|
Hans,
I'm guessing your team has been building and shipping quite a few factory built QDX-HBs.
I was wondering if you all are encountering many suffering from low power output?
Earlier your investigation pointed to the output transformer being the culprit.
I'd like to know if you've made any more headway on its cause and how to solve it.
The QDX-HB I built is in this category and I've tried rewinding the transformer without seeing much change in its output, 3W on 20m and 1W on 10m.
Thanks!
--Al
WD4AH
|
I have the same question.? About 2w out on 10m.? I ordered some transformer torroids to swap out and see if I can support Hans' theory.
Brad, AB4BA
|
Hi all
Yes we assemble 20-10m QDXs and a percentage of them seem to have low power on 10m. I didn't keep count but I'd estimate somewhere between 1-in-3 and 1-in-4 have low power. Some can be "fixed", others are still pending my attention.
I have spent a lot of time on it but yet to find a definitive solution that will work on both 12V and 9V versions all the time.
There is too much to remember off the top of my head. So I'll write up the details tomorrow of the various things I've tried, when I'm back in the lab... Maybe some of these things will work for some of you.
73 Hans G0UPL
I have the same question.? About 2w out on 10m.? I ordered some transformer torroids to swap out and see if I can support Hans' theory.
Brad, AB4BA
|
Al,
For comparison, here's what I have for QDX HB output into Hans' 50 Ohm dummy load with 8.70 VDC input (unit built for 9.0 V nominal):
20m - 3.85 watts
17m - 3.18 watts
15m - 2.43 watts
12m - 3.47 watts
10m - 1.58 watts
Tolerance note: I'm utilizing a Monitor Sensors Power & SWR Meter (130 kHz - 30 MHz, 0.00-2000 watts range), Resolution is 0.01 watts on the 0.00-19.99 watts range, Accuracy states, "This is a precision instrument with 48 calibration points and each instrument is individually calibrated at the factory. ?Expect power measurements to be better than 5% over most of the power and frequency range."
If anyone is interested, below is my output for the LB (80m - 20 m) version with the same 8.70 volts input:
80m - 3.98 watts
40m - 3.92 watts
30m - 3.64 watts
20m - 4.26 watts
Ciao! ?Peter W1DAD
|
Thank you all for the information!
I initially built the kit using the supplied parts. In troubleshooting its lower output I made a few modifications...
When I rewound T1, the output transformer, I used 24 gauge wire for the secondary, but kept 22g for the primary. I thought it might reduce any negative capacitive coupling within the transformer. It didn't seem to make much if any difference.
I rewound the 2 of the 3 LPF also using 24g wire. I wanted to make it easier to compress/expand the windings. Again, not much improvement.
I'm using SIP sockets for the BS-170s. Mill-Max brand, Mouser??I've found the to be better than the generic one's I'd used on a series 4 QDX-LB. These are 14-pin strips that I cut down to two 6x1s and were more cost effective, IMO.
Again, it's a 9V build that I mainly use now on WSPR running at 7.49V and draws ~700mA in transmit on 10m.
Any thoughts about using Type 61 mix for the binocular core?
Thanks again!
--Al
|
FWIW, I replaced the transformer torroid with a new one form kitsandparts.com, same torroid (BN43-202) as the original, same 3:3 winding for 9v.
Results are as follows:
?Original kit transformer
20m-4.2w
17m-3.4w
15m-3.5w
12m-2.0w
10m-2.5w
New transformer
20m-4.4w
17m-3.6w
15m-3.3w
12m-2.0w
10m-2.9w
So I?got a slight but not significant power improvement with a new torroid/transformer.? That being said, this past weekend with the original kit I worked 23 POTA stations on 20/17/10m.? So I am happy with the QDX HB despite a slightly low power output on 10/12m.
Brad, AB4BA
|
Hi all
So some info on where I am at on this, here.?
About 1/4 to 1/3 of high-band 12V QDX transceivers seem to suffer low output power on 10m. It seems to happen more frequently with 9V builds than 12V. A "good" high-bands QDX has approximately equal power output from 20 to 10m. A "bad" one has about 1/2 the power on 10m compared to 20m. Maybe less.?
The issue can be complicated by the fact that sometimes the LPFs require some optimization anyway. There are a great variety of winding styles. My personal feeling is that the 10m LPF cutoff might be a bit close to 30m which makes it easy to get a little higher attenuation than desired, on the 10m operating frequency; but this is anyway slight and is NOT the cause of the low power problem in general. If you do see low power, try toroid tweaking first to see if it is just a build issue (many times, it is just toroid tweaking sorts it out).?
The issue also occurs on 10m QDX-M radios so it is not a problem with the PIN diode LPF switching arrangements of the multi-band QDX.?
I have been working on this with EX0AA who has several QDXs including one that shows this problem, and has lots of good suggestions and advice. Thanks EX0AA!?
So far:
1) I took a good 12V unit and a bad 12V unit. I literally changed every component one by one, BS170s, toroids, capacitors, everything in the signal path. I was able to conclude conclusively that the problem followed the BN43-202 transformer. Not any other component. Yet - measurement of inductance of the transformer did not show any significant differences.?
2) There appears to be a weak low-Q resonance at around 10m frequency, which eats a few dB of the power output. Sometimes it appears that the resonance frequency is above 10m so it doesn't cause trouble. Sometimes below. It isn't clear what is resonating with what but the transformer is certainly one of the elements.?
3) Changing the winding style is a possible thing to try - it seemed to fix one low power high-band QDX but not others. By this I mean, whether you wind the secondary winding first, or the primary; and the primary can be wound very symmetrically by starting with the center tap and working outward.?
4) I changed to 5:5 (9V QDX) and the power output was restored on 10m however the weak low Q resonance just moved DOWN in frequency, with now 15m (and to a lesser extent 12m) power output suffering. This also supports the resonance theory.?
5) The fact that 9V QDX appear to show this problem more frequently than 12V ones could indicate that a capacitance of the BS170s is involved, perhaps the capacitance averages to a lower number at higher voltages (like a reverse biased varactor) therefore this brings the resonant frequency lower and makes it more likely to collide with 10m operations.?
6) One of the first things I found helped, is to reduce the inductance of L10 by removing two turns and increase the inductance of L4 by squeezing its turns together as much as possible; this had the effect of increasing the power output on 10m but then 12m was MUCH too high power; it was possible to control the 12m over-power situation by editing the forward diode PIN current in the Band Configuration screen for 12m, to 5mA instead of the default 30mA. Then you could get a fairly even power output from 20m to 10m. BUT, the efficiency on 10m is very poor; also, this solution was not found to work universally on every QDX. So it's not really a good solution.?
7) I tried a transformer consisting of 0.5 turns on each primary winding, and 1 turn secondary (9V operation). It showed good power output on all bands 20-10m however, the efficiency was very poor leading to overheating.?
8) I tried small capacitors from the drain of each MOSFET pair to ground. That had the effect of restoring power on 10m but moved this parasitic resonant frequency lower so power output on 15m was reduced.?
9) I tried a small capacitor between the drains and it also just moves the resonance down.?
10) Note that current 20-10m QDX are being supplied with T37-17 toroids instead of T37-10 toroids, since the latter are currently out of stock everywhere; the T37-17 are yellow/blue colour. There is no observable difference in performance between the two.?
11) Thicker or thinner wire may be used on the LPF toroids and the output transformer (i.e. 0.3mm or 0.6mm) and it does not seem to make any significant difference. Theoretically 0.6mm should lead to lower losses, higher Q, due to larger surface area; however it is harder to wind, harder to wind symmetrically and neatly, which also makes a marginal difference; for many constructors they may typically wind a neater coil with thinner wire; with less air gap between the toroidal core and the wire, which also improves things. So which is the dominant effect is hard to say. With the T37-17 toroids they need a few more turns than the T37-10 so we will supply 0.3mm wire to make it easier.?
12) I did a 2:2 transformer. That is to say, 1 turn primaries, and 2 turn secondary, for 9V operation. The center tap comes out at the wrong end of the transformer so you have to take a wire over the top; no big deal... This had good power out across the bands and also good efficiency. It SOLVED the problem for that particular QDX. However I repeated the attempt on another low power QDX and it did NOT fix that one... so again not seeming to be a universal solution.?
13) I wound a 3+3 turns primary, 4 turns secondary transformer on a T50-43 toroid using 0.60 mm wire instead of the BN43-202. The 3+3 turns bifilar, 4 turns wound over the top separately. Looked a horrible mess. Power output was low on 15m, still indicating a resonance.?
14) I wound a BN43-202 with 3 turns bifilar, 0.3mm wire, center tapped, as the primary, and 3 turns secondary. This had good power output around 4W across the bands and good efficiency. It SOLVED the problem for that QDX. However I tried it on another low power QDX and it did not solve the problem there.?
15) 4 turns bifilar on a FT37-43 with a center tap on one of the wires (2+2 : 4) at 9V operation - good equal power across all bands but low efficiency.?
16) Then tried 6 turns bifilar on FT37-43 with a center tap, and this time good equal power and efficiency improved, but still poor
17) Then went to 14 turns bifilar on FT37-43 with center tap, 9V operation: now low power again on 10m band...
18) So - could there be a sweet spot, somewhere between 6 bifilar turns which has good power but low efficiency, and 14 turns which has low power on 10m but good efficiency?elsewhere...? taking 2 bifilar turns off at a time (so as to keep the center tap "center"), unfortunately NOT... 12 turns is still low power on 10m and if you go to 10 turns you get good power but now efficiency is not good enough.?
19) Now a gross mutation transformer idea by EX0AA... which has 2 : 2 turns, bifilar wound; on one end of the transformer you cut one of the wires, and scrape the other one to make it the center tap. At the other end, you join the right pair of wires. So what you end up with is this WEIRD transformer where the wires are all bifilar wound - but the two windings are sort of 180-degrees out with each other. Kind of like imagine a circle, a loop, with the entrance to the loop at the bottom; then a second 1-turn loop, also a full turn, whose entrance is at the top (180-degrees apart). The two loops travel in parallel (in fact twisted bifilar) but their ends are at opposite sides. Surprisingly though I thought it weird, this transformer worked - but it did not solve the low power on 10m problem.?
20) it would be a VERY good idea to try type -61 material but I don't have any here yet to try. If anyone has any on hand and wants to try it, it would be very interesting to hear your results.?
So - we have some closer idea of what the problem is: some kind of parasitic resonance involving the transformer; and various solutions, some of which work some of the time, but nothing which seems to work all the time, and does so with good equal power output and without terrible efficiency.?
As I said - some low power issues are due to how the inductors are constructed and the LPF toroids need tweaking. But there is definitely some kind of parasitic resonance issue here. We have very experienced QDX builders here who have made literally HUNDREDS, and every QDX comes out identical like its predecessor. So we know there isn't a variation in build style.?
Anyone has any magic solutions, do not hesitate to share them!
Meanwhile as some people have commented, even 2W on 10m goes a long way and gives excellent QSOs.?
73 Hans G0UPL
FWIW, I replaced the transformer torroid with a new one form , same torroid (BN43-202) as the original, same 3:3 winding for 9v.
Results are as follows:
?Original kit transformer
20m-4.2w
17m-3.4w
15m-3.5w
12m-2.0w
10m-2.5w
New transformer
20m-4.4w
17m-3.6w
15m-3.3w
12m-2.0w
10m-2.9w
So I?got a slight but not significant power improvement with a new torroid/transformer.? That being said, this past weekend with the original kit I worked 23 POTA stations on 20/17/10m.? So I am happy with the QDX HB despite a slightly low power output on 10/12m.
Brad, AB4BA
|
Hans, this might sound unorthodox but it might be worth a try. For the binocular core, at least for the 3T primary and 3T secondary (9V) version, twist the primary and secondary wires together.? The secondary is now coming out on the wrong side so the secondary wires would have to be brought across the top of the binocular core to the other side.? This makes it a transmission line transformer, which will reduce any leakage inductances further and reduce any self-capacitive effects within the transformer. The leakage inductances and transformer primary self-capacitance will be absorbed in the transmission line.? ?This may help to move the weak resonance on 10 meters further up in frequency.? For the 3 turn primary to? 2 turn secondary (12v) version, wins 3 turns again, but let the third turn of the secondary still be twisted with the primary but remain unconnected at the output. A 2 turn tap would have to be added on the secondary wire.? In either case, the center tap would have to be made on the primary side of the twisted pair, making it more difficult to do.
-Steve K1RF
------ Original Message ------
Date 3/28/2023 3:21:22 PM
Subject Re: [QRPLabs] Any More Information on Low Power Output QDX-HB
Hi all
So some info on where I am at on this, here.?
About 1/4 to 1/3 of high-band 12V QDX transceivers seem to suffer low output power on 10m. It seems to happen more frequently with 9V builds than 12V. A "good" high-bands QDX has approximately equal power output from 20 to 10m. A "bad" one has about 1/2 the power on 10m compared to 20m. Maybe less.?
The issue can be complicated by the fact that sometimes the LPFs require some optimization anyway. There are a great variety of winding styles. My personal feeling is that the 10m LPF cutoff might be a bit close to 30m which makes it easy to get a little higher attenuation than desired, on the 10m operating frequency; but this is anyway slight and is NOT the cause of the low power problem in general. If you do see low power, try toroid tweaking first to see if it is just a build issue (many times, it is just toroid tweaking sorts it out).?
The issue also occurs on 10m QDX-M radios so it is not a problem with the PIN diode LPF switching arrangements of the multi-band QDX.?
I have been working on this with EX0AA who has several QDXs including one that shows this problem, and has lots of good suggestions and advice. Thanks EX0AA!?
So far:
1) I took a good 12V unit and a bad 12V unit. I literally changed every component one by one, BS170s, toroids, capacitors, everything in the signal path. I was able to conclude conclusively that the problem followed the BN43-202 transformer. Not any other component. Yet - measurement of inductance of the transformer did not show any significant differences.?
2) There appears to be a weak low-Q resonance at around 10m frequency, which eats a few dB of the power output. Sometimes it appears that the resonance frequency is above 10m so it doesn't cause trouble. Sometimes below. It isn't clear what is resonating with what but the transformer is certainly one of the elements.?
3) Changing the winding style is a possible thing to try - it seemed to fix one low power high-band QDX but not others. By this I mean, whether you wind the secondary winding first, or the primary; and the primary can be wound very symmetrically by starting with the center tap and working outward.?
4) I changed to 5:5 (9V QDX) and the power output was restored on 10m however the weak low Q resonance just moved DOWN in frequency, with now 15m (and to a lesser extent 12m) power output suffering. This also supports the resonance theory.?
5) The fact that 9V QDX appear to show this problem more frequently than 12V ones could indicate that a capacitance of the BS170s is involved, perhaps the capacitance averages to a lower number at higher voltages (like a reverse biased varactor) therefore this brings the resonant frequency lower and makes it more likely to collide with 10m operations.?
6) One of the first things I found helped, is to reduce the inductance of L10 by removing two turns and increase the inductance of L4 by squeezing its turns together as much as possible; this had the effect of increasing the power output on 10m but then 12m was MUCH too high power; it was possible to control the 12m over-power situation by editing the forward diode PIN current in the Band Configuration screen for 12m, to 5mA instead of the default 30mA. Then you could get a fairly even power output from 20m to 10m. BUT, the efficiency on 10m is very poor; also, this solution was not found to work universally on every QDX. So it's not really a good solution.?
7) I tried a transformer consisting of 0.5 turns on each primary winding, and 1 turn secondary (9V operation). It showed good power output on all bands 20-10m however, the efficiency was very poor leading to overheating.?
8) I tried small capacitors from the drain of each MOSFET pair to ground. That had the effect of restoring power on 10m but moved this parasitic resonant frequency lower so power output on 15m was reduced.?
9) I tried a small capacitor between the drains and it also just moves the resonance down.?
10) Note that current 20-10m QDX are being supplied with T37-17 toroids instead of T37-10 toroids, since the latter are currently out of stock everywhere; the T37-17 are yellow/blue colour. There is no observable difference in performance between the two.?
11) Thicker or thinner wire may be used on the LPF toroids and the output transformer (i.e. 0.3mm or 0.6mm) and it does not seem to make any significant difference. Theoretically 0.6mm should lead to lower losses, higher Q, due to larger surface area; however it is harder to wind, harder to wind symmetrically and neatly, which also makes a marginal difference; for many constructors they may typically wind a neater coil with thinner wire; with less air gap between the toroidal core and the wire, which also improves things. So which is the dominant effect is hard to say. With the T37-17 toroids they need a few more turns than the T37-10 so we will supply 0.3mm wire to make it easier.?
12) I did a 2:2 transformer. That is to say, 1 turn primaries, and 2 turn secondary, for 9V operation. The center tap comes out at the wrong end of the transformer so you have to take a wire over the top; no big deal... This had good power out across the bands and also good efficiency. It SOLVED the problem for that particular QDX. However I repeated the attempt on another low power QDX and it did NOT fix that one... so again not seeming to be a universal solution.?
13) I wound a 3+3 turns primary, 4 turns secondary transformer on a T50-43 toroid using 0.60 mm wire instead of the BN43-202. The 3+3 turns bifilar, 4 turns wound over the top separately. Looked a horrible mess. Power output was low on 15m, still indicating a resonance.?
14) I wound a BN43-202 with 3 turns bifilar, 0.3mm wire, center tapped, as the primary, and 3 turns secondary. This had good power output around 4W across the bands and good efficiency. It SOLVED the problem for that QDX. However I tried it on another low power QDX and it did not solve the problem there.?
15) 4 turns bifilar on a FT37-43 with a center tap on one of the wires (2+2 : 4) at 9V operation - good equal power across all bands but low efficiency.?
16) Then tried 6 turns bifilar on FT37-43 with a center tap, and this time good equal power and efficiency improved, but still poor
17) Then went to 14 turns bifilar on FT37-43 with center tap, 9V operation: now low power again on 10m band...
18) So - could there be a sweet spot, somewhere between 6 bifilar turns which has good power but low efficiency, and 14 turns which has low power on 10m but good efficiency?elsewhere...? taking 2 bifilar turns off at a time (so as to keep the center tap "center"), unfortunately NOT... 12 turns is still low power on 10m and if you go to 10 turns you get good power but now efficiency is not good enough.?
19) Now a gross mutation transformer idea by EX0AA... which has 2 : 2 turns, bifilar wound; on one end of the transformer you cut one of the wires, and scrape the other one to make it the center tap. At the other end, you join the right pair of wires. So what you end up with is this WEIRD transformer where the wires are all bifilar wound - but the two windings are sort of 180-degrees out with each other. Kind of like imagine a circle, a loop, with the entrance to the loop at the bottom; then a second 1-turn loop, also a full turn, whose entrance is at the top (180-degrees apart). The two loops travel in parallel (in fact twisted bifilar) but their ends are at opposite sides. Surprisingly though I thought it weird, this transformer worked - but it did not solve the low power on 10m problem.?
20) it would be a VERY good idea to try type -61 material but I don't have any here yet to try. If anyone has any on hand and wants to try it, it would be very interesting to hear your results.?
So - we have some closer idea of what the problem is: some kind of parasitic resonance involving the transformer; and various solutions, some of which work some of the time, but nothing which seems to work all the time, and does so with good equal power output and without terrible efficiency.?
As I said - some low power issues are due to how the inductors are constructed and the LPF toroids need tweaking. But there is definitely some kind of parasitic resonance issue here. We have very experienced QDX builders here who have made literally HUNDREDS, and every QDX comes out identical like its predecessor. So we know there isn't a variation in build style.?
Anyone has any magic solutions, do not hesitate to share them!
Meanwhile as some people have commented, even 2W on 10m goes a long way and gives excellent QSOs.?
73 Hans G0UPL
FWIW, I replaced the transformer torroid with a new one form , same torroid (BN43-202) as the original, same 3:3 winding for 9v.
Results are as follows:
?Original kit transformer
20m-4.2w
17m-3.4w
15m-3.5w
12m-2.0w
10m-2.5w
New transformer
20m-4.4w
17m-3.6w
15m-3.3w
12m-2.0w
10m-2.9w
So I?got a slight but not significant power improvement with a new torroid/transformer.? That being said, this past weekend with the original kit I worked 23 POTA stations on 20/17/10m.? So I am happy with the QDX HB despite a slightly low power output on 10/12m.
Brad, AB4BA
|
Hans,
Thank you for all your troubleshooting work on this issue.? The multiple partial solutions gives a good starting point for the community to continue experimenting.
Of course, the more pragmatic among us will just buy another QDX for a 75% chance at full power.? Alternatively,? just make QSO's with 1 watt on 10m and brag about your QRPp skills.? It's all good fun.
Brad, AB4BA
|
On Tue, Mar 28, 2023 at 05:11 PM, Steven Dick, K1RF wrote:
Hans, this might sound unorthodox but it might be worth a try. For the binocular core, at least for the 3T primary and 3T secondary (9V) version, twist the primary and secondary wires together.? This makes it a transmission line transformer, which will reduce any leakage inductances further and reduce any self-capacitive effects within the transformer.?
I've been winding T1's primary starting at the center tap and working out symmetrically, as Hans put it, on the QDXs I've built. This method lends itself to Brad's transmission line transformer proposal. I tried building one and had some success using 24g wire. The primary wire is cut longer than the secondary and has the center tap loop formed prior to twisting primary and secondary together. I have yet to try it in the circuit. One question, is 24g sufficient to handle the current demand of the PA FETs? Getting a 22g bifilar pair threaded through the binocular core was very difficult and resulted in stripping some enamel coating. Hans, thank you for documenting the work you all have done trying to solve this issue! --Al
?
|
Hi Steve, That's one of the things I already tried, it's one of the items on the list in my post... it didn't help...?
Hi Al, In many of the experiments I tried, I ended up scraping enamel off the thick 0.6mm wire. As long as the windings don't short to each other or another part of themselves, it doesn't matter. But I do think it's much easier doing this stuff with the thinner wire!
On Wed, Mar 29, 2023 at 12:11?AM Steven Dick, K1RF < sbdick@...> wrote: Hans, this might sound unorthodox but it might be worth a try. For the binocular core, at least for the 3T primary and 3T secondary (9V) version, twist the primary and secondary wires together.? The secondary is now coming out on the wrong side so the secondary wires would have to be brought across the top of the binocular core to the other side.? This makes it a transmission line transformer, which will reduce any leakage inductances further and reduce any self-capacitive effects within the transformer. The leakage inductances and transformer primary self-capacitance will be absorbed in the transmission line.? ?This may help to move the weak resonance on 10 meters further up in frequency.? For the 3 turn primary to? 2 turn secondary (12v) version, wins 3 turns again, but let the third turn of the secondary still be twisted with the primary but remain unconnected at the output. A 2 turn tap would have to be added on the secondary wire.? In either case, the center tap would have to be made on the primary side of the twisted pair, making it more difficult to do.
-Steve K1RF
------ Original Message ------
Date 3/28/2023 3:21:22 PM
Subject Re: [QRPLabs] Any More Information on Low Power Output QDX-HB
Hi all
So some info on where I am at on this, here.?
About 1/4 to 1/3 of high-band 12V QDX transceivers seem to suffer low output power on 10m. It seems to happen more frequently with 9V builds than 12V. A "good" high-bands QDX has approximately equal power output from 20 to 10m. A "bad" one has about 1/2 the power on 10m compared to 20m. Maybe less.?
The issue can be complicated by the fact that sometimes the LPFs require some optimization anyway. There are a great variety of winding styles. My personal feeling is that the 10m LPF cutoff might be a bit close to 30m which makes it easy to get a little higher attenuation than desired, on the 10m operating frequency; but this is anyway slight and is NOT the cause of the low power problem in general. If you do see low power, try toroid tweaking first to see if it is just a build issue (many times, it is just toroid tweaking sorts it out).?
The issue also occurs on 10m QDX-M radios so it is not a problem with the PIN diode LPF switching arrangements of the multi-band QDX.?
I have been working on this with EX0AA who has several QDXs including one that shows this problem, and has lots of good suggestions and advice. Thanks EX0AA!?
So far:
1) I took a good 12V unit and a bad 12V unit. I literally changed every component one by one, BS170s, toroids, capacitors, everything in the signal path. I was able to conclude conclusively that the problem followed the BN43-202 transformer. Not any other component. Yet - measurement of inductance of the transformer did not show any significant differences.?
2) There appears to be a weak low-Q resonance at around 10m frequency, which eats a few dB of the power output. Sometimes it appears that the resonance frequency is above 10m so it doesn't cause trouble. Sometimes below. It isn't clear what is resonating with what but the transformer is certainly one of the elements.?
3) Changing the winding style is a possible thing to try - it seemed to fix one low power high-band QDX but not others. By this I mean, whether you wind the secondary winding first, or the primary; and the primary can be wound very symmetrically by starting with the center tap and working outward.?
4) I changed to 5:5 (9V QDX) and the power output was restored on 10m however the weak low Q resonance just moved DOWN in frequency, with now 15m (and to a lesser extent 12m) power output suffering. This also supports the resonance theory.?
5) The fact that 9V QDX appear to show this problem more frequently than 12V ones could indicate that a capacitance of the BS170s is involved, perhaps the capacitance averages to a lower number at higher voltages (like a reverse biased varactor) therefore this brings the resonant frequency lower and makes it more likely to collide with 10m operations.?
6) One of the first things I found helped, is to reduce the inductance of L10 by removing two turns and increase the inductance of L4 by squeezing its turns together as much as possible; this had the effect of increasing the power output on 10m but then 12m was MUCH too high power; it was possible to control the 12m over-power situation by editing the forward diode PIN current in the Band Configuration screen for 12m, to 5mA instead of the default 30mA. Then you could get a fairly even power output from 20m to 10m. BUT, the efficiency on 10m is very poor; also, this solution was not found to work universally on every QDX. So it's not really a good solution.?
7) I tried a transformer consisting of 0.5 turns on each primary winding, and 1 turn secondary (9V operation). It showed good power output on all bands 20-10m however, the efficiency was very poor leading to overheating.?
8) I tried small capacitors from the drain of each MOSFET pair to ground. That had the effect of restoring power on 10m but moved this parasitic resonant frequency lower so power output on 15m was reduced.?
9) I tried a small capacitor between the drains and it also just moves the resonance down.?
10) Note that current 20-10m QDX are being supplied with T37-17 toroids instead of T37-10 toroids, since the latter are currently out of stock everywhere; the T37-17 are yellow/blue colour. There is no observable difference in performance between the two.?
11) Thicker or thinner wire may be used on the LPF toroids and the output transformer (i.e. 0.3mm or 0.6mm) and it does not seem to make any significant difference. Theoretically 0.6mm should lead to lower losses, higher Q, due to larger surface area; however it is harder to wind, harder to wind symmetrically and neatly, which also makes a marginal difference; for many constructors they may typically wind a neater coil with thinner wire; with less air gap between the toroidal core and the wire, which also improves things. So which is the dominant effect is hard to say. With the T37-17 toroids they need a few more turns than the T37-10 so we will supply 0.3mm wire to make it easier.?
12) I did a 2:2 transformer. That is to say, 1 turn primaries, and 2 turn secondary, for 9V operation. The center tap comes out at the wrong end of the transformer so you have to take a wire over the top; no big deal... This had good power out across the bands and also good efficiency. It SOLVED the problem for that particular QDX. However I repeated the attempt on another low power QDX and it did NOT fix that one... so again not seeming to be a universal solution.?
13) I wound a 3+3 turns primary, 4 turns secondary transformer on a T50-43 toroid using 0.60 mm wire instead of the BN43-202. The 3+3 turns bifilar, 4 turns wound over the top separately. Looked a horrible mess. Power output was low on 15m, still indicating a resonance.?
14) I wound a BN43-202 with 3 turns bifilar, 0.3mm wire, center tapped, as the primary, and 3 turns secondary. This had good power output around 4W across the bands and good efficiency. It SOLVED the problem for that QDX. However I tried it on another low power QDX and it did not solve the problem there.?
15) 4 turns bifilar on a FT37-43 with a center tap on one of the wires (2+2 : 4) at 9V operation - good equal power across all bands but low efficiency.?
16) Then tried 6 turns bifilar on FT37-43 with a center tap, and this time good equal power and efficiency improved, but still poor
17) Then went to 14 turns bifilar on FT37-43 with center tap, 9V operation: now low power again on 10m band...
18) So - could there be a sweet spot, somewhere between 6 bifilar turns which has good power but low efficiency, and 14 turns which has low power on 10m but good efficiency?elsewhere...? taking 2 bifilar turns off at a time (so as to keep the center tap "center"), unfortunately NOT... 12 turns is still low power on 10m and if you go to 10 turns you get good power but now efficiency is not good enough.?
19) Now a gross mutation transformer idea by EX0AA... which has 2 : 2 turns, bifilar wound; on one end of the transformer you cut one of the wires, and scrape the other one to make it the center tap. At the other end, you join the right pair of wires. So what you end up with is this WEIRD transformer where the wires are all bifilar wound - but the two windings are sort of 180-degrees out with each other. Kind of like imagine a circle, a loop, with the entrance to the loop at the bottom; then a second 1-turn loop, also a full turn, whose entrance is at the top (180-degrees apart). The two loops travel in parallel (in fact twisted bifilar) but their ends are at opposite sides. Surprisingly though I thought it weird, this transformer worked - but it did not solve the low power on 10m problem.?
20) it would be a VERY good idea to try type -61 material but I don't have any here yet to try. If anyone has any on hand and wants to try it, it would be very interesting to hear your results.?
So - we have some closer idea of what the problem is: some kind of parasitic resonance involving the transformer; and various solutions, some of which work some of the time, but nothing which seems to work all the time, and does so with good equal power output and without terrible efficiency.?
As I said - some low power issues are due to how the inductors are constructed and the LPF toroids need tweaking. But there is definitely some kind of parasitic resonance issue here. We have very experienced QDX builders here who have made literally HUNDREDS, and every QDX comes out identical like its predecessor. So we know there isn't a variation in build style.?
Anyone has any magic solutions, do not hesitate to share them!
Meanwhile as some people have commented, even 2W on 10m goes a long way and gives excellent QSOs.?
73 Hans G0UPL
FWIW, I replaced the transformer torroid with a new one form , same torroid (BN43-202) as the original, same 3:3 winding for 9v.
Results are as follows:
?Original kit transformer
20m-4.2w
17m-3.4w
15m-3.5w
12m-2.0w
10m-2.5w
New transformer
20m-4.4w
17m-3.6w
15m-3.3w
12m-2.0w
10m-2.9w
So I?got a slight but not significant power improvement with a new torroid/transformer.? That being said, this past weekend with the original kit I worked 23 POTA stations on 20/17/10m.? So I am happy with the QDX HB despite a slightly low power output on 10/12m.
Brad, AB4BA
|
On Tue, Mar 28, 2023 at 03:21 PM, Hans Summers wrote:
My personal feeling is that the 10m LPF cutoff might be a bit close to 30m which makes it easy to get a little higher attenuation than desired, on the 10m operating frequency; but this is anyway slight and is NOT the cause of the low power problem in general..."
Hans, is "30m" in the above quote a typo and should be 10m? Otherwise it does makes no sense to me. The lack of being able to make after-the-fact corrections to posts here can be annoying. Thanks! --Al
|
Hans,
?
I've been following this topic closely, as I seem to be faced with the same issue in my experiment to try to obtain high power output from your 50W amplifier kit on 10m band.?I have not yet been able to find a solution, but thought perhaps my results so far would add to your information.
?
For the various output transformers I've tried, even though they all put out 24W-26W on 15 m band, their performance on 10m varied widely as below:
?
FT50-43:
- Primary wound first, Secondary wound over Primary: 10.6W
- Primary and Secondary wound simultaneously as a pair'ed wire - side by side: 11.3W
?
FT50-61:
- Primary wound first, Secondary wound over Primary: 14.5W
- Primary and Secondary wound simultaneously as a pair'ed wire - side by side: 4.8W
- Primary wound on bottom half of binocular, Secondary on top half: 2.3W
?
In all cases, 4 cores are stacked, then paralleled to form a binocular core.
?
I hope you and/or someone will come up with a solution. I will keep experimenting, also.
?
73 de Aki, JO4MTH
|
Hi, I have also been playing with this and have tried a type 61 transformer and another attempt at a 43 transformer with a different core with no improvement.
I built a copy of the PA on a separate board (not a PCB - point to point connections) driven by an Si5351 set with inverted outputs and obtained similar results.
?
It is interesting to look at the gate drive voltage to the BS170 on a scope - in my case there is a big dip in the gate voltage so the BS170 are not fully turned on for the cycle.? I used a 350MHz Rigol and a homebrew active probe.? A normal 10x probe also showed the effect.? This dip appears on other bands as well but as the time period of the partial turn off is of lower percentage on the other bands the effect on output is reduced.? This dip is larger with the LPF connected to the output, but still present even when the output is connected directly to a dummy load, bypassing the pin diodes etc.
I tried removing one BS170 on each side so the one gate was driven by both outputs from the logic chip, this improved the gate waveform but the power out was low of course.
Hope this helps
Dave
M0WID
|
Could this be ground bounce?
73, Willie N1JBJ
toggle quoted message
Show quoted text
On Mar 29, 2023, at 12:03 PM, m0wid via groups.io <david.wilde@...> wrote:
Hi, I have also been playing with this and have tried a type 61 transformer and another attempt at a 43 transformer with a different core with no improvement.
I built a copy of the PA on a separate board (not a PCB - point to point connections) driven by an Si5351 set with inverted outputs and obtained similar results.
It is interesting to look at the gate drive voltage to the BS170 on a scope - in my case there is a big dip in the gate voltage so the BS170 are not fully turned on for the cycle. I used a 350MHz Rigol and a homebrew active probe. A normal 10x probe also showed the effect. This dip appears on other bands as well but as the time period of the partial turn off is of lower percentage on the other bands the effect on output is reduced. This dip is larger with the LPF connected to the output, but still present even when the output is connected directly to a dummy load, bypassing the pin diodes etc.
I tried removing one BS170 on each side so the one gate was driven by both outputs from the logic chip, this improved the gate waveform but the power out was low of course.
Hope this helps
Dave
M0WID
|
Just FYI --
I recently built Rev 5. Using a Chinese PS (haven't checked the purity) and a QRPometer, I got the following at 12 volts:
??? 20m ?? 3.24W??? ??? ??? 12m ?? 3.30W
??? 17m ?? 3.24W??? ??? ??? 10m ?? 1.21W
??? 15m ?? 1.50W
I haven't fiddled with any of the toroids, paid no attention to turn spacing.
Paul? AI4EE
|
I went ahead and installed a bifilar wound T1 on my QDX-HB using 24g (0.5mm) wire built as described by me earlier.
On my Oak Hills WM-2 QRP meter I'm reading 4W output on 20m with 8.5V drawing ~977mA. (a caveat, V/A reading from buck voltage regulator DP50V5A)
However, output on 10m is still low at 1.2W. FYI, my 10m LPF coils are L10 with 10t and L4 with 9t. I'll remove one more turn from L10 and try optimizing the LPF.
Still, I like the idea of the bifilar wound output transformer and will see how this one holds up.
--Al
|
Akira's post below got me wondering if building binocular cores out of stacked 43 or 61 cores might offer improvements.
What properties are you looking for in this transformer? There's a large difference in inductance per turns between the two ferrite mixes.
I have to ask, Hans, when your team encounters one of these low-power-on-10m units during final testing what do they do?
Thanks!
--Al?
|
These low power measurements on 10m also begs the question of what the passband looks like for 10m on both the LPF and the wattmeter. Since the LPF on the transmitter is passing all the power between 30MHz? and lower frequencies. I would see what the output spectrum looks like.....and check a "bad one" and a "good one".
73 Kees K5BCQ
|
I have had good results in the past stacking 43 and 61 cores together. yes, it can be done.
73 Kees K5BCQ
|
