Self-oscillation on RX - possible cause and fix
-
- Posts: 25
- Joined: 15 Apr 2022, 19:12
Self-oscillation on RX - possible cause and fix
Hello everyone and happy Easter.
Several days ago I assembled (tr)uSDX bought from Up Tech Store on aliexpress. Very nice piece of kit, but unfortunately it had all sorts of ringing and oscillating issues in receive mode, either SSB or CW. In fact, on CW mode the issue was even more pronounced.
Initial observation: at volume +11 very noticeable ringing with intermittent audio oscillation while on the speaker, but less so when using headphones. With the volume set at +12 it would simply oscillate and receiver performance is ruined, even when using headphones. Triggering any of the switches on vol +11 would cause short damped oscillation in the audio range. My first thought was that there was Class II MLCC in the audio path both creating and picking up vibrations from the PCB. Btw. MLCCs that are not Class I (NP0, C0G) are known also as "singing capacitors".
Cause of the issue: Investigation of the schematic diagram led me to the QSD filtering capacitors C25-C28. Here is a link to the video where I tried to record the piezoelectric effect on those capacitors https://www.youtube.com/watch?v=jImurYa7SOg .
Fixing the issue: Well, I didn't have a small enough 470nF at hand, only some 100nF 63V polyester capacitors in 5mm pitch (7.2x2.5x6.5 mm size), bundled them together in a 2x2 pack with some painter tape, used a 5 short pieces of some insulated thin wire and placed them on the unpopulated part of the Main PCB left of the opamp (if J1 is left/up).
Results: Clean audio on both speaker and headphones, even heard VU2HW on 7120k with my 1m dia magloop prototype with all sorts of European strong stations creating pile-up. AGC attack time is a bit too slow for my taste, but that's just me . Also, it seems like some aliasing is introduced, probably 100nF is not enough. The receiver is usable up to volume +13 with att2 0, too much THD with volume 14-16, I'm not even sure if it oscillates anymore. But now it's a joy to listen to SSB stations with a filter at 4k, more than listening on my ft-450
Things left to inspect: When in CW mode, there is some ringing noticeable, especially when a filter is set to 200Hz or lower. Slight tapping on C22 and C29 exhibits the same piezo effect as was with QSD capacitors, but is less pronounced. Before changing QSD capacitors I couldn't discover that. Unfortunately, checking that will probably wait for next weekend since I'm away from my shack these days.
Sidenote: Not all MLCCs have that piezo effect in the audio range. NP0 or C0G capacitors are from CLASS I and are allowed as such to be used in the audio path. There are numerous papers and application notes on that. But I don't think there are any over some 1nF even in 0805 sizes, let alone 0603 used in this transceiver. But Class II is totaly a different beast under some 30kHz, especially when in SMD format mounted on FR-4. I used Superlux HD_681EVO headphones and Yaesu FT-450 during fixing this issue.
Several days ago I assembled (tr)uSDX bought from Up Tech Store on aliexpress. Very nice piece of kit, but unfortunately it had all sorts of ringing and oscillating issues in receive mode, either SSB or CW. In fact, on CW mode the issue was even more pronounced.
Initial observation: at volume +11 very noticeable ringing with intermittent audio oscillation while on the speaker, but less so when using headphones. With the volume set at +12 it would simply oscillate and receiver performance is ruined, even when using headphones. Triggering any of the switches on vol +11 would cause short damped oscillation in the audio range. My first thought was that there was Class II MLCC in the audio path both creating and picking up vibrations from the PCB. Btw. MLCCs that are not Class I (NP0, C0G) are known also as "singing capacitors".
Cause of the issue: Investigation of the schematic diagram led me to the QSD filtering capacitors C25-C28. Here is a link to the video where I tried to record the piezoelectric effect on those capacitors https://www.youtube.com/watch?v=jImurYa7SOg .
Fixing the issue: Well, I didn't have a small enough 470nF at hand, only some 100nF 63V polyester capacitors in 5mm pitch (7.2x2.5x6.5 mm size), bundled them together in a 2x2 pack with some painter tape, used a 5 short pieces of some insulated thin wire and placed them on the unpopulated part of the Main PCB left of the opamp (if J1 is left/up).
Results: Clean audio on both speaker and headphones, even heard VU2HW on 7120k with my 1m dia magloop prototype with all sorts of European strong stations creating pile-up. AGC attack time is a bit too slow for my taste, but that's just me . Also, it seems like some aliasing is introduced, probably 100nF is not enough. The receiver is usable up to volume +13 with att2 0, too much THD with volume 14-16, I'm not even sure if it oscillates anymore. But now it's a joy to listen to SSB stations with a filter at 4k, more than listening on my ft-450
Things left to inspect: When in CW mode, there is some ringing noticeable, especially when a filter is set to 200Hz or lower. Slight tapping on C22 and C29 exhibits the same piezo effect as was with QSD capacitors, but is less pronounced. Before changing QSD capacitors I couldn't discover that. Unfortunately, checking that will probably wait for next weekend since I'm away from my shack these days.
Sidenote: Not all MLCCs have that piezo effect in the audio range. NP0 or C0G capacitors are from CLASS I and are allowed as such to be used in the audio path. There are numerous papers and application notes on that. But I don't think there are any over some 1nF even in 0805 sizes, let alone 0603 used in this transceiver. But Class II is totaly a different beast under some 30kHz, especially when in SMD format mounted on FR-4. I used Superlux HD_681EVO headphones and Yaesu FT-450 during fixing this issue.
73 de Marko, 9A6PAF
-
- Posts: 10
- Joined: 03 Apr 2022, 09:52
Re: Self-oscillation on RX - possible cause and fix
Hello, I have the same problem. Kit purchased in China from the Up Tech Store.
Krzysztof
Krzysztof
-
- Posts: 25
- Joined: 15 Apr 2022, 19:12
Re: Self-oscillation on RX - possible cause and fix
Hi Krzysztof,
If you are able and confident enough to do the modification, you may try replacing QSD capacitors with small poly capacitors by putting them as close as possible to QSD IC and with the shortest wires posible. I used 0.25 mm single core wire, with insulation it's about 0.5 mm in diameter. Otherwise, rather leave it alone until some proper solution is found. It's very easy to mess up PCB pads in 0604 sizes.
If you are able and confident enough to do the modification, you may try replacing QSD capacitors with small poly capacitors by putting them as close as possible to QSD IC and with the shortest wires posible. I used 0.25 mm single core wire, with insulation it's about 0.5 mm in diameter. Otherwise, rather leave it alone until some proper solution is found. It's very easy to mess up PCB pads in 0604 sizes.
73 de Marko, 9A6PAF
-
- Posts: 712
- Joined: 30 Dec 2021, 19:18
- Contact:
Re: Self-oscillation on RX - possible cause and fix
Thanks for your detailed insight.
We have suspected QSD Caps for a long time, and actually there has even been a lot of work done by Hans Summers (QRP Labs)
http://qrp-labs.com/qcxmini/qcxminidev4.html
I´ve ordered some 470nF Vishay THT Caps, to see what it takes to modify and if it works, then we can go from there.
Thanks again for your detailed insight.
73 Manuel; DL2MAN
We have suspected QSD Caps for a long time, and actually there has even been a lot of work done by Hans Summers (QRP Labs)
http://qrp-labs.com/qcxmini/qcxminidev4.html
I´ve ordered some 470nF Vishay THT Caps, to see what it takes to modify and if it works, then we can go from there.
Thanks again for your detailed insight.
73 Manuel; DL2MAN
-
- Posts: 1
- Joined: 07 Mar 2022, 16:09
Re: Self-oscillation on RX - possible cause and fix
Hello, I am anxiously awaiting the solution. I thought it was software so I ordered a bootloader pin kit to do the bootload and callsign download as per the instructions on web page 3 "bootloader" for (tr)usdx. It seems it is a hardware issue though. I'm going to watch your video and see if I can recreate your patch... I'm so looking forward to a working trusdx lol!
-
- Posts: 25
- Joined: 15 Apr 2022, 19:12
Re: Self-oscillation on RX - possible cause and fix
Hello Manuel,
Thank you for the webpage link, I haven't seen that before. Hans did excellent work on tracking down the issue with the button popping noise! That's the same issue that got me doing modifications on my (tr)uSDX, except that I also had oscillations.
Yesterday I changed C25-C28, C22, and C29 with poly film capacitors and the values are now according to the schematic diagram. The issue is almost completely gone, with no button popping noise on SSB mode and no oscillation on volume below 14. In CW mode there is still some feedback, but not as near as it was before. Something I noticed along the way when checking Gerber files is that PWM audio output PCB trace is going under and very close to the R16-R19 resistors and the opamp itself. When I removed C14 from the board, S-meters showed -110dB for the first time with the rf board disconnected. Then something crossed my mind.
What if internal layers are interchanged by a mistake? Ground pour from one layer would end up farther away from the opamp and that could remove opamp shielding and cause leaking of output audio to the most sensitive part of the receiver. That could easily cause instability and PWM makes things even worse. I'm not really up to removing parts of the top and bottom layers to check layer order...well, maybe
What I tried then is using U1D as an audio amplifier with the output filtered with a 1mH/470nF LC network and connected directly to the headphone socket. This reduces the current load on the uC itself and there is no current flowing through that audio trace, albeit the signal is still present there. I can't tell if there is any change in receiver operation, didn't measure if there is a change in current consumption, but it still sounds good I'll try today to completely isolate that trace from the audio signal.
Anyway, thank you and Guido very much for bringing up to life such an excellent project!
Thank you for the webpage link, I haven't seen that before. Hans did excellent work on tracking down the issue with the button popping noise! That's the same issue that got me doing modifications on my (tr)uSDX, except that I also had oscillations.
Yesterday I changed C25-C28, C22, and C29 with poly film capacitors and the values are now according to the schematic diagram. The issue is almost completely gone, with no button popping noise on SSB mode and no oscillation on volume below 14. In CW mode there is still some feedback, but not as near as it was before. Something I noticed along the way when checking Gerber files is that PWM audio output PCB trace is going under and very close to the R16-R19 resistors and the opamp itself. When I removed C14 from the board, S-meters showed -110dB for the first time with the rf board disconnected. Then something crossed my mind.
What if internal layers are interchanged by a mistake? Ground pour from one layer would end up farther away from the opamp and that could remove opamp shielding and cause leaking of output audio to the most sensitive part of the receiver. That could easily cause instability and PWM makes things even worse. I'm not really up to removing parts of the top and bottom layers to check layer order...well, maybe
What I tried then is using U1D as an audio amplifier with the output filtered with a 1mH/470nF LC network and connected directly to the headphone socket. This reduces the current load on the uC itself and there is no current flowing through that audio trace, albeit the signal is still present there. I can't tell if there is any change in receiver operation, didn't measure if there is a change in current consumption, but it still sounds good I'll try today to completely isolate that trace from the audio signal.
Anyway, thank you and Guido very much for bringing up to life such an excellent project!
73 de Marko, 9A6PAF
-
- Posts: 25
- Joined: 15 Apr 2022, 19:12
Re: Self-oscillation on RX - possible cause and fix
Hi,KI5SXE wrote: ↑20 Apr 2022, 11:30 Hello, I am anxiously awaiting the solution. I thought it was software so I ordered a bootloader pin kit to do the bootload and callsign download as per the instructions on web page 3 "bootloader" for (tr)usdx. It seems it is a hardware issue though. I'm going to watch your video and see if I can recreate your patch... I'm so looking forward to a working trusdx lol!
Let's wait for Manuel to confirm if this is a real issue, he is the boss here.
73 de Marko, 9A6PAF
-
- Posts: 25
- Joined: 15 Apr 2022, 19:12
Re: Self-oscillation on RX - possible cause and fix
I opened the bottom layer to check what was beneath it. Audio trace is visible in the first inner layer from the bottom layer. This means that opamp is not shielded from this trace with another layer. I was a bit aggressive with a diamond burr and cut that trace, but I would do that anyway if this was the case
73 de Marko, 9A6PAF
-
- Posts: 712
- Joined: 30 Dec 2021, 19:18
- Contact:
Re: Self-oscillation on RX - possible cause and fix
I call that commitment to the projectmsax wrote: ↑21 Apr 2022, 11:08 20220421_125132.jpg
I opened the bottom layer to check what was beneath it. Audio trace is visible in the first inner layer from the bottom layer. This means that opamp is not shielded from this trace with another layer. I was a bit aggressive with a diamond burr and cut that trace, but I would do that anyway if this was the case
But in fact: Below the whole OpAmp is a ground Plane, and the Signals crossing the Footprint of the OpAmp are opposite to component side. Between this Bottom Layer and the VCC Layer, where Audio is running, there´s also a complete GND Plane....
73 Manuel; DL2MAN
-
- Posts: 25
- Joined: 15 Apr 2022, 19:12
Re: Self-oscillation on RX - possible cause and fix
Ok, but on my board, it's:
- bottom plane (majority of the components)
- vcc plane
- gnd plane
- top plane (buttons, encoder, lcd...)
I checked the group buy web page and it says it's the correct layer stacking order.
- bottom plane (majority of the components)
- vcc plane
- gnd plane
- top plane (buttons, encoder, lcd...)
I checked the group buy web page and it says it's the correct layer stacking order.
73 de Marko, 9A6PAF
-
- Posts: 207
- Joined: 01 Jan 2022, 03:47
Re: Self-oscillation on RX - possible cause and fix
Replaced C22 and C29 with NP0 caps (0.1uF 50V 1%), and replaced C25-28 with film caps (0.47uF 16V 20% [measured ~0.44-0.45uF ish] acrylic) all of which should not be microphonic. Unfortunately the smallest I could find were all 1206 size. I thought perhaps I could squeeze them in and for C22&29 you just barely can. But I would NOT advise this approach for the film caps with C25-28, it is probably not worth the effort. Scraping away resist I was able to tack the left one to ground plane but not well. Other three were soldered to 3 parallel strands of 30g wire and then the two ends of the wire bundle were tacked to ground. Single wire had to be used to finish connecting the top ends of each cap. Physical stress popped off the end of some of the acrylic caps and each time started over...
The self resonance speaker squeel at increased volumes is still there though it is softened some, it does sound better but not gone. No other testing done though.
Not sure if an 805 size could be squeezed into the C25-28 spots. I kind of doubt it. But if so, then a stack of bricks of lower value film caps, or npo caps, might be less traumatic.
73
Jer aa1of
-
- Posts: 712
- Joined: 30 Dec 2021, 19:18
- Contact:
Re: Self-oscillation on RX - possible cause and fix
That´s correct, but on the Bottom Plane itself is a Groundplane below the OpAmp.
In Fact: Top, Bottom and GND Plane are filled with "GND" wherever there´s no signal Trace.
Only VCC Plane is filled with "VCC Plane" for easier routing, as GND and VCC are the most used signals.
73 Manuel; DL2MAN
-
- Posts: 25
- Joined: 15 Apr 2022, 19:12
Re: Self-oscillation on RX - possible cause and fix
Oh, OK Thank you for clarifying this for me, I was previously confused Anyway, I routed the audio signal totally out of the PCB with that trace completely unattached and didn't detect much difference. So, for the time being, the majority of the feedback issues have been resolved by switching 6 ceramic capacitors to film capacitors.
When I was awakened at about 0200UTC, I ran a quick scan on 40m using a 3m circumference mag loop placed inside the apartment, and I clearly heard stations from the US East Coast and Canada, as well as A31NN calling on SSB. On a short path, that's over 17000 kilometers. Given the strong propagation, this clearly shows that the receiver is in excellent shape.
When I was awakened at about 0200UTC, I ran a quick scan on 40m using a 3m circumference mag loop placed inside the apartment, and I clearly heard stations from the US East Coast and Canada, as well as A31NN calling on SSB. On a short path, that's over 17000 kilometers. Given the strong propagation, this clearly shows that the receiver is in excellent shape.
73 de Marko, 9A6PAF
-
- Posts: 25
- Joined: 15 Apr 2022, 19:12
Re: Self-oscillation on RX - possible cause and fix
Hi Jer,Ohwenzelph wrote: ↑22 Apr 2022, 02:36 Not sure if an 805 size could be squeezed into the C25-28 spots. I kind of doubt it. But if so, then a stack of bricks of lower value film caps, or npo caps, might be less traumatic.
I doubt that even 0805 could be squeezed into that space because it is already too crammed. I tried using 220nF instead of 470nF, and the only difference I noticed was that the audio was somewhat brighter. It could, however, be a placebo effect.
73 de Marko, 9A6PAF
-
- Posts: 10
- Joined: 03 Apr 2022, 09:52
Re: Self-oscillation on RX - possible cause and fix
Hello, does this problem occur to all (tr) uSDX owners or only we have such a problem?
Krzysztof
Krzysztof
-
- Posts: 25
- Joined: 15 Apr 2022, 19:12
Re: Self-oscillation on RX - possible cause and fix
It appears to be an isolated issue, as there are numerous transceivers operating normally. It is possible that capacitors from different manufacturers, or even different lots, have different piezoelectric characteristics.
73 de Marko, 9A6PAF
-
- Posts: 7
- Joined: 05 Feb 2022, 16:15
-
- Posts: 10
- Joined: 03 Apr 2022, 09:52
Re: Self-oscillation on RX - possible cause and fix
I can't find 604 NP0 capacitors anywhere
Krzysztof
Krzysztof
-
- Posts: 25
- Joined: 15 Apr 2022, 19:12
Re: Self-oscillation on RX - possible cause and fix
On lcsc.com, the largest capacity I could find in 0603 in C0G or NP0 is 15nF/25V. However, tantalum exists in 0603 470nF and could be stacked on top of 15nF if stacking is required.
73 de Marko, 9A6PAF
-
- Posts: 712
- Joined: 30 Dec 2021, 19:18
- Contact:
Re: Self-oscillation on RX - possible cause and fix
Since this thread is about self oscillation on RX, i want to share my observation:
I could only reproduce it with on-board speaker, so i did not think of it as a problem, since this was never meant to be a serious thing.
I tested it with an external speaker and it never happened to me. I measured the input of the active speaker with 2kohms.
It happened however with 32ohms headphones, i recently tried.
So it seems to be clearly a load issue, rather than qsd capacitor related.
73 Manuel DL2MAN
I could only reproduce it with on-board speaker, so i did not think of it as a problem, since this was never meant to be a serious thing.
I tested it with an external speaker and it never happened to me. I measured the input of the active speaker with 2kohms.
It happened however with 32ohms headphones, i recently tried.
So it seems to be clearly a load issue, rather than qsd capacitor related.
73 Manuel DL2MAN
-
- Posts: 25
- Joined: 15 Apr 2022, 19:12
Re: Self-oscillation on RX - possible cause and fix
Actually, mine would oscillate regardless of what was connected. But, literally 10 minutes ago, I simply removed C19, and it no longer oscillates. I just wanted to see what would happen...and the problem was solved?
If there is a load problem, one of the NAND gates could be used as an amplifier. I did it simply because I could and because it is a simple hack.
If there is a load problem, one of the NAND gates could be used as an amplifier. I did it simply because I could and because it is a simple hack.
73 de Marko, 9A6PAF
-
- Posts: 712
- Joined: 30 Dec 2021, 19:18
- Contact:
Re: Self-oscillation on RX - possible cause and fix
I did that, and it didn´t help
I even introduced 1 K Ohm in the path between ATMega and Input of the unused Gate -> No Change.
But C19 is an interesting finding. Going to try this now.
EDIT: Removing C19 didn´t change the issue for me. Still howling....
73 Manuel; DL2MAN
-
- Posts: 25
- Joined: 15 Apr 2022, 19:12
Re: Self-oscillation on RX - possible cause and fix
I was hoping that would help, but it didn't. OK, I'll wait for another trusdx from the PL/EU group buy to compare it to, unaltered.
73 de Marko, 9A6PAF
-
- Posts: 19
- Joined: 10 Jan 2022, 14:13
Re: Self-oscillation on RX - possible cause and fix
Digikey has 47nF metalized acrylic film capacitors size 0603
16MU473KY11608 maybe two of those stacked could be a workable solution.
16MU473KY11608 maybe two of those stacked could be a workable solution.
-
- Posts: 25
- Joined: 15 Apr 2022, 19:12
Re: Self-oscillation on RX - possible cause and fix
That's fine for C22 and C29, but it's too low for QSD capacitors. I'm currently using 220nF for C25-C28, which is the bare edge for removing signals from the neighboring SSB channel. On CW, things are even worse.
73 de Marko, 9A6PAF
Users browsing this forum: Bing [Bot] and 10 guests