Will, Gerald, Peter and all,
This has been a very interesting discussion. Just like you I see
constants published running from 1.5 to 2.0 and I don't have any solid
reason to use one over the other except to say that some well respected
people say to use this one or that one for tetrodes or triodes or that
most use 1.6 for 8877's etc.
There are several interactive calculators on line that one can use to
ball park your pi or pi-L tank circuit. For me, all of them fall short
with one thing or another missing. Some require you to already know
your Rp. Others Don't tell you the Q they design for. Most don't tell
you the constant (K) they use. None of them give you anything more than
figures for a single frequency.
I have searched the web until I ran out of Google hits and wasn't
completely satisfied with anything I found. I ran across a piece of work
done by Phil Turcotte, VE3OZZ, for pi network design. It isn't well
published. In fact I only found it on one link. The interesting thing
about his spread sheet was that he showed you all the math which would
enable you to do the same thing. I used Phil's work as a foundation for
an expanded spreadsheet in which I included six different sheets to give
the curious user a great deal of flexibility to "what if" a lot. Each of
the sheets gives you the option to change different things and get
different outputs. One thing I wanted and managed to include was that
each sheet does output data for each band, 160 through 10 meters, with
three outputs on 80/75 meters. On each sheet you can specify the
frequency you want to use for calculations. Every sheet gives you the
option to specify your own K, Q, and efficiency. They also compute the
AC line current equivalent to the plate input power. And, each sheet
will print on one page with all the info (on my little HP Inkjet).
The worksheets are arranged such that the primary inputs for the first
one are Pout and Ip (Pout-Ep), for the second are Ep and Ip (Ep-Ip), the
third is a series based on a fixed Ep and a changing Ip, ten 100 mA
steps, (Ip series), the forth is a series based on a fixed Ip and a
changing Ep, ten 100 Volt steps, (Ep series), the fifth gives you the
option to set Ep and Ip and then increases the Ep 100 Volts in ten steps
and adjusts the Ip to maintain the same Rp (Rp-Ep series) (ever wonder
what voltage/current combinations you could use to keep the same Rp, C1,
C2 and L?), and the sixth worksheet does the same thing but steps the Ip
100 mA for ten steps (Rp-IP series).
I don't claim any credit for coming up with the math. That's Phil's and
it appears it works well. All I have done is assemble it in a fashion
that might be a little more useful if you like doing a lot of that good
old "what if" like I do.
If you want to get a copy and play with it, it's here:
<http://www.geocities.com/mygs35b/pi-net.html>
It's done in Microsoft Excel.
If you have any suggestions to improve it or find something wrong with
it I'd like to hear from you. If you hate it, I'd like to hear that too.
I don't write software; I just needed a tool; so there are no fancy
graphics and I'm not headed there... ;)
73, Tony W4ZT
Will Matney wrote:
> Gerald,
>
> You got it. That's the reason why my friend Frits in the Netherlands, who I
> have corresponded with several times, came up with 1.87 from experimenting.
> That doesn't mean he is correct, just that he's comfortable with using it and
> likes the results. The reason I use 1.8 was because it was published. I have
> been speaking with another ham who writes software. We were speaking about
> this very thing and why I told him to use 1.8. If you are writing software
> and don't use a published number, you have to many starting to point fingers,
> and saying the software is wrong. By using 1.8, you can point to published
> material. I hate to say it, even if it's right or wrong. The 1.8 factor
> always did make me wonder just who actually came up with it. Now I've read
> back in time from 1991 in the ARRL handbooks to see if that was mentioned,
> and I didn't see it. I did publish what I did see which was different. If I
> recall, later than around 1980, the ARRL Handbooks Pi formulas were found to
> be
wr
> ong and were supposed to have been changed afterwards. Anyhow, after I've
> did this tad bit of research, I'm starting to take a lot of things published
> with a grain of salt. Using the tubes curves is really the best way to get
> close. The formulas were meant to just get you there in my opinion.
>
> Best,
>
> Will
>
> *********** REPLY SEPARATOR ***********
>
> On 4/12/06 at 1:33 PM TexasRF@aol.com wrote:
>
>> Will and all,
>> as you have determined, the k=1.8 is an approximation that will keep you
>> inside the ballpark at least.
>>
>> Each tube type has it's own characteristics that will cause a shift up or
>> down from k=1.8. For example, if you examine the 8877 data sheet constant
>> current curves: You will see a sharp upturn in the grid current when the
>> plate
>> potential is about 600vdc. At this point, the plate voltage swing will run
>>
>> approximately plate voltage under load minus 600vdc. If we are using 4000v
>> plate
>> voltage, the swing is then 3400vdc.
>>
>> Using the earlier RCA info from this week, and assuming 1A average plate
>> current the plate load impedance would be (3400 X 2) /3 or 2266.7 ohms.
>> Using
>> the K factor method, the required K would have to be 1.76 for the same
>> impedance.
>>
>> If we decide to use 2500v plate voltage, still at 1A plate current, the
>> plate voltage swing would be 2500 - 600 = 1900vdc. The resulting plate
>> load
>> impedance in this case then would be (1900 X 2) / 3 =1266.7 ohms. To reach
>> this
>> value with the K factor method, the required K would have to be 1.97.
>>
>> If you use a tetrode, the limiting factor for minimum plate voltage is
>> screen grid current. The knee of the curve for this is very close to the
>> screen
>> voltage used. If you raise the screen voltage from a typical voltage to
>> the
>> maximum allowed, the plate voltage swing is reduced. With a lower plate
>> voltage
>> swing, a lower plate load impedance is needed. This implies that the K
>> factor
>> would need to be higher to reach the required impedance. Of coarse with
>> higher screen voltage comes higher plate current so an additional
>> reduction in
>> plate load impedance is needed and another, even different, K factor.
>>
>> So, you can see, K is very much an approximation. Using the actual tube
>> curves would seem much more precise. Having said that, the actual
>> difference in
>> loaded Q by using k=1.8 vs 1.97 is less than one.
>>
>> 73,
>> Gerald K5GW
>>
>>
>>
>>
>>
>> In a message dated 4/12/2006 11:03:29 A.M. Central Daylight Time,
>> craxd1@verizon.net writes:
>>
>> Jim,
>>
>> I read this over again last night, and it didn't mention anything about
>> triodes only. The main reason I posted this was to show the differences,
>> plus
>> find where the illusive 1.8 came from in print. So far, the only book
>> I've seen
>> 1.8 listed in was Bill Orrs Handbook. I don't have any newer ARRL
>> handbooks
>> past the 90's as they seemed to be the same old thing, over and over,
>> with not
>> that much new.
>>
>> Here's the thing. I use 1.8 just like evryone else because it does get
>> you
>> there. I calculate the plate current though the same way it was shown in
>> this
>> volume of the ARRL Handbook, by efficiency. I seen in the old RCA
>> Radiotron
>> handbook where it said the plate current for class AB could be as much as
>> 3
>> times. How they come up with this, I don't know as efficiency is
>> efficiency.
>> Everything I've read says AB is around 60% efficient, not less. To my
>> opinion,
>> there's a lot of mis-information out there as compared to what I've seen
>> work
>> in the real world. I would like to find out though where the factor of
>> 1.8
>> was first mentioned.
>>
>> Best,
>>
>> Will
>>
>>
>> *********** REPLY SEPARATOR ***********
>>
>> On 4/11/06 at 9:04 PM jkearman@att.net wrote:
>>
>>> From: "Will Matney" <craxd1@verizon.net>
>>>> Class AB, K = 1.5
>>> My understanding is that this value of K applies to triodes, where Ep can
>>> swing nearly to zero. For tetrodes, Ep cannot swing below the screen
>>> voltage. IIRC, this has the effect of increasing K.
>>>
>>> It's useful to consider the consequences of slight errors in
>>> component-value selection. Assuming your variable controls (plate tuning
>>> and loading) have enough range to get a close match, the negative
>>> consequence would be a Q different from what you calculated. But if you
>>> give yourself enough range in tuning and loading Cs, you should be able
>> to
>>> tune for _best linearity_ (more important than best efficiency) and still
>>> get enough Q to reduce harmonics below FCC requirements.
>>>
>>> If you calculate a range of plate loads by varying K from 1.5 - 1.8, and
>>> then calculate tank circuit values based on a Q range of 12-15, you
>> should
>>> come up with tuning and loading cap values that will do the job.
>>>
>>> 73,
>>>
>>> Jim, KR1S
>>> http://kr1s.kearman.com/
>>> _______________________________________________
>>> Amps mailing list
>>> Amps@contesting.com
>>> http://lists.contesting.com/mailman/listinfo/amps
>>
>>
>> _______________________________________________
>> Amps mailing list
>> Amps@contesting.com
>> http://lists.contesting.com/mailman/listinfo/amps
>>
>>
>>
>>
>> _______________________________________________
>> Amps mailing list
>> Amps@contesting.com
>> http://lists.contesting.com/mailman/listinfo/amps
>
>
>
> _______________________________________________
> Amps mailing list
> Amps@contesting.com
> http://lists.contesting.com/mailman/listinfo/amps
>
_______________________________________________
Amps mailing list
Amps@contesting.com
http://lists.contesting.com/mailman/listinfo/amps
|