Phono Stage – my first try

[Joshua_G]

This is my first attempt at designing phono stage, or RIAA amp.
It looks to me that the circuit is quite straight forward. Each one of the 2 amplifying stages has 30dB gain @ 16mA. Accounting for the ~20dB@1KHz loss of the RIAA equalizer, makes the total gain ~40dB.

I couldn't find anywhere a way to calculate the CF cathode resistor, so I hope the value in the schematic is correct. Tubes are ordered and on their way. Also PCB design is currently in process. I have stabilized tube PSU and power transformer to do the initial tests with.

All comments are appreciated.

[kevinkr]

Hi Joshua,
I recommend a minimum of 8.2K 10K for the plate resistors as the rp of this tube is about 2K or so, and your load resistors will result in a great deal of distortion.. Better still would be a CCS load. Bias in the range of 1.2V from an IR led is also recommended. Plate current should be >10mA to achieve relatively high transconductance with the D3A. Using 10K plate resistors will require a supply of about 250V. Expect to have to trim the 22K resistor in the RIAA equalizer in order to get the equalization right. (Typically a few % one way or the other.)

Your coupling caps are probably a bit on the small side. Note also that D3A do draw some grid current and that 1M grid resistor should probably be reduced to 220K or so, if you do this you must increase that first coupling cap to at least 0.1uF.

Depending on what you are driving you may not need the CF last stage - if the input impedance of the following stage is >20K this is certainly the case.

[Arnulf]

Cathode resistor of CF stage is same as anode resistor in common cathode stage. You set its value according to your desired loadline exactly the same way.

[Joshua_G]

Hi Kevin,

Originally Posted by kevinkr

I recommend a minimum of 8.2K 10K for the plate resistors as the rp of this tube is about 2K or so, and your load resistors will result in a great deal of distortion.

Thank you.
What you say does make sense. The 1.3K Rp was chosen or 30 dB gain. With Rp 8-10 K the gain will be too high, 36 dB. How to compensate for it? D3a ri in triode mode is 1.9K.

Originally Posted by kevinkr

Better still would be a CCS load.

Like SRPP, or what?

Originally Posted by kevinkr

Bias in the range of 1.2V from an IR led is also recommended. Plate current should be >10mA to achieve relatively high transconductance with the D3A.

The present plate current is 16 mA. I remember your recommendation in another thread to set it for 10-20 mA,

Originally Posted by kevinkr

Using 10K plate resistors will require a supply of about 250V. Expect to have to trim the 22K resistor in the RIAA equalizer in order to get the equalization right. (Typically a few % one way or the other.)

Your coupling caps are probably a bit on the small side. Note also that D3A do draw some grid current and that 1M grid resistor should probably be reduced to 220K or so, if you do this you must increase that first coupling cap to at least 0.1uF.

Changing the value of Rp, all other values will have to be recalculated.

Originally Posted by kevinkr

Depending on what you are driving you may not need the CF last stage - if the input impedance of the following stage is >20K this is certainly the case.

D3a in Common Cathode triode mode Z out is 1.2 – 1.6 K, depending on Rp. Though I may get away with it, Z out of about 25 Ohm in CF mode is highly preferred. In such a case, there are no worries about any load and any length of interconnect cables.

[Joshua_G]

Originally Posted by Arnulf

Cathode resistor of CF stage is same as anode resistor in common cathode stage. You set its value according to your desired loadline exactly the same way.

Indeed, it is recommended in order to alleviate PSU burden, however, the bias may not be right for the tube. May primary concern here is to set the bias and plate current right.

[SY]

Stop. Do not pass Go. Do not collect $200.

You're trying to do something VERY difficult without having even the most basic design chops (no shame in that, a phono stage is the toughest thing to design properly). Get, read, and (most importantly) understand the preamp chapter of Morgan Jones's "Valve Amplifiers." Even if you go about things differently than he does, the basics need attention before you jump in the deep end.

[Joshua_G]

In light of some of Kevin's comments, a modified version. I had to use un-bypassed cathode resistor in order to maintain 30 dB gain with higher value plate resistor.

[kevinkr]

Originally Posted by Joshua_G

In light of some of Kevin's comments, a modified version. I had to use un-bypassed cathode resistor in order to maintain 30 dB gain with higher value plate resistor.

Not sure why 36dB of gain is an issue at all, (not that this circuit is likely to give you that much) in a world where the redbook standard of 2Vrms at 0dBFS is the normal standard for audio sensitivity the additional gain is not a bad thing. People struggled to achieve these gain levels with older passive designs - there is absolutely nothing sacred about 40dB of phono stage gain. Mine in fact has about 48dB.

You idea of using unbypassed resistors in the cathode circuit will reduce gain by > 6dB and increase your effective input noise by > 30dB which is still pretty quiet, but completely defeats the reason for using the D3A. Note also that it raises the effective rp of the tube to nearly 10K ohms - not something you really want to do. Not sure how you are modeling the D3A (whose model are you using? not mine apparently). I strongly recommend you use LED bias and not obsess about the resulting gain.. Also a CCS would normally be solid state with high performance IXYs mosfets (see the great ccs threads here, and take SY's comments to heart. He has as much experience with this tube as I do.)

What does this pre-amplifier have to drive that 25 ohms source impedance is an over-riding concern? A CF well implemented is a pretty transparent device, however here it is not really required and IMO will detract in audible if minor ways from the overall performance.

Note also that high transconductance tubes can differ very significantly in dc performance from what the simulations lead you to expect. You'll find this out right quick when you start to bread board your circuits.

Something else you should know is that you need to place the grid stoppers as close to the sockets as you can get them (< 0.75cm lead length on the grid side) I think you will also find you need stoppers on the screens - I use 221 typically here. The cathode resistor or LED leads should be very short as well. Note with the CF you should insert at least 100 ohms of series resistance in the output, placed right at the cathode - this is not a bad idea with the CC stage plate if you take from here. Very careful construction technique with short leads in a metal box is needed.

[Johan Potgieter]

Originally Posted by SY

Stop. Do not pass Go. Do not collect $200.

You're trying to do something VERY difficult without having even the most basic design chops

Now I feel better!

To chime in: This is my introduction to the worthy D3a (just shows; age and experience is no guarantee of knowledge ...).

To topic: Looking at the graphs: Ri etc vs Ia, I notice that having an unbypassed cathode resistor takes Ri from 3,2K (10mA) to about 16K. Question: Is one not then back at an undesirable Ri (apart from feedback now improving matters slightly) with a load of >10K? Knowing that Ia has to be kept up and keeping rail voltage within bounds, one wonders what a load resistor as low as Josh had and with unbypassed cathode resistor might yield in performance - sort of working as one would a pentode - Rl < ra. (Also recognising low signal amplitude.)

Honest question without suggestion - I have not done this, and you are 'well versed' working D3as.

[kevinkr]

Originally Posted by Johan Potgieter

Now I feel better!

To chime in: This is my introduction to the worthy D3a (just shows; age and experience is no guarantee of knowledge ...).

To topic: Looking at the graphs: Ri etc vs Ia, I notice that having an unbypassed cathode resistor takes Ri from 3,2K (10mA) to about 16K. Question: Is one not then back at an undesirable Ri (apart from feedback now improving matters slightly) with a load of >10K? Knowing that Ia has to be kept up and keeping rail voltage within bounds, one wonders what a load resistor as low as Josh had and with unbypassed cathode resistor might yield in performance - sort of working as one would a pentode - Rl < ra. (Also recognising low signal amplitude.)

Honest question without suggestion - I have not done this, and you are 'well versed' working D3as.

Hi Johan,
Well I think you worked it out a bit more precisely than I did (I swagged) typically rp at 20mA is in the range of 2K IIRC and probably close to 3.2K at 10mA, (I haven't checked) but the point you made about ri>ra (rp>Rp) is completely valid, and not really acceptable in triode connection. It has serious noise implications for this tube as well.