GM-70 Amplifier

[Stuart Polansky]

Ray, what Walt is saying is that if you use fixed bias, then best dynamics is with a grounded cathode. It's been said many times that cathode bias SOUNDS more dynamic even though it measures less powerful than fixed bias.

Cathode bias provides local feedback which reduces distortion.

My own thought is that because cathode biased stages are easier to drive, the driving stage has an easier load and itself is the source of the perceived dynamic difference. It's pretty easy to great the positive effects of unloading a driver stage.

I like the solution of a low impedance driver, DC coupled to the output grid.

Stuart

[Pelliott321]

just to be insane here is a circuit to have both, it seems to used on Fendor Guitar amps

Attachments

[SoundMods]

As I understand it -- cathode bias tracks the music. I wish I could find the white paper on the subject. I know it doesn't make sense.

I think the problems with cathode bias is that typically it is not done correctly. In the case of the GM-70 amps. -- David did his homework. High quality signal-path-worthy caps and big-ass cathode resistors. Why high wattage cathode resistors? Hysteresis. Using resistors as a some designers have with the wattage rating as low as 2 to 5 watts won't fry, but their values will change or modulate depending on the signal strength -- meaning that the static bias will be jumpy. Over-rated bias resistors stay "cool and collected" letting the cathode bias caps run the tubes at full gain without changing the static bias.

It was Sid Smith's preferred method for the Radio Craftsman amplifiers. I have a pair pf C400s that I hot-rodded and they rock with only 6V6's. The original bias resistors were 4-watt carbon composition types sized way beyond the cathode current. Changing them to 25-watt non-inductive wire-wound resistors was remarkable in that is was like removing layers of distortion similar to one's experience when up-grading capacitors.

It's difficult to explain, but hysteresis effects come in many forms. Like dielectric absorption and dissipation factor with capacitors.

As far as fixed bias is concerned, it works best with a grounded cathode. I played that game when I utilized the steel and transformers from a Dynaco ST-70 and did my own thing. Again another white paper discussed that effect. A lot of tube guitar amplifiers that have fixed bias have grounded cathodes.

This hobby of ours -- OY VEY! The devil is in the details -- every damn thing has an effect!

A full differential amplifier feeding a pair of fixed bias EL-34s with grounded cathodes was remarkably dynamic.

[ratbagp]

Thanks for the answers.

ray

[SoundMods]

You're welcome.

[dberning]

A combination of fixed bias and cathode bias makes a lot of sense if you don't want to get into the complexities of servo bias systems such as I use. The main reason that you don't want purely fixed bias on triodes is the low plate resistance makes them too dependent on plate supply voltage, especially problematic when using an unregulated supply. Pure fixed bias works well on class AB pentode amplifiers because the plate resistance is high and variations in supply voltage do not cause much change in plate current. Variations in screen voltage actually cause more variation in plate current, and sometimes screen voltage is regulated to good effect. Another reason fixed bias is desirable on AB amplifiers is that the average plate current varies quite a bit and cathode bias does not allow for this and the effective bias is wandering all over the place-not good. Because you have a class A amplifier, cathode bias is a good choice, but you do sacrifice power because you gobble up a lot of the effective power supply in the cathode resistor. The combination of some fixed bias and some cathode bias allows some improvement in efficiency while still maintaining some degree of dc feedback to keep the plate current more stable with respect to supply variations and tube variations. I would recommend a half-and-half approach to things--perhaps taking your cathode resistor down to 350 to 400 ohms and applying a negative bias of about half of what you now develop across your 800-ohm resistor. You should bypass the resistor as you do now. This keeps the plate resistance low and improves speaker damping. When choosing the minimum value of bypass capacitor, most people make the mistake of assuming the applicable time constant is determined by the cathode resistor. In fact, it is the parallel combination of the cathode resistor and the cathode impedance, which is non linear. I prefer to oversize the bypass and 1000 uF is practical. In the early days of amplifiers they did not have capacitors of anywhere near the quality we have today, and it does not make sense to skimp on value here. I would not worry about inductance in the cathode resistor, simply choose a rugged wire wound type as the ESL of any modern capacitor will swamp out any inductance issues of the resistor. If you are really worried about this, you might parallel a bunch of ceramic 0.1 uF rated at around 200 volts if you can find them. I say that because many ceramic capacitors have a voltage-dependent value characteristic.

Hope this helps.
David

[David McGown]

One thing about a high voltage amp using GM70s rather than, say a lower voltage 300B amp using a similar bias voltage, is that less total power is wasted across the cathode resistor. At my GM70 operating point with a plate voltage of 1025V and a voltage drop across the cathode resistor of around 85V, I lose only 85V/1025V or 8.3% of total power (voltage) available to the output tube across the cathode resistor. On a 300B amp with a 450V supply and similar bias voltage of 85v, the loss is 85V/450V or 18.9% power (voltage) loss. Therefore there is less penalty associated with using cathode bias on the GM70.

I am not in a rush to implement a mixed bias scheme, since I can currently adjust the GM70 bias with the adjustable (inductive) cathode resistor. I just have to flip a heavy amp upside down to adjust the resistor tap, and because the power supply caps are so tall (which the amp rests on while upside down), I can operate the amp with the GM70 installed. Though it would nice to just turn a pot accessible from the top of the amp to bias.

David

[SoundMods]

I was wondering when David Berning would join the dialog.

I disagree with the 1,000-ufd cathode bypass, but learned a boat load of tube tech.

Then there is my Cary CAD-805C rev.2 amps. They have 300B drivers with fixed bias that feed inter-stage transformers with capacitor bypasses across the primary and secondary windings. That scheme allows the 300Bs to drive the 845s when their grids go positive. Hence the 50-watt rating. The 845s have cathode bias with noble film cathode bypass caps taking things down to about 3-Hz. According to the specs, they run at class A up to about 20-watts output then run at class A/B up to the 50-watt max by virtue of the inter-stage transformer scheme.

It would be fun to get Dennis Had’s take on his design decisions.

These amps. have a front end based on 6SL7s run cascode (Telefunken E80CCs for some reason sound superior to even ECC35s). The CAD-805s are sweet and musical and yet scary transparent with slam.

They do not provide the usual smoozy triode SET sound. Yes I modified them – but I didn’t need to do very much as Dennis Had was there first with rabid audiophile upgrades.

So – you have the drivers run with fixed bias (50-ma has turned out to be optimum) and the power tubes run with cathode bias.

Attachments

CAD-805C ver.2 Full Mods.jpg (48.08 KiB) Viewed 37197 times

[tomp]

One thing about a high voltage amp using GM70s rather than, say a lower voltage 300B amp using a similar bias voltage, is that less total power is wasted across the cathode resistor. At my GM70 operating point with a plate voltage of 1025V and a voltage drop across the cathode resistor of around 85V, I lose only 85V/1025V or 8.3% of total power (voltage) available to the output tube across the cathode resistor. On a 300B amp with a 450V supply and similar bias voltage of 85v, the loss is 85V/450V or 18.9% power (voltage) loss. Therefore there is less penalty associated with using cathode bias on the GM70.

I am not in a rush to implement a mixed bias scheme, since I can currently adjust the GM70 bias with the adjustable (inductive) cathode resistor. I just have to flip a heavy amp upside down to adjust the resistor tap, and because the power supply caps are so tall (which the amp rests on while upside down), I can operate the amp with the GM70 installed. Though it would nice to just turn a pot accessible from the top of the amp to bias.

David

[color=#0000FF][/David you might want to make a jig out of scrap wood to hold the amp upside down. I did that for the Cherry Bomb amp that David and I built because I did not want all the weight of the transformer and chokes to bear down on the heat sinks when it was inverted. Here is a picture of that. It is make of junk wood with some felt hot glued on to prevent scratches.

Tomcolor]

[tomp]

David:

Sorry the photo did not come over the first time. I had to reduce the size.

Tom

Attachments