Yes, sure, more voltage over the Fet makes it warmer ( more noise ) and provokes some ( low ) gate current. Earlier i used 2 green Leds for 2 x 1,8V - Vbe = 2.95V. That makes the least noise but linearity suffers a bit so this is ok for a circuit with feedback. I raised that voltage recently to 4 green Leds 4 x 1.8V - Vbe = 6,55 V. Maybe i give up a dB of SN but i think for an open loop circuit this works better. See also the Lab Page on the La Boheme website where Klaus Böning puts some research into this.
I decided to drop not more then 10V over the BJT cascodes because avalanche noise can happen at higher voltages. Now you see the reason i end up at 16.5V.
I decided to drop not more then 10V over the BJT cascodes because avalanche noise can happen at higher voltages. Now you see the reason i end up at 16.5V.
Last edited:
Takes more precise work, and good check, that's all. Aspiring builders should know that non the less.
Agrees towards enough but not much voltage for medium heat and better dynamics, but not more than 8V, for low Ig. About 5-6V across I am not sure that 3 LED bias could be discernibly compressed VS 4 LED in my experience. Could be different in cascodes I use.
Yes, after discussing this for 2 years ! i think a bit less then 8V is fine.
I could try to raise the voltage over the BJT cascodes of cause until trouble sets in. On the other hand i did a lot of experiments on how high transients actually are on vinyl records.
The loudest record i could find was "March to the Scaffold" , a 45 from RR. Levels i could measure inside the circuit where much less then theorized by some here. OK, tick and pops could be a problem but i keep my records meticulous clean. 36dB of overload margin is certainly overkill although it looks fantastic on paper. The price to pay is either complexity or very high voltage.
I could try to raise the voltage over the BJT cascodes of cause until trouble sets in. On the other hand i did a lot of experiments on how high transients actually are on vinyl records.
The loudest record i could find was "March to the Scaffold" , a 45 from RR. Levels i could measure inside the circuit where much less then theorized by some here. OK, tick and pops could be a problem but i keep my records meticulous clean. 36dB of overload margin is certainly overkill although it looks fantastic on paper. The price to pay is either complexity or very high voltage.
As a proof of concept i rebuild an older stage to recent specs. It did not make all modifications. The current mirror got helper transistors but still is a cascode mirror and not an enhanced Wilson. The cascode mirror does not use feedback so has a bit more distortion but the same output impedance. This stage has only 2 plus 2 2SK170, 2SJ74. That is sufficient if you want to safe some P-Channels. It does not use the servo i proposed. The input stage is AC coupled to the buffer. I measured DC offset before the buffer and it floats around plus - minus 1V.
The buffer is still my older improved JLH design. I build a quite simple but i hope effective PSU. It uses LM317-337 with R-C pre regulation and Duttmanns enhancement trick. Inside the phone i use separate common mode coils and again onboard R-C filters.
I will fire it up now and will do a listening session.
I will post some photos and schematic later.
The buffer is still my older improved JLH design. I build a quite simple but i hope effective PSU. It uses LM317-337 with R-C pre regulation and Duttmanns enhancement trick. Inside the phone i use separate common mode coils and again onboard R-C filters.
I will fire it up now and will do a listening session.
I will post some photos and schematic later.
Ah, the Paris Hilton and Nicole Ritchie contribution.
http://www.diyaudio.com/forums/atta...improving-lm3x7-regulator-circuit-enpower.gif
(The D still walks on one leg, reminds me i haven't gotten round yet to ask Fred how he managed to fracture the other)
http://www.diyaudio.com/forums/atta...improving-lm3x7-regulator-circuit-enpower.gif
(The D still walks on one leg, reminds me i haven't gotten round yet to ask Fred how he managed to fracture the other)
Sorry, but i am not an expert on the Salas Shunt. I used it successfully earlier and this is documented on this thread. Short cables where beneficial. I did not try the remote sensing so far. On another occasion i was not so successful and ran into oscillation. I tried to suppress that with resistors. Again no luck. I should have consulted with Salas but i ran out of time. I have just setup my new company Suesskind and that required some step back from DIY. Now it is summer and i have more time. I will post a photo of the "Problem Child" and maybe Salas or others can help me to fix it.