Dynaco ST-70: Base Line Testing

In the ST-70 and the MK II/III amplifiers, the screen grid is power the same way, with the exception of different tube choices. In the SCA35, the screen is powered from the cathode of the phase inverter. This approach solves the problem of pentode tube variables, allowing the triode phase inverter section to always be properly biased: If the cathode voltage is low in the triode inverter section, then so is the screen voltage in the pentode section. This causes the pentode's plate voltage to rise, causing the triode's cathode voltage to rise, along with the screen voltage of the pentode. The value of the screen resistor used causes the arrangement to stabilize at the ideal operating voltages for the triode splitter section. It really works quite well.

I hope this helps!

Dave
Revisiting a fascinating thread for a question about the ST-35 variant and similar dual triode front ends lacking the pentode gain stage screen grid correction you describe here.
How do these stay linear with variances in the triode sections between examples, or do they?
 
Another thing, about converting the stereo 70 driver screen source to the SCA-35 arrangement?
 
as far as I understand it, a direct coupled triode pair like the ST-35 uses would have to rely on the first triode being in good operating condition for the phase inverter voltages to be proper. There is local feedback between the cathode of the phase inverter and the cathode of the voltage amp that should serve to help compensate for anything being not quite perfect.
 
The ST-70 design -- which well predates that of the SCA-35 -- could in fact be configured to operate with the DC feedback arrangement that the SCA enjoys. The ST-70 was the natural progression of combining elements of the earlier MK II and MK III designs onto a single chassis for stereo operation, and so the designs of all three are quite similar. It was also quite likely that at the time -- being the peak years of quality vacuum tube production -- the problem of wildly varying pentode characteristics just wasn't the problem it is today, now that we're getting low in the barrel. The SCA arrangement may have just been then the natural progression of design evolution by Laurent, and really considered as icing on the cake at the time. Today, with the situation being much different than it was back in the day, such a design approach may be vital, and a very worthwhile upgrade to the original design.

You would still be left with the noisy and or staticy tri-pent tubes remaining, but using DC for the heaters (to eliminate hum) and a DC FB system for the pentode screen grid (to stabilize the pentode's characteristics) sure goes a long way to make the most out of what's left in the barrel!

Triode tubes -- even the hi mu types like the 12AX7 -- will inherently be more stable from one example to the next with regards to element potentials in a given circuit. The presence of the screen grid in a vacuum tube magnifies everything over that of a triode tube: Based on basic principles, gain is much greater, frequency response is much greater, and output voltage is much greater as well, among other things. As a result then, the byproduct of all manufacturing tolerances will be much greater then, too. Going the other way, a 6SN7 triode will produce more consistent circuit performance from one example to the next than a 12AX7 does, for precisely the same reasons in reverse. Due to it's very low gain (by comparison), manufacturing tolerances will simply not have the impact that they do in a high mu triode, and certainly not to the level that they do in a high gain pentode tube.

At the time of their development, the pentode tube was heralded as a quantum leap forward, and in a great many respects, it was. But it also brought with it new manufacturing concerns as well. Today, as these tubes are more and more depleted, the results of these concerns are becoming more and more apparent.

Dave
 
If the heater filament is hotter than the cathode sheath it is heating, wouldn't the normal course be the electron fllow tendency be from hotter to cooler? So the reverse bias to counter would be to make the heater positive relative to the cathode?

Yes. This is one of the reasons why amps like the Magnavox console amps bias the heaters positive, by sitting them on the cathode voltage of the output tubes as a DC reference. For the output tubes, it makes the heaters the same potential as the cathodes (which is good enough for power tubes), and the heaters in the input tubes positive in relation to the cathodes (which reduces HK leakage).

Regards,
Gordon.
 
There is an electrical limit to that insulation though, which is where the heater-cathode voltage ratings come into play. Broadly speaking if you can minimize the difference, its to your advantage in terms of tube life. The trick with direct coupling is there is always going to be one tube that has a much higher cathode voltage than the rest, so you end up striking a compromise voltage that keeps everything within it's rating.
 
I believe that the reason to positively bias the heaters is to saturate that diode. The heater insulation is substantial, and once saturated, the current flow is DC...which is a lot quieter than the resulting AC from an unsaturated diode...

The tubes usually have h-k leakage limits in their data. The insulation isn't perfect. The best of course are the Bendix Red Bank Hy-G-300, some of which have higher h-k limits than their plates...:)
cheers,
Douglas
 
Last edited:
Revisiting a fascinating thread for a question about the ST-35 variant and similar dual triode front ends lacking the pentode gain stage screen grid correction you describe here.
How do these stay linear with variances in the triode sections between examples, or do they?

The plate voltage of a pentode has a far lower impact on current than that of a triode. In a long tail pair, with resistively loaded triodes, it is comparatively easy to get a very close match in delivered pate voltage. Take the same triode, and build a cascode with their pate voltages determined by the upper element( either a MOSFET, a BJT, a triode, or a pentode ), and you will discover what the g2 does to a pentode. Build that same LTP from pentodes, and keep their g2 voltages identical, and it is quite a bit more difficult to get a match in plate current.
cheers,
Douglas
 
The 6L6 family is fairly broad.
6L6 - metal, 360v max, 19w max
6L6G - glass shouldered tube. Same ratings as metal 6L6
6L6GA - glass straight bottle, same ratings as above
6L6GB - same as above
5881 - 400v max, 26 watts
6L6GC - 500 volts, 30 watts
KT66 - 550 volts, 30 watts
EL34 - 800 volts, 25 watts
6CA7 - 800 volts, 25 watts
KT77 - 800 volts, 25 watts



honestly I'd only really consider trying the 6L6GC downward. An ST-70 runs the plates at 410 volts and at 50ma per tube. Thats an idle plate + screen dissipation of 20.5 watts. A tube that doesn't meet or exceed that will not live a happy life. You might get by with a 5881 but its going to run it awful hard.

Or you could just run the tubes it was designed for and know nothing bad is going to happen.
 
The 6L6 family is fairly broad.
6L6 - metal, 360v max, 19w max
6L6G - glass shouldered tube. Same ratings as metal 6L6
6L6GA - glass straight bottle, same ratings as above
6L6GB - same as above
5881 - 400v max, 26 watts
6L6GC - 500 volts, 30 watts
KT66 - 550 volts, 30 watts
EL34 - 800 volts, 25 watts
6CA7 - 800 volts, 25 watts
KT77 - 800 volts, 25 watts

<snip>

Just a minor point: 6L6G is in an ST16 envelope and 6L6GA is in an ST14 envelope. IOW the GA version is “coke bottle” shaped too, just a bit smaller overall and about same size as 5V4G.
 
Yep -- and I've extended the testing to my own ST-70 as well. The 6GH8A with an adapter works just fine in the stock driver board/stock design of the ST-70. Would that it could be such an easy option for everything that used the 7199!

Dave
Dave, is the fact that the 6GH8 has a triode u of 46, yet the 7199 triode has a u of 17 any concern?
 
It shouldnt since the inverter stage runs at slightly less than unity gain. The Mu only really matters when you need gain.
 
Correct. It just means that there won't be as much loss through the inverter stage, but there will be (and always will be regardless of how great the Mu is of the triode section), a loss through the inverter section when it is configured as a cathodyne/split-load type.

Dave
 
On this driver variant tangent, there are adapter plugs available to allow use of more common tri/pent small signal alternatives to the oop & qualitatively unreliable 7199. The question is whether ideally circuit component parameters should be altered to accommodate specific replacements for best operational results, rather than leaving them as-is?
 
On this driver variant tangent, there are adapter plugs available to allow use of more common tri/pent small signal alternatives to the oop & qualitatively unreliable 7199. The question is whether ideally circuit component parameters should be altered to accommodate specific replacements for best operational results, rather than leaving them as-is?
In most cases it will work just fine.
The important think is that the plate on the pentode shoud be at app. 125V ( 1/3 of 375 'B' at power supply)
To get this it is the 1.5M to the screen that should be adjusted. The margins however are wide even
with 7199 tubes. My guess is that anything between 100 and 150V is acceptable on the plate.
It may be best to measure on the triode cathode as this is less sensitive to loading from the voltmeter.
 
I just checked mine, stock resister values in a Dynakit 6GH8A PC-3B board (stock circuit), at the triode side cathode I measured 141 volts. Additional information, I’m running with SS rectification with a 50 ohm dropping resister, 116 AC volts at the wall.
 
Last edited:
I'd still like to know how that shakes out if you were to tweak the screen resistor to restore the voltages back to Dyna specs. I'm just not all that motivated to buy an ST-70 to test it for myself.
 
Back
Top Bottom