Hi Paul, 

>Maybe 22 turns is too much so I made it 18 turns. No change. 16... 
>14... 12... 10 turns. No change. Is the core too small? I could try 
>stacking a few of these.

22 turns on one FT114-J core is more than adequate.
It gives you an XL of - j1.291k which in turn means an
attenuation of around 22dB (referenced to 50 Ohms)
for common mode currents.
So shielding is your last resort. :-)

>I still have to try building a shielded box for the PA and LPF. This 
>will take some days.

> Maybe it also needs shielding between input and 
>output circuits of the PA.

>From my experience on MW I would assume this will not be necessary
for a PA with a gain like yours (20-22dB).

73
Clemens
DL4RAJ 



>-----Original Message-----
>From: [email protected] 
>[mailto:[email protected]] On Behalf Of N1BUG
>Sent: Friday, December 29, 2017 2:19 PM
>To: [email protected]
>Subject: Re: R: Re: LF: TXing 2200m WSPR
>
>Hi Clemens,
>
>OK, I understand about capacitive coupling in transformers. So I 
>tried a simple common mode choke. At first 22 turns using RG-178 
>coax on a FT-114-JC (yes, made a typo last time). No change for 
>better or worse with the choke.
>
>Maybe 22 turns is too much so I made it 18 turns. No change. 16... 
>14... 12... 10 turns. No change. Is the core too small? I could try 
>stacking a few of these.
>
>I still have to try building a shielded box for the PA and LPF. This 
>will take some days. Maybe it also needs shielding between input and 
>output circuits of the PA.
>
>BTW I forgot to mention during the night I also tried a CM choke on 
>the DC power feed (16 turns, FT-114-JC). No change at all.
>
>If I knew for sure that a different (properly built, proven design 
>class D) PA would work here, I would not put any more effort into 
>this PA. The trouble is I can't know that another PA would be OK.
>
>Note: RG-178 on a coated core is very slippery! But I was able to 
>keep the turns tight with some tape!
>
>Paul N1BUG
>
>
>
>On 12/29/2017 07:08 AM, Clemens Paul wrote:
>> Hi Paul,
>> 
>>> I can try common mode chokes instead of isolation transformers, but
>>> will this really be better? Seems if the problem is common mode
>>> current the iso xfmrs should kill it...??
>> 
>> Basically yes.
>> But we don't know how much capacitively coupling there is 
>between primary and secondary
>> which might partially spoil the RF isolation.
>> I would just give it a try:
>> One chocke at the PA output is no big deal.
>> 
>>> I still have a couple of ideas to try: Replace 50uH choke in PA with
>>> different type core.
>> 
>> In Spice there's only a ca. +/- 0.5Vp voltage swing across the choke.
>> Hence the flux is less than 10% (3.7mT) of the max. flux for 
>this core.
>> So no problem with that choke.
>> 
>>> Use completely separate bias supply.
>> 
>> In Spice it makes no difference.
>> 
>> 
>> 73
>> Clemens
>> DL4RAJ
>> 
>> 
>> 
>>> -----Original Message-----
>>> From: [email protected]
>>> [mailto:[email protected]] On Behalf Of N1BUG
>>> Sent: Friday, December 29, 2017 11:40 AM
>>> To: [email protected]
>>> Subject: Re: R: Re: LF: TXing 2200m WSPR
>>>
>>> During the night I made a few more tests...
>>>
>>> Iso xfmr at PA output. 14T:14T on FT-114-JC core. Made the problem
>>> slightly WORSE.
>>>
>>> Iso xfmr at amp input. 4T:4T on 3 x BN73-202. Made the problem
>>> slightly WORSE.
>>>
>>> Removed iso xfmr from amp output, left iso on PA input. Slightly
>>> better than without iso anywhere, but still bad.
>>>
>>> Tried again all configurations of iso xfmrs (at antenna, at PA
>>> output, at PA input) using HP 3325B to drive the PA. No difference
>>>from using the Ultimate 3S.
>>>
>>> When I use an iso xfmr at the antenna the problem is WORSE than when
>>> I have the coax connected to the antenna ground. It's not a big
>>> change, but measurable and repeatable.
>>>
>>> Considered this might be an instrumentation problem with RF
>>> affecting the HP 1980B oscilloscope, but I can see the signal
>>> quality degrade in a receiver when the problem occurs so I don't
>>> think so.
>>>
>>> I can try common mode chokes instead of isolation transformers, but
>>> will this really be better? Seems if the problem is common mode
>>> current the iso xfmrs should kill it...??
>>>
>>> I still have a couple of ideas to try: Replace 50uH choke in PA with
>>> different type core. Use completely separate bias supply. However
>>> since the PA works perfect into a dummy load I am not sure the PA is
>>> the problem.
>>>
>>> I will try to make a shielded box for the PA. That will 
>take some time.
>>>
>>> The Ultimate 3S is in the metal case sold for it. This is not great
>>> because of the paint (or powder coat) at mating surfaces. I will try
>>> to work on that but since using the HP 3325B as a PA driver does not
>>> make any change I have doubts this will help.
>>>
>>> Paul N1BUG
>
>
>---
>Diese E-Mail wurde von AVG auf Viren geprüft.
>http://www.avg.com
>