I am just thankful for getting to build a version of the F8, using Vishay IRFP150's at 6 dollars a piece or Infineon IRFP150 for 1.30, I will never really know any difference.
If you need assistance doing some 'Spring Cleaning' - I'm sure we could arrange some help 😉
Aha
There was (is) Babelfish F8, practically blood-sucking edition, produced for nothing else than taking last penny from Greedy Boyz ......
Though, there is double trouble with blood-sucking edition:
So, after some brewing, there is increased Babelfishing level of F8, ingeniously ( as ZM always is) called Babelfish F8b
Difference in "b" explained as either:
Two main factors comprising more Babelfishing is change to N Channel input JFet..... and we really can't use anything bjt there, simply because unavoidable base current is resulting in output DC offset being uncontrollable, depending of Rout of preceding stage; so - either JFet, or much less practical (and probably worse sounding) mosfet
Principle, somewhat simplified FE:
Input JFet is nice little source follower, biased with 2mA CCS
As I want to use small PNP bjt as main input gain element, its emiter hanging from NFB divider, and said emiter need to be placed slightly bellow GND potential ( to keep output offset near 0), so I need level shifter, to further move its base for Vbe (say 650mV)
Plan is to use readily available 2SK209Y, Idss in nice range of 1.6-3mA, thus source resistor as first level shifting element is having 0 or small number of Ohms
Second level shifting element is more or less same in nature as gain bjt, but PNP connected as diode (oh yeah, 650mV again) and hopefully with opposite TempCo to gain bjt
To keep things even more funny, equally good candidates for input JFet are 2SK170 (BL, I have shovels) or - even better by current production - 2SK2145BL ........ difference being just in value of level shifting source resistor
All that shebang - source resistor and bjt as diode, possible due to biasing with small 2mA CCS and practically negligible base current amplitude of gain bjt, resulting in equally practically negligible current amplitude through JFet, source resistor and bjt as diode; so forget non-linear distortion worries
Same, with real life CCS:
There was (is) Babelfish F8, practically blood-sucking edition, produced for nothing else than taking last penny from Greedy Boyz ......
Though, there is double trouble with blood-sucking edition:
- it left sorta metallic taste in Might ZM's mouth ......... level of Babelfishing is just enough to call it Babelfish, and I was so eager to get more
- duration of penny-fishing is limited with number of unobtanium 2SK170BL I have .........
So, after some brewing, there is increased Babelfishing level of F8, ingeniously ( as ZM always is) called Babelfish F8b
Difference in "b" explained as either:
- b as bjt
- b as Blasphemy
- b as boolloocks
Two main factors comprising more Babelfishing is change to N Channel input JFet..... and we really can't use anything bjt there, simply because unavoidable base current is resulting in output DC offset being uncontrollable, depending of Rout of preceding stage; so - either JFet, or much less practical (and probably worse sounding) mosfet
Principle, somewhat simplified FE:
Input JFet is nice little source follower, biased with 2mA CCS
As I want to use small PNP bjt as main input gain element, its emiter hanging from NFB divider, and said emiter need to be placed slightly bellow GND potential ( to keep output offset near 0), so I need level shifter, to further move its base for Vbe (say 650mV)
Plan is to use readily available 2SK209Y, Idss in nice range of 1.6-3mA, thus source resistor as first level shifting element is having 0 or small number of Ohms
Second level shifting element is more or less same in nature as gain bjt, but PNP connected as diode (oh yeah, 650mV again) and hopefully with opposite TempCo to gain bjt
To keep things even more funny, equally good candidates for input JFet are 2SK170 (BL, I have shovels) or - even better by current production - 2SK2145BL ........ difference being just in value of level shifting source resistor
All that shebang - source resistor and bjt as diode, possible due to biasing with small 2mA CCS and practically negligible base current amplitude of gain bjt, resulting in equally practically negligible current amplitude through JFet, source resistor and bjt as diode; so forget non-linear distortion worries
Same, with real life CCS:
Last edited:
But, as Proof is in da Pudding, here it is, Ikebana style;
Proto made with 2SK170BL and level shifting resistor of 120R
I did found it's hard to catch hysterical beauty of chaos, with phone cam
Proto made with 2SK170BL and level shifting resistor of 120R
I did found it's hard to catch hysterical beauty of chaos, with phone cam
Attachments
More Pudding; rails and Iq as before - +/-24Vdc nominal ( 22V5 real), 1A8
SJEP down, 8R load
Ignore some weed in graph, my setup wasn't fully in mood today
SJEP down, 8R load
Ignore some weed in graph, my setup wasn't fully in mood today
Attachments
More Pudding; SJEP down, 4R load
(setup in slightly better mood )
(setup in slightly better mood
)Attachments
Last edited:
More Pudding; IRFP down, 8R load
Attachments
Few notes;
In both arrangements gain is hair bellow 6V/V for both 4R and 8R load; that's aroundish 15db
In both arrangements current through gain bjt is around 2.7mA
Forgot to measure Rout, but I plan to repeat full battery of tests as soon I get pcbs
Disclaimer - see it, if fun enough think of it, build it if your heart is tikataka
As soon I spend all available 2SJ74BL for blood-sucking kits, I'm going to introduce availability of b version kits
In both arrangements gain is hair bellow 6V/V for both 4R and 8R load; that's aroundish 15db
In both arrangements current through gain bjt is around 2.7mA
Forgot to measure Rout, but I plan to repeat full battery of tests as soon I get pcbs
Disclaimer - see it, if fun enough think of it, build it if your heart is tikataka
As soon I spend all available 2SJ74BL for blood-sucking kits, I'm going to introduce availability of b version kits
Gee ZM, I haven't received the F8 boards yet and you already have an updated design. How can we keep up with you?
I think I will be happy with the F8 based on what I have read so far - but would be very interested to read your thoughts based on listening tests between the 2.
I think I will be happy with the F8 based on what I have read so far - but would be very interested to read your thoughts based on listening tests between the 2.
Waiting for the "C" edition, and all the comments asking if the the "B" edition sounds better.
Good thing all these amps sound the same, still working on the power supply section of my "A" edition kit.
Good thing all these amps sound the same, still working on the power supply section of my "A" edition kit.
see, he knows everything needed to know:
even if by graphs F8b is better by -6db everywhere, I'm not expecting to hear that difference
what's important is - that now I'm fully happy with nicety factor level ............ and I can fish for pennies 'till the end of World
Is there something obvious that contributes to the lower distortion in F8b vs F8a?
A slight difference in NFB could do it?
A good or bad day for the test bench?
My F8a has just been released from the customs.
I will probably install Arta on my laptop.
A slight difference in NFB could do it?
A good or bad day for the test bench?
My F8a has just been released from the customs.
I will probably install Arta on my laptop.
sim results; didn't bother to write down actual test results, they were more than close to those I already had in sim
original F8:
OLG 46.83db
CLG 14.87db
NFB 31.96db
Babelfish F8:
OLG 61.84db
CLG 15.46db
NFB 46db
Babelfish F8b:
OLG 67.74db
CLG 15.5db
NFB 52.23db
I mean, nothing surprising here, lower THD in general must be result of something
As one of ideas in Babelfish F8 was to use lower ohmic values in NFB than case is with Papa's, lesser degeneration effect (smaller current envelope) for JFet, thus higher OLG; simple as that
Now, Babelfish F8b, as I chose PNP bjt as FE gain element, that draws entire other set of compromises, comparing to original F8, even to Babelfish F8, thus Babelfish F8b having slightly higher OLG
Frankly, as I was chasing few other parameters, OLG values are really just coincidental; I did took care of their values, but other things were primary
original F8:
OLG 46.83db
CLG 14.87db
NFB 31.96db
Babelfish F8:
OLG 61.84db
CLG 15.46db
NFB 46db
Babelfish F8b:
OLG 67.74db
CLG 15.5db
NFB 52.23db
I mean, nothing surprising here, lower THD in general must be result of something
As one of ideas in Babelfish F8 was to use lower ohmic values in NFB than case is with Papa's, lesser degeneration effect (smaller current envelope) for JFet, thus higher OLG; simple as that
Now, Babelfish F8b, as I chose PNP bjt as FE gain element, that draws entire other set of compromises, comparing to original F8, even to Babelfish F8, thus Babelfish F8b having slightly higher OLG
Frankly, as I was chasing few other parameters, OLG values are really just coincidental; I did took care of their values, but other things were primary