What about the big AC current flowing through the cathode bias bypass capacitor? Doesn't it increase distortion? I don't like electrolytics except where used for coupling into high impedance, moreover a cathode resistor bypass is a hot environment, potentially reducing the life of the capacitor.
Class AB1 no bypass = more distortion. The bypass capacitor can be omitted for pure class A. That's the typical case.
This thing about how bad electrolytic caps keeps coming back very often but in reality is not true IME if the capacitors are really new (not 10-20 years old) and good quality. Electrolytic capacitors show their limits in other applications like signal decoupling and first PSU filter cap but for cathode bypassing they are fine. Elna Silmic, Panasonic FK work very well. It's important that they are really new! Recently Audio Note has released a "new" kind of electrolytic which they claim to be very close to the Black Gate while they will be using the real Black Gate substitute only in their top products. This lower cost variant is the same except it doesn't use impregnated paper. Frankly I cannot see how these would be better than the Elna. The BG were too expensive to make and probably that's the reason why production ended....
If one doesn't want any electrolytic at any rate then can use MKT caps that are found in lower voltage variants up to 100 uF and are still cheap enough. Never tried this though....
EL34 triode example. Pure class A operation with 10K plate-to-plate. Anode voltage 400V and separate cathode resistors of 440R each. Cathode resistors un-bypassed. One needs 430V supply for 70 mA current per tube. This gives 14W with less than 1% THD. Very good but the voltage drive is 68V grid-to-grid which is achieved with bias shift and the price to pay is rather high as 70 mA is a lot of current for this power level. With 10K and the right anode voltage it might make some 25W in class A! If the current is reduced to be just enough with self-bias the bias shift in proportion will be more.
The very same amp with fixed bias would only require 55 mA. Don't know if pure coincidence but at 400V/55mA the bias is about 34V....
Actually should have written at least -34V. More typically at 400V/55 mA the bias of the EL34 triode is between -34V and -37V. So the Class A amp with fixed bias can achieve up to almost 16W with 33% less anode dissipation.
This thing about how bad electrolytic caps keeps coming back very often but in reality is not true IME if the capacitors are really new (not 10-20 years old) and good quality. Electrolytic capacitors show their limits in other applications like signal decoupling and first PSU filter cap but for cathode bypassing they are fine. Elna Silmic, Panasonic FK work very well. It's important that they are really new! Recently Audio Note has released a "new" kind of electrolytic which they claim to be very close to the Black Gate while they will be using the real Black Gate substitute only in their top products. This lower cost variant is the same except it doesn't use impregnated paper. Frankly I cannot see how these would be better than the Elna. The BG were too expensive to make and probably that's the reason why production ended....
If one doesn't want any electrolytic at any rate then can use MKT caps that are found in lower voltage variants up to 100 uF and are still cheap enough. Never tried this though....
You will never achieve the same with ANY pentode. The point about the 45 (or other linear triodes) is well in topic and it basically means that for class A1 or high current class AB1 operation fixed bias is almost redundant at least in PP stage....less in SE. The problem with pentodes is their highly variable g2 current regardless of class A or class AB. You will get bias shift even in pure class A operation. For this reason a common cathode resistor is only seen where economy is important and fixed bias is the most common solution for class AB operation.
EL34 triode example. Pure class A operation with 10K plate-to-plate. Anode voltage 400V and separate cathode resistors of 440R each. Cathode resistors un-bypassed. One needs 430V supply for 70 mA current per tube. This gives 14W with less than 1% THD. Very good but the voltage drive is 68V grid-to-grid which is achieved with bias shift and the price to pay is rather high as 70 mA is a lot of current for this power level. With 10K and the right anode voltage it might make some 25W in class A! If the current is reduced to be just enough with self-bias the bias shift in proportion will be more.
The very same amp with fixed bias would only require 55 mA. Don't know if pure coincidence but at 400V/55mA the bias is about 34V....
Actually should have written at least -34V. More typically at 400V/55 mA the bias of the EL34 triode is between -34V and -37V. So the Class A amp with fixed bias can achieve up to almost 16W with 33% less anode dissipation.
In post #3 DF96 writes, "Using CCS bias is even worse." So using a constant current sink like a TL783 is worse than using fixed bias? If this is true it's a dissapointment.
It seems the best way to bias would be to use a solid state device at the cathode to read the current and then adjust the negative voltage at the grid to regulate the current actively. Unfortunately my knowledge of solid state devices doesn't extend that far.
It seems the best way to bias would be to use a solid state device at the cathode to read the current and then adjust the negative voltage at the grid to regulate the current actively. Unfortunately my knowledge of solid state devices doesn't extend that far.
Last edited:
It is true. It causes dynamic distortions because bias follows by the sound signal envelope, especially when peaks approach clipping. Fixed bias is the best, but needs constant attention. Cathode resistor shunted by a cap is a good compromise. Disclaimer: I mean class A amps.
"Fixed bias", as usually understood, fixes the Voltage.
CCS fixes the Current.
So they are quite different. In particular fixed grid bias allows current to rise with signal, CCS does not.
Yes, the amp has a negative voltage applied to the grids and is adjustable for each output tube.
So if the current is fixed with a CCS at the cathode the amp will remain in class A until the tubes run out of current and then it goes to hell rather than going into class b?
And the bias voltage is only sets quiescent current. Thanks guys I learned something I didn't know.
So if the current is fixed with a CCS at the cathode the amp will remain in class A until the tubes run out of current and then it goes to hell rather than going into class b?
And the bias voltage is only sets quiescent current. Thanks guys I learned something I didn't know.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.