Amps
[Top] [All Lists]

Re: [Amps] plate bypass capacitor

To: amps@contesting.com
Subject: Re: [Amps] plate bypass capacitor
From: Manfred Mornhinweg <manfred@ludens.cl>
Date: Mon, 10 Aug 2009 16:55:58 +0000
List-post: <amps@contesting.com">mailto:amps@contesting.com>
Hi Gary, and all,

I feel an irresistible pull to add my few cents again to this thread!

> I believe that the recommended impedance of the RF choke should be 10 times
> the plate load impedance of the amp not the reactance of the plate tune
> capacitor. 

It's correct that the load impedance and not the capacitor impedance is 
the base, but the 10 times factor is purely a rule of thumb, and very 
wide deviations are possible. In any case, the choke's inductance needs 
to be cancelled out by the mathcing network.

> That is the ideal value but the ideal can not always be obtained when multi
> band operation with a single choke is desired. A larger choke will have many
> series resonances that fall in the ham bands.

Usually the chokes are broken up into smaller sections, slightly 
separated. The sections are of different sizes. So, should any section 
resonate on a given frequency, the other sections will not, and will be 
in series with the resonating one, avoiding smoke.

> When using a small value choke
> on the lower frequencies the choke uses enough capacitance to become
> parallel resonant.

It has to be kep in mind that in a parallel system like this, there 
cannot be separate parallel resonances! Instead, the whole system is 
parallel resonant at the working frequency. That includes the plate 
choke, the tuning cap, the plate capacitance, stray capacitance, the 
tank inductor, load cap, stray inductances, even the antenna reactances. 
  There is no way, for example, to have the choke parallel-resonating on 
a frequency different from the one the tank is tuned to.
Series resonance is a different matter. The choke CAN be series resonant 
  ona  different frequency than the tank's resonance. And one has to be 
careful that it isn't series resonant near the tank frequency, nor on 
any of its closer harmonics!

> What about the popular 1.8 to 30 MHz choke that is used in the ameritron
> amps and many others. It is 225 uh. It has a reactance on 1.8 MHz of around
> 2.5 K ohms. That is a long ways from 10 times the load impedance of an amp
> with a 2 k plate impedance.

Yes. It's a compromise, but works well enough. It will require a little 
additional capacitance in the tuning cap, and the RF current in the 
choke and bypass cap will be large.

> It only takes around 30 pf to resonate that choke on 1.8 MHz. why do you
> think that it would not be tuned to resonance no matter if you wanted it to
> be or not?

Indeed, the choke WILL resonate, as part of the whole tank cicuit.

> If the choke operated at 2.5 k ohms and the plate tank circuit operated at 2
> k ohms wouldn't you think that a significant amount of power would be
> consumed by the choke?

Inductances don't dissipate any power. Yes, the choke will consume some 
power, due to its resistance. But not a significant amount, if it is a 
decent one. To calculate how much power it will consume, measure the 
choke's Q factor at the frequency in question. Then calculate its loss 
resistance, from its reactance divided by Q. If the Q is 200, for 
example, the loss resistance at that frequency would be 12.5 ohms. Then 
calculate the RF current through the choke, based on the RMS RF plate 
voltage, and the choke's reactance. This RF current, squared, times the 
12.5 ohms, give the RF loss of the choke in watts.
For the whole loss of the choke, you have to add its DC loss coming from 
the DC plate current and the choke's DC resistance.

> The fact is that the choke uses enough of the plate tune capacitance to
> parallel resonate itself in the circuit and present a much higher impedance
> than it does in the non resonant mode. 

Not true! The choke still has its 2.5k of reactance, and will indeed 
take a strong RF current! But the plate tuning cap, with the slight 
capacitance added, will take an equal amount of RF current in addition 
to the one it would have taken if the choke had infinite impedance and 
the plate tuning cap did not have the additional 30pF. The net result is 
that the tube doesn't see additional RF current, but the tuning cap 
does. And the choke IS carrying the RF current, as calculated by the RMS 
plate RF voltage divided by the choke's impedance. This RF current will 
be on the order of an ampere. Which isn't so much, considering that the 
tank coil, bandswitch, tuning and load caps are all taking something on 
the order of 10 amperes!

> The choke gets tuned to resonance no differently than the coils in a pi-L
> network do. Each coil shares and uses an appropriate amount of the common
> tune capacitor.

That's correct, but still each coil does carry the same RF current as if 
it were exposed to the same voltage without the presence of a tuning 
cap! The coils don't know whether there is a cap somewhere compensating 
the current they are taking. They have their impedances, take the 
corresponding current when exposed to a certain voltage, and that's it.

Let me do a small excursion to solid state, narrow band amps, like 
typically used on VHF. There are two approaches to obtaining low loss in 
the collector choke: Make its impedance so high that it takes negligible 
RF current, or make its Q so high that it can take a large RF current at 
low loss.
The first approach is very often problematic, because it will make the 
collector choke parallel-resonate with the circuit capacitances at some 
low frequency, typically a few hundred kHz to a few MHz. And at those 
frequencies, VHF transistors have extremely high gain. Specially if the 
base choke is also made to have high impedance, low frequency 
self-oscillation is totally guaranteed! My approach to solve this is to 
use a low impedance collector choke, often having a reactance as low as 
the collector load! And making it of thick wire, so it has high Q, and 
then using low impedance, high current bypass caps, in parallel with 
electrolytic caps. That way, at low frequencies where the transistor's 
gain gets high, the collector will be essentially grounded. Sometimes I 
even use lossy decoupling at low frequencies (electrolytics in series 
with low value resistors), to further discourage ocillation. The base 
choke instead can often be eliminated, using just a resistor in its 
place, if enough drive power is available. Else, it's best in my 
experience to make the base choke have an extremely high inductance, so 
that its resonance with circuit capacitances is so low, that the 
transistor has no chance to oscillate there, because the collector choke 
is already behaving like a straight short piece of wire, and the lossy 
bypassing is in full action.

For practical implementation of some variations of these techniques, see 
the power amp of my FM broadcast transmitter, at the link below. Note 
that the collector chokes have values 500 times smaller than the base 
chokes, despite the collector and base impedances being roughly the same!

http://ludens.cl/Electron/fmtx/fmtx.html

Manfred.

========================
Visit my hobby homepage!
http://ludens.cl
========================
_______________________________________________
Amps mailing list
Amps@contesting.com
http://lists.contesting.com/mailman/listinfo/amps

<Prev in Thread] Current Thread [Next in Thread>