Re: QMX, one of op-amps self-oscillates at 25 MHz (or picks up the system clock?)
Hi Hans,
Feedback from a CMOS gate output to its input will result in oscillation at some frequency unique to each chip. It’s impossible to predict the oscillation frequency but in general, yes, it will typically be a high frequency. This is not good practice but if you want to bias a logic gate, you should use a pull-up and pull-down resistor at the input to create a half-supply bias. How is this different than using the feedback resistor? Well, the gate won’t oscillate continuously without feedback. Rather, thermal and/or electrical noise present at the input will cause the output to randomly toggle (no fixed frequency) or it may just get stuck high or low.
Using a CMOS logic gate as a high-gain amp is a risky practice. Logic gates are designed to be switches, not amplifiers for low voltage input signals. Running a 1V p-p signal into a logic gate powered at 3.3V or 5V will probably work but the totem pole input stage will carry switch-through current and may spend too much time in the linear region. This can cause unwanted high-frequency oscillation during each state transition. This is why logic gates have a maximum rise-time and fall-time spec - to minimize time spent in the linear region so the gate won’t oscillate during transitions.
A better way to implement this would be to use a Schmitt trigger biased at half supply using its positive feedback to rapidly pass through the linear region. Still, I’ve never seen a CMOS gate wired with a feedback component in my 45 years of engineering. In fact, doing such a thing would absolutely prevent passing design reviews.
Tony
On Thu, Jan 9, 2025 at 2:36?PM Hans Summers via <hans.summers= [email protected]> wrote:
The 833 Hz is low-frequency oscillation, likely being generated in IC202. R207 (470K) is providing a DC bias at the input of the gate which can cause the gate to operate in its linear region, acting as a high gain amplifier rather than a logic buffer. I’m not sure why this gate has a feedback resistor as we never do this for logic gates for the very reason we don’t want them operating in the linear region.
It's a quite common technique to effectively DC bias the gate's input at almost exactly the midpoint between it's 0 and 1 input voltage thresholds. An AC coupled small signal connected at this input is then very well "amplified" by the gate, with very high gain, converting it as perfectly as possible to a 3.3V square wave. It's very effective.?
If there is no input signal at all, the gate will oscillate at some extremely high frequency, somewhere at VHF. But as soon as even a small input signal is applied, the gate locks onto it and amplifies it near-perfectly with very high gain (creating the square wave). Which is another way of looking at it.?
You will find in QMX if you remove this resistor, the gate will probably not operate. Zero output. Or certainly not reliably. Because now the small signal from the TCXO (about 1Vpp) is centered on what? Zero maybe? You have to provide some DC bias. You could put 1M to ground and 1M to Vdd or something. It would probably also work (modern CMOS are quite symmetric). But a better way of achieving the same thing is the feedback resistor.? 73 Hans G0UPL
|
Has anyone tried using a GPS antenna as used on the Traquito just on leg of 1" that works very well on a U4B?
?
Tom
|
Re: QMX, one of op-amps self-oscillates at 25 MHz (or picks up the system clock?)
Adam,
?
You may be chasing a red herring.? In many older or even current inexpensive digital oscilloscopes, if you view the waveform at a much lower rate than its true value, you can see all sorts of sub-sampling effects and aliasing, showing what appear to be real signals, but are not really present.? This is because the scope firmware decimates the sampling rate it uses for low frequencies, causing aliasing if the signal is really at a higher rate. ?My new inexpensive digital scope does the same thing.
?
If your 25MHz clock was truly fully modulated by that low frequency 'signal' you are seeing, you would also see traces of varying amplitudes when viewed on the scope at 25MHz - but the 25MHz waveform is clean, no trace of varying amplitudes due to sinusoidal modulation, which would definitely be captured by the full rate sampling.? I think what you are demonstrating may be the effects of digital sampling in your oscilloscope, rather than a true modulating signal.
?
To verify, find someone with an analog scope and check with it.
?
Stan KC7XE
|
Re: QMX, one of op-amps self-oscillates at 25 MHz (or picks up the system clock?)
Hans Summers via groups.io <hans.summers@...> wrote: Downstream of IC202 there isn't any "almost", this is the very reason why
IC202 is there. You will have a 3.3Vpp output at IC202 output. I do. The scope shows some overshoots, but probably the probe needs to be calibrated. Can you take a similar video please for the signal at IC202 output? Yes -- posted below. I'm still having trouble trying to convince myself that the 25 MHz TCXO is
at fault here rather than - sorry to say it - some weird artefact
introduced by the measurement itself (at the TCXO output / gate input). Don't be sorry -- in the realm of high frequencies I'm not sure of anything :) HF never fails to find ways to surprise me. BTW, I find this frequency weird as well. It's very low, compared to the nominal frequency of the TCXO. It's probably way too low to result from some stray capacitances or inductances. It might be some weird artefact. But the oscilloscope never had such artefacts before (I have it for 14-15 years), and symptoms (intermittent CLK0 and CLK1 at 100 Hz, precisely the frequency of the buzzing both audible in the speaker and visible in WSJT-X's spectrum view) fit. Here's a new video with: 1. Output of the IC202 gate at various time bases 2. Signal from the TCXO with C203 (capacitor coupling TCXO with the gate) removed -- so it's a signal on unloaded TCXO 3. VDD after L201 (so the power line of the clocking subsystem: TCXO, gate, synth chip), showing some very minor ripple of the same frequency (I suspect it's the effect of the problem, not the cause of it, but at this point I'm not sure of anything)
|
On Thu, Jan 9, 2025 at 07:02 PM, Len wrote:
plus anything else that I haven't even thought of yet :)
?
?
Some lightweight pole 5-6m or taler depending on the antenna you will be using. There are some nice packable solutions that are less then 0.5m in lenght when folded.
|
On Thu, Jan 9, 2025 at 07:13 PM, Ed Kwik wrote:
Go to YouTube. Search on ham radio sota
Or simply read a good article in 30 minutes instead of loosing time watching 3 hours of videos, same ammount of info and no forced ads.
?
?
|
Re: question 50 watts amp
Consider looking closely at your 50w amplifier input SWR.? It may stress your QMX PA.? There are ways to address it.? I've written here about using the 50w amplifier multi-banded but I went a fully automated route.? You can easily swap out low pass filters on the amp output if you want a more manual approach.? Lowering the QMX power output is a good idea.? As Adam said, the amplifier input attenuator can get stressed if you do some long transmit digital modes with it.? On CW, it shouldn't be a problem.
?
Joshua
|
Len via groups.io <kc3pwb@...> wrote: Here's what I'm using: - battery 4x 18650 + linear 12V regulator. - antenna PAC-12 / JPC-12 with a longer radiating element (5.6 m). If you don't want to spend money on it, you can make a single-band dipole or an end-fed (EFHW). From my experience, end-fed works better than a dipole (it's easier to tune in the field). If you go with one of these, I also recommend a strong line and something like this (I'm not sure how it's called in English, a line throwing weight?): I chose the heaviest one I could find. You also need some hooks / carabiner clips: The line has to be strong in case your weight gets stuck on a branch (and make sure you tie it properly to clips). It happened to me several times. But after one season of playing with dipole and EFHW, I ordered PAC-12 and it made everything so much easier. No more ropes, weights, trying to find a tree, etc... The radiating angle of a vertical antenna is also better for DXing. - tuner I don't use a tuner. I tune the antenna with NanoVNA. Another thing to remember, always make sure that the VNA is charged. The same with radio batteries. - plus anything else that I haven't even thought of yet :) - speaker / headphones - voltmeter for 18650s (before the regulator) - some comfortable holder to put a key - a lightweight chair (optional, you can sit on the ground) - a lightweight table (optional, but it's easier to make notes this way) - something to write on (pen, paper) - head lamp for when it gets dark (with a head lamp you still have two hands free) - a feeder (I'm using 15m of a H155 cable + a choke on it) If you don't want to carry a chair and a table, I recommend doing notes on a smartphone instead of on paper (some notepad app, I'm using Colornote). If you want to work in a group (for example with your local club), I recommend using headphones instead of a speaker. It's easier to focus with headphones, and you're not interrupting others. Some tips when working in a group: split bands (so if you're working on 20m, it's good if someone else uses a different band), be mindful when tuning the antenna (it sweeps the band, so it causes clicks in other radios), be mindful of automatically activated PAs (my QMX enabled PTT in the PA of a guy sitting two meters from me, it was microPA 50). You might want to see my video from building a useful addon for portable work: Here's how it looks like on a field day with my local club:
|
Re: New build QMX - high noise on all bands
Just to close off this thread, following Jeff Moore's WiKi page I injected a signal of 10mV and looked at the output of the 4 Op Amps and saw the expected approximate 12kHz signal at 10 times the input signal so this did indeed seem to confirm that the PCM1804 was probably faulty. Out of interest the 4 Op Amps on my board are LT6231's instead of LM4562's presumably due to a supply issue at the time of manufacture - I suspect it is a good drop-in replacement. I plucked up the courage to replace the PCM1804 using techniques posted on this forum which is a great source of information. The noise dropped back to S0 and comfortably received a high level injected signal. All the self diagnostic plots from PuTTY now look as expected. I used WSJTX to receive WSPR signals from all over Europe and one from Australia so I think the receiver is working well now and I can start playing with my QMX.
I had actually set this construction project as a learning exercise for my son who did a great construction job. It's a shame that a surface mount chip turned out to be faulty. I had never intended to do surface mount work so this forced me out of my comfort zone which is sometimes no bad thing.
I think this is a great little transceiver and a good way to get started on the HF bands. I particularly like the built in SWR bridge allowing SWR and power measurements as well as all the self diagnostics which is great for people with not much test equipment. Well done to Hans and the team and thanks to Jeff, Ludwig, Gus and everyone who posts on this forum.
|
Re: question 50 watts amp
Many thanks Bojan for the information.
Good to hear it works fine.
I would like to make the QMX with PA ?my second station besides my IC7300.
The main reason is the full break in, i really like this.
?
73 hanz
|
Re: question 50 watts amp
Many thanks Adam for the information.
?
73 hanz
|
Here's what I would recommend assuming you're doing CW with the QMX:
?
Battery: This is all pretty low power so you can get away with as little as 3AH of power which will get you more than a couple hours of time.
?
Key: Bring what you know how to use, I will say that using a key made to be taken out in the field is a good idea. You can use a "desk" key but you might not like it.
?
Headphones: Whatever you like here.
?
Antenna: I highly recommend bringing a resonant end-fed antenna - something that will get you 10/15/20/40 or similar. I like the K6ARK if you can get them. They're easy to get one end up somewhere but remember to bring a mast or some throw line (you can use a weight you bring or a rock you find).
?
Tuner: If you do not bring a resonant antenna, you will need a tuner like the Emtech ZM-2 (a personal favorite). Don't forget a patch cable if you use a tuner.
?
Other:
- You might want to bring a short length of coax to connect your antenna to the radio. Here's my own thoughts: if using an end-fed with a counterpoise, you can either use no coax or coax with a choke at the feed point. If you don't have a counterpoise, connect a bit of coax. The QMX is less than 5W but being gentle with it is good practice.
- Bring extra connectors and adapters. You might not "need" it but if you have them with you it's very easy for you and a buddy to try out each others stuff on Field Day or something.
- Make sure you can log effectively with all your little fiddly bits up and running and a key in your hand(s), a kneeboard style writing surface is very popular.
- Test everything beforehand. Everything. Don't just connect stuff, turn it all on and use it including logging contacts. Do it in the mud.
|
Len - a PD capable powerbank 20,000 mAh will provide 36 hours of listening time. Less time if you are the object of a pile-up.
I do like my EFHW antennas, but it's a bit of work to get them high in the tree, if that is even an option. You could try starting with a cheap ground mounted vertical. Inductors can be built or purchased for this antenna...
72 es POTA on de va3rr
toggle quoted message
Show quoted text
On Thu, Jan 9, 2025 at 01:02 PM, Len wrote:
- battery
- antenna
- tuner
|
Re: QMX - Power supply off line test Process
Stan, thanks for your impressions. I will check voltages on Q110 and Q111 also on Q101 and Q102.
Osvaldo LU3MAM
Something is wrong with your 3.3V input regulator circuit.? Inspect carefully for solder blobs or shorts, near the connector pins you soldered or anywhere else - it should not operate this way.? ]
?
1. Measure the voltages on each terminal of Q110 and Q111 when lin_reg_en is grounded, and check to see if pwm_3v3 is floating high (it should be 0V) - maybe somehow Q109 is getting enabled (activating the switched mode supply) like is supposed to happen in Step 3, and you are transferring the input voltage to VDD.
?
2.? Otherwise, VDD should never be able to go above 3.3V - it is the output of a 3.3V regulator, IC101.? But it seems to be regulating fine when you did step 1, so something in step 2 is causing the input voltage to be shorted to the output.? Q101 should turn off Q102 when lin_reg_en is grounded, preventing any voltage from getting to VDD (unless Q109 is on, which it should not be).? So something is not right here.? You can check the voltages on each of the terminals of Q101, Q102, and IC101 when you ground the lin_reg_en pin.
?
Stan KC7XE
|
Re: QMX, one of op-amps self-oscillates at 25 MHz (or picks up the system clock?)
The 833 Hz is low-frequency oscillation, likely being generated in IC202. R207 (470K) is providing a DC bias at the input of the gate which can cause the gate to operate in its linear region, acting as a high gain amplifier rather than a logic buffer. I’m not sure why this gate has a feedback resistor as we never do this for logic gates for the very reason we don’t want them operating in the linear region.
It's a quite common technique to effectively DC bias the gate's input at almost exactly the midpoint between it's 0 and 1 input voltage thresholds. An AC coupled small signal connected at this input is then very well "amplified" by the gate, with very high gain, converting it as perfectly as possible to a 3.3V square wave. It's very effective.?
If there is no input signal at all, the gate will oscillate at some extremely high frequency, somewhere at VHF. But as soon as even a small input signal is applied, the gate locks onto it and amplifies it near-perfectly with very high gain (creating the square wave). Which is another way of looking at it.?
You will find in QMX if you remove this resistor, the gate will probably not operate. Zero output. Or certainly not reliably. Because now the small signal from the TCXO (about 1Vpp) is centered on what? Zero maybe? You have to provide some DC bias. You could put 1M to ground and 1M to Vdd or something. It would probably also work (modern CMOS are quite symmetric). But a better way of achieving the same thing is the feedback resistor.? 73 Hans G0UPL
|
Re: QMX, one of op-amps self-oscillates at 25 MHz (or picks up the system clock?)
Hi Adam
Downstream of IC202 there isn't any "almost", this is the very reason why IC202 is there. You will have a 3.3Vpp output at IC202 output.?
Can you take a similar video please for the signal at IC202 output?
I'm still having trouble trying to convince myself that the 25 MHz TCXO is at fault here rather than - sorry to say it - some weird artefact introduced by the measurement itself (at the TCXO output / gate input).
73 Hans G0UPL
Hans Summers via <hans.summers=[email protected]> wrote:
> Several things I don't understand. I asked about if you could connect just
> because I wanted to know if the USB connection was working; If it is, that
> kind of means that the 25MHz signal must be present.
It's working and stable. But the 25 MHz signal is present. That's why the
waveform lines are "thick" -- it's because of the 25 MHz signal, it
appears when I zoom in.
The problem is that the 25 MHz signal is modulated by this weird, ~833 Hz
one. So it's severely distorted. The scope's time base on these photos was
set to show this low-frequency modulation.
Here's a video of TCXO output (before the gate) when I switch the time
base:
Maybe it's enough for the PLL in the CPU to lock, but *almost* enough for
the PLL in the sync chip (so most of the time it's locked, but falls out
of lock here and there, and that's why CLK outputs are momentarily
disabled)?
> It looks like the vertical division on your 'scope is 100mV/div in which
> case the waveform also isn't the 3.3Vpp expected at the logic gate
> output... so are you sure of your measurement? It's confusing to me...
It's correct. I set the switch on the probe to 1:10 (to have greater
impedance of the probe), but forgot to switch it in the scope's menu, so
the scope still thinks it's 1:1.
So it shows 100 mV / div, but in reality it's 1 V / div.
In the video above it's set correctly and shows 1 Vpp (but it's TCXO
output, not gate output -- gate output is around 3.3 Vpp).
|
Re: Ultimate 3S reciever question
Ian,
So I wouldnt need bpf for each band. Unless of course I needed to filter out other radios TXing in the shack.
--
regards,
Bryan, N0LUF
|
Re: QMX - Power supply off line test Process
Something is wrong with your 3.3V input regulator circuit.? Inspect carefully for solder blobs or shorts, near the connector pins you soldered or anywhere else - it should not operate this way.? ]
?
1. Measure the voltages on each terminal of Q110 and Q111 when lin_reg_en is grounded, and check to see if pwm_3v3 is floating high (it should be 0V) - maybe somehow Q109 is getting enabled (activating the switched mode supply) like is supposed to happen in Step 3, and you are transferring the input voltage to VDD.
?
2.? Otherwise, VDD should never be able to go above 3.3V - it is the output of a 3.3V regulator, IC101.? But it seems to be regulating fine when you did step 1, so something in step 2 is causing the input voltage to be shorted to the output.? Q101 should turn off Q102 when lin_reg_en is grounded, preventing any voltage from getting to VDD (unless Q109 is on, which it should not be).? So something is not right here.? You can check the voltages on each of the terminals of Q101, Q102, and IC101 when you ground the lin_reg_en pin.
?
Stan KC7XE
|
I can tackle this but remember, YMMV.
- I am using battery packs purchased on Amazon.? I have "AA"
packs, 8 cells for 12v and 6 cells for 9v.? I am using
rechargeable lithium 1.5v cells.? The 12v pack puts out 12.1v
and the 9v pack is very very close to 9v.? The nice thing is
that these batteries maintain voltage pretty much until they go
dead.? Don't go super cheap on the rechargeables.? Look at the
reviews and spend a few more buck if need by.? Contact me
off-list if you want links to what I am using.
- 20-10 meters I use a vertical whip, ground mounted, with 5
ribbon wire radials.? Not quite the equivalent of 30 radials but
it works well even with low power.? NO loading coils and NO
tuner needed.? You merely lengthen and/or shorten for the
appropriate band. For the QMX you will want a way to check SWR
for multiple bands.? For QCX, you'll probably be fairly close if
you set up the same way each time.? You will need to determine,
at least initially, the appropriate lengths using an analyzer.?
Then you can verify at half power or so, at least on the QMX to
be sure your SWR is good.
- 60-40-30 meters choices are End Fed Half Wave (EFHW) or a
dipole.? Linked dipole will give you multiple bands.? Dipole in
inverted V configuration is more efficient than an EFHW, and
better suited for NVIS (near vertical incident skywave) for more
"local" communication at a few hundred miles.? Personally, I
rarely get on 60m or 40m and occasionally operate 30m if
conditions dictate it.? With a dipole I also recommend using a
1:1 current balun (not a voltage balun).? Common mode
currents on your coax can change your SWR every time you touch
your radio or your key (ask me how I know!)
- Leave the tuner at home.? You're working QRP and you want to
get as much signal out as possible.? Use resonant antennas that
match the frequency your are operating.
Of course one thing to keep in mind is
that asking these things opens a can of worms.? You will have as
many opinions as there are operators on the forum. ?Yes, many
other configurations will work.? You can use random wires and
tuners and they will work.? But I recommend staying away from
setups that take a very small signal and immediately reduce it by
50%.? Remember, only 3 dB of loss represents half of your output.?
A typical QRP labs transceiver will put out 4 watts, give or
take.? Why reduce that to 2 watts before your signal leaves your
site?? I speak as a very experience portable op with over 10,000
QSOs in the log in 2023 and over 8000 in 2024.? 2024 was 100% QRP,
so I think I can speak with a little authority on the matter.
BTW, the QDX and QMX are awesome rigs
for digital work in the field as well.? FT4 & FT8 for POTA
will rack up contacts pretty quickly if you do it right, even at
just a few watts.
72
Charles - NK8O
On 1/9/25 10:43, Len via groups.io
wrote:
toggle quoted message
Show quoted text
?
Firstly...apologies to all for such a newbie question, but I
only just ordered a QMX and am just now learning CW :)
?
I was hoping to put together a rig for POTA while doing AT
section hikes and was looking for recommendations for:
?
- battery
- antenna
- tuner
- plus anything else that I haven't even thought of yet :)
?
Thanks!
?
Len S
KC3PWB?
--
Len S
KC3PWB
|
Re: QMX, one of op-amps self-oscillates at 25 MHz (or picks up the system clock?)
Adam,
The 833 Hz is low-frequency oscillation, likely being generated in IC202. R207 (470K) is providing a DC bias at the input of the gate which can cause the gate to operate in its linear region, acting as a high gain amplifier rather than a logic buffer. I’m not sure why this gate has a feedback resistor as we never do this for logic gates for the very reason we don’t want them operating in the linear region.
Tony
Hans Summers via <hans.summers=[email protected]> wrote:
> Several things I don't understand. I asked about if you could connect just
> because I wanted to know if the USB connection was working; If it is, that
> kind of means that the 25MHz signal must be present.
It's working and stable. But the 25 MHz signal is present. That's why the
waveform lines are "thick" -- it's because of the 25 MHz signal, it
appears when I zoom in.
The problem is that the 25 MHz signal is modulated by this weird, ~833 Hz
one. So it's severely distorted. The scope's time base on these photos was
set to show this low-frequency modulation.
Here's a video of TCXO output (before the gate) when I switch the time
base:
Maybe it's enough for the PLL in the CPU to lock, but *almost* enough for
the PLL in the sync chip (so most of the time it's locked, but falls out
of lock here and there, and that's why CLK outputs are momentarily
disabled)?
> It looks like the vertical division on your 'scope is 100mV/div in which
> case the waveform also isn't the 3.3Vpp expected at the logic gate
> output... so are you sure of your measurement? It's confusing to me...
It's correct. I set the switch on the probe to 1:10 (to have greater
impedance of the probe), but forgot to switch it in the scope's menu, so
the scope still thinks it's 1:1.
So it shows 100 mV / div, but in reality it's 1 V / div.
In the video above it's set correctly and shows 1 Vpp (but it's TCXO
output, not gate output -- gate output is around 3.3 Vpp).
|
Ivica - YU1QRP
Joshua KJ7LVZ
hanz YL3JD
Jason KE8WSI
va3rr
Dan