Re: #qmx Don¡¯t Use USB-C PD to power your QMX?!
#qmx
On Wed, Sep 6, 2023 at 11:07 PM, Hans Summers wrote:
Dear Friendly Razvan
I did not mean anything I wrote as an offence. If it sounded like that, I beg my most sincere apologies. I understand that you are sick and tired of hearing in every second posting in this group some criticism, more or less appropriate. But you have to understand my feelings too: I have never felt so unwell switching on and off my qrp rigs. Will it boot, will it smoke, will it do or will it don't? Most likely I am badly influenced in my feelings by the readings done on this group, many of them due to ignorance, inappropriate handling and set-up. But this is also a proprietary problem of this such a heterogenous group: in it's charming way it has attracted people very much over-challenged from their soldering wits. I am soo sorry for being both too coward and too risky at the same time. I am coward related to issues I can not controll and I am? too risky in terms I think I can cope with. My reasoning does not meet the requirements needed to understand the do's and dont's of my latest soldering: seems to me that I belong to the over-challenged mentioned above. I will pack the rig into a brown bag, hide it somewhere in the cellar, take some aspirin and wait for a better mood and the time I will feel more courageous by having understood the feelings of the few (quote: Pink Floyd). Yours really, really friendly, Razvan dl2arl
|
Re: QMX new build - power output
Let's try that again with attachments this time..
|
Re: More on spike and avalanche breakdown
I have a question:
Hans seems to have focused on short and open as worst case
scenarios, as far as SWR goes. The expectation is that would
correspond to maximum current and voltage, as far as the BS170's
go...
Is that actually the worst case scenario for the PA? What happens
if there's a serious reactive component to the load? Could it
force the BS170's into a high dissipation portion of the load
line? (E.g., push it out of the high-voltage, low current / high
current, low voltage operational area of the intended load line?)
Just asking, forgive my ignorance if I'm off base!
73, Paul -- AI7JR
On 9/17/23 11:27, Tony Scaminaci wrote:
toggle quoted message
Show quoted text
Hi Hans,
I¡¯ve been following this thread with a great deal
of interest. JZ and I ran a bunch of sims on various configs a
few months back and your experimental measurements correlate
quite well with our simulation results.
I have a question regarding the drain voltages
shown on your scope shots. What is the flat-line voltage when
the FETs are turned on? I¡¯m curious how close you¡¯re getting to
0 volts, i.e., full saturation as this directly affects PA
operating efficiency.
Thanks!
Tony - AC9QY
The apparent success of
the capacitor suggests that you could remove L14
altogether. Its "constant-current" function is
rendered moot by the 100nF capacitor.
Theoretically maybe. But remember it was
something I tried during development of the PA and it
has a noticeable effect on efficiency and power output.?
The PA is now operating
in "constant voltage" mode. L14 might be
replaced by a low value resistor which would
automatically reduce the supply voltage in the
case of low Z high SWR related current draw,
further improving survivability against that
case.
I'm not feeling intuitivelt that the
resistor, having a value low enough to not too
traumatically impact power output or efficiency, would
then have a sufficiently high value to make much
difference in the Low Z high SWR case.?
You worked very hard
today and it is evening for you now. Time for a
rest!
Hi hi, I work hard every day. Started
the day with a 45 minute 40m CW session before dawn
on my QCX and 50W PA. Including US contacts Stan
W4AG and Ken K5WK. It's the 4th or 5th time I've met
Stan in the last month, around the same time of day.
Not bad considering the one useful high-up longer
leg of my OCFD antenna has detached itself and is
lying on the metal-frrame roof. None of the rigs
seem to care.
Contemplated the sunrise while
attempting fruitlessly to consolidate my ever more
tenuous grip on reality. Still here as usual to
watch the sunset too from the vaulted heights of the
penthouse R&D facility at QRP Labs towers.?
Just for your interest and
entertainment... I tried two other purely
firmware-only attempts at de-spiking the key-up. The
first one involved adjustment to the relative phase
angle between the drivers of the two sides of the
push-pull PA. See attached shaped.png. The RF
envelope shaping actually worked surprisingly well,
and of course the spike was gone, but the Drain
waveform (and peak voltage) was horrendously
horrible, perhaps less surprisingly so. No go.
The second experiment was very much
easier to code, and simply involved shutting down
one side of the push-pull a couple of milliseconds
before the other. Of course there would be no need
for it to be milliseconds, microseconds would even
be plenty of time. So in this idea, we are dumping
energy in two smaller spikes rather than one big
one. See attached shaped2.png. If energy storage is
1/2 L I^2 and running only half the push-pull halves
the current, it seems to me that 75% of the energy
gets dissipated in the first spike and only 25% in
the second spike. This is a quite convenient result
because while half the push-pull is still
operational, there is some other place for the
energy to go rather than relying on unreliable
avalanche breakdown or an uncooperative load. So the
first spike is tame, and the 2nd spike if my above
argument is correct, only has to do with 25% of the
energy. Accordingly it looks like the 2nd spike
reaches about 75V which according to my earlier
observations wouldn't practically be enough to cause
avalanche breakdown, which appeared to occur ar
around 80-85V.?
I didn't quite give up totally on
either of these software defined tricks. I'd at the
least like to eliminate ground bounce effects as a
possible cause of the spike at the end of each cycle
"on" time, which cause the apparent huge RF envelope
at the drains. I need to convince myself it's real
not measurement artefact before giving up on it; the
output RF envelope across the dummy load looks so so
nice! The stepped key-up looks like it would at
least remove 75% of the energy which wouldn't be a
bad step in the right direction particularly as it's
so simple and easy to implement it.?
I even had one more even more radical
idea which I may try just for the technical
curiosity but am too embarrassed to divulge.?
73 Hans G0UPL
|
Re: QMX new build - power output
Well I decided to start with L401. Rewound per latest manual. Had a heck of a time getting continuity, that enamel doesn't want to burn off (Hakko set at 750F and using Kester 235). I finally did get them. RF sweeps attached. A lot more gain now. No difference in TX power of course. Will go through the rest of the items tomorrow.
73 John N8JJO
|
Re: QCX Mini Loud audio clicks
Spot on, Ron
Seams to be some sort of interference.
Thank you for your help.
Best 72
Luis
CU2IJ
On 17/09/23 21:27, Ronald Taylor wrote:
Hi Luis. That sounds like interference to me. So,
maybe a dumb question but ... Is either the radio or the antenna
close to your Wi-Fi modem? Or close to a cell phone? I've heard
my cell phone data transmissions in my QCXs if the phone is
within a half meter or so of the radio. I also hear a continuous
tone (not clicking) on about 14.030 MHz that is coming from my
wi-fi modem about 2.5 meters from my operating position.
Possibly even something from a power supply or some other gadget
in the vicinity. It also sounds a little bit like an electric
arcing somewhere in the area to me. When I was a kid?we had an
electric fence and it made very similar sounds on the radio
whenever weeds grew up into it. Just some thoughts. You might
try running it on a battery to eliminate noisy power supply
and/or move it to a different location away from your house on a
temporary wire antenna and see if the issue persists. Good luck
... 73 .. Ron
Hello
all,
Please, I need your help.
I built the QCX mini 20m version initially with no problems.
Rx was good
and TX went to 5w after tweaking the filter inductors.
A few days later there are some loud audio clicks. Rx is still
working
when between clicks. Some times there are periods of a few
seconds with
good audio, then they come again.
Any clue to the origin of the problem?
In attach is a sample of the audio.
Thank you
Luis
CU2IJ
--
Cordiais sauda??es,
--
Cordiais sauda??es,
Lu¨ªs Ant¨®nio Mota Albergaria Pacheco
m¨®vel: +351 962374276
luis.a.albergaria@...
luis.a.albergaria@...
--
Cordiais sauda??es,
Lu¨ªs Ant¨®nio Mota Albergaria Pacheco
m¨®vel: +351 962374276
luis.a.albergaria@...
luis.a.albergaria@...
|
There are several cheap WWVB receivers available.? Has anyone successfully implemented WWVB as a 1pps time source rather than GPS for QRP Labs kits?? How did you do it and what are the pros and cons between WWVB and GPS?? One that comes to my mind is a clear view of the sky is not required with the WWVB 60kHz signal.? A disadvantage might be potential interference to 60kHz from LED light dimmers and such.? Which is more reliable?? Maybe have one automatically back up the other?
-Jerry AC5JM
|
Re: QCX Mini Loud audio clicks
Hi Luis. That sounds like interference to me. So, maybe a dumb question but ... Is either the radio or the antenna close to your Wi-Fi modem? Or close to a cell phone? I've heard my cell phone data transmissions in my QCXs if the phone is within a half meter or so of the radio. I also hear a continuous tone (not clicking) on about 14.030 MHz that is coming from my wi-fi modem about 2.5 meters from my operating position. Possibly even something from a power supply or some other gadget in the vicinity. It also sounds a little bit like an electric arcing somewhere in the area to me. When I was a kid?we had an electric fence and it made very similar sounds on the radio whenever weeds grew up into it. Just some thoughts. You might try running it on a battery to eliminate noisy power supply and/or move it to a different location away from your house on a temporary wire antenna and see if the issue persists. Good luck ... 73 .. Ron
Hello all,
Please, I need your help.
I built the QCX mini 20m version initially with no problems. Rx was good
and TX went to 5w after tweaking the filter inductors.
A few days later there are some loud audio clicks. Rx is still working
when between clicks. Some times there are periods of a few seconds with
good audio, then they come again.
Any clue to the origin of the problem?
In attach is a sample of the audio.
Thank you
Luis
CU2IJ
--
Cordiais sauda??es,
--
Cordiais sauda??es,
Lu¨ªs Ant¨®nio Mota Albergaria Pacheco
m¨®vel: +351 962374276
luis.a.albergaria@...
luis.a.albergaria@...
|
Re: Vacuum-tube final for QRP transmitters?
Regens using triodes work best if the tube is a low mu type.
That makes the 12AU7 better than 12AX7 or 12AT7.? Also the 12AU7 makes a
fair low voltlege regen working well at 12V plate. since there are two tridoes
so use one to isolate the antenna from the detector grounded grid will
do well for that,
Try this combo...
12AD6 as converter (crystal or self oscillating) from say 40M down to 500khz or
lower with 12AU7 as regen and audio amp.
Those two tubes will eat 300ma at 12.6V for the heaters and a few ma for the 13V plate voltage.
running a regen at low frequency makes for better selectivity and stability.
--
Allison
------------------
Post online only,?
direct email will go to a bit bucket.
|
Re: More on spike and avalanche breakdown
Hi Allison,
The scenario you describe has a mechanical analogy - spark plug timing in internal combustion engines. The goal of a rotating system is to time the ?kicks to maintain the system in a continuous and smooth rotation with minimum energy input. Timing errors cause the system to buck its rotation, lowering the efficiency of the system. I love mechanical and electrical analogies because they¡¯re sometimes easier to understand. Most of us have experience setting timing on our engines and the detrimental side effects of improper timing.
Tony - AC9QY
On Sun, Sep 17, 2023 at 3:19 PM ajparent1/kb1gmx < kb1gmx@...> wrote: The why the delay was done is RC into a 74HC14, 100 ohm variable and 47pf.
There were two one for the A and B phse (roughly 180 degrees opposed).
That and two device delay was to disallow pushing A side without a delay?
before pushing the B side.? Overall effect was a drive that was less than
160 degrees of the possible 180 for each leg.
--
Allison
------------------
Post online only,?
direct email will go to a bit bucket.
|
Re: Vacuum-tube final for QRP transmitters?
Nice 6N7 rig!? I once attempted to build a regen receiver using an 833A triode, it sort-of worked :)
Mark K9TR
|
QCX Mini Loud audio clicks
Hello all,
Please, I need your help.
I built the QCX mini 20m version initially with no problems. Rx was good and TX went to 5w after tweaking the filter inductors.
A few days later there are some loud audio clicks. Rx is still working when between clicks. Some times there are periods of a few seconds with good audio, then they come again.
Any clue to the origin of the problem?
In attach is a sample of the audio.
Thank you
Luis
CU2IJ
--
Cordiais sauda??es,
--
Cordiais sauda??es,
Lu¨ªs Ant¨®nio Mota Albergaria Pacheco
m¨®vel: +351 962374276
luis.a.albergaria@...
luis.a.albergaria@...
|
Re: More on spike and avalanche breakdown
The why the delay was done is RC into a 74HC14, 100 ohm variable and 47pf.
There were two one for the A and B phse (roughly 180 degrees opposed).
That and two device delay was to disallow pushing A side without a delay?
before pushing the B side.? Overall effect was a drive that was less than
160 degrees of the possible 180 for each leg.
--
Allison
------------------
Post online only,?
direct email will go to a bit bucket.
|
Re: More on spike and avalanche breakdown
A few years back I did a lot of work with wireless power transfer.
First order of business as I got deeper was a 300mhz with a 2ghz sample
rate DSO.? WHY?
Seems if the waveform (square wave feed) were exactly (less than 15ns)
in opposition the power output was both loaded with spikes and high current
with low efficiency.? It seems each side needed a small delay (in nano seconds)
so the two didn't overlap on the shutoff delay.?
So the waveform has to have a slight delay and early cutoff so that the waveform
is allowed to self commutate.? Why, the load included a tuned path (LPF).? You
want to kick the wheel so it does a full 360 and not drag it around while fighting
itself.? The L at the center tap helps do that.
All that said its why wide bandwidth class D is not easy as its easy to see
how things are frequency or more correctly time dependent. Compromise?
is possible but the trade is efficiency and power developed.
--
Allison
------------------
Post online only,?
direct email will go to a bit bucket.
|
Re: Newbee Q: Are there 'designated builders' for QRPLabs kits?
I have not heard of that here but I may have missed it. However, I will offer my services since I have successfully built one that works well and I have built several other kits for hams who did not want to wait, self-evaluated as all thumbs or wanted the radio but kits were not their thing!
I love building kits! The QMX was the most challenging since it really is a 5 pound radio in a 2 ounce box! Even got to replace a missing SMD capacitor I probably broke off the board while doing some required step.
So here are two posibilities if you want a QMX:
1. You get a QMX kit shipped to me and I send you mine
2. You get whatever QRP Labs kit sent to me and I build it and send it to you.
I am retired and have great time flex. Well, really, I wonder how I had time for full-time? employment! Your call.
--
73
Karl
KI4ZUQ
|
Re: How to get output to tracking page
#u4b
Hi Dave, I am new to the U4B. I have created balloon flight: Gadgetman, and it is active. I am testing at my QTH at this point. I have followed all the pointers mentioned:-
- Callsign in uppercase (in the Balloon Flight cfg)
- In my local WSJT-X cfg I am using a friends call sign with a different Maidenhead to my own, to receive the U4B TELE
- Because the HF antenna is just lying on the ground, no one else is picking up the TX, only my own local WSJT-X
- Uploaded a spot to fool Wspnet not to use the M6 spot they have memorized with my callsign.
- Set the TX to 20m, Chan 124
After many hours, the QRP Labs?tracking page has not seen me. Yet WSPR packets are recorded by WSPRNET. Output my U4B tracker program:-
19:09:51? GPS START...
19:10:01? GPS: VLD:A LCK:1 CL:0 ALT:1617.0
19:10:01? BAT:4597 TXPWR:1 TEMP:294
19:10:01? TELE START...
19:19:51? TELE DONE
19:19:51? GPS START...
19:20:10? GPS: VLD:A LCK:1 CL:0 ALT:1776.7
19:20:10? BAT:4627 TXPWR:1 TEMP:294
19:20:10? TELE START...
19:29:51? TELE DONE
Wsprnet:-
What am I doing wrong?
73 Stewart
ZR6WT
|
Re: More on spike and avalanche breakdown
Hi Hans,
I¡¯ve been following this thread with a great deal of interest. JZ and I ran a bunch of sims on various configs a few months back and your experimental measurements correlate quite well with our simulation results.
I have a question regarding the drain voltages shown on your scope shots. What is the flat-line voltage when the FETs are turned on? I¡¯m curious how close you¡¯re getting to 0 volts, i.e., full saturation as this directly affects PA operating efficiency.
Thanks!
Tony - AC9QY
The apparent success of the capacitor suggests that you could remove L14 altogether. Its "constant-current" function is rendered moot by the 100nF capacitor.
Theoretically maybe. But remember it was something I tried during development of the PA and it has a noticeable effect on efficiency and power output.?
The PA is now operating in "constant voltage" mode. L14 might be replaced by a low value resistor which would automatically reduce the supply voltage in the case of low Z high SWR related current draw, further improving survivability against that case.
I'm not feeling intuitivelt that the resistor, having a value low enough to not too traumatically impact power output or efficiency, would then have a sufficiently high value to make much difference in the Low Z high SWR case.?
You worked very hard today and it is evening for you now. Time for a rest!
Hi hi, I work hard every day. Started the day with a 45 minute 40m CW session before dawn on my QCX and 50W PA. Including US contacts Stan W4AG and Ken K5WK. It's the 4th or 5th time I've met Stan in the last month, around the same time of day. Not bad considering the one useful high-up longer leg of my OCFD antenna has detached itself and is lying on the metal-frrame roof. None of the rigs seem to care.
Contemplated the sunrise while attempting fruitlessly to consolidate my ever more tenuous grip on reality. Still here as usual to watch the sunset too from the vaulted heights of the penthouse R&D facility at QRP Labs towers.?
Just for your interest and entertainment... I tried two other purely firmware-only attempts at de-spiking the key-up. The first one involved adjustment to the relative phase angle between the drivers of the two sides of the push-pull PA. See attached shaped.png. The RF envelope shaping actually worked surprisingly well, and of course the spike was gone, but the Drain waveform (and peak voltage) was horrendously horrible, perhaps less surprisingly so. No go.
The second experiment was very much easier to code, and simply involved shutting down one side of the push-pull a couple of milliseconds before the other. Of course there would be no need for it to be milliseconds, microseconds would even be plenty of time. So in this idea, we are dumping energy in two smaller spikes rather than one big one. See attached shaped2.png. If energy storage is 1/2 L I^2 and running only half the push-pull halves the current, it seems to me that 75% of the energy gets dissipated in the first spike and only 25% in the second spike. This is a quite convenient result because while half the push-pull is still operational, there is some other place for the energy to go rather than relying on unreliable avalanche breakdown or an uncooperative load. So the first spike is tame, and the 2nd spike if my above argument is correct, only has to do with 25% of the energy. Accordingly it looks like the 2nd spike reaches about 75V which according to my earlier observations wouldn't practically be enough to cause avalanche breakdown, which appeared to occur ar around 80-85V.?
I didn't quite give up totally on either of these software defined tricks. I'd at the least like to eliminate ground bounce effects as a possible cause of the spike at the end of each cycle "on" time, which cause the apparent huge RF envelope at the drains. I need to convince myself it's real not measurement artefact before giving up on it; the output RF envelope across the dummy load looks so so nice! The stepped key-up looks like it would at least remove 75% of the energy which wouldn't be a bad step in the right direction particularly as it's so simple and easy to implement it.?
I even had one more even more radical idea which I may try just for the technical curiosity but am too embarrassed to divulge.?
73 Hans G0UPL
|
Current Limiting Adapter
#power
I have a nice voltage regulated bench power supply but I do not have any form of a current limiting device. I have designed a nice 1.25A to 50mA 10 step current limiter to further condition the voltage output of my bench power supply. Not advertising, but I obtained all components from Amazon. Hope it can be helpful to others.
73, Billy (N5SE)
N5SE
Billy Wayne Moore
7066 Shady Knoll Ln
Willis, TX 77318-6324
Phone: 936-537-2975???
|
Re: #qmx Antenna Tuner Can Kill PAs
#qmx
Thanks Ross. I noticed that after I sent it. In fact I DID use a WTST at 12 volts
?. I was switching things around and forgot that the transformer I had prewound was a 9 volt. Explains the super good output I got but nothing appears to have been harmed. I only transmitted long enough to get a power reading. Nonetheless, power
went from about 1 watt to 4.2 watts. More than can be explained by just the transformer winding.
Now I have to wind a RWTST and hope my board can survive one more transformer swap.
Tony
toggle quoted message
Show quoted text
Tony,
I hope you used the 12V RWTST and not the WTST for 9V on 12V?
73
Ross
6
On 17 Sep 2023, at 16:24, mux_folder2001 <canthony15@...> wrote:
?
Hi Hans,
Well, I was running a 9 volt build and getting reasonable power on 10 meters using the conventional winding method. Then I tried WTST and it did not make much difference. However, today I switched to a 12 volt conventional wound and had very low output on 10
meters. I then switched to WTST and got very good output on 10 meters. The WTST seems to make a very big difference with a 12 volt build and not so much with a 9 volt build, based on a sample of 1.?
Thanks,
Tony
AD0VC
Hi Tony, all
The main motivation for WTST and RWTST was that they completely resolved the parasitic resonance issue which destroyed 10m performance on a percentage of builds. Mysteriously not all, and it was never possible to identity the cause of the parasitic
resonance. These twisted wire transformers solved it. Sometimes we can be so happy about the solution we can accept not really understanding why the problem occurred.?
The bonus is that the twisted wire transformers also provide a little performance improvement (output power, efficiency) compared to the earlier straight transformer winding method.?
73 Hans G0UPL
Thanks for the tip about power measurement. I was not aware of either ALT-Right or CTRL-Left clicks.
I did no see much difference between conventional and WTST myself.
Tony
AD0VC
Tony,
You can clearly see the improvement in coupling that the twisted winding has over the conventional winding, with the expected improvement in leakage inductance as well.
I think that is especially significant at 10m since the leakage inductance has probably not been factored in to the LPF design, and at 10m is approaching the size of the filter inductors.
The negative inductance you see is the parasitic capacitance overwhelming the inductance ie. a resonance. It makes extracting model parameters more challenging.
I run with a variant of the twisted pair transformer in my QDX-HB. It is a 12V 3:2 design . The twisted pair is fully the length of the longer wire. The ratio is obtained by tapping points on the secondary, leaving wire ends to dangle. This
technically more fully realizes the transmission line properties of the twisted pair approach, although I am hard pressed to see much of a performance difference.
JZ
FWIW here is what I measured for the conventional and WTST using my nanovna. I verified the calibration when I saw the negative inductance for the WTST at 28Mhz.?
Conventional????????????????????????
??????L Wind (uh)?L Half Wind?Leak??K
7Mhz??13.56?? ? ? 3.92??? ? ? 0.107?0.996046758211066
10Mhz?10.42?? ? ? 2.98?? ? ? ?0.107?0.994852394070841
14Mhz?7.42??? ? ? 2.13?? ? ? ?0.106?0.992831448793946
18Mhz?5.2???? ? ? 1.51??? ? ? 0.106?0.989755214492647
21Mhz?3.6???? ? ? 1.16??? ? ? 0.106?0.985167780408777
28Mhz?1.26??? ? ? 0.435?? ? ? 0.106?0.957012547395809
????????????????????????
WTST????????????????????
7Mhz??12.98?? ? ? 3.72?? ? ? ?0.067?0.997415767187798
10Mhz?10.13?? ? ? 2.91?? ? ? ?0.067?0.996687504803284
14Mhz?6.97??? ? ? 2.05?? ? ? ?0.066?0.99525416175213
18Mhz?4.11?? ? ? ?1.28?? ? ? ?0.066?0.991938307476537
21Mhz?1.96??? ? ? 0.74??? ? ? 0.066?0.98301908964793
28Mhz?-1.43?? ? ?-0.18?? ? ? ?0.066?0.976650476806393
Tony
AD0VC
Tony,?
In LTSpice if you hover the cursor over a component as if to measure current through it, then press ALT and right click, LTSpice will give you a plot of instantaneous power. You can window that plot for the time period of interest. Then hover
the cursor over the plotted variable name above the plot. Press CTRL and left click. LTSpice will integrate the instantaneous lower over the window interval and open a box with the result.
Since core magnetic materials tend to show reduced permeability at increasing frequency, I like to use my NanoVNA to measure inductance as seen at the relevant frequency. This usually works pretty well unless parasitic capacitances bring a resonance
nearby.
JZ
John,
A couple of things. First, how are you getting the mosfet dissipation from the spice simulation? Are you eyeballing an instantaneous power (Vd * Id) waveform and doing a mental average??
Second, I just built a conventional 9 volt transformer and a WTST transformer and measured the leakage inductance and inter-winding capacitance. Two different cores of course. Measured at 100Khz which is as high as my LCR meter goes.
Conventional:
L winding = 25.5uh
L half winding = 7.65uh
Lleak = .505uh
Cps = 7pf
K = sqrt(1-(.505/25.5)) = .99
WTST:
L winding = 27.6uh
L Half winding = 8.12uh
Lleak = .515uh
Cps = 11.3pf
K = sqrt(1-(.515/27.6)) = .9906
Useful for the SPICE model.
For those that care about this nerd stuff.
Tony
AD0VC
Sure thing, Chris.
Simulation at 25 ohms (2:1 SWR) shows each BS170 dissipating about
half a watt, so there is some margin.
JZ
On Thu, Sep 14, 2023 at 9:26?AM Chris KB1NLW via
<chrisrey1= [email protected]> wrote:
>
> John Z
> Thanks for your response. Looks like steady state SWR of 3 will heat the BS170s to their PD limit, so I assume SWR <2 is OK.
>
> Transients from an auto tuner that can be as high as SWR of 10 will cause transient power dissipation significantly higher than the allowable dissipation.? Unfortunately we don't have sufficient thermal data on the BS170 to understand transient thermal response
at the silicon chip in order to determine allowable SWR Transient time product.
>
> Bottom line is Don't Use an Auto Tuner without reducing the transient at the QMX.? Reduced power or attenuation between QMX and tuner.
>
|
Re: #qmx Antenna Tuner Can Kill PAs
#qmx
Sorry, this is a QDX.?
Tony
AD0VC
toggle quoted message
Show quoted text
Hi Hans,
Well, I was running a 9 volt build and getting reasonable power on 10 meters using the conventional winding method. Then I tried WTST and it did not make much difference. However, today I switched to a 12 volt conventional wound and had very low output on 10
meters. I then switched to WTST and got very good output on 10 meters. The WTST seems to make a very big difference with a 12 volt build and not so much with a 9 volt build, based on a sample of 1.?
Thanks,
Tony
AD0VC
Hi Tony, all
The main motivation for WTST and RWTST was that they completely resolved the parasitic resonance issue which destroyed 10m performance on a percentage of builds. Mysteriously not all, and it was never possible to identity the cause of the parasitic
resonance. These twisted wire transformers solved it. Sometimes we can be so happy about the solution we can accept not really understanding why the problem occurred.?
The bonus is that the twisted wire transformers also provide a little performance improvement (output power, efficiency) compared to the earlier straight transformer winding method.?
73 Hans G0UPL
Thanks for the tip about power measurement. I was not aware of either ALT-Right or CTRL-Left clicks.
I did no see much difference between conventional and WTST myself.
Tony
AD0VC
Tony,
You can clearly see the improvement in coupling that the twisted winding has over the conventional winding, with the expected improvement in leakage inductance as well.
I think that is especially significant at 10m since the leakage inductance has probably not been factored in to the LPF design, and at 10m is approaching the size of the filter inductors.
The negative inductance you see is the parasitic capacitance overwhelming the inductance ie. a resonance. It makes extracting model parameters more challenging.
I run with a variant of the twisted pair transformer in my QDX-HB. It is a 12V 3:2 design . The twisted pair is fully the length of the longer wire. The ratio is obtained by tapping points on the secondary, leaving wire ends to dangle. This
technically more fully realizes the transmission line properties of the twisted pair approach, although I am hard pressed to see much of a performance difference.
JZ
FWIW here is what I measured for the conventional and WTST using my nanovna. I verified the calibration when I saw the negative inductance for the WTST at 28Mhz.?
Conventional????????????????????????
??????L Wind (uh)?L Half Wind?Leak??K
7Mhz??13.56?? ? ? 3.92??? ? ? 0.107?0.996046758211066
10Mhz?10.42?? ? ? 2.98?? ? ? ?0.107?0.994852394070841
14Mhz?7.42??? ? ? 2.13?? ? ? ?0.106?0.992831448793946
18Mhz?5.2???? ? ? 1.51??? ? ? 0.106?0.989755214492647
21Mhz?3.6???? ? ? 1.16??? ? ? 0.106?0.985167780408777
28Mhz?1.26??? ? ? 0.435?? ? ? 0.106?0.957012547395809
????????????????????????
WTST????????????????????
7Mhz??12.98?? ? ? 3.72?? ? ? ?0.067?0.997415767187798
10Mhz?10.13?? ? ? 2.91?? ? ? ?0.067?0.996687504803284
14Mhz?6.97??? ? ? 2.05?? ? ? ?0.066?0.99525416175213
18Mhz?4.11?? ? ? ?1.28?? ? ? ?0.066?0.991938307476537
21Mhz?1.96??? ? ? 0.74??? ? ? 0.066?0.98301908964793
28Mhz?-1.43?? ? ?-0.18?? ? ? ?0.066?0.976650476806393
Tony
AD0VC
Tony,?
In LTSpice if you hover the cursor over a component as if to measure current through it, then press ALT and right click, LTSpice will give you a plot of instantaneous power. You can window that plot for the time period of interest. Then hover
the cursor over the plotted variable name above the plot. Press CTRL and left click. LTSpice will integrate the instantaneous lower over the window interval and open a box with the result.
Since core magnetic materials tend to show reduced permeability at increasing frequency, I like to use my NanoVNA to measure inductance as seen at the relevant frequency. This usually works pretty well unless parasitic capacitances bring a resonance
nearby.
JZ
John,
A couple of things. First, how are you getting the mosfet dissipation from the spice simulation? Are you eyeballing an instantaneous power (Vd * Id) waveform and doing a mental average??
Second, I just built a conventional 9 volt transformer and a WTST transformer and measured the leakage inductance and inter-winding capacitance. Two different cores of course. Measured at 100Khz which is as high as my LCR meter goes.
Conventional:
L winding = 25.5uh
L half winding = 7.65uh
Lleak = .505uh
Cps = 7pf
K = sqrt(1-(.505/25.5)) = .99
WTST:
L winding = 27.6uh
L Half winding = 8.12uh
Lleak = .515uh
Cps = 11.3pf
K = sqrt(1-(.515/27.6)) = .9906
Useful for the SPICE model.
For those that care about this nerd stuff.
Tony
AD0VC
Sure thing, Chris.
Simulation at 25 ohms (2:1 SWR) shows each BS170 dissipating about
half a watt, so there is some margin.
JZ
On Thu, Sep 14, 2023 at 9:26?AM Chris KB1NLW via
<chrisrey1= [email protected]> wrote:
>
> John Z
> Thanks for your response. Looks like steady state SWR of 3 will heat the BS170s to their PD limit, so I assume SWR <2 is OK.
>
> Transients from an auto tuner that can be as high as SWR of 10 will cause transient power dissipation significantly higher than the allowable dissipation.? Unfortunately we don't have sufficient thermal data on the BS170 to understand transient thermal response
at the silicon chip in order to determine allowable SWR Transient time product.
>
> Bottom line is Don't Use an Auto Tuner without reducing the transient at the QMX.? Reduced power or attenuation between QMX and tuner.
>
|
Re: #qmx Antenna Tuner Can Kill PAs
#qmx
Tony,
I hope you used the 12V RWTST and not the WTST for 9V on 12V? 73
toggle quoted message
Show quoted text
On 17 Sep 2023, at 16:24, mux_folder2001 <canthony15@...> wrote:
?
Hi Hans,
Well, I was running a 9 volt build and getting reasonable power on 10 meters using the conventional winding method. Then I tried WTST and it did not make much difference. However, today I switched to a 12 volt conventional wound and had very low output on 10
meters. I then switched to WTST and got very good output on 10 meters. The WTST seems to make a very big difference with a 12 volt build and not so much with a 9 volt build, based on a sample of 1.?
Thanks,
Tony
AD0VC
Hi Tony, all
The main motivation for WTST and RWTST was that they completely resolved the parasitic resonance issue which destroyed 10m performance on a percentage of builds. Mysteriously not all, and it was never possible to identity the cause of the parasitic
resonance. These twisted wire transformers solved it. Sometimes we can be so happy about the solution we can accept not really understanding why the problem occurred.?
The bonus is that the twisted wire transformers also provide a little performance improvement (output power, efficiency) compared to the earlier straight transformer winding method.?
73 Hans G0UPL
Thanks for the tip about power measurement. I was not aware of either ALT-Right or CTRL-Left clicks.
I did no see much difference between conventional and WTST myself.
Tony
AD0VC
Tony,
You can clearly see the improvement in coupling that the twisted winding has over the conventional winding, with the expected improvement in leakage inductance as well.
I think that is especially significant at 10m since the leakage inductance has probably not been factored in to the LPF design, and at 10m is approaching the size of the filter inductors.
The negative inductance you see is the parasitic capacitance overwhelming the inductance ie. a resonance. It makes extracting model parameters more challenging.
I run with a variant of the twisted pair transformer in my QDX-HB. It is a 12V 3:2 design . The twisted pair is fully the length of the longer wire. The ratio is obtained by tapping points on the secondary, leaving wire ends to dangle. This
technically more fully realizes the transmission line properties of the twisted pair approach, although I am hard pressed to see much of a performance difference.
JZ
FWIW here is what I measured for the conventional and WTST using my nanovna. I verified the calibration when I saw the negative inductance for the WTST at 28Mhz.?
Conventional????????????????????????
??????L Wind (uh)?L Half Wind?Leak??K
7Mhz??13.56?? ? ? 3.92??? ? ? 0.107?0.996046758211066
10Mhz?10.42?? ? ? 2.98?? ? ? ?0.107?0.994852394070841
14Mhz?7.42??? ? ? 2.13?? ? ? ?0.106?0.992831448793946
18Mhz?5.2???? ? ? 1.51??? ? ? 0.106?0.989755214492647
21Mhz?3.6???? ? ? 1.16??? ? ? 0.106?0.985167780408777
28Mhz?1.26??? ? ? 0.435?? ? ? 0.106?0.957012547395809
????????????????????????
WTST????????????????????
7Mhz??12.98?? ? ? 3.72?? ? ? ?0.067?0.997415767187798
10Mhz?10.13?? ? ? 2.91?? ? ? ?0.067?0.996687504803284
14Mhz?6.97??? ? ? 2.05?? ? ? ?0.066?0.99525416175213
18Mhz?4.11?? ? ? ?1.28?? ? ? ?0.066?0.991938307476537
21Mhz?1.96??? ? ? 0.74??? ? ? 0.066?0.98301908964793
28Mhz?-1.43?? ? ?-0.18?? ? ? ?0.066?0.976650476806393
Tony
AD0VC
Tony,?
In LTSpice if you hover the cursor over a component as if to measure current through it, then press ALT and right click, LTSpice will give you a plot of instantaneous power. You can window that plot for the time period of interest. Then hover
the cursor over the plotted variable name above the plot. Press CTRL and left click. LTSpice will integrate the instantaneous lower over the window interval and open a box with the result.
Since core magnetic materials tend to show reduced permeability at increasing frequency, I like to use my NanoVNA to measure inductance as seen at the relevant frequency. This usually works pretty well unless parasitic capacitances bring a resonance
nearby.
JZ
John,
A couple of things. First, how are you getting the mosfet dissipation from the spice simulation? Are you eyeballing an instantaneous power (Vd * Id) waveform and doing a mental average??
Second, I just built a conventional 9 volt transformer and a WTST transformer and measured the leakage inductance and inter-winding capacitance. Two different cores of course. Measured at 100Khz which is as high as my LCR meter goes.
Conventional:
L winding = 25.5uh
L half winding = 7.65uh
Lleak = .505uh
Cps = 7pf
K = sqrt(1-(.505/25.5)) = .99
WTST:
L winding = 27.6uh
L Half winding = 8.12uh
Lleak = .515uh
Cps = 11.3pf
K = sqrt(1-(.515/27.6)) = .9906
Useful for the SPICE model.
For those that care about this nerd stuff.
Tony
AD0VC
Sure thing, Chris.
Simulation at 25 ohms (2:1 SWR) shows each BS170 dissipating about
half a watt, so there is some margin.
JZ
On Thu, Sep 14, 2023 at 9:26?AM Chris KB1NLW via
<chrisrey1= [email protected]> wrote:
>
> John Z
> Thanks for your response. Looks like steady state SWR of 3 will heat the BS170s to their PD limit, so I assume SWR <2 is OK.
>
> Transients from an auto tuner that can be as high as SWR of 10 will cause transient power dissipation significantly higher than the allowable dissipation.? Unfortunately we don't have sufficient thermal data on the BS170 to understand transient thermal response
at the silicon chip in order to determine allowable SWR Transient time product.
>
> Bottom line is Don't Use an Auto Tuner without reducing the transient at the QMX.? Reduced power or attenuation between QMX and tuner.
>
|
DL2ARL
Paul - AI7JR
Ronald Taylor
Mark K9TR
Karl
Stewart ZR1WT
N5SE