>
>
>
> This email was held on account of the picture so no picture here. Victor,
> I'll reply to your email too in a sec. I didn't want to roll it into one
> huge wall of text.
>
> Using 3mm wire for tank coil taps is a big no-no. It will add stray
> L....making 10m operation near impossible.
>
>
> Does it matter if the 3mm wire is on the load cap side of the coil?
>
> I thought only the "front" of the coil forms the first L network of the Pi
> network with the plate cap. The back side interacts with the load cap. Is
> this true, or completely false? It has huge impact on the bands switch
> placement. Because if I can have longer leads on the load cap side of the
> coil that makes this case usable. Otherwise not so much.
>
>
> Hence he can't get it to
> resonate > 25 mhz.
>
>
> The thing is it did resonate. (Victor's amp didn't). There was no
> difference about that resonance peak on the nanovna from any other - it
> wasn't wider or narrower.
>
> Likewise during tuning under power. Power peaked if I kept going it went
> down (this is with well under max drive so no grid current yet). The
> problem is it peaked at 200W instead of 1.5kW.
>
> If I couldn't resonate it would be a very simple thing to understand and
> no point to write to the list....
>
>
> Use 6mm / .25" wide copper straps.
>
>
> I used 12mm wide straps on the plate side of the coil everywhere, I
> thought there was no need on the load side. If there is ill have figure
> something out, because that's where the largest distance to cover is. Maybe
> I'll have to relocate that bands witch.
>
>
>
> You also have
> more stray L, via passing through the bandswitch rotor, to the center rotor
> hub, then onto the COM...then back to the C2 load cap. From COM of
> bandswitch to the C2 load cap should be wide, copper strap. Wide copper
> strap has miniscule uh.
>
>
> Here we are at the load side. Does 100nH extra here matter (if equivalent
> amount is subtracted from the coil)?
>
>
>
> The trick with the tiny .1uh to .3uh coil (HD coil, has a ton of current
> through it) inserted between plate block cap assy and C1 tune cap
> works..... but ONLY if the tube does NOT require a parasitic suppressor.
>
>
> The way it works with that config is..... the shunt tube C plus the tiny
> series coil forms a step down L network. That steps the plate load Z down
> to a much lower value......low enough that now a practical PI network can
> be designed and built. ( the lower the Z, the higher the C is needed for
> the tune and load caps...and the main tank coil between the caps will be
> reduced in value.
>
>
> I think this is a good approximation of this arrangement. I've played with
> the values to get it to resonate the 27pF of the tube capacitance. It did
> with 0.2uH extra. Then the load coil is shrunk to 650nH. Simnec seems to
> say coupling is 0.75 with compared to 0.63 to without.
> (here was the simnec image showing the tube capacitance, extra inductance
> and so on)
>
>
>
>
>
> You can see all of this on the Gmail account 3SEK PI / PI-L spreadsheet.
> It
> factors in all stray C, like anode to chassis C, any additional stray C,
> like from lower anode fins to chassis...and also stray C from tank coils /
> bandswitch, to side / rear walls..and also stray C from tube to rear
> /sidewalls. The software will spit out the new transformed plate load Z,
> that the PI network will be designed around. You can also change the Q on
> the spreadsheet, so you can see the increase in Q on 10m, when you are
> forced to use a smaller than normal main 10M coil. With too big a 10m
> coil, your required tune + load C will be depicted as negative values on
> the software.
>
> You can easily measure all that stray C with a digital LCR meter. Ditto
> with all the stray L
>
> Also, use a separate connection from top of plate choke to edge of anode
> fins for the B+ connection. Use a separate connection between edge of
> anode...to the plate block cap assy. And keep the B+ connection well
> away from the RF connection, like 90 degs or more. This will enhance
> stability.
>
> But since that tube you are using requires a parasitic suppressor, you
> won't be able to use the small, extra coil trick, too bad. Your only
> recourse is to minimize all the stay L and stay C as much as possible, then
> decrease the value of the main 10m tank coil tap, so the tank will
> resonate on 10m, with the C1 tune cap barely meshed.
>
> This might require the use of a vac tune cap with a low min C....or a 2 x
> section air variable, like Alpha and other's used, with a tiny 1st section,
> with low min C, used on upper HF only.... and both big and small sections
> used on lower HF.
>
> I have gone through all this mess several times before on hb amps. That's
> why I stick with the metal tubes that don't require a suppressor. That 10
> kw, 6M amp ( 3x6 triode) I designed for Scott, used the extra coil trick.
> It's the only way to get that tube to work on 6m. In that one off case,
> both 'coils' were made of 1" wide copper strap (25.4mm wide) in a 1.2
> circle shape...and both at right angles to each other. The 1/2 circle
> shape was used cuz the current will then flow on both sides. If a
> conventional tubing coil was used, it would have to be made of 1,27" od
> tubing.
>
> Very interesting, thank you for sharing this.
>
>
>
> On the latest 160-15m, 3x6 amp, the L coil trick was used...just to get the
> loaded tank Q down on 15m. Instead of a separate coil, the extra .6 uh was
> incorporated into the 20-15m coil...made from 1/2" silver plated tubing.
> The C1 cap tapped the junction of the extra .6 uh portion.
>
> Be careful when running a high Q on 10m. RF current through the 10m coil
> is aprx the plate current times the input tank Q..+ 10%. On the GM3SEK
> software, it's the overall network Q that is adjustable. It spits out the
> input Q..and also the output Q. Those 2 added = total network Q. High
> current on 10m = heat. It's just (I squared) x R...or in this case, RF
> current squared X ESR of the tank coil. (ESR = effective series
> resistance or RF resistance). Double the current in the tank coil..and the
> amount of watts dumped into the tank coil will quadruple.
>
> Watch out with modes used. FT-8 / data modes is an amp killer. Average
> DC plate current on data modes is double processed SSB. Same deal, watts
> dumped into the tank coil has now quadrupled. The heat can get so great
> that the bandswitch contacts will heat up, lose tension, RF resistance in
> the contacts will increase, and contacts will cook, due to thermal
> runaway....and the ceramic can crack.
>
>
> In the old arrl books (plus Eimac care and feeding) they just
> depicted input Q..and typ a value of '10' was used. In actuality, the
> overall network Q is '12'...which consists of an input Q of 10...and an
> output Q of 2. The newer arrl books use overall network Q..and typ use
> '12'.
>
> Jim VE7RF
>
> Thank you Jim. I u
>
_______________________________________________
Amps mailing list
Amps@contesting.com
http://lists.contesting.com/mailman/listinfo/amps
|