Re: 845 first impressions, 22dec07.



Nick Gorham wrote:

Patrick Turner wrote:


Nick Gorham wrote:

Andre Jute wrote:



Nah. Patrick's amp fits straight into the mainstream, with only the
wrinkle of the split power rail, to keep the smoothing caps and
equalizing resistors sane, being even slightly odd, and that will go
into the mainstream within a year or two, mostly without any thanks to
Patrick. Want to take a small bet on it, Iveson? A tenner that within
two years Patrick's split rail appears on at least one other
transmitting tube amp? (Or you might want to bet that it appears on a
Chinese production amp in five years or less...) Also, Patrick's amp
will give the KR 845 workalikes a big boost; we'll see more of them
now.



Well, without wanting to take anything from what sounds like a nicely
balanced amp design, I think you will find the idea of the split supply
predates this, and I doubt Patrick will suggest otherwise. the real
advantage as Mr P mentions isn't the caps or resistors (though it does
make life simpiler if the caps are exposed). Its the OP TX to ground
potential thats the real winner.


You have me wrong, I have the OPT in the +750 rail at present.

I have CONSIDERED having the OPT primary at 0V, and then having
a NORMAL +1,250V supply floating, but connected to anodes and OPT
primary
at 0V potential.
Any one is free to build any SE or PP amp this way,
and A Soundman Number One did it. ( they are a defunct Canadian amp
maker )

BUT, the problem is that you have the whole PSU including HT windings
and diodes and caps and filter choke/s
ALL riding on the anode audio signal.

THEREFORE you have to make sure the whole PSU rail parts list is WELL
insulated from
anything at OV potential, and with a view to minimizing capacitance,
because with a floating series PSU
you have a total amount of C between it and 0V in addition to the shunt
C to 0V of the OPT
plus its P winding self C. In my case I have about 2.5nF without
floating the
PSU. The low Ra of the 845 in parallel of 1.2k approx will not struggle
with such a capacitance load,
but you wouldn't want to double your troubles by having 5nF.

So one has to think imaginatively about boxing up and mounting
all the filter caps and chokes.
Ideally, the PT should have electrostatic shielding between the mains
and HT windings
to stop mains HF noise transfering to the 845 anode circuit.

So I decided the flating PSU rail was just too troublesome to get right.

A split rail allows the OPT primary to be at a sensible B+ of +750V,
and -500V for the other rail part is easy, and each rail is doable using
pairs of 400V rated 470uF caps,
which are plentiful and cheap.


Sounds good Patrick, I wonder why more don't use the el84 as a driver
valve.


!


I am using one in a 211 A2 amp at the moment, as a pentode
cathode follower (with spilt rails for the driver stage, as it happens),
its happy going way into A2, well 40v, 50v or so anyway. I am tempted to
try a similar arangement for a 572b. Keeps the room warm anyway :-).

Of course that doesn't meed the swing that an 845 requires.


I'm not so sure.

211 is biased at around -75V for 60+mA at Ea = approx 1,200V.

I am using 80ma @ 1000v, different sound to 1200v @ 60ma (thats what I
used on the last one).

Ideal load for the two conditions is quite different.

The sound may well change with loading though, no?



As a general rule for negatively biased power triodes,
and for maximum class A1 PO, RL = ( Ea / Ia ) - 2 Ra.

And the Ra changes with the EA and Ia because the curves show higher Ra
at
lower Ia Q.

For Ea 1,000V, Ia = 80mA, Ra approx 3.3k, ideal RL for max PO =
(1,000/0.08) - 2 x 3,300 = 12,500 - 6,600 = 5,900 ohms.

For Ea = 1,200, Ia = 60mA, Ra approx 3k, ideal RL for max PO = 20k - 6k
= 14k, which is over twice the
RL for the lower Ea.

But any load over say 10k for either Q condition is OK if you don't mind
a restricted range of max PO.
PO can only be a maximum at ONE RL value, and RL values either side
result in less PO.
RL values less than the value for max PO should be avoided like the
plague because THD/IMD
becomes much worses, OPT winding losses rise, and damping factor is
poorer.


But as RL is raised, the THD becomes lower, damping factor higher (
better) and it may not matter that
you have only 12 watts instead of 25 watts max if you only ever need 5
watts most days.




For widest A2 swings, the +ve travel of the grid might be 40V which adds
to the 75V,
so about +/-115Vpk??

Well, in my case -45 on the grid, so the swing is +/- 85v for a little
less than 30w output, but even 115v is less that the typical 845 point,
say 1000v @80ma, +/-150v grid, even to the 0v line, that extra 35v isn't
trivial.

Once you start using high ac impedance dc supply to the driver and
high enough Ea for the driver with swing room, getting up to 200Vrms,
280Vpk
is easy...



If you were to be at 1200v @ 60ma, and wanted to get to +40 on the grid,
the other half of the swing to -190v would require a 16k transformer,
which is opening another set of issues.

I reach the 3% point at about 27w, I could get lower, but that would
mean that the proportion of the harmonics would be a problem, with 2nd
being lower than 3rd.

The load formula for class A2 becomes more simple and like a pentode,
and RL for max PO = 0.9 Ea / Ia, so with 1,000V @ 80mA, ( 80 watts Pda
),
RL = 0.9 x 1,000 / 0.08 = 11,250 ohms.
Its a higher load than for class A1 for the same Ea/Ia because the
Va can swing much further downwards because of the A2 drive, so it must
be able to swing up as well...

Hence the 27 watts you can get, and efficiency = 100 x 27 / 80 = 34%.

But if you had 6k as an RL, there would be no point at all for class A2
because
the tube would sustain Ia cut off at a much lower maximum positive Va
swing,
and THD would become much higher than 3% after visual cut off appears on
the CRO.

For all load values though, 2H is always much higher than 3H
for the first few watts, and its the first 5 watts which are most
important.

One has to be aware of the cause of 3H. And in A2 as the the driver
stage
labours trying to provide the grid with current as the grid becomes more
positive
there is inevitable mild compression, or wave crest flattening on
positive going 211 grid voltage crests.
This combines with flattening due to cut off behaviour and thus you get
more 3H
at 27W+ than 2H.

Keep the load higher and you'll get a better result sonically, imho.

Patrick Turner.




--
Nick
.

Relevant Pages

  • Re: 845 first impressions, 22dec07.
    ... will use with an amp this expensive) it will be like having a large ... I have been thinking about "legitimate ... two years Patrick's split rail appears on at least one other ... an opposite phase at the anode, but the same phase as the LF noise ...
    (rec.audio.tubes)
  • Re: PC power supplies
    ... > I bought couple of used PC power supplies to get 12 V supply. ... Instead the 5v rail ... is assumed to be moderately 'proportional' to the 5v load, ... Most require a reasonable load (perhaps an amp or more) on the 5v rail ...
    (sci.electronics.design)
  • Re: 845 first impressions, 22dec07.
    ... Patrick's amp fits straight into the mainstream, ... two years Patrick's split rail appears on at least one other ... advantage as Mr P mentions isn't the caps or resistors (though it does ... THEREFORE you have to make sure the whole PSU rail parts list is WELL ...
    (rec.audio.tubes)
  • Peavey power amp question
    ... different topologies used for high power audio amplification. ... Peavey amp really confuses me. ... in case I some day need 1000W per channel into 2 ohms. ... emitter-resistors connected to a COMMON rail on the output-board. ...
    (sci.electronics.repair)
  • Re: Peavey power amp question
    ... Peavey amp really confuses me. ... stereo power amp. ... in case I some day need 1000W per channel into 2 ohms. ... emitter-resistors connected to a COMMON rail on the output-board. ...
    (sci.electronics.repair)
Loading