©2003, 2004 Jeremy Epstein. No reproduction, retransmission or web distibution without prior written permission.

This page has information about my
6C45P phono preamps.
diy audio
octal riaa preamp
"free lunch"
teres turntable
tuned quarter-wave pipes

6C45P RIAA phono preamps

6C45P RIAA phono preamp #2 - a DC coupled circuit

(Below you may find information about the first circuit, which uses split passive RIAA equalization.)

This new phono circuit had a roundabout path to its creation. AudioXpress magazine recently started a two-part article about a tube phono stage based on jc morrison's "Siren Song", which jc published in an early issue of Sound Practices. I thumbed through the article on the newsstand and went back to my Sound Practices archives to take a look at jc's original article. He said some things about using octal tubes that reminded me why I like them - the Siren Song is described as a "hedonistic" preamp. But I didn't build a Siren Song.

Instead, thinking about octal tubes reminded me that Bob Danielak had sent me a DC coupled 6SL7 phono preamp design he had worked out, and I had been meaning to build it for a while. I built that one, instead, and I did a nice job on it, too - I put in a pair of gas voltage regulators, and the whole audio circuit fit neatly into a little mini-chassis I bent up from a pair of Simpson's Strong-Tie plates (available at Home Depot - that's another trick Bob taught me.)

Working with the octal tubes is SO much easier than with 6C45P's - they don't oscillate, you get two triodes in one bottle so you don't have to run as much heater wiring - all in all, a pleasure to fool around with.

But the 6SL7 phono didn't sound good to me - it was very slow and ill-defined compared to the 6C45P phono I had been using, and the effort seemed like a dead end. But it did introduce me to the idea of a direct coupled phono, with one EQ stage sandwiched between two triodes, which I thought was very elegant. I sought out some information on the design of these circuits.

There are two similar, but different, networks that will accomplish the RIAA playback EQ all in one shot. Bob's design used this one, which is like the one found in the RCA RC-19 tube manual:

           |       |
          C1      R0
           |       |
           +---+   |
           |   |   |
           |  R2   |
          C2   |   |
           |   |   |

But I couldn't turn up information on designing this circuit right away, so I went with this one which I found documentation for on the Web:

          |   |   |
         C2  C1  R0
          |   |   |
          |  R2   |
          |   |   |
          |   |   |

A quick look at the two circuits shows that there is no capacitor between input and output required for the equalization. So if we are willing to figure the rest of the circuit out to get the right DC conditions, these networks can be used for direct coupling between two tubes.

The equations that cover circuit #2 are these:

r1' = r1 || R0 : the parallel combination of r1 and the following grid resistor.
Also, in reality, r1 is really (Zout + R1) where Zout=rp || RL.

But - we're doing a direct coupled circuit, so we can omit R0. This allows us to work with R1 directly (taking into account Zout) and also allows us to presrve the full gain at the plate of the tube driving the network, instead of dividing it between the series resistor and the grid resistor. This is a good thing in a phono circuit where you never want to waste any gain.

r1' x C1 = 2187uS
r1' x C2 = 750 uS
r2 x C1 = 318 uS


C1/C2 = 2.9160
r1'/R2 = 6.8774

The way to make a circuit out of these equations is as follows:

1. Pick a pair of standard capacitor values for C1 and C2 in the proper ratio. In practice this will likely mean matching up caps that are a little high and a little low from a standard value - not a problem, since you can easily get enough variation in a batch of say, 20, 5% tolerance caps. (You will also want to shave a little off of C2's calculated value to account for the Miller capacitance seen at the grid of the tube following which adds back extra capacitance in parallel with C2.)

2. Figure out what r1' needs to be in combination with the two selected capacitors, using the equations above.

3. Figure out Zout from rp and RL. Adjust r1' to find the value of the resistor for R1.

4. Find R2 from r1'.

So I went ahead and did this process, taking into account what I had on hand. Also, as I did in the split passive design below, I kept the values of the series resistors as low as possible - the high gm 6C45P doesn't need much of a load resistance and the factor that determines the minimum is really how sensitive to tube variations you want the circuit to be. I saw a suggestion to make R1 at least 7 x rp.

I selected a B+ voltage of 300 because I knew I could make a cheesy little regulator out of one resistor and two 0A2 gas regulator tubes for that voltage, plus I could work out the rest of the DC voltages well with that as the starting point. And then I built it up.

Well, it oscillated like crazy at first: my layout was not tight enough. So I took a trick from the Bottlehead Seduction phono kit and made a small ground plane strip out of a scrap of circuit board material and tightened everything up quite a bit. I found a bad solder joint or two, and things improved remarkably. I got rid of the 0A2's and made an extra pi filter to replace them - so far I haven't had much luck with 6C45P and gas regs.

Once the crazy radio nonsense stopped, it sounded very good right off the bat. The immediacy and speed of the high-gm triodes was a huge plus over the 6SL7 circuit, and there seemed to be maybe even an edge over the split passive phono which is well-liked by both my ears and others who have heard it. I'm evaluating this now, and it's sounding very fast, very spacious, and very quiet in the silences. There is a little bit of an edge which I hope will go away as the big electrolytic caps break in a bit. At first I thought it was a little tipped-up in the treble but then I let it play a while and listened to some material with good strong bass. I realized it was more an edginess I was hearing than a tonal imbalance, so I think I have the various parts values pretty well determined from what I can tell. Of course I'll measure it too, but I like to listen first.

The Neville Brothers song "Voodoo" from Yellow Moon is a great track to check the bass tonality with, by the way: there is a very live-sounding pumping electric bass sound on this recording and when you get it right it sounds just like a good club PA's bass bin, really grunty.

Here's the schematic for this one:
DC phono Schematic

UPDATE! Here's a schematic for the same kind of thing using 6GK5 triodes, in case you have a bunch of seven-pin mini sockets lying around. More gain than the 6C45P version, too: this one should be good for about 50 dB @ 1kHz.
DC 6GK5 phono Schematic

6C45P RIAA phono preamp #1 - the original circuit

I have completed my second RIAA phono preamp and it represents a significant improvement over my previous octal phono preamp, without any real increase in circuit complexity.

I built this new phono preamp because I became curious about the various low-impedance LCR phono circuits and I wondered whether their open, lush, fast sound had anything to do with the very low impedances found in their EQ sections. The ones I had heard were 600-ohm impedance circuits. Their mojo was always ascribed to the use of inductors. I thought about a design that might allow me to build a conventional RC-equalized phono circuit that would operate at a much lower impedance than the typical RC circuits I've seen. This way I could stick with a familiar, proven topology but test the idea of low series resistance and hear how it sounded.

Typical split-EQ circuits use a series resistor of 200K or more. But is this large a value really necessary? Using my favorite high-gm triode, the 6C45P, I saw I could still have a large ratio between the source impedance, Z, of the tube stage (the plate resistance rp, in parallel with the plate load resistor Rp) and the series resistor value in the RC filter. This ratio protects the filter from variation as the tube itself varies due to age etc. The rp of a 6C45P at the operating point I selected is about 1400 ohms. In parallel with the plate load of 5K I selected, the source impedance is about 1090 ohms. This is less than 8% of the total impedance that interacts with the fixed capacitor values. If a tube aged so that its rp became double the expected value, the total variation in the RC filter constant would be about 5%.

The derivation of the filter component values for a two-stage (split) passive RIAA equalization network is shown below.

I deemed 5% to be an reasonable worst-case variation, so I went forward with a design that uses series resistor values of about 12K and 17K ohms, about 15 times smaller than the values typically found in a phono preamp. I wanted to reduce the series resistor by at least one order of magnitude, in the hope that this reduction would be enough to hear if the low series resistance made a difference. This goal was met.

The gain derived from this three stage circuit is about 42 dB at 1KHz. If more gain is necessary, a 12A4 could be used or the third stage could use a bypassed cathode resistor, or both.

The sound is very fast and open, with rock-stable imaging and pure tonality. I like it very much. I wouldn't say it quite has all the lushness of the LCR phono preamps I have heard. But it shares the feeling of speed and musicality these stages have. It's possible the LCR's lushness might be approached using an output transformer instead of the cap-coupled output. I intend to try that in the future.

One problem with this circuit is that it is somewhat difficult to adjust the filter values to an accurate RIAA equalization - a very small change in resistor values makes a somewhat large change in the response and the sound. I originally used a capacitor of 5.5 nF in combination with an effective resistance of 13636 ohms to effect a 75 uS equalization filter, and I found this was very ticklish to adjust accurately. Since my 5.5 nF cap was actually 4.3 nF in parallel with 1.2 nF, I simply removed the 1.2 nF cap and raised the effective resistance to 17441 ohms. This reduces the contribution of the tube's characteristics even more and I found it was a snap to get perfect matching between channels at this value. It might make sense to convert the second section, perhaps using a convenient standard capacitor value of .15 uF (this would raise the effective resistances needed to 2120 ohms and 19080 ohms.) A project for another day.

By the way, out of frustration with the tedious process of matching up resistors, I decided just to insert some 15-turn trimmers in the appropriate places. This way, I was able to just dial in the exact resistance I wanted and this was a great help in setting things up. Sonically I hear no disadvantage to these cermet trimmers.

Still open is the question of how much of the improved detail and speed is due to the use of a high-gm triode. Probably a lot. High-gm triodes like the 6C45P are the ideal choice for a phono preamp, so I would certainly have used something similar to the 6C45P in my next phono stage anyway. Thus I am not sure yet if I've proven my point, but in the meantime, I've at least built a better phono preamp. The thing to do next would be to compare the sound of this circuit, driving my RC EQ components, to the same circuit driving a 600-ohm LCR EQ setup.

I made a number of needle drops (dubs from disc) and recorded them onto CD-ROM with a Mac G3 Powerbook for A/D conversion. I observed that the noise floor of this new preamp/computer combination was much lower than my previous iMac and phono stage combination - I can hear the difference between sources that have a little tape hiss present and sources that have none; I can hear down into the very small rumble present in my turntable. In fact I can hear a fairly large variation - even after the CD conversion - between sources, mixes, and dynamics of various LPs and singles. I regard this as one of the hallmarks of good front end gear. A good record sounds good, a bad one sounds bad. In any case the truth prevails.

My disc transfers are far better from a noise standpoint as well sounding livelier and more dynamic.

The schematic for this phono preamp is shown below, along with the power supply schematic.



(If you are interested in having me build a phono preamp for you, please contact me.)


Here are the calculations needed for a vacuum tube RIAA phono equalizer of the RC-equalized, split passive type, for a triode with bypassed, or battery-biased, or grounded cathode:

For the first section, with 75 and 3.18 uS time constants, we assume that the two filter constants are far enough apart that their interaction need not be taken into account. (In practice, we will get as close as we can with math using this assumption, and then measure the actual circuit to adjust the derived values, simultaneously compensating for filter interaction and stray and Miller capacitances.)

First find the value of rp paralleled with the load resistor Rp. Call this Zout. Then add this value to the series resistor Rseries, and finally place this sum (Rseries + Zout) in parallel with the following grid resistor Rgrid. The result is the effective resistance, Reff, seen by the filter section.

The formulas needed are:

1/Zout =(1/rp)+(1/Rp)

1/Reff= [1/(Zout+Rseries)]+(1/Rgrid)

Reff * C1 = 75 uS for the 75 uS time constant


Rcutterhead* C1 = 3.18uS for the "Allen Wright" or "Neumann" hidden 50kHz time constant

For the second section, with 318 and 3180 uS time constants, there are similar relationships between the filter components and the tube circuit elements. There is also a relationship between the two resistances, Reff and R318.

Reff/9 is the value of the other resistor in this filter section, R318.

R318 * C2 = 318 uS

This analysis assumes a coupling cap large enough to act as a short circuit for all audio frequencies, and is taken from Morgan Jones' book "Valve Amplifiers."