That amp was good for more than 1500w on all bands. There are many
commercial and homebrew amps using tetrodes that can do this.
Neutralization isn't relevant to the problem that you are having.
The Q that we have been talking about in this discussion is the the Q of
the tank circuit, not the Q of the coil. This Q is determined by the L/C
ratio of the pi-network and the output impedance of the tube, not the
X/R of the coil. Also it certainly wouldn't take 1m of tubing to make a
1 uh inductor! I think a decimal point is misplaced here!
Let me try to summarize:
The problem in your amp, as far as I can tell from your description, is
that it's hard to get a LOW enough OVERALL Q of the tank circuit because
of the minimum capacitance of your tuning capacitor plus output capacity
plus strays. One solution to this is to add a small coil between the
plate and the input of the pi-network (probably less than 1 uh). This
transforms the impedance of the tube to a lower value that's easier to
match with a practical pi-network.
It's important to keep the leads between the coils, capacitor, and
bandswitch short so that the major part of the inductance is in the
coils and not in the leads. It's possible to build a tank circuit that
will not resonate at all on 10m if you are not careful about this (I've
done it).
If you're using the G3SEK spreadsheet, there is a place to specify the
inductance of the small coil. But you have to adjust it experimentally
for best results.
If your tube needs a parasitic suppressor, you can make the coil a
little bigger and solder a noninductive resistor across a section of it.
This also requires experiment to find the best spot to put the tap.
I'll send you the page you requested. But I think all the information
you need is here.
73,
Victor, 4X6GP
Rehovot, Israel
Formerly K2VCO
CWops no. 5
http://www.qsl.net/k2vco
On 25/11/2024 11:18, flynth@gmail.com wrote:
If you could take a photo I would appreciate it a lot.
Thank you for posting the link to your amp, it is very informative. What
power output to dummy load are you getting on 10m?
After spending lots of time on this,and following with calculations I
mention below I have to call BS on the idea of such an amp with a
traditional PI tank ever delivering 1500W on 10m (with no neutralisation
to cancel anode capacitance).
To picture the difficulty, let's take a 12mm copper tube coil. You need
about 1m to make 1uH coil. Is that the thickest tube you can bend?
Congrats, this coil has a Q of only 7. 1m of such tube has 28 ohms at
30Mhz. So you'd have to offset this by using even less capacitance.
Let's take a resonant circuit at 29MHz that uses such 12mm pipe and a
coil 115mm in diameter. Even if you have at most 15pF total capacitance
(which is impossible because the tube itself in a socket has 27pF), then
you make a coil to resonate at 29MHz (2uH, 115mm diameter, 4 turns, only
1mm between turns, total length 145cm, resistance @30Mhz 39 ohm) you end
up with a Q of 9.3 with NO connecting wires. Is such Q acceptable?
According to articIes I read No.
Now make it more realistic by adding 10cm of very wide (2.4cm which will
add probably 5pF to your stray capacitance). That adds about 4 ohms
turning your Q to 8.5. What if you can't use such wide strip because of
extra capacitance and you use 12mm wide? Your Q goes down to 7.7. All
that in an impossible 15pF total scenario.
Let's take more realistic 35pF, this requires 0.85uH to resonate. Let's
make an very good coil 110mm in dia, (again 12mm tube, 1mm space between
windings) only 2.7 turns. 93cm long, 25 ohms. No leads and we're already
at measly 6.3 Q.
No wonder all these books usually have half a sentence about it. They
probably tried it, it didn't work, and they moved on to other designs.
I have yet to calculate alternative PI tank arrangements.
Victor, , thank you for pointing out the inductance. The problem in my
amp on 10m is indeed low Q. Both high residual capacitance (the tube in
socket is 27pF, the plate cap is 12pF, but this is easily swapped, the
tube can't be) and the low Q inductance are contributing to it. Without
you mentioning it I would not have realised the futility of it.
73,
F
On Mon, 25 Nov 2024, 05:42 Victor Rosenthal, <k2vco.vic@gmail.com
<mailto:k2vco.vic@gmail.com>> wrote:
I don't have an electronic version. Maybe you can get it from Google
books? Maybe you can offer to buy the book on a ham reflector. With
so many of my generation on the way out, these books should be
available. Perhaps a library has one?
I could take a photo of the section in which this is discussed and
send it to you. But there are several whole chapters on amp design
and construction that would be useful.
The Bill Orr book doesn't have as much theory, but has numerous
examples of amps that are instructive.
It doesn't "work in the opposite direction." It does lower the
resonant frequency if other things are unchanged, but if you reduce
the inductance in the tank coil to bring it back to resonance, the
result will be a lower Q which will be more efficient. Your problem
seems to be that there is too much inductance in the leads between
the coil, the band switch, and the tuning capacitor. Therefore you
can't get a low enough total inductance on 10m.
Look at the pictures of my amp at the link I sent. In the earlier
ones I used a single massive coil of tubing. This made the leads to
the band switch so long that it was impossible to tune to 10m. If
you scroll down towards the end, you'll see that I split the coil
into two sections to make it possible to have short leads.
Many amps use a completely separate 10m coil. The coil needs to be
right up against the band switch if possible.
Victor 4X6GP
On Sun, Nov 24, 2024, 23:33 <flynth@gmail.com
<mailto:flynth@gmail.com>> wrote:
Thank you. Do you by any chance have this book in electronic
format? Can you post this chapter/section, perhaps?
Alternatively I offer $30 via PayPal for the entire pdf.
For some reason these older ARRL handbooks are absolute
unobtanium online. Ebooks do not exist (ARRL's webstore
redirects to page not found if you try to buy one). And paper
copies go for many hundreds of bucks locally or take months to
ship from overseas. Absolutely insane.
There is newest edition on amazon (split into 6 ebook volumes
each for $10). I bought the volume with amp info and the whole
section you talk of does not appear.
This fragment doesn't really explain how it is supposed to work
and why it works in opposite direction when I try it. Still,
thank you for mentioning it. Perhaps one day I'll read that
entire chapter and find out the details.
73,
F
On Sun, 24 Nov 2024, 19:53 Victor Rosenthal 4X6GP,
<k2vco.vic@gmail.com <mailto:k2vco.vic@gmail.com>> wrote:
ARRL Handbook, 2010 ed., 17.9.5 (p 17.27):
"A third possibility is the use of an additional inductance
connected in
series between the tube and the tuning capacitor. In
conjunction with
the Cout of the tube, the added inductor acts as an L
network to
transform the impedance at the input of the pi-L network up
to the
2200-ohm load resistance needed by the tube. ... Since the
impedance at
the input of the main pi-L matching network is reduced, the
loaded Q for
the total capacitance actually in the circuit is lower. With
lower Q,
the circulating RF currents are lower, and thus tank losses
are lower."
They continue with calculations using an 0.5 uh inductor.
They show that
the required total capacitance at 29.7 MHz in the example is
67 pf as
opposed to 36 pf without the inductor, which is easier to
realize with a
practical capacitor.
This applies equally to a plain pi (not pi-L) network,
obviously.
73,
Victor, 4X6GP
Rehovot, Israel
Formerly K2VCO
CWops no. 5
http://www.qsl.net/k2vco <http://www.qsl.net/k2vco>
On 24/11/2024 20:08, flynth@gmail.com
<mailto:flynth@gmail.com> wrote:
> Could you let me know where is it discussed, please? (the
ham radio
> handbook by Orr? I've read it multiple times, but I don't
remember that).
> tube and
> I've added a coil in series that was stretched/compressed
between 700nH
> and 1.2uH.same as in SimSmith simulation it only moved
the resonance of
> the entire pi network LOWER. The opposite of where I need
to go.
>
> I start thinking I simply built the whole thing wrong as
there is no
> hope whatsoever of getting it to work on 10m with
connections that are
> 10cm long (one between the blocking cap and the plate
cap, the other
> between the plate cap and the load coil). I have 20~30pF
of extra
> capacitance somewhere. I should've made all these
connections 2-3cm
> long, but i didn't realise it matters at such "low"
frequencies as under
> 30MHz.
>
> The whole thing resonates at 25MHz with no plate cap at
all and with
> load coil on 0.9uH (where it should be for 29Mhz). The
tube and socket
> is 27uF. SimSmith tells me I must have another 25pF
somewhere to get
> this result.
>
> I'll probably try using thinner wire (less capacitance)
as a last idea.
>
>
> On Sun, 24 Nov 2024, 16:57 Steve Thompson via Amps,
<amps@contesting.com <mailto:amps@contesting.com>
> <mailto:amps@contesting.com
<mailto:amps@contesting.com>>> wrote:
>
> You certainly don't want a small series C anywhere at
the tube end of
> the matching, it pushes everything in the wrong
direction and you end up
> with higher loaded Q.
>
> Steve G8GSQ
>
>
> Thank you Steve, yes, it was a silly idea, I simulated it
and it looked
> nothing like it should.
>
> 73,
> F
>
>
> On Sun, 24 Nov 2024, 16:04 Victor Rosenthal,
<k2vco.vic@gmail.com <mailto:k2vco.vic@gmail.com>
> <mailto:k2vco.vic@gmail.com
<mailto:k2vco.vic@gmail.com>>> wrote:
>
> No, the coil goes in series with the input to the pi
network.
> I misspoke when I said it increases the impedance of
the tube. The
> impedance doesn't change; you are just converting the
pi network
> into one that can match the impedance of the tube to
50 ohms with a
> smaller tuning capacitor. You can think of the coil
as canceling
> some of the tube's output capacity if you wish.
> This is a well-known method of dealing with the
problem of excessive
> Q on the higher bands. It's discussed in the
handbooks and I've used
> it myself with good results.
>
> Victor 4X6GP
>
> On Sun, Nov 24, 2024, 16:48 <flynth@gmail.com
<mailto:flynth@gmail.com>
> <mailto:flynth@gmail.com <mailto:flynth@gmail.com>>>
wrote:
>
> I do understand the coil goes on the input of the
pi tank(on the
> tube side), but the series coil increases the
impedance of the
> tube as source (as seen by the pi tank). So it
goes in the
> opposite direction I need I believe.
>
> Did you mean the extra coil is connected in
parallel? This I can
> understand. This will effectively lower the tube
output
> impedance and should allow the use of higher
capacitance.
>
> I'm just trying to simulate this with NecSIM
right now. It seems
> like it might work.
>
> Is this what you were proposing?
>
> On Sun, 24 Nov 2024, 15:39 Victor Rosenthal,
> <k2vco.vic@gmail.com <mailto:k2vco.vic@gmail.com>
<mailto:k2vco.vic@gmail.com <mailto:k2vco.vic@gmail.com>>>
wrote:
>
> I'm suggesting an L-pi not a pi-L. A pi-L is
useful to
> reduce needed output capacitor size and to
reduce harmonics,
> particularly on lower frequencies. A pi-L has
an additional
> inductor at the output of the network; I'm
suggesting one at
> the input.
> I think I recall a place in the spreadsheet
for this. But I
> found it hard to do this analytically. You
can get it into
> the ballpark, but then you have to adjust the
inductance
> experimentally.
>
> Victor 4X6GP
>
> On Sun, Nov 24, 2024, 14:56 <flynth@gmail.com
<mailto:flynth@gmail.com>
> <mailto:flynth@gmail.com
<mailto:flynth@gmail.com>>> wrote:
>
> I'm not sure I'm using this spreadsheet
correctly, but
> the Pi-L values for the plate capacitance
are lower than
> normal Pi values. I'm attaching the
spreadsheet, but
> also pasting a screenshot using default
settings as I'm
> not sure if such attachments are allowed
here.
>
> On default settings the pi-L circuit also
shows much
> lower capacitance for the Pi-L network:
>
>
> 73,
> F
>
> On Sun, 24 Nov 2024, 12:07 ,
<flynth@gmail.com <mailto:flynth@gmail.com>
> <mailto:flynth@gmail.com
<mailto:flynth@gmail.com>>> wrote:
>
> Thank you for your reply. I'll
calculate this. I was
> considering adding a small (12pF)
series capacitor
> that would be shorted on all other
bands than 10m.
>
> Are there some hidden dangers with
this idea? It
> seems it would allow me to work on
10m with minimal
> modifications.
>
> 73,
> F
>
> On Sat, 23 Nov 2024, 15:32 Victor
Rosenthal 4X6GP,
> <k2vco.vic@gmail.com
<mailto:k2vco.vic@gmail.com> <mailto:k2vco.vic@gmail.com
<mailto:k2vco.vic@gmail.com>>>
> wrote:
>
> The Q is too HIGH, not low. 15 pf
minimum
> capacity on the vacuum cap is
> high, when you add in the output
capacitance of
> the tube and the strays.
>
> I suggest you think about adding
a small
> inductance (1 uh or less,
> usually) between the plate and
the pi net input,
> converting it to an
> L-pi network. This is discussed
in the ARRL and
> Bill Orr handbooks. It
> is also in the G3SED spreadsheet.
That will
> increase the output
> impedance of the tube so as to
make it easier to
> match with a practical
> pi network.
>
> If you do this, be sure to check
carefully for
> VHF parasitics, since it
> can introduce instability.
>
> 73,
> Victor, 4X6GP
> Rehovot, Israel
> Formerly K2VCO
> CWops no. 5
> http://www.qsl.net/k2vco <http://www.qsl.net/k2vco>
<http://www.qsl.net/k2vco <http://www.qsl.net/k2vco>>
>
> On 23/11/2024 15:57,
flynth@gmail.com <mailto:flynth@gmail.com>
> <mailto:flynth@gmail.com
<mailto:flynth@gmail.com>> wrote:
> > Hi,
> >
> > I've built a GU-43B based amp
roughly
> following Pa0fri design (shown here
> >
>
https://pa0fri.home.xs4all.nl/Lineairs/Frinear1500/FRI1500eng.htm
<https://pa0fri.home.xs4all.nl/Lineairs/Frinear1500/FRI1500eng.htm>
<https://pa0fri.home.xs4all.nl/Lineairs/Frinear1500/FRI1500eng.htm
<https://pa0fri.home.xs4all.nl/Lineairs/Frinear1500/FRI1500eng.htm>>)
> >
> > My modifications are: better
PSU for the
> screen(thanks to this group) , no
> > 160m band therefore no toroid
switched in
> series with the Pi tank coil. And
> > instead of a 1500pf variable
load cap I have
> 20~600pf variable plus a
> > selection of extra
capacitances on a switch.
> My plate cap is a 15~500pF
> > vacuum Jennings. Connections
between
> components are made with 0.2mm x 12mm
> > wide coil strip and where not
possible (coil
> taps to band switch) 3mm
> > copper wire is used.
> >
> > The coil is wound with 6mm
coper pipe and has
> two diameters. First is
> > approximately 40mm, has only 6
or so turns
> stretched to 80mm, then 12 turns
> > on 90mm (if I remember
correctly) quite close
> to eachother (2~3mm apart).
> > I've used my NanoVNA to set up
coil taps for
> the bands. The entire coil
> > measures 9uF at 100kHz and
works great at
> 80m. Anode voltage is 3200V
> > falling to 3050V under load.
> >
> > Here is the problem. I suspect
due to stray
> inductances of 3mm wire used to
> > connect coil taps the first
tap (under 1 uH
> if u remember correctly) is
> > only half a turn from coil
start. I didn't
> think this will cause a problem,
> > but I'm seeing this:
> >
> > On all bands up to and
including 20m if I
> increase my drive power to about
> > 10W (CW) the amplifier
consumes near 0.45A of
> current (measured with a
> > normal amp meter and a panel
meter). It puts
> out somewhere in the region of
> > 1000W. If I increase the drive
slightly it
> goes up to 0.6A and power out is
> > 1500W.
> >
> > The screen current is zero
until drive power
> reaches about 10W then it goes
> > negative to go back to zero at
about 13W. If
> I increased it more it would
> > increase rapidly and activate
the protection.
> This is on all bands up to
> > 20m.
> >
> > Today I tried 10m for the very
first time. I
> tuned normally (peaking power
> > with plate cap at very low
drive, then
> increase power until I see screen
> > current move or it gets to
target anode
> current and set the load cap just
> > below the peak power, same
place screen
> current is a little bit positive).
> >
> > But, I increase the drive, the
amplifier
> consumes the required current, but
> > output power is very low. Only
about 200W at
> 0.6A. Tuning behaves normally
> > as well as screen current. I
tried to see if
> something is heating up with a
> > thermal camera, but nothing is.
> >
> > I suspect the problem is
somewhere in my PI
> circuit. Perhaps the Q factor
> > is too low?
> >
> > Can someone, please give me
some tips where
> to look in troubleshooting
> > this? Is there some way I can
verify this
> issues as existing/resolved with
> > a nanovna?
> >
> > Also, my drive power and input
match is fine
> as my driving rig has a built
> > in ATU. I see a confirmation
of the low
> output power on my station monitor
> > (oscilloscope like device).
> >
> > Many thanks,
> > F
> >
_______________________________________________
> > Amps mailing list
> > Amps@contesting.com
<mailto:Amps@contesting.com> <mailto:Amps@contesting.com
<mailto:Amps@contesting.com>>
> >
> http://lists.contesting.com/mailman/listinfo/amps
<http://lists.contesting.com/mailman/listinfo/amps>
<http://lists.contesting.com/mailman/listinfo/amps
<http://lists.contesting.com/mailman/listinfo/amps>>
>
>
_______________________________________________
> Amps mailing list
> Amps@contesting.com <mailto:Amps@contesting.com>
<mailto:Amps@contesting.com <mailto:Amps@contesting.com>>
> http://lists.contesting.com/mailman/listinfo/amps
<http://lists.contesting.com/mailman/listinfo/amps>
<http://lists.contesting.com/mailman/listinfo/amps
<http://lists.contesting.com/mailman/listinfo/amps>>
>
_______________________________________________
Amps mailing list
Amps@contesting.com
http://lists.contesting.com/mailman/listinfo/amps
|