Pages: [1]   Go Down

Author Topic: Why is my 20m SSB Transceiver Oscillating on Transmit?  (Read 439 times)

MACLEINN

  • Posts: 10
    • HomeURL
Why is my 20m SSB Transceiver Oscillating on Transmit?
« on: April 28, 2022, 02:22:17 AM »

Hi there,

This is a kind of follow-up to my previous thread in relation to switching between Rx and Tx (see
https://www.eham.net/community/smf/index.php/topic,135367.msg1252049.html#msg1252049).  Back then, I noticed that my transceiver build was oscillating on transmit and I put that down to shabby wiring and layout.

However, since then I've cleaned up my act somewhat and the problem persists (although I have a solution - read on!).  I decided to take it to you guys for consideration and advice.  The question here is: why is this happening and why does my solution actually work.  I'll also gratefully accept other mitigation techniques, since my solution does impact the output power somewhat.

Anyway, lets get into it.  Here are some images to start with:

The transmitter chain block-->
https://imgur.com/mBprgHs

Birdseye of the overall layout (complete with mitigation hacks for the eagle eyed)-->
https://imgur.com/i1Muzqi

The offending oscillation; no audio input; 106Vpp-->
https://imgur.com/KqeJ6eL

Ok, so taking a breath, you can see the issue.  No audio and the transmitter takes off on its own.  When it does this, there is some power supply ripple and this despite all the mitigation techniques I know of: Q killing resistors, chokes, capacitors, etc.  I will say that out-of-the-box, the W6JL would seem like an angry oscillator waiting to happen, but the problem doesn't go away when I choked the DC feed to the output transformer and added additional components to stabilize the bias circuitry.

Supply ripple-->
https://imgur.com/xURKpoa
https://imgur.com/eDkL3Cw

Anyway, here is what actually worked for me-->
https://imgur.com/1aL2Kwp

A -3dB pad at the input to the pre-amp, so that's in between the 40dB pre-pre amp and the pre-amp (see block diagram point A).  Note, the attenuation is critical, it most be *at least* -3dB of attenuation.
Also interesting is that if I attenuate the signal earlier in the chain, it isn't as effective (it still oscillates, but there's an improvement).

So, here are my questions: why did this work and are there any other solutions I might try based on what you see in the images?

Additional images-->
The final amplifier: https://imgur.com/HKdmOZs
Pre-amp circuit: https://imgur.com/3LfgAbA
Pre-pre amp circuit: https://imgur.com/rJ2KqQ4
SSB audio transmission, no oscillation: https://imgur.com/Phr1KqD


Regards.
« Last Edit: April 28, 2022, 02:29:46 AM by MACLEINN »
Logged

MACLEINN

  • Posts: 10
    • HomeURL
Re: Why is my 20m SSB Transceiver Oscillating on Transmit?
« Reply #1 on: April 28, 2022, 03:21:52 AM »

Apologies, point A in the block diagram is in the wrong place.  The -3dB pad (the one inserted to cure the oscillation) is between the pre-pre-amp and the pre-amp.  Not at point A as I mentioned in my original post.
Logged

G4AON

  • Member
  • Posts: 2178
Re: Why is my 20m SSB Transceiver Oscillating on Transmit?
« Reply #2 on: April 28, 2022, 05:37:50 AM »

From the block diagram it looks like you have initially not placed attenuation between amplifier stages. I found that amplifiers aren’t necessarily as “perfect” as we think, at some frequencies their input and output impedance may be anything except 50 Ohms.

In my two home brew transmitter designs, I had to add a 3 dB attenuator between the low power amp and the 10 Watt PA.

The block diagram is:
https://www.qsl.net/g4aon/pdfs/SSB_TX_Block.pdf

73 Dave
Logged

MACLEINN

  • Posts: 10
    • HomeURL
Re: Why is my 20m SSB Transceiver Oscillating on Transmit?
« Reply #3 on: April 28, 2022, 06:14:55 AM »

Hi Dave (G4AON),
Thanks again for your input.  Actually, I credit you for the solution after your comments in my previous question, so thankyou kindly for that.  I suppose having found a solution that works I'm now interested in the cause mechanism so I can better understand the solution. 
So, to expand on what you said, let me ask some additional questions:

Is it true to say that owing to matching between amplifiers, gain and phase at frequencies other than the target frequency are such that oscillation is certain (gain of 1, phase 180)?

Is it a case that for certain frequencies, the gain is so high that it causes amp saturation?  Does this then affect the power supply (higher frequency content), which ends up affecting earlier amp stages?

If I improve the matching between stages will this help (not just attenuation)?

Are reflections at amp interface points (owing to impedance mismatches) an issue here and how would this set off the oscillation if it were true?

Anyway, you see what I'm after.  I'm hunting for the culprit, the mechanism.  Doubtless whole books have been written on this and I know it's not an easy topic, especially at the system level like we have here.

From the block diagram it looks like you have initially not placed attenuation between amplifier stages. I found that amplifiers aren’t necessarily as “perfect” as we think, at some frequencies their input and output impedance may be anything except 50 Ohms.

In my two home brew transmitter designs, I had to add a 3 dB attenuator between the low power amp and the 10 Watt PA.

The block diagram is:
https://www.qsl.net/g4aon/pdfs/SSB_TX_Block.pdf

73 Dave
Logged

G4AON

  • Member
  • Posts: 2178
Re: Why is my 20m SSB Transceiver Oscillating on Transmit?
« Reply #4 on: April 28, 2022, 06:34:03 AM »

My initial testing was to check the low power amplifier with my spectrum analyser and tracking generator. I then tested the PA in the same way. Both individually were pretty flat from around 1.5 to 30 MHz, however when connected together the response had a “wobble” in the trace. As I had gain in hand, adding the 3 dB attenuator wasn’t a problem and the resulting trace was smooth.

It is easier to add a small amount of attenuation between stages rather than worry about trying to obtain a near perfect match. You will see commercial designs tend to add modest amounts of attenuation between diode mixers, bandpass filters, etc.  2 or 3 dB helps.

73 Dave
Logged

WB6BYU

  • Member
  • Posts: 20896
    • Practical Antennas
Re: Why is my 20m SSB Transceiver Oscillating on Transmit?
« Reply #5 on: April 28, 2022, 06:49:25 AM »

You have a lot of gain at the signal frequency.  Any small
amount of feedback can cause oscillation in that case.

Not only can that feedback be caused by coupling along
the connecting wires, but also any common impedances
in the grounds.  I can’t tell from the photos whether there
is a solid copper board for a backplane that all the modules
are soldered to, but that would certainly help.

The potential problem is that any RF voltage drop across
the reactance of the ground wires is common to the input
and output of a stage, and hence provides feedback.
If two or more modules share a common ground wire,
that makes matters worse.  I’ve run into this problem
before with high-gain audio stages in direct conversion
receivers.

Even with all components mounted over a solid ground
plane, I’ve seen cases where the current distribution
in that ground plane caused feedback.

The relative phase of the input and output from each
module can make a difference:  it can change positive
feedback to negative feedback, for example, although
it may also just shift the oscillation to a different
frequency where the phase relationship is correct.
So something as simple as swapping the output leads
on a transformer might make a difference.

Specific suggestions:

1. Make sure every module has an effective ground plane
on the board.  For existing modules on perfboard, this may
involve running some copper tape or buss wire around the
perimeter and making sure all it is bonded to all ground
connections with short wires.

2. Adequate power bypassing at each stage, rather
than relying on bypassing at the power supply.

3. Direct ground connections between each board and
a common ground plane for the whole unit.  One way to
do this is to use mounting screws in each corner of each
board to connect the board ground to the common ground
plane.  You might need more for larger boards, and I’d
probably solder some jumpers as well.

4. Treat your carrier oscillator as one of the transmitter
stages.

5. Shielding of individual stages may help.


That’s not to say that there might not be other issues,
but the combination of high cumulative gain and the use
of multiple modules that might not be well grounded (it
is difficult to tell from the photo) is what attracts my
immediate attention.

W1VT

  • Member
  • Posts: 6071
Re: Why is my 20m SSB Transceiver Oscillating on Transmit?
« Reply #6 on: April 28, 2022, 08:08:30 AM »

You are assuming that the two 20 dB amplifier stages are unconditionally stable.  That they can terminated with any impedance and not oscillate.
Putting a sharp bandpass filter at the input means that noise outside the bandpass filter from the amplifier will be reflected off he bandpass filter back into the input of the amplifier, where it will amplified by 40 dB.  The 3dB resistive attenuator reduces this feedback .
Logged

MACLEINN

  • Posts: 10
    • HomeURL
Re: Why is my 20m SSB Transceiver Oscillating on Transmit?
« Reply #7 on: April 28, 2022, 08:14:51 AM »

Thanks WB6BYU.  Everything you said makes sense to me.

I tried just now to create an annotated version of the birds eye view of the board, but I gave up. It would just be too cluttered.  But, basically the board in the centre with the 3 TO-220 package regulators is the "star ground" for the supply connections.  I've tried to keep the ground wires (seen mainly twisted alongside their positive counterparts) as short a possible, but of course there's that compromise between an ideal ground (i.e., connecting all boards to a large copper ground plate) and designing things such that you can eventually place this into an enclosure.  I would hope that when I get around finally to putting this in a metal box, that the layout of the modules within will allow me to make shorter ground connections to some common ground point (not necessarily the one I have at the moment).

But, then there is the signal ground.  This is the ground that forms the outer braid of the coax connected between the modules and connects the grounds of the chained modules together.  Again, I've tried my best to keep them short, but again I'm thinking ahead to a time when these modules will be rearranged within an enclosure and I'll need the interconnects to be flexible.  I suppose as always it's a compromise?

But to address your checklist one-by-one:

Yes, each board has an effective ground plane.  The boards that are printed have an extensive ground plane as part of the design.  In some cases SMD parts are used: the two amps either side of the crystal filter are all SMD.

I believe I haven't skimped on the power bypassing, but then again, can you ever have enough?  Put it this way, my attempts to increase bypassing didn't solve the oscillation.

Direct ground connections: I've done this as best I could, but lets say I hope to do better when I get this in the box.

Treat your carrier oscillator as one of the transmitter stages.
Maybe you could expand a bit on that one?

Shielding, the enclosure design will have metal partitions between certain stages.  I may employee FT capacitors on the supply lines at that stage.

Regards. D.



You have a lot of gain at the signal frequency.  Any small
amount of feedback can cause oscillation in that case.

Not only can that feedback be caused by coupling along
the connecting wires, but also any common impedances
in the grounds.  I can’t tell from the photos whether there
is a solid copper board for a backplane that all the modules
are soldered to, but that would certainly help.

The potential problem is that any RF voltage drop across
the reactance of the ground wires is common to the input
and output of a stage, and hence provides feedback.
If two or more modules share a common ground wire,
that makes matters worse.  I’ve run into this problem
before with high-gain audio stages in direct conversion
receivers.

Even with all components mounted over a solid ground
plane, I’ve seen cases where the current distribution
in that ground plane caused feedback.

The relative phase of the input and output from each
module can make a difference:  it can change positive
feedback to negative feedback, for example, although
it may also just shift the oscillation to a different
frequency where the phase relationship is correct.
So something as simple as swapping the output leads
on a transformer might make a difference.

Specific suggestions:

1. Make sure every module has an effective ground plane
on the board.  For existing modules on perfboard, this may
involve running some copper tape or buss wire around the
perimeter and making sure all it is bonded to all ground
connections with short wires.

2. Adequate power bypassing at each stage, rather
than relying on bypassing at the power supply.

3. Direct ground connections between each board and
a common ground plane for the whole unit.  One way to
do this is to use mounting screws in each corner of each
board to connect the board ground to the common ground
plane.  You might need more for larger boards, and I’d
probably solder some jumpers as well.

4. Treat your carrier oscillator as one of the transmitter
stages.

5. Shielding of individual stages may help.


That’s not to say that there might not be other issues,
but the combination of high cumulative gain and the use
of multiple modules that might not be well grounded (it
is difficult to tell from the photo) is what attracts my
immediate attention.
« Last Edit: April 28, 2022, 08:24:42 AM by MACLEINN »
Logged

MACLEINN

  • Posts: 10
    • HomeURL
Re: Why is my 20m SSB Transceiver Oscillating on Transmit?
« Reply #8 on: April 28, 2022, 08:21:19 AM »

Bingo! That's a great observation.  I think that goes a long way towards explaining what's happening and now I'm thinking should I also terminate the filter with a pad also?  This thread has been very helpful.  Regards. D.

You are assuming that the two 20 dB amplifier stages are unconditionally stable.  That they can terminated with any impedance and not oscillate.
Putting a sharp bandpass filter at the input means that noise outside the bandpass filter from the amplifier will be reflected off he bandpass filter back into the input of the amplifier, where it will amplified by 40 dB.  The 3dB resistive attenuator reduces this feedback .
Logged

MACLEINN

  • Posts: 10
    • HomeURL
Re: Why is my 20m SSB Transceiver Oscillating on Transmit?
« Reply #9 on: April 29, 2022, 02:08:34 AM »

Unfortunately, the problem has not gone away! I decided to try for a signal report last night, just to see where I'm at.  Antenna had a measured SWR of 1.2:1. Not perfect, but reasonable.  Receiver worked well with plenty of stations received.  Clicked the transmit key and immediately I could see it was oscillating (I can see the current draw on the P.S.)

I tried adding attenuation both before the pre-pre-amp and after. This worked eventually, but my output power was wayyy down.  Clearly, the transceiver is only barely stable with 50 ohms and I need to make sure it's stable across some (narrow) range of non-ideal conditions.  Disappointing. 

I'm not sure how to set up my test environment so that it simulates a real antenna with a slightly non-ideal SWR?  If I simply attached my 50 ohm load to the end of my long-ish length of coax, would that work?

@G4AON when you were testing the frequency response of your chained amplifiers, how did you deal with the output power from the final amp?  I have a NanoVNA and a TinySA that I use for smaller signals.  I'm reluctant to go near the final stages for fear of blowing up my equipment.

Logged

G4AON

  • Member
  • Posts: 2178
Re: Why is my 20m SSB Transceiver Oscillating on Transmit?
« Reply #10 on: April 29, 2022, 03:18:20 AM »


@G4AON when you were testing the frequency response of your chained amplifiers, how did you deal with the output power from the final amp?  I have a NanoVNA and a TinySA that I use for smaller signals.  I'm reluctant to go near the final stages for fear of blowing up my equipment.

I used a Rigol spectrum analyser with tracking generator (DSA815TG), together with a 50 Watt power attenuator.

73 Dave
Logged

N7EKU

  • Member
  • Posts: 1471
Re: Why is my 20m SSB Transceiver Oscillating on Transmit?
« Reply #11 on: May 17, 2022, 06:32:41 PM »

Hi,

Just checking, your block diagram shows 14.175MHz from the VFO going into the tx mixer.  That's not where you're setting your VFO is it?

Your interstage transformers seem to show no attempt at impedance matching.  What it the winding ratio and number of turns on these?

73
Logged
Mark -- N7EKU/VE3

WB6BYU

  • Member
  • Posts: 20896
    • Practical Antennas
Re: Why is my 20m SSB Transceiver Oscillating on Transmit?
« Reply #12 on: May 19, 2022, 02:30:30 PM »

Quote from: MACLEINN

I'm not sure how to set up my test environment so that it simulates a real antenna with a slightly non-ideal SWR?  If I simply attached my 50 ohm load to the end of my long-ish length of coax, would that work?



The longish length of coax on the dummy load will still show 50 ohms,
unless you use 75 ohm coax.  In that case, the SWR will vary between
1.0 : 1 and 2.25 : 1 as you change the length of the cable.

You can calculate the result using VK2OMD's online calculator.  A 5' piece
of 75 ohm coax will give about 1.5 : 1 (59 + j21 ohms).  To get a negative
reactance with the same SWR requires about 25' of coax.

Another common approach is to put a dummy load on the output of an
antenna tuner, and adjust it for the desired load impedance.

KB1GMX

  • Member
  • Posts: 2252
Re: Why is my 20m SSB Transceiver Oscillating on Transmit?
« Reply #13 on: June 01, 2022, 02:49:48 PM »

OK...

Main problem is a layout that is RF poor and will encourage uncontrolled feedback.

First rule is inputs should be far from outputs.   So for the power amp state the
input is one side and the output should be the other side.  The fact that the input
transformer is near the output transformer is a really bad thing.

How I do it.  I cut off the Drain pin. ( I assume those are IRF510) I don't use it as
the tab is also drain.  insulating shoulder washer and solder tabs from teh drain side
with short leads to current splitting transformer and the output transformer should
be very short and on the output side (tabs point to outpout). 

The input side should also  opposing the output (other edge) and the transformer
there along with very short leads to the gates of the mosfets.  The source leads to
ground (massive area) should be as short as possible.

The driver is elsewhere and RF via coax to the output board.  It the drie ris on the
same board the board should be longer to keep with the inptu is far from output. 
Wrapping the circuit around the board to make ti fit will have ground current getting
to the driver and encourage instability.

Despite what people say the IRF510 is a VHF device and will do wuote fine at 6M
(I use them there!) so treat the whole layout as it if going to work at 100mhz or higher.
As given a chance it will!   Those that do not get power from them are not matching
the input impedance (or output )!  The IRF510 input is a large capacitor with a small
series resistance and inductance so at 20M it looks like about 80 ohms at RF (20M)
and highly capcitive.  Failure to deal with that results in low gain and instability.

IT is also reasonable to use negative feedback from drain to gate in the range of
220 to 1K ohms plus series dc blocking cap of .1uf.  Again layout and short leads.

The rest of the stages are likely suffering the same ills.

FYI the circuit looks like a copy of many marginal designs I've seen.

Allison
Logged
Pages: [1]   Go Up