| Posted by TPStatic | May 30, 2012 - 12:50 pm |
| Hi: Dave Hi Dave I'm pumped about installing this modification in my cathode based Fisher 101C with 7591 tubes. Can you suggest the proper starting values to pllace on the input side of the LM337? The schematic calls for 40V for the bias settings. Thanks large David T Hobday |
| Posted by Dave | May 07, 2012 - 11:44 pm |
| Hi Ricco -- Regarding EFB(tm) circuit action, it is true that any traditional fixed bias output stage certainly acts as a shunt regulator to its power supply as the B+ level might rise or fall. I mention this point very prominently in the article introducing the EFB concept. However, EFB acts much more like a cathode biased stage with regards to this issue. That is, as the B+ is elevated, the tendency towards increased current draw in the output stage is largely compensated for with EFB, resulting in minimal change in quiescent current flow. When the B+ is reduced, compensation again takes place to largely maintain the original current flow. In this way, the tubes are constantly being adjusted for changes in their operating environment, in much the same way a traditional cathode biased stage is automatically adjusted. However, much unlike cathode bias, the bias voltage of the output stage is quite stable with respect to changing current flow conditions due to dynamic excitation of the output stage. As a result, the stage is now operating exactly as any traditionally fixed bias output stage would. This is much unlike the action of a resistively cathode biased output stage, where increased power output demands raise the bias voltage of the output stage. Therefore, from a regulation standpoint, the voltage at the cathodes is absolutely regulated against changes produced from changing current draw demands from the tubes themselves, but automatically adjusted for changes in operating environment. It is the "Enhanced" portion of the design, where the regulator's adjust terminal reference voltage is developed from the B+ supply, that allows EFB circuit action to provide the best of both cathode and fixed bias operation. Developing the adjust terminal reference voltage from the output of the regulator would destroy this feature, causing the regulator to operate in a "dumb" fashion at a fixed voltage set point only, regardless of changes to the B+ supply. The dangers of this type of operation (regulated bias with unregulated B+) are also well covered in the article. It is (in part) because the output tube operating point is continually adjusted for the prevailing B+ conditions that such a significant reduction of distortion was achieved in spite of both channels being driven to full power output in the EFB modified amplifier. Finally, no doubt a negative voltage bias supply could be developed from one side of the HV winding as you suggest, and has been done in many a guitar amplifier design. And while that approach would have some compensating effects as the B+ conditions varied similar to the way the EFB circuit operates, it would not operate with the same level of accuracy or response as the EFB circuit does. The performance enhancements that the EFB(tm) modification has produced in the Dynaco SCA/ST-35 amplifiers is now well established, and has been verified by numerous people worldwide. I hope this helps to explain the circuit's action more clearly for you. Thanks for your comments! Dave |
| Posted by Ricco | May 07, 2012 - 03:28 pm |
| It seems the current configuration is using the regulator much like a follower to provide about 1.7V more at the cathodes than at the slider of the bias pot. It's making the output stage act more like a shunt regulator. If the line voltage goes up, it'll tend to load the supply more, stabilizing the B+ some, while shifting the operating point. It lacks the usual feedback path to actually regulate the cathode voltage. In fact, if the B+ falls some with current drain, the falling reference will reduce cathode bias voltage, increasing the current a bit more - a slight positive feedback bias situation. Wouldn't it work out better to simply replace the 360K resistor with a smaller one from the output pin of the regulator? That would also avoid most of the waste in the resistor, permitting use of a 1/4 watt part. The circuit here is a big improvement over simple cathode bias, since the bias doesn't rise excessively when power goes up. But one could get more power output avoiding the voltage lost in bias the cathode by using conventional adjustable (negative) grid bias instead. Some may have thought that difficult without a lower voltage winding for the bias supply, but It's easy to develop that voltage without a special coil or tap. A non-polar capacitive divider from one side of the tapped h.v. secondary can provide reduced a.c. to the bias rectifier/regulator. Since caps won't pass d.c. current, a half-wave bias rectifier would need an additional diode from the divided a.c. to ground (cathode to ground) to provide current flow on the other half-cycle. That approach can be very efficient since a capacitive divider doesn't dissipate power. |
| Posted by Lydia | April 10, 2012 - 04:47 am |
| Hi Dave, I have a question about the possibilty of bi-wiring an Dynaco SCA-35. I am in the process of building a set of Zigmahornets, and I want to Bi-wire them. I have taken note that the output transformers, have (original ones) 8,16 ohm connections. I am going to use a tang band 4" full range and on top or the speaker a vifa slight horn loaded soft dome tweeter. the tweeter will be facing the back wall from about 2-feet and it is on a 45 deg. angle too. The vent and full range will face the front where you will be listening and I can assure that this set up is going to be nice! when broken in!!. I want to hook up the full range to the tang band FR at the 8 ohm tap. used the 16 ohm tap for the tweeter but I don't know is what capacitor or resistor to use in that series positive line to the tweeter. any suggestions would be appreciated. P.s. I will be using your EFB eventually. but I believe the drivers being separated by the transformer itself should lend to less frequency distortion and dedicating an ohm for each driver without a crossover!. hope I can have this..... Lydia |
| Posted by Dave | March 14, 2012 - 10:22 am |
| Hi Bill -- The solution you propose is excellent. The 560K resistor will in fact be dissipating the lion's share of the heat, but it is very little heat to begin with. The original components were well overrated so as to minimize any bias drift from any heating of these components. In your case, the 560K resistor will dissipate just under .25 watt, so the 1 watt rating will be just fine. If possible, metal film units will be best, again to address maximum stability. Good luck with your installation! Dave |
| Posted by Bill | March 11, 2012 - 09:26 pm |
| Sorry, I have to say one more thing: with unequal values in parallel, I'm not sure if the 2W rating will be maintained as would be the case of equal value resistors. I think I need help on this one. |