Power out and heat in C2C

[Ulf (SM0NOR)]

My first question is: Has everyone here who built with the 3-set board bundle managed to get a full 100W on all band including WARC and 60? I'm asking because I seem to be the only one who struggles to get a more than 70 watt on 10M, 80W on 12M and 98W on 15M (other bands vary between  90 and 110).
After a hiatus of 2 months I got new energy to get the 15-12-12 band to work.
- I have completely rebuilt the amp-board and the control board to rule out any construction mistake.
- Still same result. Bad output on 15-12-10. After a while of key-down and a strange hissing sound from the board the output failed and went down to 50-ish on 15-12-10 bands

I then turned to the filter board and in particular the 15-12-10 filter. I desoldered all capacitors and checked them. And to be 110% sure I replaced them with new ones.
I noticed that the C2C 180pf had broken so that was the reason for the complete failure on these bands.
After the capacitor change and a double check of the toroids, including measuring the impedance with an LCIIB-meter I reconnected the board.
Still the same result! About 70W on 10 meter.
After a while of testing I inspected the board and noted that the 180pf capacitor had turned a bit brownish, but without breaking apart.
I have no fancy instruments so I can't measure the filters, but I can transmit on all frequencies with my HL2 so I "sweeped" the band to see where the output goes down. At about 23.5 MHZ the output goes down.
Interestingly, if I by mistake run 10M through the 6M filter I get full power  out!
All power levels are measured with 1W of drive power. Increasing drive level makes little to no difference.

Any ideas?

[jcve...@gmail.com]

Especially on the high bands (20M and up) you must make sure that the caps are C0G or NPO dielectric. X7R caps even if they have sufficient voltage will burn at high levels of RF. C2 on the amp board is also critical to getting the full power output.

If you are soldering the caps by hand be very careful not to introduce heat stresses in the ceramic material. I recommend putting solder on one of the pads without the cap present. allow the solder to cool a little and apply flux to the pad. Then put the cap near the pad with the solder, melt the solder and the flux and slide the cap onto the pad, never allowing the iron to touch the cap. then apply flux to the other end and solder the other end of the cap again the iron should never touch the cap.

If the cap is making a sound, it's more than likely due to fractures in the ceramic allowing the motion between the electrodes, failure is eminent.

Jim.

[Nick Kennedy]

I wonder what you're using to measure power with as that can be a difficult task. I've used both a homebrew diode probe with Arduino power meter as well as using oscilloscope readings and converting Vpp to power. The methods differed in results by about 10%.

I haven't used the RFPowerTools LPF set but have tried a couple of my own LPFs, on 80 and 40. I have seen up to 100 W output but generally see 90 to 95 W and 80 W on 10 meters.  I increase drive to get the desired power output, but generally back off when I get into that compression region you describe. My drive power is about 1.3 W on almost all bands.

I'll also ask if you have a solid 50 VDC supply and negligible drop in wiring and connections to the amplifier board.

Deviations in your dummy load's resistance from 50 ohms can also introduce power errors.

73

Nick, WA5BDU

[Ulf (SM0NOR)]

Nick
Thank you for your input. I'm aware of the fact the simple power meters are not accurate. So, I don't pay attention to 98 watts or 103 watts. But 70W is no doubt off from what to expect on 10 meter. And in conjunction with a warmup of C2 on the filter board, there is something that is not right here.
And about PS, I'm using the voltage converter suggested on Jim's web-page. I'm aware that it is not ideal. Better would be a real 50V supply. But the voltage is stable at 50 volt regardless of power out and band. I feed it with a 60A switched power supply from Maas.
But you're right about my DL2K dummy load from Palstar. SWR is not perfect on all bands. I'll check it...
73's - Ulf - SM0NOR  

[Ulf (SM0NOR)]

Thank you Jim!
I use parts from the BOM so they are COGs. C2 on the amp-board we have discussed in an earlier thread. I have since then per your recommendation beefed it up to a 1206 size. It seem to hold fine even under prolonged key-down.
Your remark about soldering the caps, I will take seriously. I have not been afraid of heat and I probably use to much solder. I will re-place the caps per your tips. 
Ulf - SM0NOR

[Ulf (SM0NOR)]

It took a while to get a new set of capacitors from Mouser. I changed all caps in the 15-12-10 filter and I was very careful not to heat up the capacitors while soldering

Still, C2 exploded after only a few passes of FT8. Power was 75W with 1W of drive. And when I increased to 1.5 watts of drive, I got a tad over 80 watts for a few seconds before C2 exploded just after the hissing sound.

So no, I have had three incidents with C2; twice it exploded and once it only turned brownish. 
It seem clear to me that I have a significant loss of power in the 15-12-10 line. Question is why this is happening?
What more could I try to find what is wrong?
73's SM0NOR

[Nick Kennedy]

I'll speculate a bit.  You increased drive 50% but output only increased about 8%, so I'd guess the amplifier is being driven to a much more non-linear region and the harmonic content of the output increased quite a bit. That may have been too much for the filter. Normally I wouldn't expect the filter to care but maybe this one is marginal in its ratings.

I haven't looked at the filters much though.  It would be interesting to play with the values in LTspice or some other spice version. Maybe I'll try that when I get a bit of time.

What band were you using?

73

Nick, WA5BDU

[jcve...@gmail.com]

Look at the relays associated with the 15/12/10 filter, if the top of the relay is bulging then the relay might be bad, the top bulges if it gets too hot. Also look at the bypass caps on the relay if they are bad the relay may chatter which would cause some nasty problems.

Jim.

[Ulf (SM0NOR)]

OK. I changed everything associated with the 15-10 circuit; relays, capacitor and FB:s. Still, the C2C 180pF exploded almost immediately. Since I started this project I have replaced everything; the PA-board, the switch-board, and now the components around the filter.
There is only one thing left now that has remained unchanged. And that is the DC-DC converter. I'm using the one suggested on your web. I think that the only remaining thing to do now is to get a proper 50 Volt supply. Perhaps there is some high frequency component either radiated from the DC-DC converter or superimposed on the DC-voltage? 

[jcve...@gmail.com]

Maybe it's the cap, are you always using caps from the same batch? have you tried measuring them on a LCR meter? Maybe they are missed marked.

You could also try going to a 630V or 1kV cap. 

Jim.

[Ulf (SM0NOR)]

The caps are from two different mouser orders. But they could still be of the same batch depending on their inventory.
But I'll put in a couple of 1KV and even 2KV in the next order

[Ulf (SM0NOR)]

The 1KV capacitor seem to hold up fine. No hissing sound. Power output is close to 100W on 15 and 12. On 10 around 80W with 1W drive and 90W with 2W
I think it is time to put the lid on the amp and making QSO:s

[Nick Kennedy]

I took a look at the 15-12-10 filter using LTspice. I set the voltage at 100 V peak to give 100 W. I didn't add any resistance to the L & C elements beyond what the program does. I modeled with sine wave input only, no added harmonic content. The current in C2C displayed as:

BAND           Peak      (RMS)

15                 2.25 A    (1.59 A)

12                2.74 A    (1.94 A)

10                4.37 A    (3.1 A)

Looking at the data sheets for C2C, it's hard to pin down Q or loss at these frequencies. Generally, a dissipation factor of < 0.1% at 1 MHz and 1 V rms was given.

But it seems like that's a lot of current through a 1206 part.  I think for 180 pF I'd go with two 360 pF in parallel. Or, better make that series.  Maybe two 91 pF in parallel.

73

Nick, WA5BDU

[Ulf (SM0NOR)]

This is very interesting. But why would you prefer two in series over two in parallel? The current distributes over two in parallel.

Ulf

[Nick Kennedy]

I was waiting to see if anyone would point out my error before I did it myself. Of course you are correct, there would be no value in a series equivalent.

73

Nick, WA5BDU

[jcve...@gmail.com]

Actually either way, series or parallel, the power dissipated in the caps are cut in half, parallel reduces the current, series reduces the voltage. 

However, given the PCB layout, parallel seems like the way to go.

73,

Jim WA2EUJ