got random results in terms of resulting noise. I must've sometimes burnt off all the residue and sometimes not.

So now for the long listening test....

UPDATE 2:

Mic 1 has gone for several hours with only one incident of noise: very tiny pops maybe about two hours in. I reflowed the Pad switch connectors (the only ones I didn't reflow on the switch PCB). No idea if that made a difference, but it's been quiet so far.

Mic 2 is waiting its turn.

UPDATE 3: Mic 1 has passed its listening test. There were no more tiny pops after the reflowing of the Pad switch contacts. Mic 2 is now being tested.

UPDATE 4: Mic 2 has gone for 2 hours straight without a single pop.

I think it's safe to say that both mics are now studio ready. Mic 1 is probably more suspect than Mic 2, but at least I know the fix and most important, the diagnostic process to follow.