The only phono stage I had was with the EL34 Integrated amplifier. I needed a phono capability to use with my 845 SE. My friend (and dedicated para-feed DHT enthusiast) Walter Clay kindly donated many Western Electric 416 planar triodes to the 'cause'.
416 information and implementation commentary at bottom of page.
Mk 1 Version
The phono stage was built in a case 'recycled' from a redundant Adcom 5400 power amplifier. It was a surprisingly tight fit! I would have preferred to put the PSU in a separate enclosure but I had nothing suitable and was feeling lazy about building a case especially..... I was trying to use as many components from my parts bin as possible. The Hammond transformer (B+ and rectifier supply) is really not suitable, these are very poor quality transformers which emit far too much EMI. I had to resort to unbolting the Hammond and turning the unit until the hum was minimised. I wound up with 2mVp-p with the cartridge connected which is a bit better than 60dB on 1Vrms out. I am quite pleased with this result given that the PSU is in the same case. The heater transformer was another parts bin item. It is actually quite good. The electrostatic and electromagnetic fields are well contained by core and winding shields. It is much better quality than the appearance would suggest. The massive heat sink has the heater rectifier and a power Darlington transistor used as a pass element, controlled by an LM317. The current is 4.2A at 6.3V. NOTE: The 416 is susceptible to heater turn on surge failure. I prevented this by incorporating a time constant in the LM317 'adjust' pin circuit such that the heater voltage takes some 30 seconds to reach full voltage.
The topology is very simple: 416B as a common cathode input, Lipshitz passive RIAA Eq and 5687 SRPP output. You might notice that the B+ filter choke is in the negative side. The reason for this is to ensure that the reservoir capacitor repetitive charging currents cannot flow in the ground circuit. The series pass element in the positive side acts to prevent such currents flowing in the positive side.
I used a pair of double 9-pin turret boards from Antique Electronic. The 416B is a perfect fit into the hole for a 9 pin socket. I am very pleased by the sound quality.
Mk1a Version
The MK1 now belongs to a friend. After a lot of use we decided that it would be worth some development to further explore the potential of the design and particularly, of the 416 triode. In particular, the soundstage was a little limited. Furthermore, we both feel that the ECC99 is a better sounding tube than the venerable 5687 and that the extra gain would be useful. I felt that the soundstage limitation probably stems from the use of a single regulator and so the regulator was re-built as a double unit, the EL34 pass tube being replaced with a 6H30. The front end (416) of each channel is decoupled from each output stage (ECC99) using a 10H choke and 25µF oil cap. I took a lesson from the MK11 and re-worked the RIAA section to take advantage of the low Rp of the 416 to use a much smaller value of R1 (input series resistor) for the Lipshitz equalisation network. In the MK11, this feature resulted in much faster dynamics. It also reduces the equalisation loss from around -6dB to around -1dB (@1kHz). The resulting increased sensitivity (58dB to 66dB) prompted me to play (again) with the (Hammond) B+ transformer orientation. I ended up mounting it sideways on a wedge as close to the front of the case as possible. Signal to hum ratio improved by around 12dB. Hum is now only barely audible with the level on his (horn) system full up, at which point the sound level is way too loud. Altogether most satisfactory. Incidentally, I did try a potted military transformer. It was worse than the (200 series) Hammond. The 416 anode studs get very hot (even though they are run quite gently) and the solder connection to the studs seemed to be degrading with time so I made a pair of copper 'fan-like' coolers (see pics below).
The results of the modifications are extremely rewarding. Very quiet with a hugely improved soundstage and sense of location. The sound is rich yet detailed and 'fast'. I prefer it to the MK11 and so there will very likely be a MK11a sometime.... By the way, the cartridge I use is a Benz micro L2. This is my first true 'high end' cartridge. I shudder at the prices yet having put so much effort into phono-stage development, to be overly parsimonious with the cartridge is perhaps a little self defeating. I located an almost new unit on Audiogon at $550 which is less than half retail. Was it worth it? Oh yes, very much so.
View
showing (left) power transformer mounted on a wedge, (centre) B+ regulators and
(right) B+ input filter, also heater transformer.
View
showing overall layout. The decoupling filters from each front end are at
top left while the large heatsink is for the heater supply series pass
transistor. This still uses a LM317 to control a 2N3055. The LM317
has a long time-constant connected to the 'adj' pin to obtain a slow ramp-up to
protect the 416 heater which is susceptible to current surge.
Underside
of each channel, 416 at top, Eq in middle and ECC99 at bottom.
Top
view of phono boards showing specifically, the 416 anode coolers. The
Sowter (1:10) input transformers are also visible.
Closer
view of coolers.
Mk11 Version
A friend of mine liked the Mk1 version so much that he let me swap it for a Sota Star Sapphire turntable with the vacuum platter and 'electronic flywheel' Not a bad deal, however, this left me with a turntable and without a phono stage! So, enter the Mk11:
The basic topology is the same with three differences:
1/ Each stage is fed from a separate shunt regulator,
The shunt
regulators are of the active type, using a triode shunt element control by a
triode error amplifier. The significant difference between my design and
others is that I have used a -150V negative rail, (stabilised using an 0A2) to permit the error amplifier,
X2 to
operate 'below ground' as it were. The error voltage is fed back to the
cathode of the error amplifier while an 0G3 (85A2) is used to provide a
reference voltage to the grid of each error amplifier. This
approach permitted me to connect the cathode of the shunt triode, X1 directly to
ground. That is; the shunt triode is connected directly across the B+ to
ground, with
no other devices at-all in the shunt path. This design permits an
exceptionally wide and flat phase response. It is almost impossible to
design a passive power supply that meets this requirement. The idea
was to see how it would sound with no bypass caps across the B+ supplies.
If you study the picture of the phono stage, you will notice a bank of PIO
condensers in the centre. These are bypass capacitors for each of the 4
stages, arranged to be switch-able in and out without creating a 'bump' I
cannot hear any difference between the sound of the unbypassed shunt regulators
and that of the bypassed shunt regulators, so I think that my design concept has
met my goals.
2/ All stages are choke loaded,
3/ I decided to take better advantage of the low plate resistance of the 416 to reduce the RIAA series resistor from the common 270k region to 49k. The idea was an attempt to bring the dynamics of the R-C passive RIAA network closer to that of an inductive RIAA Eq network. Comparing the two designs, the MK1 is a little sweeter, laid back compared to the MK11 which is quicker with more resolution and definition - perhaps too much, it does bring out recording issues such as tape hiss: I have been asked "if LPs are so good, how come you cannot hear the tape hiss?" Well, on some LPs this is now clearly audible. I am using a Benz Micro LP2 cartridge.
A further difference between the Mk1 and the Mk11 is that I used heater current regulation for the 416s in the Mk11 to prevent switch on surge.
416 Information
The pin-out of the 416B is:
Looking on the base, put the location pin at 12 O'clock; the cathode is at 3,
heater at 6, heater at 10. Heater is 6.3V at 1.1A. (Be careful, at least
one of the variants is 6.1V.) The top pin is the anode and the threaded
ring is the grid. I soldered directly to make connections. There
were variants A through D, D having the greatest maximum dissipation, 6W.
At such high dissipations, I think that a heatsink may have been necessary for
the grid.
Typical operation (for the B) was 185-200V with Vg +1.5V and Ia 20-25mA
(operation was telecom mux at 4000GHz, some 19,000 simultaneous calls). I
suspect that the configuration was to apply bias voltage to the grid and the
input to the cathode thereby using the grid as a Miller screen.
I found that the units I have were used and most likely removed at the end of
normal service life. This means that the emission is low and thus the Ra
high.
One of the characteristics of high mu triodes is the tendency to develop
'contact potential' in between the grid and cathode which will self bias the
tube if the anode current is limited by a largish load resistance. I used
38.1k (this seems overly large but the Ra of the units I have appeared to be
nearer 10k than the spec 3.9k) and a supply voltage of 300V. The bias
settles at around 5mA. I get a gain around 44dB. I included a 60.4R
cathode resistor bypassed with a 1500u Blackgate to provide a degree of cathode
bias. This has a switch to allow me to switch the resistor in and out of
circuit (short the resistor). Listening, I found that I could hear
'further in' to the music with the cathode resistor in circuit.
With a normal moving-magnet input grid resistor of 47k, a small amount of grid
current flows, reducing the bias (the contact potential increases reducing Ia).
With the low dcr of a cartridge, the grid will be (nominally) at ground
potential and in this case, I believe that the grid current tends to zero but I
cannot properly substantiate this other than hours of use do not seem to have
affected my cartridge (Sumiko high-output MC). I intend to include an
input transformer (such as Sowter 6495). Being mu-metal cored, an LF
extension test will reveal if there is grid current or not. (Since
first writing this, I have incorporated the 6495 transformers, the LF extension is not
affected thus the grid current does indeed seem to tend towards
nil.) The use of the input transformer permits operation with
either MM or MC cartridges. In the case of a MM cartridge, the input is
applied across the secondary which serves as a grid choke. To facilitate
this, I fitted RCA sockets across both the primary and secondary of each
transformer. For MC operation, the secondary sockets are used to plug in
load resistors. E.G. the transformer is 1:10 so if I need a 1k load, I put
100k across the secondary. Similarly, for MM cartridges connected across
the secondary, I use the primary sockets to plug in 470 ohms to reflect a 47k
load to the secondary. Sonically, the
416 seems to be a terrific choice for low-level phono signals permitting
absolutely astounding resolution and I am very pleased with it.
However, I did have to select units for gain, current and noise. There was
a great deal of variation in the dozen units I have. THE CAVEAT IS:
They were used, unused units may/should be much better. Since then I have obtained some new
416Cs and they are, indeed, much better.