DCAudioDIY.com

Interesting. You are using fixed bias with the tubes, despite RCA's admonition against doing so. So is Atma-sphere. How do you get away with it? Very low grid circuit resistance? You mentioned DC coupling from the driver to the O/P grid. That makes sense, avoiding blocking in the caps and so forth, I assume this helps? I don't know why the tube isn't considered suitable for fixed bias?

Thanks again.

Stuart (the ignorant!)

Stuart,

The 6AS7 is designed for voltage regulator service and as such there is feedback to control the operating parameters of the tube. In my ZOTL I use servos to keep the tube where I want it to be, and this allows me to have dc coupling. Operating this tube (dc) open loop is not recommended as it is a low-impedance device. This is especially bad with an unregulated power supply. Some cathode dc degeneration is needed if a servo is not used. You can either protect things with fuses and hope they don't blow too often, or you can use a substantial cathode resistor and cap bypass it if you like. The larger the resistor the more stable it will be. IF you idle 42 mA as I do, you probably should have a cathode resistor of around 300 ohms in each cathode to give a reasonable degree of stability. I don't believe the commercial OTLs that use low-value resistors are stable, and they may have some success by selecting and matching tubes, but I would not do it.

David

http://www.atma-sphere.com/Products/#M-60


Text from the ad copy:




If they had only been around for a few years, the copy might be taken as worthless, but they've been around since 1976. It seems intuitive that an unstable amp would have failed in the marketplace many years ago. But maybe not. Audiophiles are like Mercedes Benz customers: pay no attention to the man behind the curtain. The same Mercedes failures occur repeatedly, yet the customers flock back and remark at how reliable they are. (I have a friend who made LOTS of money as a service manager for a MB dealer).

_________________
Walt

Let me clarify where one point of instability comes from. Let's take characteristics from the RCA tube manual which has the plate resistance at 280 ohms with 125 mA current (one plate) supply voltage 135 V, cathode resistor 250 ohms. Now change to fixed bias with no cathode resistor. Now the actual plate voltage is 103 V after subtracting the voltage otherwise generated across the cathode resistor. Now suppose the plate voltage increases by 10 % because the power supply is unregulated as is true of the commercial OTLs. Because the plate resistance of this tube is a very low 280 ohms, a delta 10 % is 10.3 volts which is a delta 36.8 mA increase in plate current!!!!!

Now what also tends to happen with tubes in general is that the additional heating causes further increases in plate current due to the presence of gas. So a runaway condition can occur. This is why fixed bias operation is not recommended. The folks that designed these tubes knew better. I have heard lots of stories over the years from dealers and others who have owned or sold some of these commercial OTLs about destroyed speakers and other things.

David

Not sure where my math is wrong, but I could very well be.

If the 6AS7 passes 0.125A when 103 volts is applied, then the static plate resistance at that point would be 824 ohms. If the plate voltage is increased 10% to 113.3, then cathode current should be 0.1375 amps, or an increase of 12.5mA (or 10%). I realize that the apparent plate resistance will change a bit, but it shouldn't be that much? Wrong?

Now, three other things:

1) There is a small 5.0 ohm cathode resistor, not zero, but still very small, so admittedly feedback due to degeneration across the resistor will be very small.

2) Not only is the B+ unregulated, so is the driver B+/-. So, if B+ increases by a percentage due to line voltage fluctuation, so does B-.

3) The transformers in use have a very low DCR and ample current capability, more than twice the peak current that the amp can use at maximum output. So, I think that a 10% voltage variation is an unrealistic scenario, unlike when using a Dynaco power transformer

Bottom line is, I will find out for myself, because I'm going to take the plunge.

My favorite power supply lack of regulation is this comparison.

Many moons ago I rebuilt a Dyna ST70 (what a piece of junk, except for the OPT). I solid state rectified the beast.

Voltage output, no load, >614VDC!!!!!!!
Voltage output with 30mA x 4 EL34 cathodes, plus under 20mA of input circuitry, 440V!
This pretty much defines junk iron.


Harman/Kardon Citation II. I use a small 2A transformer, 24V, to kick up B+. The power supply is a doubler. DCR should be a real killer in a doubler, I think.

Voltage output, no load= 540 VDC.
Voltage output with ~50mA of input circuitry and 400mA of idle current (I biased the KT88s to the stock 100mA rating, just for the test, normally run them at 60-65mA) =526 VDC.
Now THAT is good iron and great "regulation" for a non regulated supply!

Where's the Like button...?

Stuart,

One must take the small signal dynamic impedance from the tube transfer characteristics. This is where the 280 ohms comes from. I did not make this up it is the tube manual spec for the given operating conditions. By the way, the plate dissipation in the above case goes from 13 watts to 18 watts!

Yes, the grid voltage if unregulated and allowed to track the unregulated plate voltage compensates somewhat but not fully. Somewhere I did the estimate and you get back about half for the particular case I considered. With some clever use of zener diodes you could probably get the grid voltage variation to cancel the plate current variation with respect to line voltage variation. But there is still the question of stability due to tube gas/ temperature.

I also looked at some 300B curves I have and found that the same issues prevail.

David

A few years ago (well, 10 or 12) I built a PP 6538 amp using direct-coupled CF driver. with fixed bias. I stole an idea from Bruce Wenger (anyone remember him?) and stacked the output tubes in an SRPP configuration. 5K p-p load, about 350 VDC. A friend of mine still says it was the best thing I ever built, and it did sound great, very transparent and sweet. The 6538 only needs about 40 volts drive, as opposed to the 6336. Wish I had still had my 6538s, don't know what I did with them when I moved. It was probably 10-12 watts output. I'm thinking about gathering a few more and trying something like that again.

For a while Bruce was marketing an SE amp with 6AS7 or 6538 in this stacked configuration--about 6 watts. He used a parallel 12AX7 input tube. Very simple, and really good-sounding.

Don't know if this helps or not.