Hi,
I´ve tried increasingly higher values for the McMillans up until 39k. The Gain was significant until about 27k (with one pair of 2SJ74) but not in the bass, more in the highs. Must be somewhere in this thread. I think it is important to match them very closely to keep the common mode input to the diff pair as low as possible.
With 3 output fets it is probably better to use two pairs of SJ74 and get a higher bandwidth. Since my speakers are 100dB/Watt I disconnected 4 output fets (8 left) and just used one pair.
I had a DC servo for a while but took it out again because the solution with a PTY was good enough for me (max 1V of absolut offset from startup, still dependent on the rail (mains) voltage. I did not hear any negative effects with the DC-Servo though.
Hope this helps a bit,
William
Hello,
thanks a lot!
Since I can get 20 pcs of 2SJ74, I decided to use 3 in parallel. And to use a DC servo which enables me to eliminate all resistors at the input-CCS and keep the CMRR high. 2 CCS on drain side of the JFets to rise the open-loop-gain (at least for low frequencies) will be tested when its running.
This will be the amp for a friend. For me, I have decided to switch to annother project: A balanced amp like the Aleph-X, but with a 240/9240 output stage as source follower and the complete gain in a complete-symmetric JFet (4 parallel, cascoded K170/J74) input stage, loaded with several CCS to keep enough open-loop-gain. Will try out if it can beat my Aleph-X and my collegues Aleph-JX...
Regards, J-C
thanks a lot!
Since I can get 20 pcs of 2SJ74, I decided to use 3 in parallel. And to use a DC servo which enables me to eliminate all resistors at the input-CCS and keep the CMRR high. 2 CCS on drain side of the JFets to rise the open-loop-gain (at least for low frequencies) will be tested when its running.
This will be the amp for a friend. For me, I have decided to switch to annother project: A balanced amp like the Aleph-X, but with a 240/9240 output stage as source follower and the complete gain in a complete-symmetric JFet (4 parallel, cascoded K170/J74) input stage, loaded with several CCS to keep enough open-loop-gain. Will try out if it can beat my Aleph-X and my collegues Aleph-JX...
Regards, J-C
Pass DIY Addict
Joined 2000
Paid Member
Waaaaaay back in the beginning of the thread someone reported on the sound difference between the 044 and the 240. The impressions are that the 044 sounds a little "darker" - less treble - than the 240 that was mainly attributed to the higher gate capacitance of the 044 when compared to the 240. The 044's have a high current and temperature rating than the 240's do.
William: I never was able to reclaim my "lost" output power when I moved my front end from 9610's to SJ74s. I've kind of given up on that and I'm paying attention now to ambient temperature differences and how that affects my offset. Sometime over the summer, I'll put a PTY solution in place.
I think I last adjusted offset during the summer, but I want to wait and re-measure things later to be sure. Right now, I get the following:
Jan, Feb: Min outside temperature near -18c
Cold sink temp 12c - amps live in unfinished area of basement
After 2+ hours of operation:
Front sink temp 44-45c
Rear sink temp 51-53c
Offset voltage 1.9v to 2.2v
Eric
Hi,
surprisingly the data sheet from 044 shows not much more capacitance than the 240-s. I will stay at the 044-s.
But why has there not been found a relieable solution for the absolute DC-offset during so many years? I think a solution w/o feedback resistors will improve sound quality. Its only needed to control the CCS depending on the DC. Isnt there a proven solution? If my idea will work, and there is an interest, will post it.
Regards, J-C
surprisingly the data sheet from 044 shows not much more capacitance than the 240-s. I will stay at the 044-s.
But why has there not been found a relieable solution for the absolute DC-offset during so many years? I think a solution w/o feedback resistors will improve sound quality. Its only needed to control the CCS depending on the DC. Isnt there a proven solution? If my idea will work, and there is an interest, will post it.
Regards, J-C
Hi,
a DC-servo will work fine (0,00V). A PTC will work fine too (<1V) with a lot less parts. If the mains voltage is more or less stable the second solution is OK. If not I would go for a dc-servo.
I did not hear any degradation of the sound quality with my dc-servo which injected the current upstream from the IRF9610 and was heavily filtered.
Sound quality gains by getting rid of the McMillan resistors will depend on the values you are using. If they are low (4k7) gains will be high, if not (33k) they will be very subtle.
I think there will be interest for your idea! Just post it when you´re ready.
William
a DC-servo will work fine (0,00V). A PTC will work fine too (<1V) with a lot less parts. If the mains voltage is more or less stable the second solution is OK. If not I would go for a dc-servo.
I did not hear any degradation of the sound quality with my dc-servo which injected the current upstream from the IRF9610 and was heavily filtered.
Sound quality gains by getting rid of the McMillan resistors will depend on the values you are using. If they are low (4k7) gains will be high, if not (33k) they will be very subtle.
I think there will be interest for your idea! Just post it when you´re ready.
William
Pass DIY Addict
Joined 2000
Paid Member
How about using not-so-low-noise-but-cheaper fets like 2sj103 in the input stage? Will it be an improvement over irf9610?
My Aleph-x has single outputs and ~15V, so a single fet in the input could be enough? Eric's page talks about just replacing the irfs with jfets, what about the ztx550s?
My Aleph-x has single outputs and ~15V, so a single fet in the input could be enough? Eric's page talks about just replacing the irfs with jfets, what about the ztx550s?
2Sj103BL** is having much lower xcconductance then , say usual , 2SJ74BL
if you have proper triples of 2SJ103 , put them instead of one 2SJ74BL
in that case , just toss them in (two triples) , instead of IRF , leaving same current through LTP
** they must be BL current range
edit : what about ZTX ?
if you have proper triples of 2SJ103 , put them instead of one 2SJ74BL
in that case , just toss them in (two triples) , instead of IRF , leaving same current through LTP
** they must be BL current range
edit : what about ZTX ?
Last edited:
yup - Yfs
LTP , as input pair , arranged in
regarding ZTX ..... I really don't know which exact schmtc you're referring to
Eric's replacement:
http://www.facstaff.bucknell.edu/esantane/movies/JFET-input.jpg
also seen here a few pages before
Finalising my Aleph-X, spkr protection and servo's
I'm just finalising my Aleph-X prototype before building my final form mono-blocks with wife acceptance factor included.
I have a few problems that seem to already have been discussed;
I need to drive my amplifiers from a single ended source and tried to use a single to balanced converter (based on DRV134) - all worked well until one day the DRV134 gave up and put one output at v+ rail level. Not a problem because I only had a load resistor connected across the speaker outputs but I would have cried my eyes out if I had popped a driver in my speakers - A speaker protection circuit would seem the obvious answer but I have some concerns about it tripping because it will be comparing DC from the speaker outputs to ground but the amplifier outputs are referenced to each other rather than to ground. The DC offset is also very temperature dependant (which brings me back to the servo circuit). Has anyone implemented speaker protection on an Aleph-x?
Is there a preferred method (i.e. KTY81/110 temperature sensor(s) vs a servo?)
I'm just finalising my Aleph-X prototype before building my final form mono-blocks with wife acceptance factor included.
I have a few problems that seem to already have been discussed;
I need to drive my amplifiers from a single ended source and tried to use a single to balanced converter (based on DRV134) - all worked well until one day the DRV134 gave up and put one output at v+ rail level. Not a problem because I only had a load resistor connected across the speaker outputs but I would have cried my eyes out if I had popped a driver in my speakers - A speaker protection circuit would seem the obvious answer but I have some concerns about it tripping because it will be comparing DC from the speaker outputs to ground but the amplifier outputs are referenced to each other rather than to ground. The DC offset is also very temperature dependant (which brings me back to the servo circuit). Has anyone implemented speaker protection on an Aleph-x?
Is there a preferred method (i.e. KTY81/110 temperature sensor(s) vs a servo?)
http://xen-audio.com/documents/f5x/XEN Balanced Speaker Protection Public 131003.pdf
There is an updated, improved version (JFET input) with a PCB layout.
But I have no time in the next 6 months to verify and publish.
So you should try this one first.
Patrick
There is an updated, improved version (JFET input) with a PCB layout.
But I have no time in the next 6 months to verify and publish.
So you should try this one first.
Patrick