Member
Joined 2006
Skorpio,
the circuit from post #66 had to be severely redesigned to make it work with +/-12V. R4, R5 should be replaced with CCSs and R7, R11 should be 6k8. Usable output swing will be much lower (up to 5V_peak max) and the original idea compromised. I wouldn't do it, although it can be made to work. Try it - you might like it...
-------------------------------------------------------------------------------------
MASantos,
I would use the amp in post #118 with preamp giving 20V_peak output swing and Laterals working at +/-24V with 1.2A bias which will produce about 25W_avg at 8 Ohms load. We may push it to 30V_peak swing and use two pairs of Laterals (supplying them with +/-35V, 0.8A bias per pair) for total of about 55W at 8 Ohms. For lower impedance load we would go lower with voltage and raise the current. Of course, something like F4 or some other power follower is a viable option, it's only that I prefer Laterals in Source Follower arrangement.
Anyway, when you know what you need/want, the most appropriate solution will present itself.
-------------------------------------------------------------------------------------
CFT,
I listened both amps on all of my speakers (Canton CT-1000, Rogers LS3/5a, Epos M5, Cabasse MiniDoris, Dynaudio DIY 3-way sealed+variovent) and the only things I could differentiate out of realm of very subtle nuances is a tiny bit of added gentleness in presentation of female voices on LS3/5a with amp from post #118 of this thread, as well as a tad of more resolute dynamics presented in complex orchestral passages with VFET_PP amp on Canton CT-1000. Otherwise, both amps present the same naturalness, transparency, tonal and spatial balance.
I'm sure that someone else might have a different impressions - I give you mine...
the circuit from post #66 had to be severely redesigned to make it work with +/-12V. R4, R5 should be replaced with CCSs and R7, R11 should be 6k8. Usable output swing will be much lower (up to 5V_peak max) and the original idea compromised. I wouldn't do it, although it can be made to work. Try it - you might like it...
-------------------------------------------------------------------------------------
MASantos,
I would use the amp in post #118 with preamp giving 20V_peak output swing and Laterals working at +/-24V with 1.2A bias which will produce about 25W_avg at 8 Ohms load. We may push it to 30V_peak swing and use two pairs of Laterals (supplying them with +/-35V, 0.8A bias per pair) for total of about 55W at 8 Ohms. For lower impedance load we would go lower with voltage and raise the current. Of course, something like F4 or some other power follower is a viable option, it's only that I prefer Laterals in Source Follower arrangement.
Anyway, when you know what you need/want, the most appropriate solution will present itself.
-------------------------------------------------------------------------------------
CFT,
I listened both amps on all of my speakers (Canton CT-1000, Rogers LS3/5a, Epos M5, Cabasse MiniDoris, Dynaudio DIY 3-way sealed+variovent) and the only things I could differentiate out of realm of very subtle nuances is a tiny bit of added gentleness in presentation of female voices on LS3/5a with amp from post #118 of this thread, as well as a tad of more resolute dynamics presented in complex orchestral passages with VFET_PP amp on Canton CT-1000. Otherwise, both amps present the same naturalness, transparency, tonal and spatial balance.
I'm sure that someone else might have a different impressions - I give you mine...
Member
Joined 2006
Yes. I can only confirm what Mr. Pass and others already said - the feedback loop is just a tool, like a fork. You can use it to eat easier or you can stab your self in the eye with it...
When you have a nice, linear circuit, the feedback loop is welcome to define/limit the gain, bandwidth and lower the Zout a bit. Using it to correct the obvious errors made in previous process won't bring much happiness.
The same goes for active devices of different types - using them in accordance with their properties and the with the well chosen aim in mind is what brings fine results. They don't have the 'sound' of their own - it's just the way people use them.
It's just my experience and few grains of knowledge collected during the years that tells me so...
When you have a nice, linear circuit, the feedback loop is welcome to define/limit the gain, bandwidth and lower the Zout a bit. Using it to correct the obvious errors made in previous process won't bring much happiness.
The same goes for active devices of different types - using them in accordance with their properties and the with the well chosen aim in mind is what brings fine results. They don't have the 'sound' of their own - it's just the way people use them.
It's just my experience and few grains of knowledge collected during the years that tells me so...
I have played with a balanced LTP and balanced complimentary version of this circuit and really like what I have heard. I followed that trail after listening to the Salas RIAA that uses a folded cascode as the gain stage. Wonderful sound. I am trying to decide between the balance version and the balanced version of the BA3 FE. Both can sound very, very good.
Member
Joined 2006
Thank you for the answer
My need is a linestage with little gain, and the 12V supply was only to make it easyer (already have one)´.
But I will make a new with +30V to give the JFETs optimum performance....
It's not all about gain, but the available output voltage swing too - if you are OK with 5V, go for it (+/-12V PS voltage).
Adjust the circuit accordingly - reserve about 4V for CCS (meaning about 8V of Vds for JFETs), set the Id through JFETs to 80-90% of Idss and account for 8-10mA through BJTs and you'll have a winner.
Hi Juma, I think polarity of C2 on post #118 should be reversed
Yes, also C5 at post #101.
Hi Juma, I think polarity of C2 on post #118 should be reversed
True, my drawing mistake, thanks for spotting it.
Although it's not a big deal - at such a low voltage (about 1V) elkos don't care much about polarity.
Uffff, it's good that that I'm more careful when soldering than when drawing stuff...Yes, also C5 at post #101.
Last edited:
To drive the F5 make R9 = 1k
That way you'll lower the gain (to about 12dB) and Zout too so you won't need the buffer between the preamp and F5.
BF862 preamp is very nice, transparent and somehow sweet and tame in the same time but you can't expect from it more than 5V_peak of clean output swing, so its use is limited.
I also enjoy this one, it's really great, dynamic, truthful and rich sounding but its use is limited by relatively low Zin when configured for high gain. Also, although I'm not bothered with that, some are repelled by the use of both input and output capacitor.
"Preamp ideas for F5 post 53" is OK but not in the league of the previous two.
OTOH this LSKpre is something else, it really plays wonderfully and it can be whatever you want it to be - in the poetic language of great Mayakovsky: a cloud in trousers...
Hi Juma,
The bjt alternatives you posted are really interesting since jFETs are becoming harder to find.
However, I have been following discussion on the VSSA thread and the feedback elco seems very critical sonically and also very hard to find, especially down here.
Which option do you think is better, use 4.7uF MKP to bypass C1 and C2 on post #118 with or modify input to a 4 bjt diamond buffer Hiraga style?
The bjt alternatives you posted are really interesting since jFETs are becoming harder to find.
However, I have been following discussion on the VSSA thread and the feedback elco seems very critical sonically and also very hard to find, especially down here.
Which option do you think is better, use 4.7uF MKP to bypass C1 and C2 on post #118 with or modify input to a 4 bjt diamond buffer Hiraga style?
... use 4.7uF MKP to bypass C1 and C2 on post #118 with or modify input to a 4 bjt diamond buffer Hiraga style?
It's simple enough to try both and see what you like better.
Adding more active devices bothers me more than a cap or two..
Thanks for the tip Juma. 
True enough that it is simple to build both, but for me, it usually takes several months of listening, optimization and measurement to get satisfactory result for one topology, I simply don't have the time I want for this hobby. A tip coming from experience like yours hopefully leads me to a less bumpy road, saving valuable little time that I have.
True enough that it is simple to build both, but for me, it usually takes several months of listening, optimization and measurement to get satisfactory result for one topology, I simply don't have the time I want for this hobby. A tip coming from experience like yours hopefully leads me to a less bumpy road, saving valuable little time that I have.
Is there any reason this version or the version in post #66 couldn't drive the uncrippled F4 directly without the buffer?