Re: Kikusui ORC-27A tubed oscilator mods.



Alex wrote:

"Patrick Turner" <info@xxxxxxxxxxxxxxxxxx> wrote in message
news:49940A5B.E274FA68@xxxxxxxxxxxxxxxxxxxxx
Recently I built a wien bridge oscillator using an opamp to get 5 ranges
and from 2Hz to 200kHz at 2.5V maximum output.

But I have not got a square wave function yet for that, and even with
output clamping diodes there is still some danger of a stray input from
a tube amp anode supply damaging the unit if it gets in the input RCA
lead.

I found I had Kikusui oscillator model ORC-27A made in Japan probably
in the 1950s. It has a radio twin gang tuning cap with each gang about
40pF to 400pF and with a resistance range of 20k to 20M for the 4
ranges, 20Hz to 20kHz, although it actually goes from 19 to 250 on the
dial which is accurate within 1% and not bad for a simple cheap thing
made in Japan at that time.
There is a 6AV6 input tube ans following 6AQ5 for the oscillator, then a
12AT7 for square wave, able to be switched in, and a final 6AQ5 CF
output buffer. Rectfier is a 6X4. Not many repairs had been done by
others over the years although the 10k output pot had been replaced. The
output tube is a 6AQ5 pentode wired as a triode CF with the 10k pot as
the dc carrying cathode load, and the wiper of the pot goes to a 20uF
electro and then to the output. I don't like the use of pots with dc
flow through them so I replaced this carbon track pot with 10mA dc with
a 12mAdc CCS using KSE340 between cathode and 0V. The output from the
cathode is now via 10uF rated for 450V to a 10k log pot for the high
output range level taken from the wiper to an RCA socket instead of the
old 4mm banana socket that was there originally. Then I have a 1k log
pot in series with the bottom of the 10k pot and 0V and wired so the low
-20dB approx level can be taken out to a second low level output RCA
socket. This arrangement gives nice fine adjustments of 2 outputs
regardless of the levels, and the 27A can sure pump out the voltage,
some 8Vrms even at 250kHz.

The CCS use means means the CF works with only the cap coupled load of
the two output series pots totaling 11k, so its THD is fairly low.

The 27A does not use a NFB lamp globe for the dynamic adjustment of the
NFB loop. Ah no, the Japs were somewhat brighter than the average nerd
in the US from say HP and they have what looks like a lamp bulb in what
looks like the wrong position in the NFB R network.
In the 27A, there is a thermistor mounted within a glass lamp bulb and
suspended between two very fine wires that look like a lamp filament.
This thermistor screws in to a socket just like a 6.3 dial light bulb,
and its resistance at room temp is 100kohms. At operation in the circuit
it has 8Vrms across it and by calculating voltage and currents in the
circuit I figured its R = 5k approx. As the V0 rises, the thermistor
current rises and heats it up and its resistance falls thus increasing
the amount of NFB. There is a 6AV6 input triode driving a 6AQ5 in
pentode with 4.7 dc anode load. The 6AV6 cathode R = 2k2, and the load
looking at the cathode with NFB is 2.5k effectively so that the NFB
network forms a load of 7.5k with 5k thermistor added. So with the 4k7
dc load the total ac & dc loads in parallel on the 6AQ5 = about 2.9k.
The thermistor gives far better NFB regulation and freedom from bounce
compared to a special lamp globe used in the bottom or cathode end of
the NFB network.

For square waves, The 27A has a 12AT7 set up to operate a grossly
overloaded cascaded pair, and the peak output voltage with square wave
is close to the same as the peak sine wave voltage, which is as it
should be. The square wave isn't too good at 200kHz, because the highest
F present is about 250kHz. But this is still the case if you set the F
to 5kHz, and then the square wave looks nice and square and its just
right to test tube audio amps for instability and overshoot/ringing
because the ring F will always be less than 250kHz usually.

The 27A also has a "complex" wave which is the original sine wave plus a
mains F added in nearly equal amplitude, so that gross intermodulation
artifacts could be seen if one needs to. An amp tested with such a two
tone signal need only have a following simple multi section CR HP filter
so that all you see on the CRO is the HF of interest. If the HF
amplitude becomes visibly amplitude modulated then you have about 2%
IMD, and you can make all your own conclusions about the performance of
the amp. IMD suddenly becomes very visible when the amp begins to clip,
and gross envelope distortions develop. 27A could easily be adjusted so
the mains F signal taken from one side of the HT winding and fed into
the 12AT7 while *not* being over driven is such that the ratio of HF to
LF signal = 1:4, and this complex signal is in fact the old standard two
tone signal for measuring IMD percentage. In music, bass signals
commonly average 4 times the amplitude as treble signals and thus bass
produce 16 times the load power as do mid&treble signals. Therefore we
should be acutely interested in seeing just what effect on mid&treble
tones is caused by the amp simultaneously amplifying large amounts of
bass energies.

The use of the 12AT7 to mix two signals so the output is a much smaller
voltage than the normal oscillator sine wave output of 8Vrms means that
the 12AT7 does not generate huge amounts of IMD itself. Usually, while
an amplifier operates under the clipping level, the IMD at the onset of
clipping is all we want to know and then we can estimate the maximum IMD
at low voltages proportionately to output voltage.
It is very rare that the IMD would be high in a tube amp at low levels
where class A action prevails, but with class AB there could be a peak
in IMD (and for THD) as the output level is reduced towards zero. Crook
OPTs don't help either.

Where one wants to find out what are the F and voltages of IMD artifacts
at all levels then much more sophisticated methods should be used and
with purer tones to begin with.

The 27A THD at 1kHz isn't extremely low because the open loop gain of
6AV6 plus 6AQ5 is only 57, and reduced by the NFB to about 3. And the
wien bridge positive FB network can pass considerable 2H and so we will
see maybe 0.15% in the output. The H above 2H are much more attenuated
by the WB network. If H above 2H do appear at the output then 1/3 of
them appear at the NFB input port without much of them also appearing at
the PFB port, so the NFB action then works to reduce the higher H (and
any LF noise below the fundamental Fo).

The opamp oscillator I made has far lower THD of less than 0.01% simply
because it has vastly more open loop gain than a pair of simple tubes,
and THD reduces according to the formula THD with FB = THD without FB /
( 1 + [gain x ß] ).

In WB oscillators, ß, the fraction of the output fed back is always
0.33. So the more gain, the less distortion.

For ordinary bandwidth and power measurements and general functions, the
20Hz to 200kHz range is enough for most folks.

But I like to see what happens **below** 20hz and its often down here
that am amp under test will shown a nasty peak in its response due to
the poor arrangement of stabilizing & phase compensation networks, and
it may oscillate at LF if left without a load connected. Hence the need
to test right down to 2Hz if possible. I often test up to 2MHz, but
usually 250kHz is enough.

Anyway, the Kikusui with a diode clamp to protect its CCS at the output
will be able to withstand occasional blats from the RCA lead
accidentally falling against 24V mains or onto anode voltages. Even a
lightning strike will be tolerated.

But anything with SS chips is always prone to gradual or sudden total
fuct-uppness from HV. That's how my Topward gene finally became too much
carbon to be able to fix anymore. I found I did have another SS function
gene and I have tried to repair it. It was made in Oz by BWD Instruments
P/L in Melbourne, their model no 160A, from about 1985 maybe. BWD seems
to have gone from the scene. It has oddball chips. One seems to be an AM
generator, P250 NPD 8303, and the other is a sine/square/triangle
generator FUA 7600 MQB 7906.

I searched these numbers in Google and drew a blank. Maybe the letters
on the chip, FUA stand for "*** youse all". The power voltages are
present but the chip is dead, nothing burnt anywhere else, and the 4
regs and 10 other discrete bjts all seem OK and the other several
opamps.

Bah, all that SS stuff just to make a bit of a signal to test a tube amp
is not really required by the serious audio experimenter.

When you look into the under chassis things in the Kikusui, you see
-40dB less parts than in the SS genies. And my old Topward had twice the
parts of the BWD!.

On my way around the web, I found a program for converting a PC into an
oscilloscope called 'winscope', and a free download as well.

http://madan.wordpress.com/2006/06/25/pc-based-osciloscope-winscope-251/

But the actual virtual oscilloscope screen size allowed when using it is
tiny, and a bigger display would be nice.

Does anyone know of a better ***free*** one?

Patrick Turner.

From what you wrote it sonds that the Kikusui generator was relatively
*smartly* designed. Apparently that engineer thought it through.
My impression with the japanese electronics of the 60's is that it was
neatly assembled, but brainlessly designed.

We must recall that the Japs were able to get aircraft airborne that
could do wondrous things in WW2, and all with far less weight and
complexity than the US planes that shot them down if they didn't get
shot down first.

A lot of crap came out of Japan in the 50s and 60s, heck, their country
had been ravaged by 1945 and its a miracle they were able to sell any
exports. The crap factor didn't matter because it was so much cheaper
than the Australian made crap. And lemme tellya, there was a lotta Oz
crap masquerading as fancy electronics.

One example is the Trio shortwave radio 9R-59DS. Imagine the circuit of the
LO (up to 30MHz) where the inductor is about 18cm away from the tuning
capacitor and these components are connected by an insulated sloppy wire,
tied in a loom with other conductors (!) to make it look tidier...

The crap floods in from China now; ppl love it, its so
CCCCHHHHEEEEAAAPPPP.

Always buy two items if made in China because you'll need one for
tommorrow while the other one is being repaired.

Patrick Turner.



Regards,
Alex
.

Loading