Patrick Turner wrote:
Ian Bell wrote:
Patrick Turner wrote:
Ian Bell wrote:
Patrick Turner wrote:
Ian Bell wrote:
The 6SN7 is said to be a good tube for audio and the 6CG7 is apparently
its B9A direct equivalent. Looking for sources of the 6CG7 on the
internet I found more examples of its supposed equivalent the 6FQ7 than
of the 6CG7 itself. Looking at the data sheets the two seem extremely
similar and the 6FQ7 even has slightly less plate to grid capacitance.
So, is the 6FQ7 a worthy alternative to the 6CG7? Are there any real
differences between the two? Presumably there must be some difference
else why have a different number.
Cheers
Ian
The 3 tubes have virtually the same gm, µ, and Ra wherever you find
them.
Many varieties exist, with Siemans NOS probably the best. The CG7/FQ7
have slightly lower Pda ratings than the 6SN7.
I found the Siemans 6CG7 to have a slightly higher µ.
Anode and grid size varies with varieties, but as long as you keep the
relative distances inside the tube about the same the
gm, Ra and µ stay about constant.
Patrick Turner.
I simulated a 6sn7 in mu follower circuit and it gave very good
distortion figures - best I've seen yet - this was at 8mA Ia. So I
ordered some from Colomor valves in the UK so I can try them out. They
are the 6FQ7 version made by Matsu****a - didn't know they made tubes.
Does Matsu****a have a good or bad reputation for its tubes??
Japanese. I doubt the japs made worse sounding tubes than anyone else.
Its not going to cost you much to find out for yourself.
The secret to the µ-follower is that you have a large enough R between
top cathode and bottom anode.
If the Ia = 8mA, and that would be only good if output voltage swing is
smallish, then the Ea can be say 100V for top and bottom tube
and if the B+ = 300V, then you can have 100V across the R between the
two tubes, so R = 12k.
If the top tube internal gain is 18, then the bottom tube has an anode
load = 18 x 12k, plus any biasing R.
Its still only 216k effectively, or about 25 x Ra. But you should get
less than 0.1% THD at 10Vrms output,
which means less than 0.01% at 1V output, which isn't much, eh.
The design I simulated is very similar to that but with a 285V supply
and R = 10K resistor and 100V across each triode. I did not want to up
the supply much beyond 300V because at present the top cathode sits at
over 180V so I'll need to raise the heaters to about 100V because the
max dc heater positive to cathode voltage is only 100V according to the
data sheet.
The idea is to have the heaters biased at Vdc about 1/2 way between the
two cathodes.
From what I can gather there are three factors that govern the
distortion in mu followers. The first is anode current - higher anode
currents give lower measured distortion.
This is true with resistance loading also. But the higher Ia, and lower
the Ea, the lower the range of Vo because of grid current.
The second, as you say is the
effective Rp determined by the top tube mu and the resistor between.
This essentially reduces the effect of varying ra on distortion.
The higher the triode RL becomes, the flatter the load line becomes when
you plot it across the Ra curves.
The Vo swings don't intersect Ra which changes value. If you have an RL
20Ra, then for 6SN7 that'd mean about 200k,
and if the maximum Vo = +/- 33pk then load current change is only +/-
0.16mA if the 200k is the only loading present.
So you can have the Ia lower than with a resistance load and still get a
linear outcome. Put it this way, if the load approaches a CCS, it
becomes pointless to have 6SN7 Ia higher than 4 mA providing 4mA is
adequate for the wanted Vo into whatever other load you may have, or
Miller C.
The
last factor is the tube type and from the tests I have done so far this
seems to make the largest difference. You can try almost any of the
regular modest mu double triodes (6DJ6, 12AU7, 6N1P) and you will get
between 0.5% and 1% THD at 20V rms and about a tenth that at 2V rms.
However, use a 6SN7 (or one of its derivatives) at the same current and
the 20V rms THD drops to little more than 0.1%.
I was curious why this should be and then I came across an interesting
web page about the internal construction of tubes and its effect on
linearity. You can see for yourself he
http://myweb.tiscali.co.uk/g8hqp/audio/valvedist.html
I have no idea if what he says is true but it would explain why the 6SN7
is so well liked in audio circles.
There are many varieties of 6SN7 and the ECCxx range of double triodes.
ECC32 is a particularly linear twin triode and the 6SN7 is of the same
evolutionary "family".
Indeed the 6SN7 is a more linear tube than 12AU7, or 12AT7.
The 6SN7 often had tubular shaped anodes, grids and cathodes, and to my
mind is a more perfect way to make a tube.
But take a look at the range of 6CG7 around. Many had the 25mm tall box
section anodes of the 6SN7 crammed into the smaller bottle.
Damned linear though. Many 6CG7 had overall smaller anode/grid/cathode
structures, but the internal part of the anode was a square section.
The Siemens 6CG7 are like that and are regarded as the gold standard in
twin triodes. But then take a look at a 300B. OK, its a power tube, but
its amoung the most linear of all.
What I do know is that any variety of 6CG7 and many other medium µ twin
triodes are amoung the most linear devices ever invented and which don't
need to depend on external loops of NFB to reduce the distortion.
The lowest possible THD in a triode occurs where you have a CCS anode
load, and zero Ia change when Va changes.
The only way to exploit this fully is to have a CCS loaded gain triode
direct coupled to a cathode follower buffer.
Then you have a CCS cathode load for the CF, and the only load might be
a 100k gain pot. The majority of the THD generation is then in the CF
but its reduced by the series voltage NFB inherent in the follower
connection; ie, if you had 0.1% THD at 10Vrms output with a load = 100k,
then the follower connection with 6SN7 would reduce that 0.1% by 1/18
approx, or down to 0.0055%.
Even with a horizontal load line, ie CCS load only on the 6SN7, the
spacing of the Ra lines for various Eg1 bias values does show that
you'll get some THD; the spacing is *not* even between Vg1=0 and Vg1=-5,
when the Ea change would have a range of +/-50Vpk.
Often when you test a given triode carefully by measuring the THD you
will conclude that the people who drew the Ra lines for a given tube
were obviously using gear with distortion in the test gear which was of
course all tube based and not necessarily more linear than the tubes
being measured. The curves drawn so impressively in the best data books
such as those produced by STC back in the 1950s are a cause for me to
smile, because the nice large pages so well bound and produced are full
of what are only approximate Ra curves, and curves which look simply
impossible, or implausible, because of their irregular spacing that if
true would produce many more artifacts than the usual bit of mainly 2H
with all the other harmonic garbage -20dB below the 2H.
So hence the breadboard test with accurate measuring gear using test
signals with THD 0.002% is essential if you want to bathe your mind
with the purity of the truth on vacuum tubes.
You cannot simulate the taste of a fine chocolate cake.
So don't depend too heavily on a simulator; its only a guesser's tool.
I won't argue with that!
The Matsu****a 6cg7 tubes have arrived and I have tested them in the mu
follower circuit at 8mA. The results were disappointing. All four gave
about 0.7% THD at 20V rms and 0.07% at 2V rms which is no better than I
got with an ECC88 or a 6N1P at the same 8mA current.
Looking at Morgan Jones test circuit I notice he uses two LEDs in the
bottom cathode leg for biassing which sets Vg at about -3.4V and from
the curves means the plate voltage is about 140V. He uses a 10K
intermediate resistor as I do but his pentode CCS has a 390V supply. My
circuit uses a cathode resistor to set Vg to about -1.34V which gives Ea
about 117V (measured) and I have a 320V supply. So I thought I might try
moving the operating point closer to Jones's, because the curves look
less bunched in that region but if I operate both triodes at the same
point they have 140V each across them plus 80V across the 10K meaning I
need a 360V supply which I don't have. I must admit I assumed both
triode halves should be operated at the same point but I am not sure
that is necessary assuming the bottom triode is the principal source of
distortion. So an alternative would be to run the top CF at the same
current but a Vg of -1.2V so it drops just 100V which would fit with my
320V supply. It's an easy mod to make so I think I'll try it.
You also mentioned the 6SN7 has 25mm boxed anodes and some 6CG7s cram
these into a 9 pin bottle. The Matsu****a ones I have measure about 19mm
so there is clearly a difference there but the ECC88 and 6N1P I have
measure only 9 and 10mm respectively so the 6CG7 are twice as tall and
the box is definitely fatter than the ECC88 etal.
Cheers
Ian
Patrick Turner.
Cheers
Ian
Patrick Turner.
Cheers
Ian