Keyboard Shortcuts
ctrl + shift + ? :
Show all keyboard shortcuts
ctrl + g :
Navigate to a group
ctrl + shift + f :
Find
ctrl + / :
Quick actions
esc to dismiss
Likes
Search
QMX bad Audio and RF Sweep, no Image rejection
|
I've successfully built a low- and a highband QMX without problems, today I built the midband, but this time something is wrong.
After uploading the firmware, I started with the hardware tests. Audio filter, image and RF sweep are bad on all bands, see excamples below.
?
LPF and SWR are good on all bands.
?
I checked all soler joints on the torroids, checked again for shorts, but couldnt find anything suspiciuos. Touched up most of the solder joints, but no luck.
?
Has anybody an indea what to check?
?
?
Audio filter sweep:
?
Image sweep:
https://roebert.selfhost.eu/nextcloud/index.php/s/9tGfiRMKzx7T2QT
?
RF filter sweep:
https://roebert.selfhost.eu/nextcloud/index.php/s/ztDdwLooTRfZH8Q |
|
I was on vaccation last week, so it took a little longer to check the QMX.
?
I've removed the trifiliar transformer T401, checked it for correct pairs and shorts, cleaned everything and soldered it back in. --> No improvement.
?
?
?
Then I did the same for L401 A and B, counted the windings, checked for shorts, and checked all capacitors. --> No improvement.
?
?
|
|
Your solder joints look good in those pictures. I’m still inclined to look at a problem within one of these inductors before looking elsewhere. L401 an and b were tested end to end for continuity separately and then together in place? ?If trifilar is paired correctly and your soldering is good bottom side I’m persuaded it’s in one of those two inductors. Without getting very specific about what you tested and how.?
happy to chat on the phone too if you like. Pm me and I’ll send my number.? |
|
That looks like a very nicely wound T401.? Improper pairing of leads or shorting in this transformer would cause the symptoms we see, so I was hoping you would find it there - darn.
And with the L401a/b pair, if you didn't get the 0th turn connections both together in the correct mounting hole, would likely have caused similar issues.? Since this part is a departure from the other two kits you did, I was also hoping you would find the problem there - drat.
?
If you have an oscilloscope and signal generator, here is what I would try next (taken from a separate post I recently made).? This is basically to check to see if you are getting the 2 pairs of quadrature signals to the D/A chip (PCM1804, IC407).?
If you have a signal generator that can put a sinewave signal (abt -50dBm or so) at the antenna jack, you can use that to trace the receive signal path with a scope. I used my Elecraft XG3 for this, tuned to a 30m frequency (I chose 10.120MHz).?? A tinySA will also work, as will other signal generators Tune your QMX to the same frequency in CW mode.? If you use an oscilloscope to measure at the input pins 4 and 5 on the PMC1804, you should see a sine wave at about 12.5kHz, about 300mVp-p. ?You have to use AC coupling on your scope, since these pins are DC biased up. Then do the same for the 4th and 5th pins down the other side of the chip. [Alternately these measurements can be done at pins 1 & 7 of the two op-amp chips? If any of these does not have your sine wave, you need to trace back (via the chips and T401 and L401) to find out why.? It is even better if you have a two-channel oscilloscope, and can look at both sides at the same time: they should be offset by 90 degrees (1/4 wave).? Note that the signal goes back to RF on the other side of the mixer chips, so it will no longer be at 12.5kHz, and it will be weaker since it is before the buffers.? If all 4 of those pins do have a reasonably good looking sine wave at 12.5 kHz, properly offset, then the problem is with the PCM1804 or with its outputs or supporting circuitry / connections. |
|
Thanks for the inputs so far!
?
As I have removed, checked, doublechecked and tripplechecked T401 and L401, I'm pretty confident they are good.
?
I have an oscilloscope, but no real signal generator. Maybe I can use my nanoVNA with attenuators to get the signal down to -50 or -70dBm. The nanoVNA doesn't have a sine output, its more like a rounded square wave, ist this a problem for these tests? |
|
Stan,
?
Your post (included below) is an excellent reference for those digging into trouble with the receiver.?
I intend to try it out on my working receiver someday when I have time, just to reinforce the lesson.?
According to this calculator, -50dBm into the receiver is 707 microvolts RMS.
And according to this chart, that S9+23dB, which sounds reasonable:
?
That RMS voltage would be the starting point when I design the attenuator, bringing it down from a 3.3v pk-to-pk square wave.
?
There's an awful lot that must be correct for the received signal to pass through the transmit LPF's, perhaps it would be better to inject the signal into node RX_IN?? Or perhaps into IC402-pin9?? Perhaps IC403 pins 7 and/or 9 before T401 is installed?? And do this with a signal already available on the board, perhaps the 25mhz TCXO going through a specified network of attenuating resistors and a series cap and DC offset.? Perhaps the next rev of the QMX kit can add those resistors? and testpoints, and provide a jumper cable to go between testpoints.? That would be a very cheap addition, and save many a lot of headaches.
?
Regarding your later post in this thread:
?
> On Fri, Nov 8, 2024 at 09:17 PM, Stan Dye wrote:
> The square wave output is not a problem for these tests, but the level is: nanovna-H4 puts out about 420mV (CW mode), and you need something on the > order of 500microV or less.? So that's about a 30dB attenuator on the nanovna, if I did my quick math correctly.
?
That's a slightly smaller signal than the earlier 50dBm=707 microvolts, but still fine.
However, a factor of 1000 in voltage is a factor of 1000*1000 in power,
since power into a given impedance is proportional to the square of the voltage.
So you need 60dB of attenuation, not 30dB.
?
I find it hard to think in dB and dBm, and often need a refresher.
For those not math phobic, here's some fundamental equations and identities where "**" is for exponentiation.
I find python excellent for doing calculations in electronics (especially if complex impedance is involved).
?
Watts = Volts*Amps = Volts*(Volts/Amps),? therefore power rises with the square of the voltage.
?
Power-in-milliWatts = 10**(dBm/10)
Examples:
1 dBm =? 10**(1/10) = 1.26 mW
-50 dBm = 10**(-50/10) = 0.000010 mW
?
VoltsRMS = sqrt(Ohms * Watts)? =? sqrt(Ohms * mW/1000)
Examples:
At 1 dBm:? ? sqrt(50ohms * 1.26mW/1000) = 0.251 Volts RMS
At -50 dBm:? ?sqrt(50ohms * 0.000010/1000) = 0.000707 Volts RMS
?
This webpage ? https://www.giangrandi.org/electronics/radio/smeter.shtml
states: "each S-point is defined as a 6 dB change in signal strength"? ?and
"for frequencies?below 30?MHz,?S9 is defined as a voltage of?50?μV?over 50 Ω at the receiver antenna connector"
?
Jerry, KE7ER
?
?
?
On Tue, Nov 5, 2024 at 05:30 PM, Stan Dye wrote:
|
|
Correction:
Watts = Volts*Amps = Volts*(Volts/Ohms),? therefore power rises with the square of the voltage for a given resistance.
?
?
?
On Sat, Nov 9, 2024 at 08:51 AM, Jerry Gaffke wrote:
|
|
Here's a quick explanation of what dBm and dB are.
I had to think hard, as it's been a few years since I last used this stuff.
?
dBm is an absolute power level, defined as
dBm = 10 * log10(power/1mW)
and a 1mW signal is exactly? ?10*log10(1mW/1mW) = 0 dBm
?
Reversing that equation, we find that? dBm = 10 * log10(milliwatts/1mW)
?
So 1 milliwatt is 10*log10(1/1) = 0 dBm
2 milliwatts is 10*log10(2/1) = 3.01 dBm
4 milliwatts is 10*log10(4/1) = 6.02 dBm
10 milliwatts is 10*log10(10/1) = 10.0 dBm
100 milliwatts is 10*log10(100/1) = 20 dBm
?
dB (decibel) is the difference in dBm (decibel milliwatt)? or dB between two powers.
For example, if an amplifier has 4 milliwatts going in and 100 milliwatts coming out,
it has a gain of? 20dBm - 6.02dBm = 13.98dB
A 13.98 dB attenuator with 100 milliwatts going in will have 4 milliwatts coming out.
?
A more direct way to calculate dB is from the ratio of the two powers:
? 10*log10(powerA/powerB)? ? ? ?Example: ? 10*log10(100mW/4mW) = 13.98 dB
?
These are power levels in Watts, not voltages.
If we are dealing with voltages in the equations for dB and dBm we must know
the impedance at which the voltages are measured in order to know the power level.?
The impedance is often assumed to be 50 ohms resistive, and left unstated.
?
The reason for the use of dB and dBm is that the result of a series string of amplifiers and attenuators
can be calculated by adding the dB gain or loss of each stage instead of multiplying numbers with many many zeros.
Since I don't use dB and dBm daily, I often have to convert to Volts or Watts before it means much to me.
?
Jerry, KE7ER |
|
The equation for dBm is:? ?dBm = 10 * log10(milliwatts)
Reversing that equation we get:? ?milliwatts = 10**(dBm/10)
?
On Sat, Nov 9, 2024 at 10:09 AM, Jerry Gaffke wrote:
|
|
Thank you, Jerry.? Good info.
Please note that since I posted that, I was referred to this article in the Wiki /g/QRPLabs/wiki/37111
which (near the end) suggests this same approach - but that op used a significantly higher input of 10mV, and he notes that the value at the 1804 chip inputs would be about 10x this value.? So perhaps when I did it I had my source set closer to a 3mV value.? So lesser attenuators would not be a problem.
?
Since the inputs to the PCM1804 are at the 12kHz (?) IF bandwidth, and the sine wave input occupies only a small part of that bandwidth, a high level signal is necessary in order to detect it on most of our cheap scopes...
?
And I agree that injecting the signal at other points can be helpful - and is easily done by putting a .01uF capacitor (for DC blocking) on the signal source output, and probing it where you want it.? My original intent was to simply verify first whether the signals were getting to the 1804 - if they are not, then more spots can be checked to trace it.
?
Stan KC7XE |
|
When debugging the receiver, I believe we could safely inject a signal up to 3.3v pk-to-pk
without doing any damage. ? The op amps have a gain of around 10, so they would saturate,
feeding the PCM1804 with 5v square waves.
Safer might be 1v pk-to-pk or lower, which is still easily seen on a cheap scope.
?
Injecting an appropriate S9+ signal into the BNC for receiver tests as you suggest
is the best first step for debugging the receiver on a completed rig.
But a completed rig could have any number of errors in the build,
a progressive build would make finding those errors much easier.
?
If we drive at 100khz or less into RX_IN and short across a receiver BPF
with a cap,? debug could be done with a $30 scope: ?
https://www.amazon.com/Oscilloscope-Handheld-Portable-Automotive-Bandwidth/dp/B0CD1WKP33
This assumes we can get sufficient energy through T401 at such a low frequency.
?
I would like to create a complete set of debug instructions for a progressive build:
?
To expand on that for the receiver:
First step might be to take the raw QMX+ board as delivered, use alligator clips to supply
a current limited 7vdc to the holes for the power connector, a USB connection to a host
to control the rig and catch the receiver audio, check out Vdd and Vcc.
Inject RF into IC403 pins 7 and/or 9, verify results on the USB host.
Next step would be to add T401, short across one of the BPF's,
inject a signal into RF_IN, should now get single signal recepetion.
Third step is install one of the BPF's and feed RX_IN with an appropriate signal.
From there, move on to the transmitter, and then the front panel.
?
At each step the builder can verify it works, and if things suddenly fail
they can inspect only what they just worked on and hopefully find an error.
?
I hope the forum didn't find my rant on dB and dBm annoying.
It helps me understand stuff like that if I explain it in full,
and once I write it up I figure I may as well post it.?
Might be of use to somebody.
?
Stan wrote:
"I was referred to this article in the Wiki /g/QRPLabs/wiki/37111"
?
As that article mentions, there've been a number of failures of the PCM1804.??
Would be good to know if that is still happening since the last firmware upgrade improved the SMPS's.
Easy to imagine that the PCM1804 might fail before the 5.6v zener at D108 manages to clamp Vcc.
?
Jerry, KE7ER
?
?
?
?
On Sat, Nov 9, 2024 at 10:56 PM, Stan Dye wrote:
|
to navigate to use esc to dismiss