RIAA phono stage with E180F and ECC802S tubes

[ivg]

Hello guys!
I'm interested to add an analogue source to my music setup. It will be a vinyl. 🙂
Not bough yet a turntable, but it will be with MM head. I want to build my own tube phono stage just because I recently got some interesting tubes.
I've spent a few weeks in web research trying to arrange my tubes and select appropriate. 🙂 I believe the time was not wasted. I read many articles and reviewed many schematics...
Finally I decided to develop some mix of all ideas that I've found. Output stage is a mu-Follower, I choose it because of low output impedance.
Here is a schematic of one channel:

Also I did a PSU simulation:

Hum rejection ration is about 90dB as you can see in this screenshot. Of course in real device the numbers will be smaller.
Any suggestions, enhancements?
Thank you in advance!

 

[ivg]

There are some measurements, of course virtual for now:
THD at 100Hz:

THD at 1kHz:

THD at 10kHz:

RIAA curve:

Also an archive with LTSpice schematics

 

[ivg]

I’m not experienced in tubes, just general knowledge from university. Maybe anybody tube guru can review my paintings? 🙂

 

[bondini]

I like your design and the sims look good (nice RIAA response etc). I have a few questions:

• What is your target gain at the 1kHz nominal mid-band frequency?
• What was the simulated input signal for each of the distortion sims?
• Could you re-post your sim circuit showing quiescent currents and voltages?
• Is the quiescent current through each of the QLTP-690Cs enough to ensure low dynamic resistance? I'm not familiar with these LEDs - looking at the data sheet the dynamic resistance looks to be 5 or 6 Ohms providing the current stays above 10mA.

One of the reasons I like your design is the use of a pentode as the first gain stage - with a bit of work, they can give high gain, low distortion and low noise. But it does take work, especially because the pentode amplifies all the noise from the power supply, plus it contributes some partition noise. I would recommend downloading and reading:

• Frank Blohbaum's article on his "BestPentode" configuration (here: https://linearaudio.net/new-low-noise-circuit-approach-pentodes); and
• Merlin Blencowe's re-joinder (here: https://linearaudio.net/partition-noise-and-bestpentode-revisited).

If you stick with the pentode (and the E180F/6688 is a nice pentode) you may want to consider:

• the low-noise E186E/7734 variant;
• Frank Blohbaum's Best Pentode configuration (I have had good results); and
• ensuring the power supply is in a separate enclosure.

 

[ivg]

Thank you, @bondini !
Few quick answers:
1. The target gain I planned to achieve was 40dB or more. Actually got a 49dB
2. All posted measurements are at 5mV (amplitude) input signal
3. I will try to add voltages and currents to schematic. Pentode G2 voltage is 150V, current is about 1,2 mA; anode current about 6,5mA, voltage 140V; grid bias is 1,8V. But I could be wrong about anode voltage because writing this from my memory
4. These LEDs are just red LEDs with 1,8V forward voltage drop. I didn’t find any other suitable model in standard LTSpice library
5. I’ve limited to use E180F pentodes, I have them already 🙂 Also I have E83F, but it looks not very good in this place
6. Because first stage has a high gain and will amplify any noise from power supply I tried to reduce it in PSU and of course heater will be powered by the regulated DC (LD1084)

 

[Chris Hornbeck]

The circuit certainly looks fine. I have no idea why the 100Hz simulation would have such weird high frequency spikey noise/harmonics. Was it run with an "ideal" power supply, or with the proposed supply? Maybe it's complaining about the LED bias diodes? Maybe you could re-run the sim with resistors in place of the diodes, just to narrow the search.

A fun variation would have another E180F as the second stage, and very short connecting wires going out.

All good fortune,
Chris

 

[ivg]

@Chris Hornbeck the simulations ran with ideal PS. I will try to make same with proposed PSU.

 

[Chris Hornbeck]

I asked because the proposed PS could maybe/somehow have generated the spikey high frequency artifacts, but that wasn't in the sim. Because these are simulations the cause can't be normal real-world issues of noise contamination or RF interference or such, so is very puzzling.

The only thing not "as ordinary as dirt", well known forever and well understood, is the LEDs themselves. Running the sim with resistors as cathode bias, maybe paralleled with very large capacitors, would eliminate (or not) the LEDs as the culprit. Maybe all this is some strangeness within the sim, but it's interesting to try to find the cause.

All good fortune,
Chris

 

[Mark Tillotson]

You can add these Spice options to improve numerical accuracy:
.options plotwinsize=0
.options numdgt=10

These turn off lossy compression (which creates some simulation artifacts) and forces double float usage for higher dynamic range.

 

[ivg]

@Chris Hornbeck @Mark Tillotson weirdness on 10Hz plot was caused by lack of data. I just increased a number of samples and got few new plots.
BTW, I've replaced LEDs by resistors (large cap in parallel) and there are no significant changes - all plots looks same as with LEDs.

And here are same plots for phono+PSU combined: