LCR Phono preamplifier Part 3

...SOLD...

Power supply

A good sounding preamp or amp always means a well calculated power supply, this is the circuit heart. Too often neglected or subject to approximations because calculation is time consuming and most hobbyist do not evaluate the real impact on music restitution. A supply is NOT an oversized transformer plus a huge amount of filtering capacitors.

Tech corner with useful reminders about time constant and ripple attenuation.

1- ADSR and audio signal
Picture from the Net. Courtesy of Landr Journal

To understand the great importance of time constant we first need basic information about audio signals. These are totally different of the sin waves we use to measure power, distortion or phase in our amp and preamp.
An audio signal is a very complex association of a fundamental and many harmonics of different amplitude and duration. For our supply it is a succession of non-symmetrical transients that form the signal envelope or ADSR of music. To make it short we must deal with attack transients that are in the 15 to 20 milliseconds for the fastest instruments (piano, guitar, drums), which means the supply must recover in that interval of time before the next attack.
The first LC or RC cell time constant should not exceed that value.
In the real life it can't be that precise due to the components tolerance, especially electrolytic capacitors, thus it is a safe practice to calculate a first TC not more than 10 milliseconds (but not under 5 milliseconds or sound becomes harsh).
Note: For proper transient response, TC should increase after each decoupling cell.

2- Ripple
For this preamp I choose a CLC filter.
Ripple with an input capacitor depends upon C, total current in circuit and mains frequency. It can be appreciated by the relation

Vr = Itot / C x f

Itot Total current in circuit (Ampere), C input capacitor (Farad), f ripple frequency (twice the mains with a full wave rectifier)

Note: for a choke input filter, ripple is

Vr = Vout / 6π² f² LC √2

3- Smoothing cells
LC and RC networks are voltage dividers for the AC current that comes from the rectifier. They are necessary to smooth out the ripple.

Let's calculate the preamp supply step by step.

I have a 240V / 320V_0_320V power transformer followed by a GZ32 rectifier and a 24µF C₁ capacitor in a circuit drawing 60mA @ 375V.

Vr = Itot / C x f

Vr₁ will be 60 10^-3 / 24 10^-6 x 100 = 25V or 6,6%

4- Time constant depends upon C and Rs the total series resistance. To calculate Rs we must:

1- measure primary and half secondary (for a full wave rectifier) transformer winding resistance (respectively 12Ω and 120Ω) to determine the total equivalent resistance Req
2- calculate transformer voltage ratio n = Vs/Ve = 320/240 = 1,333
3- calculate Req = Rsec + n² Rprim = 120 + (1,777 x 12) = 140Ω
4- determine rectifier internal resistance Ri (~260Ω)            O.H. Schade method
5- calculate Rs = Req + Ri = 400Ω
6- calculate TC = Rs x C₁ = 400 x 24 10^-6 = 10 milliseconds

Note: a GZ34 rectifier with its lower internal resistance (~100Ω) will give a
6 milliseconds TC, on the other hand a GZ37 (~500Ω) something around 18 milliseconds. This to say that changing the rectifier will affect both TC ratio and DC voltage, reason why sound also changes (dry and bright with a GZ34, warm and mellow with a GZ37). Too many people just make a plug, play and listen test without considering this factor and recalculating the whole supply as it should be done.

a first cell (L = 10H/160Ω and C₂ = 100µF) makes a voltage divider where

Vr₂ = Vr₁ Z_C2 / Z_L + Z_C2

      Z_C2 = 1 / 2 π f C₂ (~ 16Ω @ 100Hz)             Z_L = 2 π f L (~ 6,3KΩ @ 100Hz)

      Vr₂ = 25 x 16 / 6300 = 63,5mV / -75dB and TC = 16 milliseconds

and so on... until I reach at least -150dB ripple rejection and proper TC to feed my preamp.

PSU schematic



Parts and construction

Quality parts are chosen for their reliability and sound qualities. In addition to the MP/MKV and F&T electrolytic I usually favor, the filtering bank will be fitted with excellent vintage SAFCO-TREVOUX 15µF PIO.

These military caps are incredibly rugged, well made and heavy. They are new and fresh as if they just came out of the factory, still in their original 1959 packing. These caps have one of the lowest loss tangent I measured on paper in oil capacitor and give a smooth and clear sound. This upgrade is at the price of a cumbersome supply weighing about 20 lbs (the hermetically sealed transformer could power a 6V6 push pull), hence the two handles to ease manipulation.
Be very careful during wiring, especially pins selection, ground returns and heaters. Remember that input signal is in the same magnitude than heater/cathode leakage noise and it can be a nightmare to get rid of a low frequency buzz.

most of the wiring is solid

Power supply with sub chassis to ease construction

Tubes I used

EC88
I have Siemens (Germany) and TFK (Philips Heerlen plant) on hands. See part 2
Tests were conducted with Philips E86C as output tubes.
To say the truth both 88's are great sounding tubes with very detailed music, lot of life and great sound stage. Maybe I will keep the Philips Heerlen that appear to be the quietest tubes... by a very slight margin.

EC86
Replaced the Philips E86C (Mullard made) by TFK EC806S and I entered in a different world. The Telefunken's are audibly better by any means. Deep, wide, alive. They are THE tubes to go! Incredible punch with sweetness. Accuracy is astounding. With the Philips the overall sound is very pleasant but by comparison music appears slightly out of focus, like behind a curtain. Amazingly had the same feeling with the E80CC Philips vs Tungsram in the SRPP preamp.
So, I made another test with the 88's and the best association is E88C Heerlen + EC806S TFK, no contest. Perfect balance, articulation and clarity.





Some vinyls for this test, out of my 5000+ LP's collection, mostly jazz

LCR Phono preamplifier Part 2

...SOLD...

In the last post I described a constant current coupled stage with a gain of 53 and low output impedance.
For my project I need a gain of 2800 to compensate the LCR setup insertion loss (0,158_see part 1), means about 52 (2800/53) for the first stage.

Only a high Gm triode or pentode will give such an amplification on the low Tamura load with a wide frequency response. Not so many candidates when you must find tubes with inherent low noise & distortion. Among usable triodes my choice went to the E88C. complete data here.

Almost unknown, this tube is a dream when you have to deal with very small signals. Primarily developed for UHF use where high gain and low noise are mandatory, they share these qualities with a few others tubes like the EC8020 and the EC8010, but have one of the lowest distortion & noise figure you can expect from a triode. This emission noise is hard to measure. With a differential amplifier coupled to my HP 3561A dynamic analyzer I got a -107dB (4,5µV) noise on an open grid tube.

Means the micro details from the vinyls won't be hidden by thermionic emission. Moreover, the very low Miller capacitance (1,2pF, about half the EC86 one) will insure extended highs.

Telefunken (Philips made) & Siemens
.




Its medium ρ makes this tube an excellent candidate for transformer coupling. With a 20K load @ 10mA it works in a very linear region and gain is high (~51).
It is a super quality frame grid tube with outstanding construction whatever the manufacturer.





Preamp overall gain will be 51 x 53 x 0,158 = 427 or +52,5dB.
The 5mV out of my MC transformer will produce 2,10 volts, just as expected.

Full schematic

Note: LCR network can be either feed or loaded by 600 ohm, in any case you have to remove one 600 ohm resistor ( R4 or R12 ), this ensures network to work properly.

Alternate driver

For those interested I performed some tests with a C3g pentode and got good results, way better than in triode mode, the E86C balancing its natural dry and analytical sound.
I finally preferred the E88C because final gain was too high but this can be favorable with very low output carts. The G3g will even give lower distortion than the E88C with no Miller's effect. To get it usable, cathode is unbypassed and gain is reduced by a factor of ϒ where

                                                ϒ = 1 / 1+Gm x Rk          (~ 0,29)

the first stage gain will be       Gϒ = Gm x RL x ϒ          (~ 82)

and the preamp overall one 82 x 53 x 0,158 = 686 or +56,5dB.



A 5787 voltage reference tube makes an unusual feature to insure steady Vg2 in place of the decoupling capacitor. It is a minimalist shunt regulator that draws 3 time the grid current. In that way, with an unbypassed resistor, any cathode voltage change due to grid input signal won't affect voltage setup and Vg2 will remain constant.
I set the current to 6,5mA, but any value between 5-25ma may be used taking in consideration that regulator noise is proportional to current flow and may vary from one maker to another. Had best results with Cifté (Mazda) and Raytheon WA series.

Next time, power supply and tests

LCR Phono preamplifier Part 1

...SOLD...

Building a phono preamp is a difficult exercise in the way we have to deal with very small signals that must be greatly amplified without any loss and/or unwanted noise.
If, like me, you enjoy the qualities of moving coil cartridges to extract the quintessential parts from your beloved vinyl’s the expected signal out of a quality transformer or other step up device is roughly in the 5mV range. This signal must be amplified to about 2V to accommodate the usual low gain line amplifier. This means an overall amplification of 360/400 or +50/52dB.

To reach that goal most of the schematics found in literature use 3 or 4 stages.
In my personal conviction ʺ the fewer, the better ʺ I expect this high gain with 2 stages to minimize sonic coloration from the tubes to be used. To add difficulties, I also want a low output impedance to drive any kind of load.

Before spending lots of time with math (load lines drawing, gain calculation, bandwidth and so on), I had to determine which kind of correction would be the best for an accurate and neutral transcription of signal.
The most popular and cost-less designs use active feedback with all the inherent difficulties and compromises due to its implementation like phase shift, distortion, output impedance and eventually oscillations. A better alternative is a passive RC network (split or not) but to work properly it imposes a low impedance source and a load at least 50 time greater than the first series resistor.
I prefer, by far, the LCR network despite its cost. Very precisely wounded chokes, close tolerance resistors and capacitors can give an equalization within ± 0,1dB.

However, there are two drawbacks using such a RIAA correction

1 – Needs to be driven with a low and constant impedance within the audio range. The most practical way is the use of two matching transformers, one on each side. Such a setup is cost no object but worth the investment.
For that purpose, I can use some good Tamura transformers I have on hand (A8713 20K/600 and TKS50 600/50K).

In that way the future preamp architecture will be:

                  Input tube II Transformer II LCR EQ II Transformer II Output tube

2 – The LCR EQ introduces a -20dB (X 0,1) signal attenuation, means these
two stages must have something like 70/72dB gain.
To simplify calculation of these two stages, it is convenient to consider the transformers & LCR units as a single device with a gain of 0,158 or -16dB.

Going backward in my design is also helpful as I exactly know what my final requirements are: 2V with low output impedance.

Final stage

The best way to get the low output impedance and wide bandwidth I need is a cathode follower but alone, with a gain of 0.95/0.98, it would be useless for my project. A nice solution is a constant current direct coupled stage. It consists of a classic anode follower DC coupled to a cathode follower and sharing the same supply. This setup was widely used in professional equipments to source a volume or a tone control.
The tubes I choose are the EC86 / EC806S, renowned for their warm, detailed and dynamic sound. The Russian 6C3Pi was another possible choice, cheap but very well made it is very linear too, unfortunately a different pin-out makes a quick comparison test impossible. Too bad, I previously used this tube in a DAC buffer stage and it was very rewarding. Gave a detailed and silky smooth sound.

Back to my E86C's. An 18K load will give a gain of 54 and a switch to un-bypass V1 cathode give me the possibility of a lower gain (~35) if necessary.
A 150 ohms bias resistor sets current at 10/11mA. These tubes sound better when drawing current and generate lesser noise.
The cathode follower with a gain of 0,98 will set the overall one at 53 and its AC load will be around 13K, considering the input resistance of the following line preamp. High enough to give very little distortion, characteristics being almost vertical (dashed lines).
Output impedance will quite low at 70 ohms.





The 1Nxxxx diode is necessary to protect V2 cathode from stripping when applying HV on a cold tube.

more to come...

C3g Line preamplifier Part 2

...SOLD...

Power supply

From the beginning this preamp was intended to be an exercise in the 6J5 way of simplicity and I calculated bias such a way they could both share the same power supply. The two units are very similar in terms of HV and current requirements and it helps me to make a quick switch for comparisons.

Parts and construction

Even if the circuit is as simple as the 6J5 line preamp, implementation is different. All parts take place on a thick CNC aluminum sub chassis that fit in a much larger cabinet mainly due to transformers size.
High quality L pad attenuator (Siemens silver contacts rotary switch, Dale RL7R resistors) and other selected components to match these superb transformers. Loctal sockets are original Siemens ones and input rotary switch a Grayhill 44H.

Bias is set by Dralowid vintage German carbon resistors, they have a slight grain and a laid-back sound I like very much.

During my tests I noticed better dynamics and lower distortion increasing RL to 160 Ω. Also the C3g works better without feedback.
Be aware of possible oscillations with high Gm tubes. I used grid and anode ferrite beads soldered next to the socket lugs. Wires should be as short as possible and close to chassis (connected to ground and acting like a shield).

Construction details

pictures are worth a long story

almost done, still waiting for the silk screened front plate

On the bench for tests

Low distortion (0.08%@1Vrms) and good squares (100Hz and 10KHz) for a transformer coupled tube without feedback.



Listening report

The rectifier plays a considerable role in the result and among the many ones I stock I prefer, for this preamp, the Mazda GZ32. This is a great sounding tube and still a bargain for the price. My all-around favorite one is the above mentioned plus Philips 5R4GYS, Belvu 5U4G, RFT EYY13, Mullard GZ30, STC 5Z4GY/S depending upon use.

This preamp needs some burn in time to reveal its own character.
Sounded thin and obscure at the beginning, with “thump” bass, but after 2 weeks listening to different kind of music it gained in transparency and revealed a surprising analytical capability with lot of dynamics as said before. It is now evident that this preamp has a completely different character and appears to be detailed to the extreme (it makes me feel the 6J5 sound coarse, which is not). This restitution, like under a microscope, can be an inconvenient on poor or just average sources. I have some CD's, mainly AAD's, I can't listen to because I hear the surface noise of the original material! The C3g is an incredible information catcher but with very little soul.

This, to my ears, puts the preamp on the cold side of music reproduction and I will reserve it for string music, piano or small baroque formations. For me it is not an all-around preamp, but I have to say that my trials were made with the 801A amp/La Scala speakers, a very “clear” combo and probably not the best suited for this tube. I would certainly have a different perception using my Wharfedale Super 12 RS/DD and MC1/60 amp but with a very small listening room it is not possible to have the two systems at the same time.

Another interesting point is a great 3-dimensional restitution, almost like the E80CC SRPP. This peamp has a wider / deeper sound-stage that gives the sensation of being surrounded by music. The 6J5 is flat by comparison. If you like that kind of sound the C3g is the tube to go, taking in consideration that it can be the fact of the tube, the transformer or both...
However, ambiance is not rendered accordingly, and I miss the very living like 6J5 restitution, its "rough" sound gives a more human and pleasantly textured music.
Once again it is just a matter of taste and not a definitive judgment.

some CD's for this test