[Amps] Autotuner for Amp Input

Fri Oct 8 10:18:08 PDT 2010

Thanks, Jim W7RY. I did in fact find two used copies on Amazon, but the 
prices were up there with new engineering books. Perhaps the "eBay 
syndrome" for books?

Also looks like I missed Jim VE7RF's comments, somehow. ? Anyway, I was 
only looking for a 'rule of thumb' for adequate flywheel effect, not 
trying to use that to design the network. I use W5FD's pi-net equations. 
I don't know what Ian's spreadsheet uses... I'll pop over there and take 
a look when I get a few minutes.

Thanks for the info, interesting stuff.  In the specific case of my 
YC156 project, things are made a little more complex (but *maybe* easier 
overall) by the inclusion of a switchable 0/3/6db attenuator in the 
input line before the pi-net. It's purpose is ostensibly for drive 
attenuation, but it will be interesting to see how the "buffering" 
affects drive characteristics, IMD etc, if at all.

73,
Jim N7CXI
(too many Jim's around here... ;-)

On 10/8/2010 8:35 AM, Jim Thomson wrote:
> Date: Thu, 7 Oct 2010 17:54:01 -0700
> From: "Jim W7RY"<w7ry at inbox.com>
> Subject: Re: [Amps] Autotuner for Amp Input
>
> You should be able to pick up a 23rd edition on Amazon.com.
>
> 73
> Jim W7RY
>
> ##  I found 38 x mistakes in that book, and am only 1/2 way through it.
> Most of em are typo's and stuff, so be careful. Some of it is silly stuff,
> like calling it C1 and C1, when they meant C1 and C2.    In his red covered
> book, they have C1 and C2 reversed on the tuned input, with C1  being
> closest to the cathode and C2  facing the xcvr.  That caused Jon Ogden
> no end of grief with his 4-1000 tuned input.
>
>
>> Playing with the models shows that designs between 25 and 70 ohms
>> cathode impedance meet the 13pf/meter criteria using a pi-net input with
>> a Q of 4. Above 70 ohms not so much.
>
> ## 13 pf per meter is not even close.  I found the better way to do it is
> to use GM3SEK's  PI spreadsheet.   That works out dead on for tuned
> inputs every time. Beware, it will not accept 2 x equal Z's... like
> 50 ohms in and 50 out, or 70 in and 70 out.  We have to 'fool' the
> software a bit..since it was made for high power pi nets.. and factors
> in stray tube C, and plate choke Z, etc.    If the tube input Z is say 40 ohms,
> you enter  50 ohms for the 'plate load Z'.. and  40 ohms  for the output Z.
> IE; ur trying to match 50 ohms to 40 ohms.  Enter 999999  for the value of the
> plate choke, and enter 0  for all stray C and stray L.    Enter a Q =3.3  for the
> operating Q.    It will  spit out the correct values.   The input Q  will be something like
> 2...and the output Q  will be  1.3    The sum of the input and output Q = total
> network Q.    The C1 cap is based on the input  Q of 2  in this example.
> [same deal with conventional hi-power pi nets in kw amps, these days  we call it
>   a loaded Q of 12... in the old days they called it a loaded Q of 10.   The network
> Q is actually 12..... 10 on the input, and 2 on the output. ]
>
> ##  don't get the Q  too high..esp on the higher bands.  On my 3CX3000A7, input
> Z is aprx 50 ohms.    I tested the tunable pi net by putting bird wattmeters on both
> sides of it..plus heath dummy load on the output.   With 200w  applied to the input,
> I was only getting 160w  out on 20-17-15m. [and 198w on lower bands]   This was after
> tweaking the C1 +C2  caps  for flat swr.  I had to increase the coil taps by just 1/4 turn
> on each of the higher bands  [ and decrease the value of C1 +C2] .  Ok, now I was getting
> 195 watts out on the upper bands.
>
> ##  I even tried a Q of 5+...and the solid 6 ga cu wire that the 4 uh tapped coil was made from,
> would actually start getting warm with 200w cxr applied.   200w=2A into 50 ohms.  With a Q=5,
> you end up with 10A  of circulating RF current.  I verified that by inserting an RF ammeter on input
> end [ low bands].   Point here is, get the Q  too high, and you lose power in the tuned input coil..esp
> with typ 16-22 ga material.  Here I thought I was putting 200w into the cathode.. and was burning up
> 40 w !   After the fix... then it all goes to the cathode.  The result is  more grid current/plate current..
> and way more power out.
>
> ##  I didn't want to mess  with 9 x tuned inputs, and design them to each handle 200-300w... so opted
> for a tapped 4 uh coil  made from 6-8 ga wire  [ or .125" cu tubing].   Coil  consisted of 17 turns  on a
> 1.5" ID.   2 x 2160 pf broadcast variables used.  Each was  4 x sections... and each section is 17-540pf.
> All 4 x sections  simply strapped in parallel.   Each broadcast cap is padded on 160m....with  4 x 500 pf
> doorknobs [HT-50/58 type].  4" ball drives, [with 6:1 redux vernier]  made by jackson bros was used on each
> cap.  They are marked 0-100, in fine increments, across 180 deg of the skirt.   4"  diam = 12" circumference,so
> the 0-100 scale is marked over 6" .    Tune it once for each band, then write down the numbers.
>
> ## Then you are guaranteed a flat input swr, across the entire band..on every band.   For a laugh, we tested
> the tuned input, with the bird wattmeters on both sides, into the dummy load..with a 800 w cxr... and it
> runs just fine, zero heating, all bands.  But the Q can't be too high.  Run it up till the PO starts to drop
> off, then back the Q down a hair.  Ok, now it's dialed in correctly on the high bands.    On the low bands
> you can get away with higher Q, [4-6] but it doesn't buy you much.  It's  fine for a manual tuned input, since
> you can have 2 x settings.. CW  and phone band.   For fixed PI nets, it would be a pita...as swr will rise
> on band edges.
>
> Jim  VE7RF
>
>
>
>
>
>>
>> If I ever find a copy of Orr's 23rd edition at a reasonable price I'll
>> pick it up. I see quotes from it zip past here every now and then.
>>
>> 73,
>> Jim N7CXI
>>
>>
>> On 10/7/2010 11:58 AM, Gary Schafer wrote:
>>> Bill Orr recommends around 13pf per meter for the capacitance at the
>>> cathode
>>> of a GG amplifier.
>>>
>>> 73
>>> Gary  K4FMX
>>>
> _______________________________________________
> Amps mailing list
> Amps at contesting.com
> http://lists.contesting.com/mailman/listinfo/amps
>