Topics

QCX 30m #2911 low output power #qcx #lpf

Steve
 

I also have low output on 30m. I'm getting about 2 watts out with 12v, and this is after playing with the windings on L1, L2 and L3. I was getting 1.5w before. I think I removed 2 turns on L3 that gave me the extra half watt.

 

Some people have reported some success by changing LPF capacitors, but I've not tried that. I'm okay with the 2w, but my 40m QCX is putting out almost twice that with no adjustments.

 

Steve, N4IAG

Greg
 

Hi Adrian,

I too have low Pout- about 1W- and started to do some troubleshooting.  I did find my rx problem... a resistor and capacitor I somehow got in the wrong places in the 200 Hz filter area.  I'll get back to the tx problem soon.

I'd say your insertion loss is too high, for one thing.  I assume you zeroed your spectrum analyzer and tracking generator.  Attached png (hope you can see that format) shows my 30m QCX LPF response to 100 MHz.  Noise floor of the specan is down about -90 dB, so flat noise floor at -50 dB is probably leakage on the QCX board.  I have another LPF image similar in span to yours, but not on this PC.

I've also assembled the Ultimate Relay-Switched LPF Kit and found several performance issues at the corner frequencies- 60 & 40 m were flat and sharp, 30 & 20 m were sloping to the corner with several dB of IL, and 15 & 10 m were sloping more than 60 & 40 but not as bad as 30 & 10 m.  All capacitors were in the right spots and measured close to nominal, and all toroids had the specified number of turns.  So perhaps the QC of the toroid material is poor.

73,
Greg  WA7IRW

Adrian Scripcă
 

Hi Steve, Greg,

@Steve: that's pretty much the same behavior I am getting, perhaps even lower power. From the pictures I first attached to the email, I get some 20.4V peak to peak signal at output, which mean ~1.05W into 50Ohm load. In order to define the LPF as culprit I would still need to know what signal amplitude do people usually see on FET drains.

I will attempt to remove some turns from all LPF inductors to see whether that improves anything.

@Greg: unfortunately I cannot see the picture you have attached to the email; it seems like a failed draft attachment on groups.io website. Perhaps you can re-post it?
Yes, I also considered that ~2.3dB of IL is too high for the LPF, especially when looking at Han's website for LPF response images; there the passband seems flat with an IL of something lower than 0.5dB. It's still a mistery to me what causes that high insertion loss; it might be a mistake I did or some slipped design detail, which is why I would appreciate others to provide info regarding the numbers they see on 30 meter versions.

I am keen on finding out the cause of the reduced power in the hope that it will help other builders as well.

Thank you es 73!
Adrian, YO6SSW



On Wed, Feb 28, 2018 at 11:37 AM, Greg <groush-reno@...> wrote:
Hi Adrian,

I too have low Pout- about 1W- and started to do some troubleshooting.  I did find my rx problem... a resistor and capacitor I somehow got in the wrong places in the 200 Hz filter area.  I'll get back to the tx problem soon.

I'd say your insertion loss is too high, for one thing.  I assume you zeroed your spectrum analyzer and tracking generator.  Attached png (hope you can see that format) shows my 30m QCX LPF response to 100 MHz.  Noise floor of the specan is down about -90 dB, so flat noise floor at -50 dB is probably leakage on the QCX board.  I have another LPF image similar in span to yours, but not on this PC.

I've also assembled the Ultimate Relay-Switched LPF Kit and found several performance issues at the corner frequencies- 60 & 40 m were flat and sharp, 30 & 20 m were sloping to the corner with several dB of IL, and 15 & 10 m were sloping more than 60 & 40 but not as bad as 30 & 10 m.  All capacitors were in the right spots and measured close to nominal, and all toroids had the specified number of turns.  So perhaps the QC of the toroid material is poor.

73,
Greg  WA7IRW


Ryuji Suzuki AB1WX
 

Adrian,

When you see the waveform at the drain, can you do DC couple and see if the swing reaches down to zero volts when the FETs are conductive? If not, the drive is not hard enough, and the drain RF impedance is higher than 50ohm, but the LPF input is assuming 50ohm.

With a class-E amp with parallel resonance load, the drain impedance is about 50 ohm when driven by 12V and RF output of 4W. If the output power and impedance is inversely related, and the impedance and drive voltage squared are proportional. Also, class E amp requires the drain tank circuit to be slightly inductive at the operating frequency (or resonant frequency a bit above the operating frequency).

Greg
 

Hi Adrian,

Have you checked the QCX troubleshooting guide at  qrp-labs.com/qcx/qcxtrouble.html#ic3  ?  The image shows a 40Vpk swing similar to what you show.  It looks like the same image as in the assembly manual.

Not sure what happened to the first filter response image.  I see it OK up above.  I used the Add Pictures button, but with a png format.  This try is with a jpg...   The darn s/w sticks it full size right in the message, not as a thumbnail.  Must not be doing it right.  Help is no help.

73,
Greg

Adrian G4AZS
 

Hi Adrian,

I have just completed a 30m QCX, and I have the same problem as you with low RF output. I have also improved it slightly by taking a couple of turns off L1, 2 and 3, but still only 1W O/P.

I have looked at the waveform on the mosfet drains, and it appears correct according to the manual. It seems that the problem is still in the low pass filter, and I'm wondering if you have made any more progress in diagnosing it?

This seems to be a common problem, and I think others will also be interested.....

Everything else is working on my QCX, and I have managed one QSO!

Thank you and 73,
Adrian G4AZS

Steve
 

On Wed, Mar 28, 2018 at 05:11 am, Adrian wrote:
I have looked at the waveform on the mosfet drains, and it appears correct according to the manual. It seems that the problem is still in the low pass filter, and I'm wondering if you have made any more progress in diagnosing it?

Adrian YO6SSW mentioned in another thread that replacing C25 & C26 with polystyrene caps helped him. I also did this on my 30m QCX (after reducing turns) and my power went from 2W to 3W on 12v, for a nice 50% increase. Obviously there is some problem with the supplied caps, at least on 30 meters.

Steve N4IAG

Colin Weaving
 

For what it’s worth, let me report my experience with low power output on 20M QCX.

My tests have all been with a 12V good quality current limited power supply into a 50 ohm load. My power measurements are with a homebrew QRP Power Meter that has been checked / calibrated against Vpp readings on a Tektronix scope and the output from my K3 10W rig.

1 Built as per Manual with components supplied. Pout = 0.9W,

.2. I then reduced the turns on L1,L2 and L3 one turn at a time from 16 to 14 turns and found the best Pout to be with L1, L2 and L3 all at 15 turns. (there may be a case for reducing L3 to 14 turns !). Pout now 1.1W

3. I then replaced the supplied 390pf ceramic capacitors with High Quality ? Kemet
ceramic capacitors. Pout now 1.75W

4. The Kemet capacitors had measured about 370pf so I added 18pf ceramics in parallel to approximate 390pf. Pout now 1.85W

5. The Kemet capacitors were replaced with 390pf silver mica. Pout now 2.1W

At this point Rx current was 0.11A and Tx had gradually increased to 0.38A as output had increased.

Probably, I should revisit the L1,L2 and L3 turns again to see if that can be further tweaked with what appears to be better capacitors. The PC Board has stood up well to quite extensive rework, but I fear that I must be close to removing pads etc. Therefore I will wait for feedback and/or Hans recommendations.

My voltage reading at the BS170 output is 38Vpp and waveform similar to Manual. After L3 it is still 38Vpp and after L2 40Vpp (?). Then after L1 it drops to 30Vpp. It seems a large drop if these readings are real.

That’s my experience. I am not too unhappy as 2W is about what I use /P.

Comments welcome

Colin M3WCK





rig

Alan G3XAQ
 

Colin,

With a 12v PSU (and thick leads) I see on my 20m QCX:

40v peak at the drains of the PA FETs (and it goes to zero when they are turned on)
57.5v pp after the first LPF coil L3
44v pp after the second coil L2
35.8v pp across the 50 ohm load (= 3.2W)

So yes, I think you need to tweak the coils. It would be good if you invested in an LC meter so you are not adjusting the coils in the dark. I found L3, the first LPF coil, had the biggest effect on power output (I noted this in a previous message). I can easily get over 4W output if I turn up the supply voltage but I'm not keen on thrashing the PA FETs.

73, Alan G3XAQ

On Wed, Mar 28, 2018 at 07:15 am, Colin Weaving wrote:


For what it’s worth, let me report my experience with low power output on
20M QCX.

My tests have all been with a 12V good quality current limited power supply
into a 50 ohm load. My power measurements are with a homebrew QRP Power Meter
that has been checked / calibrated against Vpp readings on a Tektronix scope
and the output from my K3 10W rig.

1 Built as per Manual with components supplied. Pout = 0.9W,

.2. I then reduced the turns on L1,L2 and L3 one turn at a time from 16 to 14
turns and found the best Pout to be with L1, L2 and L3 all at 15 turns. (there
may be a case for reducing L3 to 14 turns !). Pout now 1.1W

3. I then replaced the supplied 390pf ceramic capacitors with High Quality ?
Kemet
ceramic capacitors. Pout now 1.75W

4. The Kemet capacitors had measured about 370pf so I added 18pf ceramics in
parallel to approximate 390pf. Pout now 1.85W

5. The Kemet capacitors were replaced with 390pf silver mica. Pout now
2.1W

At this point Rx current was 0.11A and Tx had gradually increased to 0.38A as
output had increased.

Probably, I should revisit the L1,L2 and L3 turns again to see if that can be
further tweaked with what appears to be better capacitors. The PC Board has
stood up well to quite extensive rework, but I fear that I must be close to
removing pads etc. Therefore I will wait for feedback and/or Hans
recommendations.

My voltage reading at the BS170 output is 38Vpp and waveform similar to
Manual. After L3 it is still 38Vpp and after L2 40Vpp (?). Then after L1 it
drops to 30Vpp. It seems a large drop if these readings are real.

That’s my experience. I am not too unhappy as 2W is about what I use /P.

Comments welcome

Colin M3WCK





rig

g4sra
 

On 28/03/18 13:31, Steve via Groups.Io wrote:
On Wed, Mar 28, 2018 at 05:11 am, Adrian wrote:
Obviously there is some problem with the supplied caps, at least on 30 meters.
I disagree, it is not obvious at all.
Have you measured how much of that extra 1W of power is actually
contributed by the wanted frequency ?

Please post 'before' and 'after' spectrum analyser sweep images ;)


Steve N4IAG
When I see concrete proof, I will consider changing mine.

73 Charlie

Steve
 

On Wed, Mar 28, 2018 at 08:18 am, g4sra wrote:
I disagree, it is not obvious at all.
Have you measured how much of that extra 1W of power is actually
contributed by the wanted frequency ?

When I see 2 watts on my wattmeter, then change the caps to a different type (same value) and the meter now reads 3 watts out, that’s all the proof I need. I don’t have a lot of test equipment but this makes sense to me. Sorry that you don’t agree.

 

Steve N4IAG

Colin Weaving
 

Alan

Thank you for informative reply. The Vpp readings from L3 to L1 are particularly useful. It gives me something to aim at, and the significant difference between our readings after L3 suggests that toroid is the place to start. One of the tests I did suggested 14 turns on L3 rather than 15, but I could not repeat it.

Yes, I have a capacitance/inductance meter. I have some concern about accuracy but 15 turns on the toroid gave me close to 77uH for L1 and L3, so it’s in the ballpark. L2 should be 90uH (16 turns with my meter) but my earlier work showed 15 turns to be better!! This needs another look with the new capacitors.

Good point about any voltage drop in the power supply cables. They are reasonably robust and under Tx conditions about 50mV drop to the QCX on board terminal. Not enough to worry about.

Last point is that I attempted to use an offboard low pass filter that could be experimented with to save wear and tear on the QCX board. I disconnected C29 from L3 and connected the output from C29 to the output BNC connector, obviously disconnecting L1 from the BNC first. My first and only attempt had similar low power output, but significantly higher current drain ( 0.7V against 0.35V). The low pass filter used different size toroids but with appropriate turns to match inductance. Discouraging , so I have not pursued it further. Any thoughts?

Thank You Colin M3WCK

On 28 Mar 2018, at 15:50, g3xaq <alan@...> wrote:

-