An EFB controlled output stage is most useful in a two channel amplifier where both channels are powered from one common power supply, and where there is significant power supply droop or sag from idle to full power when both channels are driven, and where the output stage is sensitive to power supply droop.
Power supply droop becomes greater the more load the power supply is asked to support. In a two channel configuration, when both channels are driven to full power, it can cause the power supply voltage feeding the output stages to sag to such a degree that it causes a shift in output stage operating point. This leads to increased distortion.
One way of course to address this is to regulate plate, screen, and bias voltages. In a well designed regulated power supply, these three voltages are held rigidly in place at any power output level. Thus the operating point of the output stage is fixed regardless of power output level. Many commercial amps use regulated power supplies, and certainly it works well, but it can be costly and can generate excessive heat.
EFB achieves similar results as full-on regulation. It does it by allowing the power supply voltage to sag as it will and allows bias and screen voltages to "track" power supply voltage sag. Thus even though there is a power supply sag as the amp is pushed to full power, the relationship between power supply voltages (plate, screen, bias) are maintained. Thus from the perspective of the output stage, it can't tell the difference between traditional regulation and EFB regulation since in both cases the voltage relationships are held constant. Thus you get the same low output stage distortion with an EFB regulated output stage as you do with a traditionally regulated power supply.
Another major benefit of EFB regulation is the cost to implement is significantly less, as well the components to operate the EFB regulator run much cooler than a traditional linear regulator.
Yet another benefit to EFB is you can typically tune the output stage to idle the output tubes cooler while still achieving the lower distortion at full power output. There is typically a distortion null that can be tuned for by adjusting bias (fixed bias) of the output stage. This can preserve output tube life.
I have discussed EFB in the sense of tracking B+ "sag," but it also tracks B+ boost, or said another way: mains fluctuations.
What kind of distortion improvements can be expected with EFB regulation vs. the same amp with no regulation? Mid band improvements of a typical 60 watt dual channel KT88 push-pull amp with your average CLCRC power supply design range from about 200% to 300% improvement. Real numbers at 1 KHz: approximately 0.4% THD without to 0.15% THD with, both channels driven to max power. At the frequency extremes, improvements are not quite as pronounced, but an EFB controlled output stage should see < 1% THD across the board from 20 Hz to 20 KHz at max power, both channels driven, IF the output transformers are of top quality. Certainly that's not as fabulous as the world class amps (like McIntosh), but the cost to implement EFB is quite a bit less than traditional regulation or unity coupled output stages.
Now on the other hand, if you build a tank of an unregulated power supply, say with choke input filtering, where the power supply sag isn't going to be as pronounced, you will see little to no benefit from adding EFB, the same way you would see little to no benefit of traditional power supply regulation.
Another question one might ask is: Does EFB add its own sonic signature? In other words, while EFB reduces output stage distortion, does the amp have the same sonic signature with EFB as without EFB? That's a harder one to answer since it would require building two identical amps, one with EFB and one without, and a partner to do double-blind testing with you to make that determination. In my experimentation, I don't think it colors the sonic presentation at all.
Power supply droop becomes greater the more load the power supply is asked to support. In a two channel configuration, when both channels are driven to full power, it can cause the power supply voltage feeding the output stages to sag to such a degree that it causes a shift in output stage operating point. This leads to increased distortion.
One way of course to address this is to regulate plate, screen, and bias voltages. In a well designed regulated power supply, these three voltages are held rigidly in place at any power output level. Thus the operating point of the output stage is fixed regardless of power output level. Many commercial amps use regulated power supplies, and certainly it works well, but it can be costly and can generate excessive heat.
EFB achieves similar results as full-on regulation. It does it by allowing the power supply voltage to sag as it will and allows bias and screen voltages to "track" power supply voltage sag. Thus even though there is a power supply sag as the amp is pushed to full power, the relationship between power supply voltages (plate, screen, bias) are maintained. Thus from the perspective of the output stage, it can't tell the difference between traditional regulation and EFB regulation since in both cases the voltage relationships are held constant. Thus you get the same low output stage distortion with an EFB regulated output stage as you do with a traditionally regulated power supply.
Another major benefit of EFB regulation is the cost to implement is significantly less, as well the components to operate the EFB regulator run much cooler than a traditional linear regulator.
Yet another benefit to EFB is you can typically tune the output stage to idle the output tubes cooler while still achieving the lower distortion at full power output. There is typically a distortion null that can be tuned for by adjusting bias (fixed bias) of the output stage. This can preserve output tube life.
I have discussed EFB in the sense of tracking B+ "sag," but it also tracks B+ boost, or said another way: mains fluctuations.
What kind of distortion improvements can be expected with EFB regulation vs. the same amp with no regulation? Mid band improvements of a typical 60 watt dual channel KT88 push-pull amp with your average CLCRC power supply design range from about 200% to 300% improvement. Real numbers at 1 KHz: approximately 0.4% THD without to 0.15% THD with, both channels driven to max power. At the frequency extremes, improvements are not quite as pronounced, but an EFB controlled output stage should see < 1% THD across the board from 20 Hz to 20 KHz at max power, both channels driven, IF the output transformers are of top quality. Certainly that's not as fabulous as the world class amps (like McIntosh), but the cost to implement EFB is quite a bit less than traditional regulation or unity coupled output stages.
Now on the other hand, if you build a tank of an unregulated power supply, say with choke input filtering, where the power supply sag isn't going to be as pronounced, you will see little to no benefit from adding EFB, the same way you would see little to no benefit of traditional power supply regulation.
Another question one might ask is: Does EFB add its own sonic signature? In other words, while EFB reduces output stage distortion, does the amp have the same sonic signature with EFB as without EFB? That's a harder one to answer since it would require building two identical amps, one with EFB and one without, and a partner to do double-blind testing with you to make that determination. In my experimentation, I don't think it colors the sonic presentation at all.
Last edited:
In one case, I upgraded the PSU of an ancient AB2 amp which had DC coupled output valves. This requires a more stable supply than normal.
You could create the conditions above simply by using the original choke input filtered anode supply but swapping out the rectifier to silicon, using gas regulators for the screen supply, and large zener diodes with a long time constant for the bias supply return for the ground supply.
It wasn't expensive, and it generated less heat then the original.
It also reduced distortion by a quantum leap from about 4% at 30W to below 0.5%, doubled the clean output power of the amp and forced up the maximum output power from a very dirty 60W to well over 85...
Thanks for the info - I am planning an EL84 PP UL type build, but I think I will just go with a typical fixed bias arrangement, and a heavy duty power supply.
Going to use Transcendar Z565's
Possibly the Tubecab circuit with 6DJ8+LED biasing on the input stage cathode (I am assuming it is better than the Dyna ST-35 circuit?)
I am also considering a lower gain version of the Mullard 5-10
Going to use Transcendar Z565's
Possibly the Tubecab circuit with 6DJ8+LED biasing on the input stage cathode (I am assuming it is better than the Dyna ST-35 circuit?)
I am also considering a lower gain version of the Mullard 5-10
A reasonable plan. You can always change the output stage biasing later (add EFB or whatever).
I've never used the Transcendar Z565's. Note they are not clones of the Dynaco iron though. The Dynaco Z565's or the clones are excellent iron.
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.