|
I recently assembled a QDX rev 6, its the third QDX built, I have a
reasonable amount of building experience including multiple QRP labs
kits.
What I have not been able to resolve, I've researched online
troubleshooting and followed through, but it hasn't helped.
When I do the self test (verified using external signal generator
and not the self testing) my image rejection is non-existance varying
from band to band from between -6 to +10 db.
So far, inspected transformer T2, removed and verified the
windings. Finally with a new core T2 was re-wound an installed. The
A-A, B-B, C-C connections have repeatedly been checked, and that area
of the board gone over a couple of times with a jeweller's loupe for
poor solder or bridges, no fault found. Inputs and outputs of IC4 have
had a cursory check, same for the surface mount components, no obvious
fault found.
Next troubleshooting steps should be?
Thanks,
Ron VE8RT
--
Ron VE8RT <ve8rt@...>
|
One more pieceof information, Sideband suppression is very good, within the published specifications. Image rejection is pretty much zero. Ron VE8RT VE8TEA On Sun, 25 Aug 2024 11:49:28 -0600 "Ron VE8RT" <ve8rt@...> wrote: I recently assembled a QDX rev 6, its the third QDX built, I have a
reasonable amount of building experience including multiple QRP labs
kits.
What I have not been able to resolve, I've researched online
troubleshooting and followed through, but it hasn't helped.
When I do the self test (verified using external signal generator
and not the self testing) my image rejection is non-existance varying
from band to band from between -6 to +10 db.
So far, inspected transformer T2, removed and verified the
windings. Finally with a new core T2 was re-wound an installed. The
A-A, B-B, C-C connections have repeatedly been checked, and that area
of the board gone over a couple of times with a jeweller's loupe for
poor solder or bridges, no fault found. Inputs and outputs of IC4 have
had a cursory check, same for the surface mount components, no obvious
fault found.
Next troubleshooting steps should be?
Thanks,
Ron VE8RT
--
Ron VE8RT <ve8rt@...>
-- Ron VE8RT <ve8rt@...>
|
What is the image frequency?
What was the QDX tuned to when observing that image?
?
--
Allison
------------------
Post online only,?
direct email will go to a bit bucket.
|
Thanks Allison, I was going to do some more reading on the description of operation today if it was slow at work. It wasn't, GPS database updates took most of my time, and our latest and greatest acquisition arrived yesterday, a two place amphibious Air Tractor 802 Fireboss water/retardant bomber new from the manufacturer. I had to check it out, there's work for me to do on it before next fire season. Back to the question, the image rejection figures are unacceptably poor on every band on the QDX using the internal test. I did verify it with an external signal generator, the image shows up on the waterfall. The desired frequency (QDX set to USB, the dial frequency plus audio, in this case 1,500 Hz) waterfall intensity is about the same when I move the signal generator (without changing the RF output level) to 24,000 Hz plus 1,500 Hz (25,500 Hz) below the dial frequency. It makes little difference what band its on or what frequency it is set to. Attached is a screen capture of an image test for 40M. Ron VE8RT On Tue, 27 Aug 2024 08:42:49 -0700 "ajparent1/kb1gmx" <kb1gmx@...> wrote: What is the image frequency?
What was the QDX tuned to when observing that image?
--
Allison
------------------
Post online only,
direct email will go to a bit bucket.
-- Ron VE8RT <ve8rt@...>
|
On 28/08/2024 02:45, Ron VE8RT via groups.io wrote: the image rejection figures are unacceptably
poor on every band on the QDX Ron, Causes for no image rejection starts at T2. The signal is split into two. Bad soldering or windings in wrong holes can give wrong phase or just one output.. T2 is most likely but one of the signals may be missing in the path to IC10. 73 Alan G4ZFQ
|
Thanks Alan, after several tests of T2, removing it and re-installing it, verifying every solder connection, finally installing a newly wound replacement, and verifying the phasing and solder connections and that there were no shorts, I'm confident that I can rule out T2 as the problem. Using an oscilloscope I took a look at the inputs and outputs of IC10, all seem present, checked the components in the path to the inputs and outputs, so far the problem has not been found. What I do not quite understand yet, how do I have good unwanted sideband rejection while having no image rejection? A couple of hours ago I tried to post to the list with a very small screen capture graphic of the image test, but I haven't seen my own post in my inbox yet. Ron VE8RT On Wed, 28 Aug 2024 08:42:41 +0000 "Alan G4ZFQ" <alan4alan@...> wrote: On 28/08/2024 02:45, Ron VE8RT via groups.io wrote:
the image rejection figures are unacceptably
poor on every band on the QDX Ron,
Causes for no image rejection starts at T2.
The signal is split into two.
Bad soldering or windings in wrong holes can give wrong phase or just
one output..
T2 is most likely but one of the signals may be missing in the path to IC10.
73 Alan G4ZFQ
-- Ron VE8RT <ve8rt@...>
|
For clarity,
?
In IQ SSB RX image is the same as saying opposing unwanted sideband.
So if a signal is tuned for 1KHZ tome the image is always dial frequency
- +1khz or 2khz away.? If the QDX QSD (IC4) translates the spectrum
down to two audio channels one with a 90degree offset. So the audio
applied to ADC(IC10) has no sideband suppression as that is done in
software.? This remains true at the 12khz IF.? ?There are many questions
in my mind about the internal processes.
?
So if there is poor or no opposing sideband suppression one or the other
audio channel is not working (IC4, IC7and IC9 or even IC11) and of
course paths between them.??
?
The sweep is inconclusive as the scale is unclear.? if its HZ then opposing
sideband is bonkers.? IF it Khz something else is going on.? The manual
is unclear as to what that screen shows.
?
In the sweep the image (opposing sideband)? is not suppressed.
?
FYI even if not suppressed the SSB will sound good, however when
tuning across a carrier with both sideband present it will tune down
to? zero and then up.
?
In this case we are talking about a spurious frequency aifits ~24khz away.
We call that a spur.
?
The spur can be related to digital noise from PS, Oscillating opamp,?
and even a harmonic or sub-harmonic of the A/D sample rate. or
other external signal.? ?FYI the sample rate is 48ksps.
?
--
Allison
------------------
Post online only,?
direct email will go to a bit bucket.
|
On Wed, Aug 28, 2024 at 10:39 AM, ajparent1/kb1gmx wrote:
The sweep is inconclusive as the scale is unclear.? if its HZ then opposing
sideband is bonkers.? IF it Khz something else is going on.? The manual
is unclear as to what that screen shows.
?
I thought the manual was pretty clear this was khz? From the 1.10 manual: "In order to be able to visualize and measure the image rejection, firmware 1_05 introduced the
¡°Image sweep¡± screen, which sets the VFO to the band center frequency as defined in the Band Configuration screen, then sweeps the signal generator input from -30.5kHz (below) to +7kHz (above). "
?
-Nate
N8BTR
|
Firmware is 1.10 from the sweep screen image.
?
Don't have the 1.05 manual.?
Must have 1.05 operation manual 1.05A.
?
With that the image is NOT suppressed as it should be.
?
?
Allison
------------------
Post online only,?
direct email will go to a bit bucket.
|
I'm not making progress with this problem, other than I think that if it were a component or construction issue then I'd also see poor unwanted sideband rejection too, is this a correct assumption? If its a firmware issue with the SDR then it is beyond my capability. With no path forward I think this, my third QDX, will sit in a box until its needed as a parts donor to repair another one or for another project. Ron VE8RT VE8TEA On Wed, 28 Aug 2024 09:07:00 -0700 natereik@... wrote: On Wed, Aug 28, 2024 at 10:39 AM, ajparent1/kb1gmx wrote:
The sweep is inconclusive as the scale is unclear.? if its HZ then
opposing
sideband is bonkers.? IF it Khz something else is going on.? The manual
is unclear as to what that screen shows.
I thought the manual was pretty clear this was khz? From the 1.10 manual: "In order to be able to visualize and measure the image rejection, firmware 1_05 introduced the
¡°Image sweep¡± screen, which sets the VFO to the band center frequency as defined in the Band Configuration screen, then sweeps the signal generator input from -30.5kHz (below) to +7kHz (above). "
-Nate
N8BTR
-- Ron VE8RT <ve8rt@...>
|
Hi Ron and all,
?
I think I have the same problem, no image rejection.
?
So, I've built a high-band QDX, Rev 6. Transmission is good on all bands. Filter sweeps look good (apart from that the peaks are at somewhat higher frequencies than the target bands).? Like Ron, I desoldered T2, checked it and soldered it back, no change.
?
Putting the QDX in IQ mode, and looking at the audio on the computer using an oscilloscope program (AudMeS on Linux), I see a signal only in the right channel, with the left channel showing exactly 0 (even when I zoom in vertically). This makes me think the ADC chip is not working, as there should probably be some noise in each channel otherwise. @Ron, could you try the IQ mode and see if you have any signal at all in both channels?
?
Another issue I see is a strong, broad? noise peak in the lowest few 100 Hz of the RX audio, see attached screenshot.
?
Does anyone have suggestions for how to continue debugging? The radio is quite usable already, I've had several QSOs with it, but I'd really like to fix the receiver problems.
?
Best,
Fredrik
|
You need to Un Check "Flatten" in WSJTx.?
Ron
On Thu, Oct 10, 2024 at 10:30?AM Fredrik (OH1HSN) via <fjansson= [email protected]> wrote: Hi Ron and all,
?
I think I have the same problem, no image rejection.
?
So, I've built a high-band QDX, Rev 6. Transmission is good on all bands. Filter sweeps look good (apart from that the peaks are at somewhat higher frequencies than the target bands).? Like Ron, I desoldered T2, checked it and soldered it back, no change.
?
Putting the QDX in IQ mode, and looking at the audio on the computer using an oscilloscope program (AudMeS on Linux), I see a signal only in the right channel, with the left channel showing exactly 0 (even when I zoom in vertically). This makes me think the ADC chip is not working, as there should probably be some noise in each channel otherwise. @Ron, could you try the IQ mode and see if you have any signal at all in both channels?
?
Another issue I see is a strong, broad? noise peak in the lowest few 100 Hz of the RX audio, see attached screenshot.
?
Does anyone have suggestions for how to continue debugging? The radio is quite usable already, I've had several QSOs with it, but I'd really like to fix the receiver problems.
?
Best,
Fredrik
|
Thank you @Ronald , unchecking "Flatten" did help against the low-frequency peak in the spectrum, so that turned out to be a separate problem.
?
The lack of image rejection remains. Some more research on that: I connected a signal generator (tinySA) to the QDX input, and tuned it near the reception frequency. With an oscilloscope I measured an audio-frequency signal at both sides of C58 and C60, with a 90 degree phase shift between the two capacitors.This indicates that T2 is installed correctly. I also confirmed that the audio signals reach the ADC chip.
?
In IQ mode with a sound card oscilloscope program, I see a nice sine signal in the right channel and zero in the left.
?
Two possibilities come to mind:
- the ADC chip is broken so the left channel is dead
- some pins of the ADC chip are not properly soldered so that the left channel doesn't work
?
I'm considering trying to re-solder the ADC chip by dragging a ball of solder along the pin rows with a soldering iron, then use a solder wick to clear any bridges created.
?
Did anyone else run into this problem?
For the future, would it be possible to add a test in the firmware showing the two audio channel amplitudes?
?
?
?
|
In my QDX, damage to the mixer chip IC4 FST3252 led to the fact that I received with the same volume stations on the main frequency and 24 kHz higher. That is, the mirror channel of reception was not suppressed. Replacing the chip solved the problem.
--
73, Oleh, UY2ZA
|
Randomly dumping new solder on fine pitch parts can cause more trouble than it cures.
Unless you can get the solder flowing exactly right, it's better to inspect first and use the soldering iron sparingly.
?
Get lots of light and as much magnification as you can, then look for solder bridges
A bridge could be a very fine whisker, invisible without magnification.?
Poke at each pin with an x-acto knife blade tip to see if any are loose.
Perhaps gently draw the knife tip between suspect pins to take out possible whiskers.
?
For magnification consider these, listed in order of cheapness:
1)? Strong reading glasses
2)? Desk lamp with magnifying lens
3)? Binocular headband magnifier
4)? USB microscope ($20 or less, has a cell phone imager plus a plastic macro lense, led lights)
5)? A solderscope (a homemade binocular microscope) ?http://www.qsl.net/k9ac/solderScope.html
6)? A $200 binocular microscope such as the AmScope-SE400-Z
7)? A $1000 name brand binocular microscope
?
We had #2 and #7 at work, I used them a lot.??
Bought #6 for use at home, it's been very handy.
If I had the time then, I would have gone with #5 instead.
?
The USB microscopes can be pretty good, but binocular vision helps more than you might think.
Most of the time I just reach for the reading glasses.
?
Jerry, KE7ER
?
?
?
On Sat, Oct 12, 2024 at 02:14 AM, Fredrik (OH1HSN) wrote:
I'm considering trying to re-solder the ADC chip by dragging a ball of solder along the pin rows with a soldering iron, then use a solder wick to clear any bridges created.
|
On Sat, Oct 12, 2024 at 10:32 AM, Jerry Gaffke wrote:
The USB microscopes can be pretty good, but binocular vision helps more than you might think.
I agree 1000%, good binocular magnification (it doesn't have to be a lot, I usually end up at 8X or so) and LOTS of light, preferable coaxial with the microscope objectives, is the way to go.? I use an LED ring illuminator and 2 LED desk lamps to thoroughly illuminate my work area.? I find USB 'scopes highly lacking.
?
Good #7SA tweezers (antimagnetic is important as many SMT components are magnetic due to nickel barrier plating) are a must; buy at least 3 pair as you need two and you will inevitably drop one, causing the "elf shoes" phenomenon.
73, Don N2VGU
?
|
Have been getting by with a bright desk lamp or two since the low end
AmScope just has a dim gooseneck light that wasn't worth messing with.
But on Don's hint I did a web search for "ring illuminator", there's lots of
aftermarket devices to choose from?and some are quite inexpensive.
?
Other hints:
Put a plastic baggie over the eyepieces when not in use to keep the dust off.?
And be aware those binocular microscopes can tip over when swung to the side.
?
On Sat, Oct 12, 2024 at 01:22 PM, Donald S Brant Jr wrote:
I use an LED ring illuminator
|
Yes. I have a ring illuminator attached to my 10X power binocular microscope.
The bottom part of the microscope is rectangular and the illuminator has three screws to fix it into place, so I attached three, small, stick-on rubber feet. They provide a solid mount for the illuminator.
Have been getting by with a bright desk lamp or two since the low end
AmScope just has a dim gooseneck light that wasn't worth messing with.
But on Don's hint I did a web search for "ring illuminator", there's lots of
aftermarket devices to choose from?and some are quite inexpensive.
?
Other hints:
Put a plastic baggie over the eyepieces when not in use to keep the dust off.?
And be aware those binocular microscopes can tip over when swung to the side.
?
On Sat, Oct 12, 2024 at 01:22 PM, Donald S Brant Jr wrote:
I use an LED ring illuminator
|
On Sat, Oct 12, 2024 at 06:12 PM, Jerry Gaffke wrote:
And be aware those binocular microscopes can tip over when swung to the side.
Not with this stand, the reach of which lets me cover most of my assembly bench. ?
73, Don N2VGU
|
Thanks Oleh! I'm quite confident my problem is further in the chain, as I see nice I and Q signals after the audio amplifiers = at the inputs of the ADC (attached screenshot, measured over the capacitors after the OpAmps)
?
Thanks Jerry and all for the magnification and soldering advice!
I have a desk lamp with magnifying glass to solder under, and I also have an USB microscope. I agree, the USB microscope has some issues. For me mainly that I cannot work well under it, as there is little space. Also there is a bit of lag in the image, making my hands feel strangely heavy when I move a tool under the microscope. It seems it's time to start looking for a binocular microscope. This time, I've used the USB microscope to inspect the soldering.
?
To return to the image rejection issue.
I've measured the ADC chip pins with an oscilloscope and found this:
- the reference and common pins all have a good-looking voltage near 2.5V, same between left and right channel
- there is a good audio signal on each Vin
- simultaneously measuring LRCK and data out, there is data in the right channel but only zeros in the left. (attached image, red is LRCK, yellow is data. Don't mind the amplitudes, scales are different and one of the probes was on 10x accidentally)
?
Frustratingly, once while testing I got a good result on the image rejection test, then it got back to the old behavior of no image rejection and the left channel being silent.
This feels like a loose pin sometimes making contact. But the measurements above make me think the ADC chip is broken. I did try to reheat the chip pins associated with the left channel, with no change.
?
I think I'm stuck here, unless I start replacing the ADC chip. Not expensive but I'm not 100% confident I can solder it... and the radio is decently usable? as it is.
?
@Ron the original author: sorry for hijacking the thread. I don't know if you have the same problem as me. I'd suggest one way to find out is to measure the audio signal
after theOp Amps with an oscilloscope, with a signal source attached. If you loose one of the channels earlier in the circuit like Oleh, you would see it there.
?
Best,
Fredrik
?
?
|
Fredrik (OH1HSN)
Ronald Taylor
Donald S Brant Jr