Rerurn to Romy the Cat's Site

peter wrote:
By "dynamic viscosity" do you mean the reluctance of an amplifier to be dynamic, i.e., the degree to which an amplifier sounds like it is bogged down, usually as a result of a poor rise time on transients, i.e., poor HF response, or a tube which is badly biased and loaded wrongly which is typical of many tone control stages in generic junk tube amps of the 1950 / 1960 era ? 
 
Romy replied:
Actually, the Milq has no problems with dynamic; in fact it is the most dynamic amplifier I even heard. By the "dynamic viscosity" I did not mean the dynamic itself by rather the dynamic of dynamic, or the dynamic of changing the dynamics. I mend sound do not "dive" into nothingness and do not rise to some kind of dynamic absolute but rather it moves across the dynamic range according own some kind of pattern. Sometime it has "sticky" dynamic of dynamic and sometime aggressive dynamic of dynamic. It has own brain and has own reasoning. Interestingly that this own "reasoning" make sound much more like acoustic sound.

---

peter wrote:
The 6E5P is a russian tube and perhaps the 6CL6 is a near equivalent so instead of 6E5P could 6CL6 be used or even EL84 with equally good results ? 
 
Romy replied:
6E5P, in the given application, with the given biasing method is the hart of the amp. I am sure you might fine a lot of tubes that might swing the same voltage and fulfill other electrical requirements but it will have a very little to do with sound. 6E5P and the way in which it was used was use explicitly because the very specific sonic characteristics. Would the 6CL6, EL84/6N14P have the same sonic qualities? Whatever it will be it will be a totally different amplifier about which I have no business to comment.

Peter writes:
My understanding had been that biasing method is a method of setting up a given tube so it works with the wanted anode current at idle with the wanted anode voltage and these conditions are met by somehow applying a correct grid DC voltage and that properly implemented biasing could always be done several different ways and the tube would have exactly the same harmonics and sound during normal AC signal operation.

---

peter wrote:
It appears that the amplifier does bias a 6E5P with two gas tube regulated positive and negative voltages so that a negative voltage can be derived from a resistance divider and applied to the input tube grid.  While some holy audio priests say electro caps are forbidden ( unless perhaps thay are Black Gate or some other expensive brands) there is not much wrong with a Nichicon 1,000 uF cap to bypass cathodes.  At 100Hz they have 16 ohms of reactive impedance and there does not seem to be much evidence that their sonic signature can be detected in any AB tests. 
 
Romy replied:
I religiously against Black Gate capacitors and religiously against the Sound that Black Gate create. Furthermore I have very-very high prejudices against people who feel that Black Gate capacitors are useful for sound. From my point of view those people juts have no idea what to listed while they are listening.

Peter writes:
Sorry Romy, perhaps I was not clear in what I wrote but I wish to understand the engineering if that is possible of how one well implemented method of biasing such as grid leak biasing produces a different sound from an amplifier over another well implemented method of biasing such as using a Nichicon 1,000 uF cap to bypass cathodes.

---

peter wrote:
Did you consider instead using an electrolytic and R used to provide cathode bias ? 
 
Romy replied:
To ground cathodes was the very primary objective of the entire design. I see no reasons to lift cathodes from ground.

Peter writes:
My understanding had been that cathodes are considered by many audio and radio engineers to be grounded when they have an R&C circuit between cathode and 0V.  The R allows a DC voltage to appear across itself according to Ohms Law, V = I x R.  But at audio signal F the C forms a low impedance path for current.  3,200 uF has 0.5 ohms of impedance at 100Hz.  At 0.001Hz, getting close to DC, the cap impedance is quite high, and is 50,000 ohms.  At 2 kHz where the ear is most sensitive the 3,200 uF has Z = 0.025 ohms, or extremely low in comparison to the other circuit impedances where it is located.  However, in the past, C used for the C in the R&C signal bypass circuits were often only 25uF across say 500 ohms of R, so at LF there was always some unwanted current FB with a phase shift that reduced the bass response slightly. But even so, where ZC = R is at 12.8 Hz.  With C = 1,000 uF the pole between 500 ohms and 1,000 uF is at 0.32 Hz, so bass response roll off due to R&C bypassing cannot easily be measured and the amount of cathode current FB at signal F is entirely inconsequential.

---

peter wrote:
The thread mentions:
It was 10PM and it was like an instant bliss. It immediately reimbursed me with the exact sound that I was visualizing for quite along time. It slowed down the brutal and unreasonably fast dynamics, it harmonized and coordinated everything and set the correct relation and reasoning between the pitches, it created completely different phenomenal (!) bass and it basically made up the Milq sound in a way in which it is know today.  I can understand that by not using an input coupling cap the amplifier can produce quality bass.
 
Romy replied:
Nope, you misunderstood it. We never even consider using the input-coupling cap. Input input-coupling cap would be necessary to support the language of amplification ceremony but not for the purpose of amplification.

Peter writes:
My understanding had been that input caps are quite okay at amp inputs if they do not reduce the response wanted from the amps.  Some people who want the best bass response possible, believe they can achieve that by using all solid state amps which have no cap couplings within and have a squillion dB of NFB which is greatest at DC.  Some other people are happy with -3dB response at 10Hz, so then using a 0.47 uF cap to feed a bias R of say 200k will produce a "pole", i.e., -3dB point at 1.7Hz, which is so far below the 10Hz that at 10Hz the additional C-R coupling will only make a level change of a fraction of a DB and for human beings it would be quite inaudible.

---

peter wrote:
Do you still believe that the brutal and unreasonably fast dynamics were corrected by the method of biasing the input tube? 
 
romy replied:
I would not use word "corrected". The current Milq's biasing did change the sound very dramatically and in very right direction. Was it "correction" or not I do not know. I personally do not think so because a "correction" presumes the something was initially right and then got screwed. The Milq was not created by "correcting" the wrong things but rather by attempt of creating of the initially right things.

Peter writes:
My understanding had been that there would be no engineering explanation for why well implemented biasing methods in class A signal tubes should change the AC operation and distortion character of the tube.

---

peter wrote:
My question:
If so, then could the same result be produced by using a standard 0.47 uF coupling cap with 150k bias R then -3dB = 2.2Hz, with no perceivable sonic difference. 
 
Romy replied:
Why do you need a coupling cap? The topology, the superstructure of the amlifers, requires a coupling cap to blocs DC from inputs. Does Sound require this coupling cap?

Peter writes:
My understanding had been that when connected to a preamp with DC at its output, this DC could be amplified, and many speakers do not like DC.  Also, slow moving signals below say 10Hz including DC can be regarded by many people as nuisance garbage so a cap is put in at the front end to keep such stray signals out.  Less stray LF noise would perhaps improve the sound.  As long as the pole at the amp input is low enough the cap has not much effect.  Most tube amps are bandwidth limited devices which means that because of their OPT and coupling caps, the F range is limited. So most good tube amps will go from 5Hz to 65kHz at low levels, but some would be quite distressed with a high level signal of 5Hz at full power because of OPT saturation effects; even at 1/10 of full power, 5Hz will cause saturation in some OPT. But luckily, music F rolls off fast below 30Hz.

---

peter wrote:
My question:
Now that you have some time of experience with the amplifiers, have you found this to be a reliable biasing method ? 
 
Romy replied:
Peter, why do you have a question about the reliability? I personally did not experience any problems but what kind problems even theoretically could be? The positive bias tube goes down? So what, when a source is connected then the positive bias tube serves no purpose and the DC voltage is grounded on the source output. You actually could pull the positive bias tube out of the socket while the amps is paling with no problems. The negative bias tube goes down? So what, the tube goes down and stops to regulate. The negative bias supply become form 148V 195V that will raise the fist tube bias from minus 3.4V to minus a few volts higher, effectively closing down the first tube. The raise of the bias will not be anthem too strong to rise above the max bias voltage for the tube. I have many-many time during the assembling and testing when I did all imaginable things with the biasing tubes and I did not detect any issues with reliability.

Peter writes:
My understanding had been that tube aging does not rest well with grid leak biasing and that grid leak biasing was confined to AM radio makers who used it in preference to other methods primarily for cost purposes.

---

peter wrote:
My lack of understanding:
I am not sure of the output impedance for the amplifier, however, if we have a triode with Ra = 80 ohms, its load would be about 320 ohms at the anode, so the RL/Ra = 320/80 = 4.0, so the damping factor is 4. If the load match was for 6 ohms, then the OPT has a Z ratio = 320/6 = 53.3:1, so the source R = 80ohms/53.3 = 1.50 ohms, so if the speaker is 6 ohms, it is being fed from a source of 1.5 ohms, so if the speaker Z fell to 3 ohms at some F the output voltage would fall considerably.  The price for no FB is high output source resistance unless the OPT ratio is high. Higher than standard OPT ratios make an amp inefficient, but the higher the load used with a low load match the lower the THD/IMD becomes, and fidelity is improved.

Peter wrote:
Do you think that it could actually be this lower THD/IMD that has lifted the veil so to speak and simply the improved fidelity that is making you so happy with the results? 
 
Romy replied:
I disagree that there is such a thing as standard OPT ratios. More loaded tube is slower, has more harmonics and more blunt parabola with a tone rolls to it's peach. More idle tube has shaper sound; more steep the parabola profile and fewer harmonics. (Let disregard the inefficient as something that is not important). I would not know what "lift the veil" would mean. The sound after amplifier should have (as close as possible) a correct (acoustic-like) harmonic content, across the spectrum and across the dynamic range. All the rest is the business of the loudspeakers. Of course me, like anyone else made a number of tests how to load the plates of the 6C33C to my specific drivers. In my case, if I am not mistaken the HF/MF and Upper bass channel is loaded to 1200Ohm (single plate of 6C33C) and LF channel to 625 (Full 6C33C).

Peter writes:
Okay, about 600 to 1,200 ohms for 1/2 a 6C33C will do, so about 400 ohms for a whole tube, and that's not far away from a guess of 320 ohms.  400 to 1,000 ohms would give the better fidelity, but less maximum power, which is okay if we have horns where max power is irrelevant.

Peter writes:
My understanding had been that OPT ratios for triodes RL should exceed 4 x Ra and for pentodes and tetrodes which have loads = 0.9 Ea / Ia, where Ea is the idle anode voltage, Ia is the idle current, RL << Ra.

---

Regards,
Peter.

Rerurn to Romy the Cat's Site