RCA Phono Stage Redux

[Eli Duttman]

My math says the gain of the "classic" RCA phono circuit is 45 dB. That's not far off the "magic" 50 dB. that raises MM carts. to CDP level.

I have some thoughts on squeezing the extra 5 dB. out and I'd appreciate feedback.

Make the B+ rail 300 V. CCS load (1 mA.) the 1st gain block and use the same Agilent LED as employed in the Bottlehead Seduction for bias. Increase the 2nd gain block's load resistance to 120 KOhms. Buffer the 2nd gain block's O/P with a DC coupled Zetex TO92 case MOSFET, as described in MOSFET "Follies".

MOSFET "Follies"

Bass extension will be improved by switching the 2nd gain block to contact (grid leak) bias. The cathode will be grounded and a 20 MOhm Caddock resistor replaces RCA's 680 KOhm value in the grid leak position.

[jaudio]

This was my first phono amp project. I never could get the bass quite right,maybe 45db explains it. I hope you get it to run properly. I might try it again.

[hacknet]

HEY!

the first think i thought of when i read mosfet follies was to do exactly this!

[hacknet]

the article claims that miller effect doesn't affect the follower, i did a simple simulation with the a voltage source with 35k output impedance and a mosfet with 360pf gate capacitance. i get quite abit of roll off even at 1khz.

am i missing something?

[Eli Duttman]

Originally posted by hacknet
the article claims that miller effect doesn't affect the follower, i did a simple simulation with the a voltage source with 35k output impedance and a mosfet with 360pf gate capacitance. i get quite abit of roll off even at 1khz.

am i missing something?

Look here. Mike Preamp Thread Cogsncogs made the KEY point. While the source does follow the gate, transfer capacitance interacts adversely with a high Rp tube to roll HF off.

The capacitances of TO220 case FETs are large. OTOH, the capacitances of the little TO92 case Zetex FET are much more reasonable and have a smaller impact.

If push comes to shove, a cap. coupled JFET as the buffer will get the job done. However, I'd like to be DC coupled at the FET's gate.

Someone more clever than I am and really knowlegeable about RIAA EQ might rework the network to take advantage of HF roll off when the transfer capacitance is "high".

[hacknet]

hmm.. yeah. i can't find a source for the zetex mosfets. i think i might do the jfet approach. i'll look around and see..

is there any software out there that can do distortion analysis?

[Brian Beck]

In the MOSFET Follies link the author says: >>"But wait!" I hear you yell. "The MOSFET has a big gate-source capacitance. Won't that suck all the high frequencies out of the signal?" No, it won't. The big gate-source capacitor is there, OK, but when the device is hooked up as a follower, the apparent capacitance seen by a circuit driving the MOSFET gate is reduced to a very small level by the fact that the source is following the gate almost perfectly, so the apparent capacitance is reduced by the local feedback to an inconsequential level. Even a high impedance 12AX7 plate is not affected in the audio range by this capacitance. Even the big power, big capacitance MOSFETs do this one OK. I have done this, and what's more, done the necessary measurements to find the effect on the response of the composite stage - and there is no detectable change in frequency response below 30kHz. None.<<

Yeah, but… There WILL be phase shift in the audio band due to this capacitance even if here is “no detectable change in the frequency response below 30KHz” It's not just the absolute value of the FET's capacitance that we need to consider, it's also the nasty variability of that capacitance with signal swing. Even if the absolute value has little effect on frequency response (magnitude) at 30 KHz, there will be in-band phase shift due to this capacitance this is modulated with signal swing, even with the bootstrapping effect of a follower. The worst way to drive such a capacitance is with a high Z source, say a 12AX7 plate. The result will be phase intermodulation distortion, the worst kind. This is the prime reason why tubes sound better than transistors, IMO. A little bit of PIM adds a false edge to music that some might find appealing on first listen. IMO, it’s best to avoid this approach.

[Eli Duttman]

Brian,

You make a good point. A low gate capacitance JFET cap. coupled to the anode of the 'X7 will work fine as the buffer, but I'd like to salvage the DC coupled gate.

OK, make the 2nd gain block a 12AT7 section. CCS load the 'T7 at 3 mA. CCS loading maximizes both gain and linearity. Gain should be about 35 dB. Given its high gm and low Rp, the 'T7 is better suited to driving the little Zetex MOSFET than the 'X7 is.

[ThorstenL]

Konnichiwa,

Originally posted by Eli Duttman
My math says the gain of the "classic" RCA phono circuit is 45 dB. That's not far off the "magic" 50 dB. that raises MM carts. to CDP level.

Just a small note, 50db is too much gain.

Here is why.

The CD has a full scale output level at 2V. The average level is usually 14db lower or 400mV.

LP is defined with 5cm/S as 0db but allows per IEC 25cm/S level, in other words a 14db headroom. So, a Phonostage to match CD needs to produce 400mV with a nominal input of 5cm/S, thyphically MM level is reckoned to be 2.5mV...10mV with a "design center" at 5mV for 5cm/S.

So, 5mV need to produce 400mV or 38db gain. It is desirable to have 6db more gain (possibly switchable?) than that which places our nominal input at 5mV for 5cm/S and at least 12db headroom on top of the 14db headroom designed into LP for that nominal input.

So we need 26db headroom on 2.5mV input voltage and 44db Gain for a "bogey" MM Phonostage unless really quiet Grado's are used. Output referred we hence need the ability to produce 8V RMS with low distortion with 44db Gain. Better would be a Phonostage with a selectable MM gain at 32db, 38db and 44db.

Then adding a Transformer with a 1:5/10/20 option makes sure we can accomodate a wide range of MC Pickups as well, with 1:5 giving nominal 0.5mV input with 0.25mV and 0.125mV @ 5cm/S as nominal input for 44db gain selected.

Sayonara

[Eli Duttman]

Thorsten,

Thanks for your remarks.

After the losses in the EQ network, the net gain of a CCS loaded 'X7 section will be < 20 dB. Let's say the gain is 18 dB., for purposes of discussion. That leaves 26 dB. for the 2nd triode to produce. To drive the Zetex MOSFET buffer, a high gm/low Rp type is indicated. The gain of a CSS loaded 6922 is approx. 30 dB. That's 4 dB. too much, but for several reasons the 6922 just might be best, in spite of its "high" gain. I don't know if the mu 22 ECC99 is quiet enough and if it will work well with contact bias.

This is getting really interesting. A NOS TFK ECC808 could provide the 1st gain block for both channels and a JJ E88CC could provide the 2nd gain block for both channels. The internal shields of both types should help channel separation a little.