Wintersweet HighVividity Amplifer finalized

[HighVividity]

Dear all,

I am so happy to announce that the Wintersweet Highvividity Amplifier is finalized. It took me more than 1 year to reach this point. The website is http://www.wintersweet.com/highvividity, I hope you enjoy.



Best,
Highvividity

[DougL]

Unusual design.
Can you give me some Idea what your design goals and tradeoffs were?

Doug

[HighVividity]

Thank you for your interest. My goals was listed on the website. When used as is, The trade-offs are,

1. Multiple voltages. The 3V supply can be derived from the 6.3V AC that is readily available. The -36.5V supply may require another power transformer. The 820V supply, the floating filament, the floating screen grid all require their own transformers.
2. High power plate resistor. I used 8 x 3.3K, 3W resistors in serial. Maybe you can use another EL34 in place of those resistors, but that requires another set of filament power supply. If you use a loading inductor to replace the plate resistor, well, I have no idea whether the circuit will be stable. You may try.

If it is used as driver stage of 2A3, 300B, 845, etc, because of the reduced dynamic range requirement, the 820V supply can be reduced and replaced by the main power supply. The filament can be non-floating (just to be raised to a proper voltage), if you chose a tube allowing high cathode-filament voltage. The screen filament can be non-floating too, making the cathode-follower stage triode-connected. So the only effort needed is to generate the extra -36.5V supply.

The gain is a perfect cathode-follower amplifier, or a perfect driver stage for output stage. I believe it can beat every existing driver stage design, from every aspect.

[leadbelly]

Originally Posted by HighVividity

I believe it can beat every existing driver stage design, from every aspect.

Oh, is that all?

[kevinkr]

One thing that comes to immediately to mind is that it appears you are violating the input common mode range of the 5532 which does not really extend to the positive supply rail.. ( Positive cmr is typically VCC-2V for this device. Another vendor's 5532 might not function at all in this circuit.)

I am also wondering about HF stability, I would not be surprised if there was some residual low level oscillation due to the inclusion of the EL34 inside of the op-amp feedback loop. Have you considered turning the op-amp into an HF integrator?

I guess I would also question your contention that an op-amp is equivalent (sonically or in any other sense) to a transformer in this circuit. (Have you actually tried a transformer?) Using it to servo the operating point is clever, however I would expect its ac performance to be a somewhat different matter given the reliance on feedback margin for linearity. Also wondering what drove the choice of the 5532, the LM4562 might be worth a look as well.

Don't take my comments too critically as I am noted for my aversion to sand in the signal path of a tube amplifier, irrationally perhaps, but I barely consider a solid state mosfet CCS acceptable in the audio path of an amplifier - in fact there are none in any of my amplifier designs. (I live with op-amps in my dac, and that is about it..)

[HighVividity]

Kevinkr,

Thank you for your comments. Sorry that I read the document wrongly, regarding the maximum input rating (rail to rail) as the common mode range.. The chosen of 3V was for using 6.3V AC with 3 terminal regulator, with resistors I have in hand. I think increasing it to 4V or even 6V will solve the problem.

HF stability is not a problem, I used 5PF capacitor to add a pole at around 160KHz. I think this is what you mean with "HF integrator". This made the loop stable. The waveforms are very clean, without any sign of oscillation.

I did not say an op-amp can be sonically equivalent to an inter-stage transformer. This is only an driver, and it serves its purpose well. if you want the sound of transformer, I have a pre-amp for that purpose.

The feedback margin is not tight. I did not test removing the 5P capacitor though. but 5P with 200K gives me 160KHz. I think that is high enough.

There might be better choice than 5532. This is just a demo; you can choose whatever good op-amp you want.

You comments are very appreciated.

[HighVividity]

I have updated my schematic so that the power supply is raised to 4V. I will make the corresponding change to my amplifier later.

[DougL]

Looking at the design, I have some questions.
The pentode cathode follower has the additional load of the floating supply.
I was wondering if you measured / listened to it in triode mode?
With a Mu of 10 and an Rp of 800 ohms, distortion should be low, and output Z should be low. I also noticed you used a 3Rp transformer, which TCJ says is "correct".

Does the CF output stage suffer from blocking due to the cap coupling?

In your response, you noted that you chose a resistor loaded driver because you were concerned with the stage stability of a choke load. I was curious about your choice of an EL34 driver. Wouldn't a 6V6 or EL84 have been lower distortion with less current and voltage?

Doug

[HighVividity]

Dear DougL,

Thank you for your questions. Let me answer them one bye one.

Yes, the floating power supply follows the cathode voltage. But I think it can be thought as just a (small) capacitor to drive.

No, I did not measure or listen to it in triode mode. I don't think the distortion would be anywhere near big, but I designed this circuit as a demonstration, and wanted the distortion to be as small as possible. Also when the filaments are floating, I need two separate 6.3V winding anyway, and as I use two separate transformers for that, the two high voltage windings are just there for screen grids.

Actually the transformers are 5K:8. But my speakers are 4Ohm ones, I just labeled the transformer as 2.5K:4. The plage-cathode voltage divided by working current (350V/70mA) gives 5K, so a 5K load may maximize the output power.

I used cap couping to attenuate the low frequency a little bit, to avoid EL34 saturation. The choice of 0.033UF gives flat cathode current down to 20Hz. The voltage response ( I think, forgot how much ) is -3dB at 20Hz.

I think a choke load may be stable, just I did not test. I use transistor load for its ease of analysis. I chose EL34 because of it's high plate-cathode voltage rating. 6V6 or EL84 have <400V (maybe 300V) rating but EL34 is rated at 800V. The 300V might be actually 600V, since 300V could be transformer loaded plate voltage, but EL34 is safer to use for me. With reduced supply voltage, 6V6 or EL84 may be good too. I don't concern about distortion, because the 60dB open loop gain of NE5532 and 50 times gain of EL34 ensure negligible close loop distortion. I chose 15mA DC driver current rather arbitrarily, only because it has plate resistor power dissipation I can handle. A smaller current might work as good, but a higher current has additional dissipation difficulty to handle.

Highvividity

[HighVividity]

Lower driver stage DC current also requires lower grid voltage. Currently it is at -22.6V. If it is lower, I might have dynamic range problem for the lower side. I could not make the negative supply of 5532 lower. -37V is what my LM337 is rated; and 44V is the maximum supply voltage difference of 5532.

Choke load may require additional circuit to stabilize the DC operating point. So I chose resistor load. Easier to understand and analyze.