I was planning to use the 50nF caps etc as per the GEC 30W schematic
http://www.jacmusic.com/KT88/SCHEMATIC3.gif
I will use the amp with a 24dB/Oct crossover @ 120Hz so it will not see much signal below 100Hz. Despite this I do want to build an amp that can be used full range when neccessary.
Keeping (one of) the interstage caps small seems to be the best way of avoiding "motorboating". I don't see how an input LF pole will help, it's outside the f/b loop and if something is going to oscillate it will "find" a signal from somewhere to kick it off.
Thinking aloud...
Why not introduce a modest imbalance in the o/p stage to deliberately introduce some 2nd harmonic? This will "fill in" the even harmonics so that the spectral pattern fits the smooth decay which is considered desireable.
See here:
http://www.google.co.uk/url?sa=t&rc...sg=AFQjCNFQOawnNr_1IPcpXMS9UQItBTpcGw&cad=rja
Probably best to do that in the driver stage of the amp. You have better control of it, particularly keeping it at the lower harmonincs, while keeping the output stage simple and optimized for output.
This is common misconception.
1) Double C cores can be easily assembled with micro air gap.
2) Toroids tolerate much higher flux density, thus compensating extra sensitivity to DC imbalance. I have a test pair of output toroids which handle 10% of imbalance without any problem.
The AC global nfb loop creates a closed loop characteristic; (similarities to a PLL damping transient Q; get the timeconstants wrong and it oscillates);
The conditions for negative feedback imply a phase difference of 180° which equates to 12dB/octave; however a simple R/C creates a 90° phaseshift of 6dB/octave.So the closed loop of the amp has to designed in that no more that 10dB per octave though the unity gain point....(this is the tricky bit that many don't understand) then the amp is stable.
If the amp is a "close design" it could be modulated into oscillatory mode by a hefty input signal that can just tip it off the stability boundary; the receipe of gain,phase and frequency being the key components of a Bode plot.
Radiotron Hd bk 4th Ed chapter 7;Neg feedback page 315 ++.
The output transformer has a frequency reliant pole, cut off droop as do all the coupled R/C stages....this is where Williamson had the problems. SO it becomes impossible to govern a 4 stage R/C amp with strong global NFB throughout the whole frequency range, so it's reduced at both extremes with the interstage step cap values. Hence the loop gain curve of the 200W amp drops at the upper and lower ends.
Understanding stability and Loop filter timeconstants is difficult, but a simple squarewave can uncover a raft of problems.
A trip into Switchmode power design requires an instant knowledge of what I've outlined, so my approach to tube amp loops and their problems becomes quite familiar.
If a specific input signal of either LF or HF triggers a closed loop amp into oscillations then reworking is required.
We all have different design approaches; traditional GEC designs using the Williamson diff driver uses the famed 6SN7 which is a very steady tube giving a gain of around 23dB as a diff driver, I use 12BY7A's as diff driver triodes with CCS, which offer a much lower Miller capacitance, this being important leverage into reducing the Miller capacitive loading of the anode section of the concertina which, yes, due to strays causes HF misbalance; an important issue with wide b/w output transformers. The subsequent improvement in HF audio balance strips any standard triode configuration., the 28dB gain reduces the front end requirments but remember doing this tack, a flighty interstage will also suck more front end noise. One can get around this snag by using a pentode as triode for the front end and using a step up input transformer.
More thoughts ? The after-effect of global negative feedback with the output transformer included in the circuit, pushes the output stage to be AC current sourced....not really desirable.
richy
The conditions for negative feedback imply a phase difference of 180° which equates to 12dB/octave; however a simple R/C creates a 90° phaseshift of 6dB/octave.So the closed loop of the amp has to designed in that no more that 10dB per octave though the unity gain point....(this is the tricky bit that many don't understand) then the amp is stable.
If the amp is a "close design" it could be modulated into oscillatory mode by a hefty input signal that can just tip it off the stability boundary; the receipe of gain,phase and frequency being the key components of a Bode plot.
Radiotron Hd bk 4th Ed chapter 7;Neg feedback page 315 ++.
The output transformer has a frequency reliant pole, cut off droop as do all the coupled R/C stages....this is where Williamson had the problems. SO it becomes impossible to govern a 4 stage R/C amp with strong global NFB throughout the whole frequency range, so it's reduced at both extremes with the interstage step cap values. Hence the loop gain curve of the 200W amp drops at the upper and lower ends.
Understanding stability and Loop filter timeconstants is difficult, but a simple squarewave can uncover a raft of problems.
A trip into Switchmode power design requires an instant knowledge of what I've outlined, so my approach to tube amp loops and their problems becomes quite familiar.
If a specific input signal of either LF or HF triggers a closed loop amp into oscillations then reworking is required.
We all have different design approaches; traditional GEC designs using the Williamson diff driver uses the famed 6SN7 which is a very steady tube giving a gain of around 23dB as a diff driver, I use 12BY7A's as diff driver triodes with CCS, which offer a much lower Miller capacitance, this being important leverage into reducing the Miller capacitive loading of the anode section of the concertina which, yes, due to strays causes HF misbalance; an important issue with wide b/w output transformers. The subsequent improvement in HF audio balance strips any standard triode configuration., the 28dB gain reduces the front end requirments but remember doing this tack, a flighty interstage will also suck more front end noise. One can get around this snag by using a pentode as triode for the front end and using a step up input transformer.
More thoughts ? The after-effect of global negative feedback with the output transformer included in the circuit, pushes the output stage to be AC current sourced....not really desirable.
richy
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.