Re: NFB101 part 3



Ian Thompson-Bell wrote:

Patrick Turner wrote:

Ian Thompson-Bell wrote:
Part 3 of NFB101 is now available for download as a pdf here:

http://www.ianbell.ukfsn.org/nfb101/nfb101.pdf

Comments welcome.

Cheers

Ian

After a very quick read of your work I concluded you are on a good
track,
but have some comments/questions.

How about making all your terms the same as in RDH4?

Models for ß x Vo, the fraction fed back and how it is derived looks
different to samples
used in 99% of cases in amps made now, ie,
the FB voltage is from a simple resistance divider, so there isn't any
reason
why the basic block drawing of FB application doesn't show the divider.


Except I would argue that in 99% of cases it ends up NOT being just a
simple resistor divider. Most power amps have a compensation capacitor
across the feedback resistor and every RIAA stage that uses feedback is
not a simple resistive divider.

Using FB for frequency attenuation is not common unless used in an phono
amp.
But for all other amps to get a linear F response, the resistance
divider
is almost always used and it does convey the simple idea.

Where you have a block diagram with amp stages, why not have all stages
shown
as a triangle with +/- input and always assumed + output as in most
textbooks?


You may be right, but I thought real circuits with tubes in them would
be more familiar to tube heads than op-amp like triangles.

First come the principles with triangles, then comes the complexity
of what is within the triangles.

just MHO.

With regard to tube amps and Rout after NFB is applied, the
reduction of of Rout due to series voltage NFB is not just the same
amount
as the gain reduction.

I am not clear what you mean here - series derived feedback would be
current feedback. Voltage derived feedback would be shunt. Please clarify.

Series **voltage** NFB is the case where FB is applied frok an R divider
from Vo to say an input tube cathode. The applied NFB is "in series
with"
the applied grid voltage.

A cathpode follower is a case of local series voltage NFB.

Local series **current** NFB is where you have an unbypassed Rk.



RHD4 explains it all, and categorizes the types of NFB and PFB and
gives the effects of each type of FB in a table.

The figure for gain used in Rout calculations after Global NFB FB is
applied
is the ( open loop gain of input and driver stages of an amp )
all multiplied x ( µ of the output tubes ) all divided by
the turn ratio of the OPT.

The figure used for gain for Rout calculations should be the open loop
gain. I don't know enough about PP tube stages to know if that is the
same as the formula you gave above.

The load affects the output tube open loop gain, so
as load reduces, so does gain.

So the amount of applied voltage NFB in dB also reduces as load reduces.

But the Ra of a single tube or Ra-a of a pair of PP tubes doesn't
change.
The Ra is the real source of the amp's internal resistance.

Nor does the µ or gm.

So you have to consider the Rout as a separate issue to when the amp is
loaded.

So assume the amp isn't loaded at all.

Output tube gain will be close to µ.

Open loop gain = input stage gain x output tube µ.

And its this product that has to be used in Rout calcs when series
voltage NFB is used.




The Rout without NFB is taken as the Rout at the secondary of the OPT,
which equals the total reflected P and S winding resistance of the OPT
plus the ( Ra-a ) / ( ZR of the OPT ).
Ra-a is twice the Ra of one tube of a PP pair, and the Ra
must be determined for the idle Ea and Ia condition of the output tubes.


Agreed.

Basically, what I am saying is that Rout after GNFB is reduced by MORE
than the closed loop gain / open loop gain.


Is it? Why?

Because µ is always greater than the output tube gain.

Next time you carefully measure a tube amp and
include the allowances for the OPT winding resistance, you WILL find
that
the reduction of gain due to say 12dB of NFB means you'll need 1V of
input
instead of 0.25V with no NFB and you may find that Rout has been reduced
from say 20 ohms to 3 ohms, not to 5 ohms as your idea would suggest.

It's all in RDH4.

Rout is very simply measured.

Set the amp so if makes 2.00Vrms output with NO LOAD.

Then connect a load of say 8 ohms.
Make sure no serious distortion occurs.

Say the amp output voltage at the load falls from 2V to 1V.

Rout = change in amp voltage / change in amp current.

Change in amp voltage is due to loading.
amp current unloaded = 0.0 amps.
Change in current = difference between loaded and unloaded.

In this example, Rout = ( 2V - 1V ) / 0.125A - 0.0A = 1 / 0.125 = 8
ohms.

So the change in Rout is easily measured with or without NFB.

The OPT winding resistance might total 1 ohm measured at the secondary.
Its harder to measure. I don't have time to fully
explain what you must read in RDH4.

So if the amp has Rout = 8 ohms, and Rw = 1 ohm, the
Ra-a measured at the sec = 8 - 1 = 7 ohms, and if the OPT ratio
was 5k : 8 ohms, then ZR = 625:1, and Ra-a = 625 x 7 = 4,375 ohms

This may indeed be the case for a UL amp using a pair of KT88 and which
has no NFB applied.

You could place 10 ohm resistors in each cathode circuit and measure the
change
in signal tube current to find out Ra-a as well. That's another story.


Cheers

If one set sails to write a text book, one really should read a few that
have been written
before you got onto the subject, IMHO.



Patrick Turner.



Ian
Perhaps you could include a simplified explanation
which conveys the concept of just how NFB works rather than
just describe math.

See http://www.turneraudio.com.au/tube-operation3.html

Best Regards,

Patrick Turner.
.

Relevant Pages

  • Re: Best tap for NFB?
    ... it by reducing NFB. ... cuts gain by the resistor ratio. ... work) and the same increase goes into the driver tubes. ... by using the shelving network before the balanced amp, ...
    (rec.audio.tubes)
  • Re: Best tap for NFB?
    ... it by reducing NFB. ... cuts gain by the resistor ratio. ... coupling cap roll-off. ... by using the shelving network before the balanced amp, ...
    (rec.audio.tubes)
  • Re: Signal to Noise Ratio
    ... In many a tube power amp, the SNR can decline as power output is ... Yes because the common tube amp has a power supply that is not all ... All such tubes are OK in a power amp and most line stages. ... In a preamp with gain, the attenuator should be after the gain tube ...
    (rec.audio.tubes)
  • Re: KISS 141 by Andre Jute: Just the Facts, Mam: The Complete Tube Transfer Function
    ... At "KISS 141 by Andre Jute: Just the Facts, Mam: The Complete Tube ... except a simple expedient way to stabilise an amp in 5 minutes. ... Sure if tranny loads and C shunts exist then the mid F gain won't hold ... Many 3 stage amps with an OPT will have more than 45d phase shift at the ...
    (rec.audio.tubes)
  • Tube amp project - continued
    ... Because putting another gain stage for FX loop recovery makes the amp ... reasonable output will be able to drive the power section to hard-clip ... No real need for gain there. ... Lose a preamp tube, you can pull the FX loop and carry on, ...
    (alt.guitar.amps)