Phono Stage Shoot Out and Conclusions

For some time I've had the feeling that there was something not quite right about the sound I was getting out of my most recent phono stage. The phono stage in question had never been compared to anything else, so it was difficult to understand what was going on. To add to the complications I use a Grado Reference Platinum cartridge that has a reputation for being rather laid back - so I was not really sure if the often boring sound was due to the cartridge or a problem with my phono stage design.

The design is a two stage with Lipshitz based passive eq., high transconductance type tubes, and gyrator loads. The tube complement is triode connected D3A in the front end, fixed biased 5842 output stage and gyrator loads based on the 5687.

The problem seemed to center around my choice of cathode bias in the first stage with the D3A. This resulted in a 60 ohm resistor and a 1000uF electrolytic cap shunted by a variety of other caps, after a long succession of problems I thought I had the ideal combination.

Something about the laid back, uninvolving sound quality told me otherwise. Last week I hosted the fourth annual Quincy Audio Fest and had the opportunity to compare this phono stage to one designed and owned by Arthur Loesch - this provided the required reference point..

I'm going to quote from an email I sent one of my friends (an attendee) about the whole experience:

"The phono stage A/B just adds ammunition to my contention that electrolytics have no place at all in the signal path of tube electronics. I have used diode and LED biasing in the past and honestly thought that having some dc current feedback because of the very high slope of the transconductance curve of the D3A would be a good thing. (I.e. tiny changes in effective grid bias voltage result in large changes in the operating point) Turned out that despite multiple iterations with ever more expensive electrolytics that I was wrong.. The diodes were clearly much better, and I should have chosen operating points that allowed the use of a single led instead. I don't generally use current sources as a load and using fixed bias whether by diode/LED or actual grid bias is a mixed bag because of the very high transconductance and the tolerance on it - no two tubes bias up the same and if the load resistance is not high enough gain mismatches will then ensue. Turns out that the supposed benefit of a small amount of dc cathode degeneration came at a very high price from an AC performance perspective.

It was an interesting lesson in why I should take my own advice.. LOL Obviously now I should really be making sure that I match the tubes over some range about the intended operating point or go to bonafide current sources as loads.

The Grado while not the ultimate cartridge was clearly not entirely to blame for the lack luster sound I was hearing. I had that nagging suspicion, but before ripping the phono stage to pieces blind I needed a reference point. Now I've got one."

The cathode bias resistor and bypass cap were replaced by a pair of 1N4148 which provide the proper bias level and a low dynamic impedance. A LED would be better, however I would have needed to set up the input stage slightly differently to have the proper operating current, note that because I do not use CCS I do have to select tubes that match and reach the target plate current of ~15mA. Fortunately these tubes are rather long lived so I should not have to do this again soon, gain matching between the two channels is better than a couple of tenths of a dB.