On Tue, 20 May 2008 22:35:46 GMT, Paul <paulguy@[EMAIL PROTECTED]
> wrote:
>On Mon, 19 May 2008 23:08:15 -0500, flipper <flipper@[EMAIL PROTECTED]
> wrote:
>
>......snip!......
>>
>>I'm beginning to get the feeling I could talk till I'm blue in the
>>face so a better idea might be for you to pull out RDH4 and read
>>Chapter 7, Section 1 (v) "Combined positive and negative feedback at
>>the mid-frequency" and Section 2 "Practical feedback Circuits" (xi)
>>"Combined positive and negative feedback."
>>
>>Figure 7.51a is similar (principle is the same), but not identical, to
>>the one I employed . Mine has two less resistors but takes a bit more
>>work balancing idle currents and gain.
>>
> OK, I pulled that old book out and checked it out, and I see the
>light.
> I haven't opened that book for a number of years, but I was
>surprised to see how well they covered feedback issues, that dozens of
>other engineering texts that I have don't even suggest.
Yeah. In the tube heyday it wasn't practical to plop down an infinite
number of tubes so there was a lot of ingenuity employed to get the
most from what you've got and that's one of the things that attracts
me to tube design.
As I mentioned, that particular circuit tickles my fancy because it's
essentially 'free', component wise. A very clever 'trick' by Harmon
Kardon.
The bootstrap as a semi constant current source is another one I take
a fancy to and I used that in the PC Speaker Hybrid amp (solid state
power stage).
In the early days of discrete transistors some of the same techniques
were used but it soon passed as they became cheaper than dirt and it
wasn't too long before the Japanese were selling '15 transistor'
radios with half of them used as diodes just so they could advertise a
large transistor count.
And nowadays they're just another microscopic dot on the die.
> In my words, you are generating extra (local) gain using positive
>feedback, to increase your (global) loop gain so as to reduce
>distortion in another amplifer section (usually the final stage).
Yep, there ya go.
I say free "component wise" because it isn't free function wise. You
lose bandwidth, as one would expect, but you got most of that back
when you close the loop. But an interesting problem that may not be
obvious is if you DC couple the gain triode to the phase splitter bias
drift/variation from tube to tube is multiplied as well because the
phase splitter's idle current goes through the common Rk. That's why I
AC coupled this one: it breaks the DC PFB.
In my sub watt guitar amp I used an NPN phase splitter and that
'required' DC coupling so I used a different approach. I put a second
'bias' resistor under the NPN, to reduce the idle current, and AC
bypassed it into the bottom load resistor. That let me make the DC PFB
small while using a larger amount for AC.
The same thing works with DC coupled triodes but if you're going to
bypass the bias resistor you might as well break the DC PFB entirely
with AC coupling.
The 'really' really clever one is also from Harmon Kardon. It's like
the PC Speaker Tube Amp except the gain triode's load resistor is
taken to the phase splitter cathode, instead of B+, so it's
bootstrapped to a 'near infinite' impedance (CCS). The tradeoff there
is low plate voltage on the gain triode and I didn't like it that low
on the 6AW8 so I used the B+ version. Works good with a 12AX7, though,
and I use it on my 12AX7/6GK6 PP amp (essentially the same thing
Harmon Kardon did with 12AX7/6BQ7s but I bias the phase splitter like
the PC Speaker amp whereas.they used grid leak).
> This was an enlightening argument!
>Thanks,
You're welcome. I enjoyed it too.
>-Paul
|
