03-11-2016, 01:01 AM
1. An issue with a 1M pot that hasn't been mentioned yet is hum pickup. Depending on the physical layout, a high value pot might mandate a screen cable that would not be needed with 50k, say. Ultimately though, there is nothing wrong with Andy's choice of a 50k pot given his design objectives.
2. LF rolloff is a complex issue, and one I was avoiding until the other problems have been resolved. But in short, it's usual to define the required -3dB at one part of the chain. All other "poles" would be an octave or more lower in frequency so that they are spaced away from this point, meaning that the LF rolloff is a nicely defined "Gaussian" function formed by that one pole. For what it's worth, I did write some words about this subject some 15 years back - http://www.markhennessy.co.uk/preamp/ana...search.htm - where I show that setting the main LF rolloff with the input capacitor is quite feasible given an anticipated wide range of source impedances. At the same time, I also show that setting the HF rolloff point is not feasible at this point - any HF filtering at the input should be regarded as ultrasonic/RFI filtering, and the HF rolloff proper needs to be defined elsewhere.
So here, while I'm happy to reassure Andy that his understanding of how to calculate Cin was good, I would probably change it to 100n (~32Hz) or 47n (~68Hz), depending on what the output transformer dictates. The other time constants in the circuit are in the sub-10Hz region, so probably OK, but increasing those caps a little wouldn't hurt. As an aside, it's well-known that using electrolytics for filtering is bad for distortion, so it's normal to "oversize" them (totally academic here, naturally - but the idea of positively defining the LF -3dB point with a nice film capacitor at a known part of the circuit and spacing all the other poles well away is common practice and a good principle to learn).
As I said at least once earlier, some rail decoupling between stages would be a good plan.
2. LF rolloff is a complex issue, and one I was avoiding until the other problems have been resolved. But in short, it's usual to define the required -3dB at one part of the chain. All other "poles" would be an octave or more lower in frequency so that they are spaced away from this point, meaning that the LF rolloff is a nicely defined "Gaussian" function formed by that one pole. For what it's worth, I did write some words about this subject some 15 years back - http://www.markhennessy.co.uk/preamp/ana...search.htm - where I show that setting the main LF rolloff with the input capacitor is quite feasible given an anticipated wide range of source impedances. At the same time, I also show that setting the HF rolloff point is not feasible at this point - any HF filtering at the input should be regarded as ultrasonic/RFI filtering, and the HF rolloff proper needs to be defined elsewhere.
So here, while I'm happy to reassure Andy that his understanding of how to calculate Cin was good, I would probably change it to 100n (~32Hz) or 47n (~68Hz), depending on what the output transformer dictates. The other time constants in the circuit are in the sub-10Hz region, so probably OK, but increasing those caps a little wouldn't hurt. As an aside, it's well-known that using electrolytics for filtering is bad for distortion, so it's normal to "oversize" them (totally academic here, naturally - but the idea of positively defining the LF -3dB point with a nice film capacitor at a known part of the circuit and spacing all the other poles well away is common practice and a good principle to learn).
As I said at least once earlier, some rail decoupling between stages would be a good plan.
