Topics

Advice wanted: QCX-mini capacitor microphonics


Hans Summers
 

Hi all

Advice sought, from the good folks on the list many of whom are > 80dB more knowledgeable than I am.

There are some microphonics involved in the QCX-mini. It is not catastrophic but it would be nice to eliminate this if possible since it is not a problem on the QCX/QCX+. I noticed initially an audible click whenever the rotary encoder changed the frequency, and started suspecting PCB layout issues with LCD digital signals inducting noise into RF signal paths etc., even though as usual I have taken the greatest care over the PCB layout.

However as I investigated more, I discovered this is not the cause. I was able to connect the controls PCB on 20cm cable and the problem disappeared. It is solely a problem of MICROPHONICS, the mechanical noise is feeding through the physical structure to microphonic components. Tapping a pen on the PCB causes the same effect. 

Upon scratching all the SMD components with the sharp broken blade of my former glorious ceramic bladed miniature screwdriver, I was able to determine that the only significantly microphonic components are:
  • C43/44/45/46, the 0.47uF QSD sampling capacitors
  • C21/C22 the 1uF AC coupling capacitors in the audio amp (these are a very much smaller problem than the QSD sampling capacitors). 
On researching this issue this afternoon, I found this is a known and common problem, with high-K dielectrics used in high density MLCC (Multi-Layer-Ceramic-Capacitors). Being soldered directly to the PCB, the problem is multiplied compared to through-hole components (which is presumably why we don't see this problem on the QCX/QCX+).

The 0.47uF capacitors in my prototype are from UK, Farnell part number 2627455 and sure enough have a high-K Y5V dielectric. 

I note that there are NP0 (low-K dielectric) 0.47uF capacitors but the cheapest Digikey show is $1.83 each even in quantity 1000... and we need FOUR for every QCX-mini! OUCH! They are also relatively huge, in 2220 packages - so this is clearly not an option. 

Other types of capacitors could be a possibility e.g. Digikey list "Film Capacitors" for 0.47uF at $0.20 (qty 2000) but still in a larger 1206 package, that may be hard to fit in.

I also read of other possible strategies - apparently if two identical capacitors of half the value are mounted symmetrically in parallel on opposite sides of the PCB then the peizoelectric (microphonic) effect somewhat cancels out. Another thing they suggest could help is a little slit in the PCB between the pads but I am not sure how practical that is from a PCB manufacturing point of view, for 0603 components. 

Another option is abandoning SMD for these four capacitors and using the usual 0.47uF through-hole types for the QSD capacitors, same as in QCX/QCX+ where they don't cause problems. Actually I believe these are also SMD MLCC capacitors exactly the same, just with two wires soldered to them and encased in a little yellow ceramic blob... but presumably they are not affected because they are sticking up out of the board, not soldered directly to it. I could investigate this but of course, space is at an absolute premium in QCX-mini...

SMD MLCC capacitors are used throughout but I believe these four sampling capacitors are a particular issue due to the relatively high value capacitance, and the fact that they are right at the start of the signal chain where signal levels are tiny; any piezoelectric effect generated noise is therefore amplified with the full force of all the gain stages which come later. 

I should emphasise that the QCX-mini is still a joy to use, but, I think these clicks etc are going to annoy people so I do need to find a solution...

Well... ya learn something every day, and this is a most unexpected problem. 

All comments, suggestions and experience will be welcomed!

73 Hans G0UPL
http://qrp-labs.com


Ronald Taylor
 

Hi Hans. Just a random thought on this. The "plates" of an SMD capacitor are parallel to the board when installed and the plates of a leaded part are perpendicular to the board. Perhaps ... the vibrations on the board being more pronounced in the vertical direction would have more of an effect on the plates of a horizontally mounted part. An easy test of this would be to mount an SMD part vertically on one pad and connect the top of it to the other pad via a short wire... Maybe ???

Ron

On Thu, Sep 24, 2020 at 8:06 AM Hans Summers <hans.summers@...> wrote:
Hi all

Advice sought, from the good folks on the list many of whom are > 80dB more knowledgeable than I am.

There are some microphonics involved in the QCX-mini. It is not catastrophic but it would be nice to eliminate this if possible since it is not a problem on the QCX/QCX+. I noticed initially an audible click whenever the rotary encoder changed the frequency, and started suspecting PCB layout issues with LCD digital signals inducting noise into RF signal paths etc., even though as usual I have taken the greatest care over the PCB layout.

However as I investigated more, I discovered this is not the cause. I was able to connect the controls PCB on 20cm cable and the problem disappeared. It is solely a problem of MICROPHONICS, the mechanical noise is feeding through the physical structure to microphonic components. Tapping a pen on the PCB causes the same effect. 

Upon scratching all the SMD components with the sharp broken blade of my former glorious ceramic bladed miniature screwdriver, I was able to determine that the only significantly microphonic components are:
  • C43/44/45/46, the 0.47uF QSD sampling capacitors
  • C21/C22 the 1uF AC coupling capacitors in the audio amp (these are a very much smaller problem than the QSD sampling capacitors). 
On researching this issue this afternoon, I found this is a known and common problem, with high-K dielectrics used in high density MLCC (Multi-Layer-Ceramic-Capacitors). Being soldered directly to the PCB, the problem is multiplied compared to through-hole components (which is presumably why we don't see this problem on the QCX/QCX+).

The 0.47uF capacitors in my prototype are from UK, Farnell part number 2627455 and sure enough have a high-K Y5V dielectric. 

I note that there are NP0 (low-K dielectric) 0.47uF capacitors but the cheapest Digikey show is $1.83 each even in quantity 1000... and we need FOUR for every QCX-mini! OUCH! They are also relatively huge, in 2220 packages - so this is clearly not an option. 

Other types of capacitors could be a possibility e.g. Digikey list "Film Capacitors" for 0.47uF at $0.20 (qty 2000) but still in a larger 1206 package, that may be hard to fit in.

I also read of other possible strategies - apparently if two identical capacitors of half the value are mounted symmetrically in parallel on opposite sides of the PCB then the peizoelectric (microphonic) effect somewhat cancels out. Another thing they suggest could help is a little slit in the PCB between the pads but I am not sure how practical that is from a PCB manufacturing point of view, for 0603 components. 

Another option is abandoning SMD for these four capacitors and using the usual 0.47uF through-hole types for the QSD capacitors, same as in QCX/QCX+ where they don't cause problems. Actually I believe these are also SMD MLCC capacitors exactly the same, just with two wires soldered to them and encased in a little yellow ceramic blob... but presumably they are not affected because they are sticking up out of the board, not soldered directly to it. I could investigate this but of course, space is at an absolute premium in QCX-mini...

SMD MLCC capacitors are used throughout but I believe these four sampling capacitors are a particular issue due to the relatively high value capacitance, and the fact that they are right at the start of the signal chain where signal levels are tiny; any piezoelectric effect generated noise is therefore amplified with the full force of all the gain stages which come later. 

I should emphasise that the QCX-mini is still a joy to use, but, I think these clicks etc are going to annoy people so I do need to find a solution...

Well... ya learn something every day, and this is a most unexpected problem. 

All comments, suggestions and experience will be welcomed!

73 Hans G0UPL
http://qrp-labs.com


Conard Murray
 

Hello Hans,

All you can do is try stuff and see what happens. Maybe slots around the caps would encourage the shock to go around the caps?

Maybe use nylon spacers and screws … and a rubber washer under the standoffs and the board? Maybe mount the encoder in a rubber washer? How about glopping silicone goop over the caps or the encoder?

I think I would try the leaded versions first and see what happens.

Looking forward to the mini!

73,

Conard, WS4S

 

 


Nick Norman
 

On 24/09/2020 16:29, Ronald Taylor wrote:
Hi Hans. Just a random thought on this. The "plates" of an SMD capacitor
are parallel to the board when installed and the plates of a leaded part
are perpendicular to the board. Perhaps ... the vibrations on the board
being more pronounced in the vertical direction would have more of an
effect on the plates of a horizontally mounted part. An easy test of this
would be to mount an SMD part vertically on one pad and connect the top of
it to the other pad via a short wire... Maybe ???

Ron
Or even 'on edge' IYSWIM

73
Nick


--
The sky spun again as Marco turned the ship so that 'down' was where
long tradition had always put it, in the region of the feet.
(Strata)
16:45:01 up 1 day, 7:16, 10 users, load average: 6.16, 6.41, 6.62



--
73 Nick M0HGU


Richard AG5M
 

I have no technical expertise as you have, but my comment is to permanently solve the problem whatever it takes. A few more dollars to eliminate microphonics is not going to stop anyone from buying the Mini. Will only add to its value, which is still the best value anywhere in the QRP world! Get it right the first time. 73, Richard AG5M

On Thursday, September 24, 2020, 08:06:29 AM PDT, Hans Summers <hans.summers@...> wrote:


Hi all

Advice sought, from the good folks on the list many of whom are > 80dB more knowledgeable than I am.

There are some microphonics involved in the QCX-mini. It is not catastrophic but it would be nice to eliminate this if possible since it is not a problem on the QCX/QCX+. I noticed initially an audible click whenever the rotary encoder changed the frequency, and started suspecting PCB layout issues with LCD digital signals inducting noise into RF signal paths etc., even though as usual I have taken the greatest care over the PCB layout.

However as I investigated more, I discovered this is not the cause. I was able to connect the controls PCB on 20cm cable and the problem disappeared. It is solely a problem of MICROPHONICS, the mechanical noise is feeding through the physical structure to microphonic components. Tapping a pen on the PCB causes the same effect. 

Upon scratching all the SMD components with the sharp broken blade of my former glorious ceramic bladed miniature screwdriver, I was able to determine that the only significantly microphonic components are:
  • C43/44/45/46, the 0.47uF QSD sampling capacitors
  • C21/C22 the 1uF AC coupling capacitors in the audio amp (these are a very much smaller problem than the QSD sampling capacitors). 
On researching this issue this afternoon, I found this is a known and common problem, with high-K dielectrics used in high density MLCC (Multi-Layer-Ceramic-Capacitors). Being soldered directly to the PCB, the problem is multiplied compared to through-hole components (which is presumably why we don't see this problem on the QCX/QCX+).

The 0.47uF capacitors in my prototype are from UK, Farnell part number 2627455 and sure enough have a high-K Y5V dielectric. 

I note that there are NP0 (low-K dielectric) 0.47uF capacitors but the cheapest Digikey show is $1.83 each even in quantity 1000... and we need FOUR for every QCX-mini! OUCH! They are also relatively huge, in 2220 packages - so this is clearly not an option. 

Other types of capacitors could be a possibility e.g. Digikey list "Film Capacitors" for 0.47uF at $0.20 (qty 2000) but still in a larger 1206 package, that may be hard to fit in.

I also read of other possible strategies - apparently if two identical capacitors of half the value are mounted symmetrically in parallel on opposite sides of the PCB then the peizoelectric (microphonic) effect somewhat cancels out. Another thing they suggest could help is a little slit in the PCB between the pads but I am not sure how practical that is from a PCB manufacturing point of view, for 0603 components. 

Another option is abandoning SMD for these four capacitors and using the usual 0.47uF through-hole types for the QSD capacitors, same as in QCX/QCX+ where they don't cause problems. Actually I believe these are also SMD MLCC capacitors exactly the same, just with two wires soldered to them and encased in a little yellow ceramic blob... but presumably they are not affected because they are sticking up out of the board, not soldered directly to it. I could investigate this but of course, space is at an absolute premium in QCX-mini...

SMD MLCC capacitors are used throughout but I believe these four sampling capacitors are a particular issue due to the relatively high value capacitance, and the fact that they are right at the start of the signal chain where signal levels are tiny; any piezoelectric effect generated noise is therefore amplified with the full force of all the gain stages which come later. 

I should emphasise that the QCX-mini is still a joy to use, but, I think these clicks etc are going to annoy people so I do need to find a solution...

Well... ya learn something every day, and this is a most unexpected problem. 

All comments, suggestions and experience will be welcomed!

73 Hans G0UPL
http://qrp-labs.com


namerati@...
 

On Thu, Sep 24, 2020 at 06:06:12PM +0300, Hans Summers wrote:
On researching this issue this afternoon, I found this is a known and
common problem, with high-K dielectrics used in high density MLCC
(Multi-Layer-Ceramic-Capacitors). Being soldered directly to the PCB, the
problem is multiplied compared to through-hole components (which is
presumably why we don't see this problem on the QCX/QCX+).
I do notice the issue on my QCX as well.

Other types of capacitors could be a possibility e.g. Digikey list "Film
Capacitors" for 0.47uF at $0.20 (qty 2000) but still in a larger 1206
package, that may be hard to fit in.
When we were designing audio amplifiers we were firmly instructed to avoid any ceramics in the signal path, under penalty of death, and use film capacitors instead. Microphonics was one of (several) reasons given.

Depending on the dielectric it may be possible to find film capacitors with a similar (or better!) volume to capacitance ratio than ceramics, but it has been a while.

Tantalum was another dielectric we were forbidden from using, but as it also offers high capacitance per mm^2 it may be worth considering.

See also:
https://sh.kemet.com/Lists/TechnicalArticles/Attachments/62/2007%20CARTS%20-%20Reduced%20Microphonics%20and%20Sound%20Emissions.pdf


Ron Carr
 

I wonder if using lower value sampling capacitors would be useful.  Even if made of the same material, a 0.1 uf would have 1/5 of the microphonic material.   My understanding, which may be wrong, is that using lower value capacitors would just increase the bandwidth of the detector and the effect would be hidden by the 200 hz audio filter.

Ron   K1URC


Steven Dick
 

Wow you never know what problems arise from left field even though the same schematic has been reproduced in different versions.  Reminds one that components have real physical properties that can adversely affect system performance based on their physical characteristics and environment they are in.  With the goal of keeping cost down and tight pc board space, my recommendation would be:

1. Just resolve the problem with  C43/44/45/46.  That's where most of the microphonics have the biggest negative effect.
2. Stick with the high K dielectric caps.  Other caps with low microphonics are cost-prohibitive. 

Therefore:
1. The simplest and recommended solution is to use through-hole caps if PCB space permits, though no doubt very challenging to fit them.
2. Film caps in a 1206 package are not recommended as they raise the cost and don't help much size-wise except in height.
3. Two half-value caps mounted on both sides of the board is unduly complex from a manufacturing standpoint and is iffy at best as providing much improvement. - Not recommended.
4. Super tiny slits in the PCB is unduly complex from a PCB manufacturing standpoint . Not recommended.

One other possibility is to stick with SMD components but try to acoustically isolate them from the main pcb in some way. One possibility is to mount the four SMD parts on a subassembly which mates to the main PCB with wires similar to the TCXO for the QCX+. Adds complexity and some cost but may do the job while taking less space than the through-hole caps.

-Steve K1RF

------ Original Message ------
From: "Hans Summers" <hans.summers@...>
Sent: 9/24/2020 11:06:12 AM
Subject: [QRPLabs] Advice wanted: QCX-mini capacitor microphonics

Hi all

Advice sought, from the good folks on the list many of whom are > 80dB more knowledgeable than I am.

There are some microphonics involved in the QCX-mini. It is not catastrophic but it would be nice to eliminate this if possible since it is not a problem on the QCX/QCX+. I noticed initially an audible click whenever the rotary encoder changed the frequency, and started suspecting PCB layout issues with LCD digital signals inducting noise into RF signal paths etc., even though as usual I have taken the greatest care over the PCB layout.

However as I investigated more, I discovered this is not the cause. I was able to connect the controls PCB on 20cm cable and the problem disappeared. It is solely a problem of MICROPHONICS, the mechanical noise is feeding through the physical structure to microphonic components. Tapping a pen on the PCB causes the same effect. 

Upon scratching all the SMD components with the sharp broken blade of my former glorious ceramic bladed miniature screwdriver, I was able to determine that the only significantly microphonic components are:
  • C43/44/45/46, the 0.47uF QSD sampling capacitors
  • C21/C22 the 1uF AC coupling capacitors in the audio amp (these are a very much smaller problem than the QSD sampling capacitors). 
On researching this issue this afternoon, I found this is a known and common problem, with high-K dielectrics used in high density MLCC (Multi-Layer-Ceramic-Capacitors). Being soldered directly to the PCB, the problem is multiplied compared to through-hole components (which is presumably why we don't see this problem on the QCX/QCX+).

The 0.47uF capacitors in my prototype are from UK, Farnell part number 2627455 and sure enough have a high-K Y5V dielectric. 

I note that there are NP0 (low-K dielectric) 0.47uF capacitors but the cheapest Digikey show is $1.83 each even in quantity 1000... and we need FOUR for every QCX-mini! OUCH! They are also relatively huge, in 2220 packages - so this is clearly not an option. 

Other types of capacitors could be a possibility e.g. Digikey list "Film Capacitors" for 0.47uF at $0.20 (qty 2000) but still in a larger 1206 package, that may be hard to fit in.

I also read of other possible strategies - apparently if two identical capacitors of half the value are mounted symmetrically in parallel on opposite sides of the PCB then the peizoelectric (microphonic) effect somewhat cancels out. Another thing they suggest could help is a little slit in the PCB between the pads but I am not sure how practical that is from a PCB manufacturing point of view, for 0603 components. 

Another option is abandoning SMD for these four capacitors and using the usual 0.47uF through-hole types for the QSD capacitors, same as in QCX/QCX+ where they don't cause problems. Actually I believe these are also SMD MLCC capacitors exactly the same, just with two wires soldered to them and encased in a little yellow ceramic blob... but presumably they are not affected because they are sticking up out of the board, not soldered directly to it. I could investigate this but of course, space is at an absolute premium in QCX-mini...

SMD MLCC capacitors are used throughout but I believe these four sampling capacitors are a particular issue due to the relatively high value capacitance, and the fact that they are right at the start of the signal chain where signal levels are tiny; any piezoelectric effect generated noise is therefore amplified with the full force of all the gain stages which come later. 

I should emphasise that the QCX-mini is still a joy to use, but, I think these clicks etc are going to annoy people so I do need to find a solution...

Well... ya learn something every day, and this is a most unexpected problem. 

All comments, suggestions and experience will be welcomed!

73 Hans G0UPL
http://qrp-labs.com


Virus-free. www.avast.com


The Crunchbird
 

Hans, you seem to have answered your own question. SMD ceramic capacitors are a really bad source of microphonic problems. You may have to replace some of them with non ceramic types. Try tapping each capacitor with an insulted stick such as a piece of plastic knitting needle to identify which ones are causing the problem. One other very serious limitation of some types of ceramic SMD capacitors is that they change value rather dramatically with age. I have seen equipment that worked fine after it was produced that started to fail about a year out. The problem being that a capacitor or capacitors had changed in value. The fix is to reheat them but no consumer of course would do that. Dave. N2SN 


Graham, VE3GTC
 

Hans,

I was building a smd 40m receiver of similar design and struggled with microphonics in the audio chain. 

I tried tantalums since all signals were positive with respect to ground and that gave an improvement but the real fix however was using film capacitors.

Sometimes you just have to spend a bit more and do it right.

You have quite likely seen this web page but I am sure others have not ( for some light bed time reading )

https://www.edn.com/signal-distortion-from-high-k-ceramic-capacitors/

cheers, Graham ve3gtc


 


Henning Weddig
 

Hans,

my advice: never use ceramic caps of high dieelctric K like X7R or similar in coupling appliances or time keeping units e.g. in the QSD or as filter caps in lowpass or similar filters including PLL loop filters:

apart from the high microfonic behaviour (piezo effect) they suffer from

    high temperature sensitivity vs capacitance,

    high dependace of capacitance versus overlay supply voltage (DC voltage)

I had good experience using SMD film caps from Panasonic

73

Henning Weddig

DK5LV

Am 24.09.2020 um 18:34 schrieb The Crunchbird:

Hans, you seem to have answered your own question. SMD ceramic capacitors are a really bad source of microphonic problems. You may have to replace some of them with non ceramic types. Try tapping each capacitor with an insulted stick such as a piece of plastic knitting needle to identify which ones are causing the problem. One other very serious limitation of some types of ceramic SMD capacitors is that they change value rather dramatically with age. I have seen equipment that worked fine after it was produced that started to fail about a year out. The problem being that a capacitor or capacitors had changed in value. The fix is to reheat them but no consumer of course would do that. Dave. N2SN


Dave
 

Many years ago I was commissioned to recap two high end tube MacIntosh audio amplifiers, replacing the many ceramic capacitors with Polycarbonate film capacitors to eliminate the microphonics and distortions the ceramic caps were causing. For good measure the carbon composite resistors were replaced with metal film resistors.

The results were astonishing according to the audiophile owner and worth every cent he paid. My ears, sadly, were not as sensitive to the fine details as his.

Dave

On Sep 24, 2020, at 13:21, Henning Weddig via groups.io <hweddig=aol.com@groups.io> wrote:

Hans,

my advice: never use ceramic caps of high dieelctric K like X7R or similar in coupling appliances or time keeping units e.g. in the QSD or as filter caps in lowpass or similar filters including PLL loop filters:

apart from the high microfonic behaviour (piezo effect) they suffer from

high temperature sensitivity vs capacitance,

high dependace of capacitance versus overlay supply voltage (DC voltage)

I had good experience using SMD film caps from Panasonic

73

Henning Weddig

DK5LV

Am 24.09.2020 um 18:34 schrieb The Crunchbird:

Hans, you seem to have answered your own question. SMD ceramic capacitors are a really bad source of microphonic problems. You may have to replace some of them with non ceramic types. Try tapping each capacitor with an insulted stick such as a piece of plastic knitting needle to identify which ones are causing the problem. One other very serious limitation of some types of ceramic SMD capacitors is that they change value rather dramatically with age. I have seen equipment that worked fine after it was produced that started to fail about a year out. The problem being that a capacitor or capacitors had changed in value. The fix is to reheat them but no consumer of course would do that. Dave. N2SN




Hans Summers
 

Hello Henning, all

Thanks for all the thoughts so far. Soldering SMD capacitors standing upright with a wire to one terminal, or on their side isn't possible because SMD machines can't do that sort of weird thing, humans can... but all SMD will be done by machine. 
 
my advice: never use ceramic caps of high dieelctric K like X7R or
similar in coupling appliances or time keeping units e.g. in the QSD or
as filter caps in lowpass or similar filters including PLL loop filters:

In a perfectionist scenario, yes. In QRP Labs kits all the RF capacitors in the filters are NP0 types by Vishay, supplied via Digikey. But these are sub 1000pF. The cost of NP0 ceramics gets very high as you go up in capacitance. QCX, being a direct conversion architecture, does a large amount of the signal processing at audio where larger capacitances are involved. It would be prohibitively expensive to substitute all those with film capacitors etc. 

In general, QCX does not suffer from any noticeable microphonics. But uses MLCC throughout including in the QSD 0.47uF sampling capacitors. Unfortunately the difference in QCX-mini is that they are soldered directly to the PCB which just acts as a soundboard collecting all the vibrations in existence and coupling the vibrations right across the capacitor. This probably affects all the capacitors in QCX BUT, the QSD capacitors are in a particularly sensitive place of the circuit and also have a high density being 0.47uF in that 0603 package. Any piezo-electric voltage generated is directly added in parallel to the tiny signals we are trying to detect, then amplified up to around 100dB through the QCX signal path. After the first pre-amp (IC4, +40dB approx) any such noise is automatically of much lower significance. 

I just tested my 20m QCX+, on the dummy load (much easier to hear anything in the quiet, compared to when the antenna is connected). I tapped all the capacitors. I found that the QSD 0.47uF capacitors cause a pop in the audio when you tap them with a screwdriver, and so does C21. There were some others too in the audio circuits. However, no matter how hard I tap the PCB next to them, you can't get any pop. 

FYI I have confirmed my suspicions about the through-hole capacitors used in QCX. They really are just a plain SMD capacitor, MLCC, with two wires soldered to it, and immersed in a yellow blob of something, with "474" (0.47uF) printed on the outside. So there is absolutely NO difference in the capacitor itself. It's a difference in how mechanical vibrations are coupled into it. 

Attached, a series of four photos. Here I took a 0.47uF capacitor as supplied in the QCX/QCX+ kits, and sacrificed it in the name of science. The size is 0603. A totally standard 0603 MLCC capacitor, SMD, packaged up as a through-hole component. 

So it seems the easiest way forward, is to try replacing those four caps with the usual QCX through-hole ones. If that solves the problem then I can see if I can make a permanent solution on the PCB. That would be an acceptable solution! 

Thanks for all the suggestions! 

73 Hans G0UPL
http://qrp-labs.com



Paul AI4EE
 

Maybe that explains why some of my very old 105 caps are so much larger than the new ones - no SMD inside.


On 9/24/2020 2:09 PM, Hans Summers wrote:
Hello Henning, all

Thanks for all the thoughts so far. Soldering SMD capacitors standing upright with a wire to one terminal, or on their side isn't possible because SMD machines can't do that sort of weird thing, humans can... but all SMD will be done by machine. 
 
my advice: never use ceramic caps of high dieelctric K like X7R or
similar in coupling appliances or time keeping units e.g. in the QSD or
as filter caps in lowpass or similar filters including PLL loop filters:

In a perfectionist scenario, yes. In QRP Labs kits all the RF capacitors in the filters are NP0 types by Vishay, supplied via Digikey. But these are sub 1000pF. The cost of NP0 ceramics gets very high as you go up in capacitance. QCX, being a direct conversion architecture, does a large amount of the signal processing at audio where larger capacitances are involved. It would be prohibitively expensive to substitute all those with film capacitors etc. 

In general, QCX does not suffer from any noticeable microphonics. But uses MLCC throughout including in the QSD 0.47uF sampling capacitors. Unfortunately the difference in QCX-mini is that they are soldered directly to the PCB which just acts as a soundboard collecting all the vibrations in existence and coupling the vibrations right across the capacitor. This probably affects all the capacitors in QCX BUT, the QSD capacitors are in a particularly sensitive place of the circuit and also have a high density being 0.47uF in that 0603 package. Any piezo-electric voltage generated is directly added in parallel to the tiny signals we are trying to detect, then amplified up to around 100dB through the QCX signal path. After the first pre-amp (IC4, +40dB approx) any such noise is automatically of much lower significance. 

I just tested my 20m QCX+, on the dummy load (much easier to hear anything in the quiet, compared to when the antenna is connected). I tapped all the capacitors. I found that the QSD 0.47uF capacitors cause a pop in the audio when you tap them with a screwdriver, and so does C21. There were some others too in the audio circuits. However, no matter how hard I tap the PCB next to them, you can't get any pop. 

FYI I have confirmed my suspicions about the through-hole capacitors used in QCX. They really are just a plain SMD capacitor, MLCC, with two wires soldered to it, and immersed in a yellow blob of something, with "474" (0.47uF) printed on the outside. So there is absolutely NO difference in the capacitor itself. It's a difference in how mechanical vibrations are coupled into it. 

Attached, a series of four photos. Here I took a 0.47uF capacitor as supplied in the QCX/QCX+ kits, and sacrificed it in the name of science. The size is 0603. A totally standard 0603 MLCC capacitor, SMD, packaged up as a through-hole component. 

So it seems the easiest way forward, is to try replacing those four caps with the usual QCX through-hole ones. If that solves the problem then I can see if I can make a permanent solution on the PCB. That would be an acceptable solution! 

Thanks for all the suggestions! 

73 Hans G0UPL
http://qrp-labs.com



Steven Dick
 

One more comment on this - problems coming out of left field sometimes require solutions from left field.  How about a simple lump of putty on top of  caps C43/44/45/46 to greatly dampen vibrations and reduce microphonics.  Simple enough to try.

-Steve K1RF

------ Original Message ------
From: "The Crunchbird" <pulsedevil@...>
Sent: 9/24/2020 12:34:01 PM
Subject: Re: [QRPLabs] Advice wanted: QCX-mini capacitor microphonics

Hans, you seem to have answered your own question. SMD ceramic capacitors are a really bad source of microphonic problems. You may have to replace some of them with non ceramic types. Try tapping each capacitor with an insulted stick such as a piece of plastic knitting needle to identify which ones are causing the problem. One other very serious limitation of some types of ceramic SMD capacitors is that they change value rather dramatically with age. I have seen equipment that worked fine after it was produced that started to fail about a year out. The problem being that a capacitor or capacitors had changed in value. The fix is to reheat them but no consumer of course would do that. Dave. N2SN 

Virus-free. www.avast.com


Arv Evans
 

Steven

Way back in the bad old days of vacuum tubes and microphonics we found 
that microphonics arising from audio feedback sometimes could be solved 
by placing the speaker outside of the main chassis.  Using headphones  
also helped resolve the problem.

Arv
_._


On Thu, Sep 24, 2020 at 2:22 PM Steven Dick <sbdick@...> wrote:
One more comment on this - problems coming out of left field sometimes require solutions from left field.  How about a simple lump of putty on top of  caps C43/44/45/46 to greatly dampen vibrations and reduce microphonics.  Simple enough to try.

-Steve K1RF

------ Original Message ------
From: "The Crunchbird" <pulsedevil@...>
Sent: 9/24/2020 12:34:01 PM
Subject: Re: [QRPLabs] Advice wanted: QCX-mini capacitor microphonics

Hans, you seem to have answered your own question. SMD ceramic capacitors are a really bad source of microphonic problems. You may have to replace some of them with non ceramic types. Try tapping each capacitor with an insulted stick such as a piece of plastic knitting needle to identify which ones are causing the problem. One other very serious limitation of some types of ceramic SMD capacitors is that they change value rather dramatically with age. I have seen equipment that worked fine after it was produced that started to fail about a year out. The problem being that a capacitor or capacitors had changed in value. The fix is to reheat them but no consumer of course would do that. Dave. N2SN 

Virus-free. www.avast.com


Torbjorn Skauli
 

Isn't there a way to dampen the mechanical vibrations from the encoder? Could you try to hold the PCB against some rigid massive object, such as a screwdriver handle, in different locations, to create a quieter zone for vibrations and see if that suggests a mechanical fix? How about loosening the screws that attach the controls board to the main board? Or loosening the encoder from the board? Or mounting it with longer wires? Can the controls be hole mounted in the enclosure instead, for those who get the enclosure?

Torbjorn, LA4ZCA


Hans Summers
 

Hi all

Thanks for all the suggestions. 

Ron, Nick: Edge-on or standing-on-end SMDs wouldn't be practical from a manufacturing perspective. 

Conrad: I am already using nylon spacers and screws. Rubber washer, not... but I don't have clearance to add even a fraction of a mm... same applies to the encoder... Slots, there is not a lot of space, I would be very hesitant to start mechanically mutilating my board. 

Blu-tac, putty to dampen vibrations: yes I can try that, it will take only a moment. I do not expect positive results. It may gain a few dB but that won't be enough. This is my intuitive feel. I think it's like a loudspeaker; if you put something heavy on the loudspeaker enclosure you change the sound a bit, muffle it a bit. You change a few dB but not a lot. It still makes a lot of noise. Or a mobile phone. When it vibrates, notice how much louder it is if you put it on a table! If you stand something else on the table yes you can knock it down a few dB but I don't think damping is enough. Just my feeling, but yes I will try it for the laughs. IF I can find any blu-tac. But the kids have that sort of thing so I can go raid their rooms. 

QCX - I have not noticed any microphonics; if you scratch some of the capacitors you can get crackles yes. But nothing that is transmitted by ordinary vibration through the enclosure and PCB. So there is no problem in QCX/QCX+ I think. 

Avoiding any ceramics in the signal path, is impractical entirely - since the cost would escalate to the sky and it would also make no difference, since there is in general no problem except these specific QSD capacitors. 

Ron - lower value sampling capacitors e.g. 0.1uF do widen the audio bandwidth, as you say the CW filter would fix that... but, still there is some theoretical performance impact because strong interfering signals get through the detector and can theoretically cause intermodulation problems downstream. I thought about this too, changing them to 0.1uF but I am quite reluctant to go down that route since I don't really want to have to go through a whole new round of measurements to see if there are any other impacts and what they are - I am quite keen to avoid many changes to the schematic, so that this project does not consume large R&D time. 

Steve K1RF, yes, all agreed... actually two half-value caps on either sides of the board would be possible because anyway the QCX-mini has SMD components on both sides of the board (see the photos); but, this would also consume greater board area for the caps, twice as much in fact. Slits wouldn't be practical either as you said. And film caps are larger anyway. 

> One other possibility is to stick with SMD components but try to acoustically 
> isolate them from the main pcb in some way. One possibility is to mount the 
> four SMD parts on a subassembly which mates to the main PCB with wires 
> similar to the TCXO for the QCX+. Adds complexity and some cost but may 
> do the job while taking less space than the through-hole caps.

Yes, I was also wondering about this too... It could actually also be an easy solution because I could fit a little sub-assembly into the cut-out space on the top PCB. So there wouldn't be any cost increase. If this ends up a smaller physical space requirement, and if there isn't a way to fit through-hole caps in, this could be a good option. 

Overall it seems the best option and simplest is to try the through-hole caps. As I mentioned, really they are also 0603 SMD caps with wires soldered on and a yellow blob of something, then call them "through hole"... but at least the dielectric is not rigidly fixed to a soundboard (the PCB), evidently the wires are enough deterrent for that. So, I will try this first, see if it fixes it, and if I can fit it on the PCB. Ideally I need to try and do some experiments with the existing prototype PCBs, and ideally avoid another iteration of prototyping which will cause weeks of delay and more time and back ache assembling them. 

Thanks again for all the great suggestions, you guys are the best... the only thing in the world better than QRP Labs kits, is the QRP Labs community! 

73 Hans G0UPL





Steven Dick
 

Hans, One more comment. Kemet came out with the KPS series SMC cap which is essentially an SMD cap or caps mounted on a shock isolating lead frame for acoustic isolation, minimizing microphonics.  The smallest size they come in though is 1210.  But maybe the mechanical concept could be applied to standard caps? Or 4 caps on a substrate with a lead frame for surface mount installation.  Attached are papers on the microphonics problem and Kemet series SMC data sheet.

-Steve K1RF

https://sh.kemet.com/Lists/TechnicalArticles/Attachments/62/2007%20CARTS%20-%20Reduced%20Microphonics%20and%20Sound%20Emissions.pdf

https://www.mouser.com/pdfdocs/KemetKPSJune302010.pdf

-Steve K1RF





------ Original Message ------
From: "Hans Summers" <hans.summers@...>
Sent: 9/25/2020 3:25:00 AM
Subject: Re: [QRPLabs] Advice wanted: QCX-mini capacitor microphonics

Hi all

Thanks for all the suggestions. 

Ron, Nick: Edge-on or standing-on-end SMDs wouldn't be practical from a manufacturing perspective. 

Conrad: I am already using nylon spacers and screws. Rubber washer, not... but I don't have clearance to add even a fraction of a mm... same applies to the encoder... Slots, there is not a lot of space, I would be very hesitant to start mechanically mutilating my board. 

Blu-tac, putty to dampen vibrations: yes I can try that, it will take only a moment. I do not expect positive results. It may gain a few dB but that won't be enough. This is my intuitive feel. I think it's like a loudspeaker; if you put something heavy on the loudspeaker enclosure you change the sound a bit, muffle it a bit. You change a few dB but not a lot. It still makes a lot of noise. Or a mobile phone. When it vibrates, notice how much louder it is if you put it on a table! If you stand something else on the table yes you can knock it down a few dB but I don't think damping is enough. Just my feeling, but yes I will try it for the laughs. IF I can find any blu-tac. But the kids have that sort of thing so I can go raid their rooms. 

QCX - I have not noticed any microphonics; if you scratch some of the capacitors you can get crackles yes. But nothing that is transmitted by ordinary vibration through the enclosure and PCB. So there is no problem in QCX/QCX+ I think. 

Avoiding any ceramics in the signal path, is impractical entirely - since the cost would escalate to the sky and it would also make no difference, since there is in general no problem except these specific QSD capacitors. 

Ron - lower value sampling capacitors e.g. 0.1uF do widen the audio bandwidth, as you say the CW filter would fix that... but, still there is some theoretical performance impact because strong interfering signals get through the detector and can theoretically cause intermodulation problems downstream. I thought about this too, changing them to 0.1uF but I am quite reluctant to go down that route since I don't really want to have to go through a whole new round of measurements to see if there are any other impacts and what they are - I am quite keen to avoid many changes to the schematic, so that this project does not consume large R&D time. 

Steve K1RF, yes, all agreed... actually two half-value caps on either sides of the board would be possible because anyway the QCX-mini has SMD components on both sides of the board (see the photos); but, this would also consume greater board area for the caps, twice as much in fact. Slits wouldn't be practical either as you said. And film caps are larger anyway. 

> One other possibility is to stick with SMD components but try to acoustically 
> isolate them from the main pcb in some way. One possibility is to mount the 
> four SMD parts on a subassembly which mates to the main PCB with wires 
> similar to the TCXO for the QCX+. Adds complexity and some cost but may 
> do the job while taking less space than the through-hole caps.

Yes, I was also wondering about this too... It could actually also be an easy solution because I could fit a little sub-assembly into the cut-out space on the top PCB. So there wouldn't be any cost increase. If this ends up a smaller physical space requirement, and if there isn't a way to fit through-hole caps in, this could be a good option. 

Overall it seems the best option and simplest is to try the through-hole caps. As I mentioned, really they are also 0603 SMD caps with wires soldered on and a yellow blob of something, then call them "through hole"... but at least the dielectric is not rigidly fixed to a soundboard (the PCB), evidently the wires are enough deterrent for that. So, I will try this first, see if it fixes it, and if I can fit it on the PCB. Ideally I need to try and do some experiments with the existing prototype PCBs, and ideally avoid another iteration of prototyping which will cause weeks of delay and more time and back ache assembling them. 

Thanks again for all the great suggestions, you guys are the best... the only thing in the world better than QRP Labs kits, is the QRP Labs community! 

73 Hans G0UPL





Virus-free. www.avast.com


@CurtisM
 

As you are leaning, let's go with those 4 or 6 caps being leaded. It's a great cause. Fine to stop development here. Nice progress.

Curt