I have a VAC 7070 here for repairs, upgrades, and
for instalation of active protection circuits.
I have just about completed the circuit changes which included
removal of output stage FB capacitors to "neutralise" the 300B input C,
re-working of the values of RC used in the 6 way switchable NFB networks
to give unconditional stability, including phase/gain shelving networks
for LF and HF, and including 8ohms plus 0.22uF across the 8 ohm outlet
to load the amp at HF. Finally the amp was able to
give a very nice stable response
from 7Hz to 84kHz with 6dB NFB, and Rout of about 1 ohm.
The noise the amp makes with 4 x 300B in PP for each channel was the
main obstacle left.
I started by rectifying the heater supplies for the two 6SN7 used for
input, CPI,
and balanced amp. The circuit used is virtually identical to the
original Williamson amp from 1947.
I added a CCS to the tail of the balanced amp which much reduced the
common mode 2H which otherwize appears at each phase of output.
The loading on the balanced amp 6SN7 is 25k dc supply R and cap coupled
100k
grid bias R for the following output stage.
I got the driver stage to make 91Vrms at 3%, and its about as good as
anyone could
with R loading instead of choke + R loading as seen in schematics at my
website.
67Vrms is needed to drive each of the output grids where the Ek = 93V,
Ia at idle is 59mA,
and Ea = +416V, so the 300B are very comfortably set up to give a lot of
class A PO,
with each 300B dissipating about 25 watts.
Anyway, I finally got the amp with zero NFB to give a full BW and
operate without low level LF instability, and to give hum&noise less
than 1mV.
The original VAC had the 300B idling at an unsafe 37 watts.
Before I was asked to intervene, one 300B turned white, and just quit,
then another after a red hot episode,
and the amp is sent to me for a fix. Two others were dodgy.
I discover one 300B develops a slightly positive grid, and hum appears
at the output
when it is used, so I apply 27 + 27 ohms as a divider to centralise the
hum on all the heaters.
Each output tube's dc heater supply is made by rectifying
a Vac winding with a bridge and 33,000uF to make 5Vdc with 66mV of hum.
I try another of the original 8 x 300B and find one shorts from anode to
grid
when heating power is applied, Hum sure increases, and 450Vdc is seen
across the 10W x 1k5 cathode R.
I turn it off, label the 3rd crook tube, and plug in the last I can
select, and then the amp
stops humming. For how long?
The anode supply is an SS bridged rectifier feeding 4 x 330uF in a
series/parallel
set up to give a total C of 330uF, and the ripple is about 1.5Vrms at
the OPT CT.
If one 300B is left out of the circuit, hum at the output increases
alarmingly, because Idc is very unbalanced.
So the amp depends on its cathode bias and output tube bias stability
for low hum. During the many tests over a few hours I performed, the hum
varied considerably.
One would need to only use speakers with sensitivity below 89dB/W/M if
one expects
silent operation. For horn speakers, amps need to be very quiet indeed.
The VAC would be totally unsuitable for horns, but a single 300B with
squeaky clean PSU
would be fine.
The 300B were Sovtek types, and very microphonic.
Tapping them lightly made them ring audibly like a bell,
and this created wave forms on the output in excess of 20mV very easily.
( Is the subtle actual microphony in tubes responsible for the different
"sound" we hear
with different brands of tubes? )
The TRIODE output stage means Ra is very low, and with a
low impedance power supply the 100Hz hum voltage at the OPT CT creates
a common mode hum current in each output tube of about 5mA, and
when inspecting the cathodes at the top of each 220uF cathode bypass cap
there is 22mV of 100Hz hum, and at low level use this intermodulates the
incoming signals, and when observing the THD of the amp at low levels
normally used for listening
I found there was slightly more amplitude modulation of the distortion
wave form than the
actual harmonic distortion itself.
This would not be the case if the PSU was better filtered.
So the voltage measurement of whatever harmonic or IMD was present
showed that
it rose faster and sooner than I would ever like to see between 0.0 Vo
and 2Vo,
and this is supposed to be the really clean region of any amplifier
working because
most listening is done with signals below 2Vrms.
So, would I bother ever using 300B for anything I'd sell anyone?
Not likely, because one MUST better filter the heaters to reduce hum <
10mV
and prevent its modulating Ia, and better filter the B+ applied to the
OPT CT.
The filtering of the 5Vdc cathode suplies is most effectively done using
CLC,
and where you have an amp with 8 x 300B, its a lotta gear you have to
use.
The VAC has 4 small 12VA transformers each with 2 x 7Vac windings to
make 8 windings.
There are 8 bridges and 8 x 33,000uF caps.
It looks like they didn't want to do it right and have 8 windings for
300B heaters
on the main large PT. This would have left room for the filter chokes
for the B+.
Using KT88, 6550, or better still, KT90
would be cheaper, easier, less microphonic, and give longer tube life
without what appears to be the fragility of the 300B.
The amp owner has given me a fresh set of Sofia 300B output tubes that
do look nice,
and I hope they live up to their looks.
At least they won't be idled at levels which invites failures,
and if one fails, the amp will politely shut down
without smoke or burning out cathode resistors and caps and starting a
fire
and scorching a ****tion of the 4mm thick fibre board used with turrets
for connecting point to point.
I hate the use of boards in all tube power and preamps.
They just don't belong.
There isn't any access to the blind side of the turrets, and although
the board
is held in the amp with a few screws, its a terribly horrible job to
remove the board if one has to.
Accessing the tube sockets between the board and top metal part of the
chassis is an awkward
job.
Sockets should always be bolted to the metal plate that forms the
chassis toplate.
Terminal srips which can be turrets if you like should be on strips
of 8mm x 8mm fibreglass rods held away from the chassis to allow wires
to pass under.
The board actually reduces volume in which parts can be mounted.
In the VAC, I was stuck with the damn board, because the alternative of
chucking all VAC's work under the chassis into the bin and starting all
over
was much more work than replacing all the GE capacitors with Wimas.
Just why so many 300B failed isn't clear, but I have replaced all
coupling caps
in case some fail again soon. Several had already been replaced by some
other tech before me.
Last year I repaired a pair of Sun Corp amps with 2 x 2A3 per channel
there was ac heating,
and a hum nulling pot was used for each heater circuit which has cathode
biasing.
I gave the owner a pair of headphones to use while adjusting hum to
minimum levels every few weeks.
Where you might have 4 x DH triodes per channel and ac heating,
adjusting hum
on the 4 tubes would not be reliable, and not easy, because the
adjustment is interactive, and because of the
problem with IMD with incoming signals.
Hence octal tubes with IDH cathodes are much better, unless you reduce
hum on the cathode
Vdc supply to really low levels. This could be achieved CLC as I said,
and the L doesn't need to be very large, but otherwise done with a
regulator
circuit using bjts, but if anything goes wrong, they can get zapped a
lot easier
than a choke.
Eventually, I will get the VAC to be as good as it can be, and I may yet
have to
put in a pair of small chokes for the B+, 1H at least.
So far, the VAC with 4 x 300B hakes almost as much maximum power as my
SE845
with a pair of 845 giving 60 watts max, A1 PO.
The 845 SE amps have 7dB global NFB and VAC now has 12dB max global NFB,
but the
VAC makes twice the THD at 5V into 8 ohms than the SE amp.
Just because an amp is a really gee whiz super duper push pull class A
triode
concoction, it does not mean that it will measure better in *any* way at
listening levels
compared to many SET amps.
it takes 12dB GNFB to get Rout of the 4 x 300B PP to be just undr 0.5
ohms.
Its 2.4 ohms without GNFB.
The SE845 has Rout = 0.35 ohms with only 7dB NFB and 0.9 ohms without
NFB.
So no wonder the SET amp which has equal max PO to the max PP amp PO
sounds better. There is less distortion of all kinds.
When we listen to the utter drivel Trevor Wilson sprays about the Net,
we can all see just how utterly incorrect he is,
as is born out again by my comparitive observations of fundamentally
good amps like the VAC
and including ones I design and build myself.
I do like the 300B; I am fond of it, it does well with audio signals.
The VAC sounded very well just before it came to me, with at least 2 of
8 tubes
not operating, or operating poorly.
But you do have to look after the 300B well to get it to perform at its
best,
and that means providing voltage supplies like those from good
batteries.
Use piles and piles of capacitance, and chokes where appropriate.
In Quad-II amps, the B+ is applied to the OPT CT filtered only by 16uF,
and thus the Vripple = 17Vrms at the OPT CT. ( Quad-II is a chic toy
thinge from 1955 )
In my measurements, the IMD caused by interaction of riple voltage and
signal
means total distortion measured with a sine wave signal about equal to
the THD alone.
But the KT66 has a high value of Ra, and thus common mode ripple voltage
current flow in the
tubes isn't great. It would be if you run the output tubes as triodes.
So I don't recommend it unless you upgrade the PSU with CRC or CLC B+
filtering
all as shown at my website.
Patrick Turner.
|
