Your damned right homemodder, must have been reading my thoughts! Inspired by the memory distortion article. Maybe fully complementary - it simultates very well.homemodder said:
Here are some RMAA distortion measurements, 33R dummy load. Maybe they reflect my listening impressions. The standard JFET input amp has slightly higher OLG because of less source degeneration, and it runs at 18V and the others at 15V. I think that's why the THD is a little lower.
BTW, I gave up on active current sources on top of the LTP. While it simulated extremely well, I never liked the sound of them. I tried both a standard bipolar current mirror and two independent JFETs.
Attachments
When a I use jfet/bjt in LPT, circuit was unstable or sound was sharp/unpleasant, maybe layout /biasing is clue
Cascodes sometimes sound sharp/ unpleasantbut cure is smalll srtopper and physical low C layout (base leg not to PCB but in air via resistor to voltage reference source in case BJT cascode)-great J Carr tip
Cascodes sometimes sound sharp/ unpleasantbut cure is smalll srtopper and physical low C layout (base leg not to PCB but in air via resistor to voltage reference source in case BJT cascode)-great J Carr tip
Nelson,
I always pay much attention to your listening impressions, as I`m not interested in any nonsensical distortion figures.
The evaluation would be more truthful and the differences more apparent with no feedback applied.
The greatest advantage of CFP is in the output stage (the weakest point in this amp), but it is not suitable here.
I always pay much attention to your listening impressions, as I`m not interested in any nonsensical distortion figures.
The evaluation would be more truthful and the differences more apparent with no feedback applied.
The greatest advantage of CFP is in the output stage (the weakest point in this amp), but it is not suitable here.
The JFET/bjt is the least unpleasant of them all. It's very smooth and "buttery", so I don't agree with your findings. I have tried this in several amps now, and the result has been the same - a warm and sweet tone, yet with all the details (even in very messy perfboard amps). The cascoded input stage has a slightly unpleasant sound, and this is also in concordance with earlier findings. Speaking of unpleasant, current mirror or active current sources on top of the LTP - that's unpleasant.Uchi Deshi said:When a I use jfet/bjt in LPT, circuit was unstable or sound was sharp/unpleasant, maybe layout /biasing is clue
Cascodes sometimes sound sharp/ unpleasantbut cure is smalll srtopper and physical low C layout (base leg not to PCB but in air via resistor to voltage reference source in case BJT cascode)-great J Carr tip
Lumba, is it essential for an output stage to be single ended, or is there a way to make a good complementary output stage? I was thinking of the CFP output you've recommended in other threads with bipolar master and MOSFET slave. I've always found transistors to sound very transparent in common collector configurations in general and the diamond buffer in particular, but I've never heard a CFP output stage. I've tried it once but it oscillated like hell, and I haven't dared to try it again.
Nelson I started using jfet, bjt cfps back in around 1993, this was recommended by linsey hood in some letters he sent to my varsity electronics prof, he didnt mention thd figures as such but said it was a very good sounding combination and the reason I tried it. In reality THD figures are very good too, Ive used it in some higher powered amps to good effect but it has voltage limits where and one has to use a cascode. At lower voltages I prefer without the cascode, there are some high voltage jfets, I use 2sk223 which give very good results. These parts are in the same league than 2sk170 but even better as their capacitances are lower.
As the diamond buffer is operating in class A you shouldnt have any problems with oscilation using the cfp, in class AB I know it has some issues. In some of Hoods notes he has mentioned his preference for BJT mosfet cfp outputs, again stating that they sound very good. This view is mentioned in hifi books like the one by Ben Duncan and the one by Sinclair showing some of Hoods circuits.
As the diamond buffer is operating in class A you shouldnt have any problems with oscilation using the cfp, in class AB I know it has some issues. In some of Hoods notes he has mentioned his preference for BJT mosfet cfp outputs, again stating that they sound very good. This view is mentioned in hifi books like the one by Ben Duncan and the one by Sinclair showing some of Hoods circuits.
Homemodder, it's good to hear I'm not the only one thinking this CFP input arrangement sounds good. In general I find the mix of bipolar and JFET input amps make a more natural blend compared to using just one of the types.
I understand AKSA use CFP input in his top of the line amp, but it's probably all bipolar. I think roender also uses CFP input in a "secret" amp. The sound is too good to be ignored.
It would be interesting to try a CFP output, but then I need to lay out a new board.
I understand AKSA use CFP input in his top of the line amp, but it's probably all bipolar. I think roender also uses CFP input in a "secret" amp. The sound is too good to be ignored.
It would be interesting to try a CFP output, but then I need to lay out a new board.
Nelson,
class A biasing is essential, so the transistors carry the entire signal waveform but the stages do not necessarily need to be single-ended, a balanced topology can be chosen, for instance in order to eliminate dirty grounds. (By definition, the push-pull stage is balanced as well, however I would avoid it). For a flawless operation, the CFP parts should be selected with care, biased in class A . When done right, it`s superior anything else.
A class A headphone amp can be built without significant inconveniences but considering your fastidious sonic taste and my even more fastidious principles, you will need to find a big fat box. It won`t be ideal for a portable application.
class A biasing is essential, so the transistors carry the entire signal waveform but the stages do not necessarily need to be single-ended, a balanced topology can be chosen, for instance in order to eliminate dirty grounds. (By definition, the push-pull stage is balanced as well, however I would avoid it). For a flawless operation, the CFP parts should be selected with care, biased in class A . When done right, it`s superior anything else.
A class A headphone amp can be built without significant inconveniences but considering your fastidious sonic taste and my even more fastidious principles, you will need to find a big fat box. It won`t be ideal for a portable application.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.