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+ Power Increasing Question
|
Three days ago I completed the build on my QMX+ and I'm happy to report it is working fairly nicely. My only disappointment is the low power output on 20(2.9w), 17(2.8w), and 15(3.3w). Those measurements were made in the terminal application, AFTER having tweaked all the low pass filter toroids. Well, actually 10 meters isn't much to write home about either, providing about 1.9w out. But my primary focus here is on the 20, 17, and 15 meter bands. I've spread the turns on L515 and L516 as much as possible. That gave me a bit of improvement from where it started but being right at 3 watts or less isn't what I was aiming for. So, I'm leaning towards removal of one or two turns.
?But, I've got questions. Are those inductors connected in series, where the inductance adds? Do I remove a turn from each toroid, or two turns from one of them? ?Which one - does it matter? -- Jim / K7TXA Eagle, ID SKCC 10447T BUG 301 |
||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
Hello Jim
I have also just set up a QMX and measured with the terminal application. My QMX runs on 12V as termination I have the dummy load from Hans/Qrplabs I have measured the following power 160m 6,8W ; 80m 6,2W ; 60m 4,9W ; 40m 5,3W ; 30m 4,1W ; 20m 4,2W ; 17m 2,7W ; 15m 4,4W ; 12m 2,7W ; 11m 2,5W ; 10m 2,4W and 50m 2,8W It is clear that the power drops continuously towards the high bands, but I have a dip in the 17m band I wound the output transformer as an RWTST according to the instructions 73 Bernd sorry for my bad englisch |
||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
Without running an electronics course I’d suggest you confine modification initially to the following?
Get a good external SWR /power meter and cross check readings with it Take detailed photos /notes of what you have now (power and winding placements ) Get some plastic sticks and on TX use them to gently compress or stretch the winding and see if things move in the right direction or not? you don’t need to add/remove turns until you got to fully squashed or fully stretched winding layouts and it was still improving as you did? |
||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
Hi Bernd, all On the QMX+, I caution against being too upset if some of the bands have lower power output than others. It's a design of engineering compromises, and not everything is perfect. There are 6 LPFs. Except for 160m and 6m, the other bands all share multiple bands to one LPF. The LPFs are optimized as far as possible; the design has a push-pull PA with very low even harmonic output content, to reduce the requirements on the LPFs. Still there's impedance match to consider and you'll generally find that where a LPF is shared between two (or more) bands, you can play with spreading and squeezing turns, and sometimes you can get a bit of improvement across the board, but you can only get a certain way with it; in the end optimizing one of the shared?bands will impact another.? There are also software things that can (and will) be done. Right now the PA envelope shaping modulator is run a little further below 12V than you'd like; It's a P-channel MOSFET and could get a lot closer to the input voltage (there isn't a 0.6V drop across the transistor as there would be for a bipolar transistor). The reason it only gets to 90% of max (for example) is because the gain factor of that PA modulator's amplification from the DAC output, and the max DAC value in the firmware, is chosen rather conservatively to take into account component tolerances.? What I will do in future is implement a self-calibration feature so that the QMX+ can check the PA modulator gain by measurement and then set itself to use more of the full range, get closer to the input voltage in other words. That will increase power output everywhere. It will be best to make a calibration tool which can also REDUCE power output if it's over 5W (or some target value), so for example on 160m run the PA at a lower voltage.? In this way we should get both higher output power but also more equal power across?the bands.? Also as always do bear in mind that the difference in watts is what we most think of, but what actually matters for communications purposes is the dB power. A given watts' difference is much less in dB so generally we tend to worry too much about maximizing the output power. S-points are in dB (1S = 6dB) so to gain a 1 S-point advantage you need to multiply by four. For example, from 4 Watts of power to 16 Watts is 1 S-point. From 4 Watts to 5 Watts is under 1dB, less than 1/6th of an S-point... In other words, we should worry less.? 73 Hans G0UPL On Mon, Jun 24, 2024 at 10:21?AM Bernd DL3BLA via <b.langner=[email protected]> wrote: Hello Jim |
||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
I recently completed a QMX+ kit.? The firmware loaded, switch on current 90mA.? I used the puTTy terminal programme to look at all the parameters/values that the QMX+ should exhibit. I found that power output was as follows: ? 160m???????????????????? 5.85W (calculated – oscilloscope across dummy load) 80m?????????????????????? 4.66W 60m?????????????????????? 1.9W 40m?????????????????????? .75W 30m?????????????????????? .36W 20m?????????????????????? 0.2W 17m?????????????????????? .07W 15m?????????????????????? .2W 12m?????????????????????? .04W 11m?????????????????????? .03W 10m?????????????????????? .04W 12m?????????????????????? 0W ? I first suspected that the filter toroids or the transformers may have been incorrectly wound but subsequent checking (firmware sweeps, rewinding T501 and T507, attaching a Nanovna to each RF & LPF ?showed they were fine). My concern rose when I looked at some voltages.? The CLK voltages were all ok but on measuring the outputs of IC’s 501, 502 and 503 from CLK2 input gave some interesting results.? These were p-p voltages with a 100MHz oscilloscope. ? IC501 (74AHC1G86D) ??????????????????????????????? 1.8MHz???????????????? 14MHz????????????????? 28MHz? ??????????????? 50MHz Clk2 input??????????? 4.7????????????????????????? 4.8????????????????????????? 4.9????????????????????????? 5.0????????? (Pin1) Pin4 output??????? 4.2????????????????????????? 3.5????????????????????????? 2.8????????????????????????? 2.7 ? IC502 (74AHC1G86D) Clk2 input??????????? 4.7????????????????????????? 4.8????????????????????????? 4.9????????????????????????? 5????????????? (Pin1) Pin4 output??????? 4.2????????????????????????? 3.5????????????????????????? 2.8????????????????????????? 2.8 ? IC503 (74ACT08D) Pins2,5,10,13????? 3.3????????????????????????? 3.3????????????????????????? 3.3????????????????????????? 3.3????????? (From TX 3.3V) Pins 8,11????????????? 7.4????????????????????????? 4????????????????????????????? 1.6????????????????????????? .4??????????? (gates of Q503/Q505) Pins3,6????????????????? 7.4????????????????????????? 4????????????????????????????? 1.7????????????????????????? .6??????????? (gates of Q504/Q506) ? Q503/Q505 Drain???????????????????? 63V??????????????????????? 20V??????????????????????? 4V?????????????????????????? 0 ? Q504/Q506???????? Drain???????????????????? 65V??????????????????????? 19V??????????????????????? 6V?????????????????????????? 0 ? There are no passive components in this part of the circuit so my conclusion is that IC’s 501, 502 and/or 503 do not have the High Frequency response they should. Are the IC’s faulty or is there something else I should check.? Some expert help here is needed. ? Greg, ZL4GW |
||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
Bi Ludwig
Thank you for the reply.? The probe is a PP-150 which came with the oscilloscope.? I have used it for a few years. Its bandwidth (@10:1) is 100 MHz.? The probe compensation allows the square wave calibrator to show a good squarewave with no ringing. The input impedance is 10Mohm/18.5pF. It is the output voltages from the ExOR gates and the AND gates that concerns me as the lower the voltage output, the lower the RF output from the BS170's.? The measured RF output is within a few 10ths of a Watt as shown on the puTTY Terminal.? Both measurements decrease at the same rate to 0 @ 6m. Any help you can give would be appreciated. Regards Greg |
||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
On Tue, Jul 2, 2024 at 02:32 AM, Greg Walsh wrote:
It is the output voltages from the ExOR gates and the AND gates that concerns me as the lower the voltage output, the lower the RF output from the BS170's.?Greg, it looks like an increasing load to IC503 produced by the PA. The higher load may affect also the load to IC501 and IC502. Could you show the waveforms at G and D of each group of BS170 (with DC, not only AC) at let's say 1.8 MHz and 28 MHz? 73, Ludwig |
||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
I am seeing similar indications out of IC503 with my just completed QMX+.?
It was working fine for a couple of hours on 20m at 4.3w. Tried tuning up on 10m and power out went to 0w. Switched back to 20m and still no output. Having been through this and with a handful of spare BS170s, I just clipped them out and started looking at IC503. I'm getting a wonky output from one pair of the gates (C&D) and only 2v P-P. Good and bad traces attached.. The output of IC501 is also a bit weird compared to IC502 but I'm thinking this may be caused by a bad gate within IC503. I will be replacing IC503 tomorrow. If all looks well I'll plug in a new set of BS170s and see how it goes. -- Randy, N4OPI |
||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
Success (so far). I replaced IC503 and now have a nice 5v TTL level signal out on both sides. BS170s not installed yet. Waveform attached.
This was on 20m. I tried 6m and it was bit lower amplitude at 4v. But still much better than before. The IC501 and IC502 XORs also have a nice clean output now. Onward.. -- Randy, N4OPI |
||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
Hi Greg,
fine pictures except the content. The waveforms looking strange to me. Lets have a look at 1.8 MHz. The gate voltage is going down to -2V! Why negative? I guess not from the IC503. The voltage is >= 1V for around 70% of the period. The part of the waveform with voltage around 1 V is wrong. The drain voltage is low when the FET is ON. The duty cycle is around 70% for both groups. It looks like the FET are switched on by gate voltages >= 1V, so for 70% of the period. An overlapping ON is bad. For 28 MHz the FET are between ON and OFF all the time. The gate voltage stick at a medium value and couldn't go to LOW and HIGH. Something is blocking "large" voltage alterations. So I have some more questions:
73, Ludwig |
||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
Hi Ludwig.? Thank you for the analysis.? To answer your first 3 questions:
The GND lead was clipped to the bnc output connector The ground lead is approx 150mm The original BS170's are in circuit. Unfortunately while looking for the waveforms you asked for, I must have shorted the +ve at the BS170's as they have no Drain voltage so I think Q507 has been destroyed as there is 11.5V on its Source, nearly 11.5V on its Gate and 0V on its Drain.? So at this stage I have ordered a replacement AOD403 from Digikey and it will probably take a week or so to be delivered. Thank you for your help to date and I will update when I have replaced the AOD403. Regards Greg |
||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
Greg, sorry to read about the damage.
But you could check now the gate waveforms without working BS170. (Of course the gate capacity is "working".) Now there is no possible feedback from drain current to the gate circuit. From my post before: The voltage is >= 1V for around 70% of the period. The part of the waveform with voltage around 1 V is wrong.It would be interesting to know if the input duty cycle of IC503 is equal to the wrong output duty cycle. But maybe without drain voltage the wrong duty cycle is not wrong anymore? 73 Ludwig |
||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
Hi Ludwig
Your reply inspired me to make the gate measurements - The first 4 meaurements were with the FET's in circuit but for comparison I removed Q504 and Q506 and remeasured them.? The 1.8 MHz waveforms were the same as first measured (approx 5V pk), so have not included them, but the 28 MHz waveforms were certainly different in amplitude.? (The smoother waveform at 28 MHz may be partially due to the 'scope and probe's frequency response). I hope the file names are sufficiently descriptive. I'll have a look at the duty cycle shortly. Regards Greg |
||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
Hi Greg,
interesting news. So for 1.6 MHz no more negative voltage at the gates and 50% duty cycle. This looks fine. So something was caused by the switched drain voltage / current. But how came this into the gate circuit, by the chanel to gate capacity??? Or is there a wrong coupling between the drain and the gate circuit??? Questions but no answer. But for 28 MHz it's terrible, both with all FET and with two FET removed. Nothing to be seen what looks like a 28 MHz signal. Going back to your post from Jul 3th. The drain voltage was going up to around 61V. Assuming the supply voltage was 12 V this was 5 times of the supply!!! Checking my QMX the drain voltage was in the region of the supply voltage. Your high value looks like produced by a resonant circuit with a high real part of the impedance, not near 50 ohm. Could you check
It would be also fine to remove all BS170 and check the signals from IC503 for different bands. The step next is a visual check of the capacitors from "+12V" to GND and from VCC to GND. To get a better impression of your system:
Sorry for this lot of work. 73, Ludwig |
||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
Hi Ludwig
Sorry for the delay but family matters over the weekend.? I've done many measurements and am reasonably certain either the 'scope or the probes are limited in bandwidth.? See my notes and findings below... Could you check
It would be also fine to remove all BS170 and check the signals from IC503 for different bands.? I removed all BS170's and made the following measurements:
The step next is a visual check of the capacitors from "+12V" to GND and from VCC to GND.? I checked all caps on both rails visually and with a DMM where possible and all appear ok. To get a better impression of your system:
Regards, Greg. |
||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
On Mon, Jul 8, 2024 at 06:37 AM, Greg Walsh wrote:
Sorry for the delay but family matters over the weekend.Hi Greg, this is important. And I needed some time to think about your last post. The outputs from IC503 are better. How about the waveforms, something like rectangular (with smooth slopes)? Is the TX frequency the largest part of the waveform? Decrease of the output may point to the frequency response of the probe or the measurement situation or it could point to a load connected to the output. Who knows. Could you check please some resistance values:
About the photos attached and with some markers. Some of the solders are looking a bit strange, far away from a smooth surface. BS170 pins_answer1: red - unwanted tin ball? blue - for exact measurement cleaning would be good (washing the flux away) yellow - a circle structure around the soldered wire, cold solder? T501 end 1_2_answer1: red - the enamel wire comes with 45° from the PCB. It seems to be very close to the GND plane. T501 end 2_answer1: yellow - see about strange solder above red - the wires are twisted. Was there enough room in the hole to put the twisted wires through the PCB? The solder looks wide, maybe to wide? Could it be the tin is touching the GND plane only isolated by the thin green silk screen? Btw I've found no information about your power supply. Good luck and I hope you will find something to improve. 73 Ludwig |
||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Bruce Akhurst
to navigate to use esc to dismiss