excuse the hand drawn circuit presentation of schematics
and large file size.
One channel shown with 2 x 6GW8, UL, classAB
Two triode phono stage NFB eq 1 x 12AX7, feedback RIAA eq
Three triode phono stage NFB eq 1 x 12AX7, feedback RIAA eq, buffer
Three triode phono stage Passive eq 1.5 x 12AX7, passive RIAA, buffer
Phono Amp PSU Schematic Low Power supply for 2 and 3 tube preamps
10 tube Preamp April 2000 5 x twin triodes per channel
1 x j-fet 2SK369 simple pre-preamp A test circuit showing THD for a single 2SK369 j-fet
Test filter, Reverse RIAA eq A simple test filter with discrete RC components
there was once what was called the Playmaster Amplifier series
which were published in the old Electronics Australia magazine, long since closed
down. This circuit is a development of the original and this was prepared for a client
who had a sample of one of these Playmaster amps on a very poorly home made
chassis, with lots of what must have been 1930's resistors and capacitors taken
from old radios. Amazingly, it worked a bit, not very well, but was dangerous,
and was a mess to look at.
transformers were put onto a new stereo chassis, made large
for a source select switch, volume and balance control. I included simple 12AX7 triode
phono stages in case he ever got a TT. It packs more punch than any other ten
watter that's ever been heard, and shames Krells, and big clunky stuff like that.
No pictures available, the project was supplied to a happy client in 1998,
well before I could type, or produce a website.
transistor voltage regulator used to prevent LF instability when
stage is used.
The circuit is
very simple, and uses a concertina driver stage for the output
and a triode gain stage for input stage. Less than 20 Db of Global NFB is employed.
Three Triode phono amp, Passive RIAA eq.
The above phono preamp schematic is designed
for ease of construction, minimize
component numbers, allow a fairly accurate RIAA eq with the aid of a reverse RIAA
network, make use of a well known common tube type, have low distortion, a warm sound,
high gain, and a low output impedance for somebody who may want to use it to
transfer vinyl to CD or run long cables to an integrated preamp or power amp well
away from their listening chair.
What more could you want?
The calculations for passive RIAA eq are
simpler than trying to use NFB
triode distortion and noise is negligible at such tiny voltages involved with a triode phono
stage, then why use any NFB? Well some people would say they want extremely low
distortion, lower than lowish. I leave the experimenter to decide which sounds the best.
Without NFB, some care has to be taken with
the power supply, and I have shown above
an absolute minimum shunt regulation arrangement using a 10k resistor, 1 x 470 uF cap,
and 3 x 75 volt x 5 watt rated zener diodes, and I have assumed you will have a +320 volt
Making assumptions never got me anywhere in
life, so for a basic power supply, go to
Phono Amp PSU Schematic further down this page.
The power supply shown should suffice for such
a basic circuit, with some room for
use with tubes which draw a little more anode current. Since we only want 5.2 mA
of anode current supply for 3 x 12AX7 tubes for two channels, and then need 6 mA of
current flow in the shunt regulator zener diodes, the total anode power will be only
11.2 mA x 320Vdc = 3.6 Watts, using a 240 volt secondary winding on a power transformer.
Such a power tranny would at least also have a heater winding to provide 6.3 volts x 0.9 amps
= 5.67 watts, so the total power use is only 9.27 watts. In practice, we would purchase a
30 VA transformer, because they are likely to be more rugged. It could have a 120 volt sec,
which could be used in a doubler circuit to make the +320 and the heater winding of at least
6.3v could be used in a doubler as shown to make +16 volts at 0.45 amps, which can be
filtered down with R and C to provide a noise free DC heater supply. A12V shunt regulator
zener diode and series diode is used to make to make 12.6Vdc and they would draw 0.3 amps
The schematic calls for the heater supply to
be biased at +50Vdc, to relieve the pressure on
insulation between the cathode and heater in the cathode follower, only rated for 90V.
This is easily done by making a voltage divider in the power supply voltage output bypassed
to ground via the 100 uF, or else just taking a 100k from the top of the first 50V zener diode
to the floating heater circuit, and bypassing to ground with 100uF.
Where heater voltage is 12.6Vdc, the lower side of the 12.6V should be at +50V.
Bias isn't critical, one could have bias between +50V and +70V.
The limit for
low distortion output voltage production from a cathode follower
into a low RL
is the cut off of current in the tube. In this case when the grid voltage goes negative enough
to cut off the tube current, the 10k ac RL and the 100k follower dc RL form a divider, and
the peak negative going voltage is limited to [ 10k / ( 100k + 10k ) ] x Ek. Ek is the cathode
idle Vdc which will be about 90Vdc, hence the 10k peak swing is [ 10 / 110 ] x 90 = 8.2 volts.
In fact even if we had a 5k0 RL, we could get 4.3 peak volts.
Under normal use, the output voltage won't be
this high, as the average 2mV
input from a MM
cartridge will produce an average 0.8 volts output. We could even afford to use a lower gain
second gain tube, such as a 12AY7 or 12AT7. If we wanted more voltage swing into a low
value ac RL, we would be better off using 1/2 12AU7 for the cathode follower, so we could
have an idle current of 4mA, and the dc RL would be 22k, and with a 5k ac RL, we could get
a peak voltage swing of 18V.
I would add that the gain tube before the
cathode follower could easily produce up
to about 45Vrms of signal output, so you won't ever overload such a preamp with a phono signal.
The output impedance of the follower is about
700 ohms, so cable capacitance and low input
impedance will not result in high frequency losses.
Triode Phono amp, RIAA eq with NFB
The above phono preamp schematic is virtually
as simple as the two triode phono stage
with NFB eq but has an additional cathode follower buffer output. It should sound well.
It is unknown if this topology sounds any
better or worse than the passive eq method, and
I leave arguments about the sound to others, but my experience is that the NFB eq method
compared to the passive eq method can be very good indeed, if the values of the FB network
are chosen carefully, and the network doesn't impose too low a load on the tubes driving it,
even at HF. The values of R&C in the RIAA feedback network should be trimmed with the
aid of a reverse RIAA network used to equalize the input signal in the same way as a a record
The final buffer stage is where the feedback take off point is for the NFB
network. Any error
signal created by the following cable or impedance will be subject to NFB correction and will
have less effect than with any other normal follower stage output stage.
A normal 1/2 12AU7 cathode follower as used in the above 3 triode amp with passive RIAA eq will
have Rout = 600 ohms approx. The Rout will be further reduced by the loop NFB for the RIAA eq.
The amount applied varies with frequency and is only about 6dB at 10Hz so Rout at 10Hz at the
cathode of the 12AU7 will be about 300ohms but at 1kHz it will be 30 ohms and even much lower
There is a slight danger that with so much NFB that HF oscillations could be a problem if cable
capacitance was high so to safeguard the amp against any possibility of oscillation a series
R of 270 ohms is connected between the cathode and output terminal.
The 5k horizontal R in the NFB network imposes
a HF time constant well above 20
kHz to prevent
the likelihood of oscillations at HF when the cathode of the follower becomes virtually directly
connected to the input tube cathode at HF, via the capacitors in the NFB network.
We don't need the NFB to keep increasing above 30kHz, and we can level off the amount applied
with this trimmer resistor. It is unlikely that any recordings have had any significant HF content added
above 25 kHz, ( unless the recording was for the short lived experiment with quadrophonic sound
using a multiplexed signal containing a 50 kHz modulated carrier tried back in the 1970s. )
Indeed the cutting head amps would have filters to prevent HF oscillations,
so if the supersonics were not emphasized, there is no need to de-emphasize what was never
The above amp
will have less gain than the 3 triode passive preamp, since the
FB is active at
LF, but still the gain will be more than adequate for all concerned.
supply schematic at Phono Amp PSU Schematic.
Note, the PSU shows simple zener diode shunt regulation of the B+. Clever people will
use a series pass element voltage regulator to avoid any wasted B+ Idc current.
I do not have strict rules for power supplies,
other than insisting the voltage they supply be from
a substantially low impedance at low frequencies, and free of noise from the rectifiers.
I do not subscribe to the school of thought which stipulates that paper in oil caps must be used,
along with chokes. Use them if you must, but modern high value electrolytic capacitors allow
RC filters to be used.
For the DIY constructor, feel free to build your supplies any way you wish but remember phono
stage amps need well filtered supplies.
I have done supplies with tube rectifiers, tube regulators, and they can all be made to work effectively
and reliably at such low levels of power. I quite enjoy seeing a nicely tubed power supply for a preamp,
and I have built a couple for customers on separate chassis.
But in power amps I build, the use of tube rectifiers would cause excessive inefficiencies, and never
improve the sound quality.
It can't get much simpler than this.
You will never be able to buy suitable PT with a 240Vac sec for the HT for B+ of +300Vdc
for a tube preamp from your local electronics parts dealers except EVATCO in Qld,
who may stock a suitable Hammond PT imported from Canada.
Other local electronics dealers with many stores around Oz are are RScomponents, Jaycar,
Dick Smith, wescomponents.
Jaycar do have some 1.4VA to 60VA PTs which have 240Vac primary
and up to 30Vac secondaries with multiple taps to suit most low voltage needs for most
people's solid state projects.
See the Jaycar online catalog page for ac - ac transformers at
Then scroll down to the 240V:15V transformers,
catalog number MM2004,
240V : 15V at 2 amps max, 30VA rating. Sec taps are at 6V, 9V, 12V and 15V. I've always found the sec
voltages are a little higher.
1. Buy TWO
of these trannies, and each is priced at $18.95 ( 2012 price
Use one of them with normal mains 240V primary. Its becomes
The secondary is not grounded, as it will be biased at about +60Vdc. From the Com to 9Vac
tap you can produce a rectified voltage of about +12Vdc using one 1N5408 diode and a
4,700uF cap. Then use an R&C filter to get a wanted 6.3Vdc at up to 2Amps dc for the
cathode filaments. If you have 4 x 12AX7, you need 1.2Amps, so the RC filter will be 4.7
ohms rated for 10W, plus 2 x 4,700 caps rated for 25Vdc. This gives 50Hz ripple of about
85mV, low enough. The RC filter could two RC sections each 2.2r + 4,700uF, and ripple
slightly lower at about 60mV.
Use the second transformer, it is T2. It is connected "back to
front" with its 15Vac winding
connected right across the T1 15Vac winding of the first transformer. There will be approximately
240Vac produced across the T2 240Vac winding which can be used to produce up to about
+320Vdc at 15mAdc. This 240V winding is doubly isolated from the mains.
Don't allow the B+ Idc to be higher than 15mA, its about 5 Watts of power.
CHECK TOTAL POWER DRAWN. The first 30VA tranny must
not draw more than 30VA from mains.
There MUST be a mains fuse, I suggest 0.25Amp,
slow blow type 2AG.
There must be a 3 amp slow fuse in filament
There must be a 1 amp slow fuse in T2 15Vac
Secondary of T1 loads can be:-
filaments = 9Vac at 1.7A = 16VA
Max load of T2 sec = 240Vac x 20mAac gives 320Vdc x 14mAdc = 5VA, then add 10% losses.
load at T1 sec = 5.5VA, so current = 0.37Aac.
Total maximum current in T1 sec = 1.7 + 0.37 = 2.07A which is OK as winding has 2A rating.
VA used = filaments, 16VA, HT = 5.5VA,
TOTAL VA = 21.5VA which is less than the 30VA rating, so OK.
you won't make clouds of smoke as the fuses should protect you
your mistakes, which I confidently predict that you will make.
Get a tech to check out your work before turning anything on.
tube Preamp, April 2000
of one preamp channel was used in my own "prototype" preamp,
shown in the
picture at the Preamps1 page, and was firmed up in April 2000. It is a good performer with
phono provision for moving magnet phono replay.
From left to
right, the first tube is a 12AX7 mu-follower with gain = 88, (
39 dB ),with a low Rout
to feed a passive RIAA filter network for vinyl use. This suits MM cartridges with outputs down
to about 1.0 mV. Second tube is a 12AT7 SRPP to provide extra gain of 40 times ( 32 dB )
to complete the phono stage, to give a low Rout.
The gain of the phono stage is 350 times at 1 kHz, or 51 dB.
stage is a SRPP stage with a 12AX7 to provide deletable tone
control for high and low
frequency eq. It uses a Baxandal network in a variable shunt feedback network, and has a gain
of exactly unity, or 0.0dB.
This allows for seamless insertion into a signal path for accurate comparison of with or without
tone control. Although tone controls have lost favor amongst audiophiles, they are very useful
for checking a speaker's basic bass, mid, treble balance,
and correcting some bad old recordings, ( and some lousy new ones! )
stage is another 12AT7 SRPP stage which is also deletable,
to provide a gain of 8 times
( 18 dB ), for the line level inputs. There is shunt feedback used to reduce the 12AT7 gain to a
sensible level, and to reduce the Rout to feed the balance controls and a pair of dual ganged
gain pots, seen in series, which allows the tube to see a favorable loading.
The effective total line stage gain is thus 4.5 times, (13 dB).
allow or disallow the use of tone control or line stages. In
these stages are rarely used.
stage is a pair of cathode followers using a 12AU7 which buffers
from effects of combined losses from interconnect cables and any following power amp
input circuit. Gain is about 0.92 times, or -0.7 dB.
The output impedance of the preamp is thus only 600 ohms, and it will power any loads
down to a few k ohms, and the preamp will be compatible with any SS equipment.
Both left and
right channels have two separately variable pairs of outputs so
systems can be set up with different sensitivity power amplifiers to allow their comparison at
an exactly similar level of signal output.
It would be easy to use such a preamp to provide filtered outputs allowing the use bi-amping.
stage input suits moving magnet type of cartridges. The use of
cartridges would require a pre-preamp, or a step up transformer, or an additional amp stage
built into the circuit.
The 'Preamps1 page' gives recently developed details for cascode circuitry in the phono stage
using a j-fet for low level MC inputs.
The tube line up includes
Phono, 2 x 12AX7, 2 x 12AT7, Line gain, 2 x 12AT7, Tone control, 2 x 12AX7, Buffer outputs, 2 x 12AU7.
anode current supply required for the 20 triodes involved will
be approximately 25 mA.
I use a transformer capable of 50 mA. The B+ supply needs to be about +340v to +370v, so about
8 watts should be allowed. In my prototype, I used a separate potted B+ transformer rated for 30 VA.
supply for 20 triodes requires 12.6 volts at 1.5 amps, so about
19 watts is needed,
but I use a 30VA transformer, also potted.
DIY folks should allow for future changes to signal tubes such as the 6CG7, or other octal tubes,
which need twice the heater currents as the above listed tubes, so in fact the heater current
rating of the transformer should be for 3 amps.
The two transformers of my amp are mounted inside a mild steel on the chassis which is
500 mm long. There is thus sufficient distance between the sensitive phono amp tubes,
and the power supply.
If the one
power transformer is chosen, it should be rated at 50 VA, and be
made using a Bmax of
no more than 0.8, have GOSS cores, and should be potted.
I used all DC
for the heaters, and I had a LM350 regulator, which is a TO3
device to remove
the ripple from the heater supplies after the rectifier. There were no chokes in the preamp.
The transformer heater windings, rectifiers, and regulator are not connected to 0V directly,
but are "biased" up at +70v to allow the use of the SRPP and cathode followers without exceeding
the 90 volt ratings for the heater to cathode insulation. Then to keep hum out from stray transformer
capacitance from the mains primary, the whole heater is bypassed to 0V with 100 uF.
There was no hum in my prototype amp.
There also has
been attention paid to 0V paths in the amp, and the schematic
the optimum 0V routing. The chassis is connected directly to the earth wire coming from the
wall socket, but the 0V line is only connected to the chassis via a 5 watt10 ohm resistor.
Thus it is difficult to get earth loop hum problems, despite the un-balanced circuitry used
pre-preamp using 2SK369 j-fet.
This test schematic could be used for a pre-preamp ahead of a normal MM phono
amp for MC use. But it was used to get some idea of how linear a high gm j-fet such
as the 2SK369 (or 2SK147) really is, and the answer is that above 0.1Vrms output the
THD becomes excessive even with nearly 12dB of NFB applied. The THD can be reduced
by near doubling of Id to 5mA, when the gm rises to 40mA/V thus increasing the open loop
gain to 88, and thus increasing the amount of applied NFB. The THD reduction is only
marginal and compared to a triode the j-fet has about 25 times more THD for
the same output voltage.
For MC use, I
recommend the drain RL be reduced to 1.2k so Id would be raised
5mA. The same 100 ohms for Rs could be retained and gain overall would then be about
9.4x which is enough to raise the 0.3mV from a typical MC at 1 kHz to 2.9mV, with THD
less than approx 0.005%. The result will not be as quiet as a properly made cascode circuit,
but may be quite sufficient.
Output resistance will be determined by the RL of 1.2k and noise from this R should be just
66dB unweighted below the signal level of 2.9mV. The fet will not have troublesome
microphony that a tube would have in this application.
fets like bjts are very prone to stray magnetic fields, along
with any input wiring.
Magnetic screening for the enclosure, well filtered rail supplies and well routed earth paths
and short leads around above amplifiers all all essential for low hum levels.
Such pre-preamps should be set up well away from any other equipment with a power
transformer within including innocent looking cd players and tuners and tape decks.
I tested small
signal SE BJTs in the same application and found them all to
give little better
THD measurements and the spectral content of the thd contained greater amounts of odd
numbered harmonics. Noise was worse. And for lowest noise in a typical small TO92
package bjt the collector current is usually about 0.2mA, so the use of 5 parallel bjts is needed
to get Ic up to 1mA, and then the maximum output voltage is limited to 0.6mA rms, or 0.6V rms for
a 1k load.
BJT Input impedance is low and the biasing is more difficult and
I just refuse to build any pre-preamps with bjts.
Test filter, Reverse RIAA eq
This filter is
a great way to test the frequency response of a phono amp. The
all frequencies above very low frequencies in the same manner used during the record
cutting process. The filter profile is equal to the RIAA profile used for all LP records.
The filter can
be easily assembled on a small piece of plastic board, and fixed
to the inside
of a metal container, so that it remains well shielded, and fitted with RCA input and
generator which has at least 100 ohms output impedance, or lower
connected to the filter input, and the amp to be tested is connected to the filter output.
oscilloscope and wide bandwidth voltmeter is placed at the
output of the phono amp
being tested, and the signal gene set for 1 kHz, and the level adjusted for a 1 Vrms output at
the amp. If the RIAA filter components in the amp are of the correct value, and have not
drifted over time, then the amplitude of all frequencies between 100 Hz and 10 kHz should
appear to be equal, with a roll off of only 2-3 dB at 20 Hz, and 20 kHz.
signal should be able to be changed to 'square wave' and the
output wave when
the gene frequency is 1 kHz should appear as a substantially square wave form, without
peaks or troughs along its horizontal parts. This will show that the eq is phase correct,
and confirms the accuracy of the amplitude response.
Using a filter
like this is far easier than laboriously trying to measure the
amplitudes of test signals accurately set to spot frequencies along the audio band.
And any errors in the amplitude of the signal gene can be neglected if the square wave
test is used, although I do like to use the sine wave test. A signal gene with a swept
frequency facility will also show the response well.
There is a
more scientific approach to reverse RIAA filters at
To Index Page.