Probably belongs in some F4 thread, but I've been considering building another M2 to go biamp on my Klipsches. (Forte II not biampable, but they need some work so I figured I'd mod them for biamp during surgery).
I want the low and high amp to have similar characteristics, as there is an overlap between the woofer and mid horn.
I think M2 will drive the F4
. Better to add another M2 or hang an F4 off of the one that's already there?
I don't think the bi-amping idea will work. When the J2 sees the 4-ohm tweeter load it can't drive enough voltage, and the F4 ends up putting out about the same wattage as the J2.
However, if I drop bi-amping and just run the J2 through the F4, then I get better results. Running both the J2 and F4 at 28.5V rails I can get the J2 to swing 52V, which drives the F4 to a nice 80W Class A.
At those rail voltages the J2's output FETs are dissipating 46W at idle. There's only a pair of them, so the heatsinking requirements are modest.
The F4's are dissipating 23W, but I've got 4 pairs there. Total wattage I need to sink is 185W/channel, which (in my climate) is within the remit of something like the 5U chassis.
However, if I drop bi-amping and just run the J2 through the F4, then I get better results. Running both the J2 and F4 at 28.5V rails I can get the J2 to swing 52V, which drives the F4 to a nice 80W Class A.
At those rail voltages the J2's output FETs are dissipating 46W at idle. There's only a pair of them, so the heatsinking requirements are modest.
The F4's are dissipating 23W, but I've got 4 pairs there. Total wattage I need to sink is 185W/channel, which (in my climate) is within the remit of something like the 5U chassis.
I think I understand the optocoupler in the M2, where it's being used as a bias spreader. However, that's a push-pull design; single-ended designs don't need bias spreading.
So is the optocoupler in the J2 being used as sort of an Aleph-y current source (ie: to modulate the current provided by tracking the output current)?
So is the optocoupler in the J2 being used as sort of an Aleph-y current source (ie: to modulate the current provided by tracking the output current)?
So some searching turned up this seemingly legit company offering unmatched SJEP120R100 for 44$ a piece (worth is weight in gold). Unmatched on non-specific builds are too much for me, but this could be a decent little treasurer trove for some of you.
meanwhile I'm crossing my fingers from the next run of matched LSJ74 for a Ba-1 fe
meanwhile I'm crossing my fingers from the next run of matched LSJ74 for a Ba-1 fe
Hmm.
So there's no conventional bias spreader in the M2? It also uses an Aleph-y style bias that tracks the amount of current needed (rather than being a more-or-less fixed voltage that the input oscillates around)?
Me thinks I'm going to have to go back and read everything again. At some point it will have to sink in....
So there's no conventional bias spreader in the M2? It also uses an Aleph-y style bias that tracks the amount of current needed (rather than being a more-or-less fixed voltage that the input oscillates around)?
Me thinks I'm going to have to go back and read everything again. At some point it will have to sink in....
As some may know, I like to get into trouble and I'm obsessed with cascode/bootstrap quimeric organization replacing singles, in all stages.
Take a look at this article about SiC-cascode units:
Hybrid Cascodes Simplify SiC Adoption in Popular Power Circuits | Electronic Design
When I have time and energy, I will mod the J2 with "bootstrapped compound-CFP" input section and then will try to cascode the output section.
The first mod is expected to have significant improvement in sound quality.
Cheers,
M.
Take a look at this article about SiC-cascode units:
Hybrid Cascodes Simplify SiC Adoption in Popular Power Circuits | Electronic Design
When I have time and energy, I will mod the J2 with "bootstrapped compound-CFP" input section and then will try to cascode the output section.
The first mod is expected to have significant improvement in sound quality.
Cheers,
M.
... I will mod the J2 ...
The first mod is expected to have significant improvement in sound quality.
Cheers,
M.
Gulp! You are the epitome of a fearless amp builder!
I don't have the courage to mess with an amp that I consider just about perfect! I can't wait to read about your results.
Gulp! You are the epitome of a fearless amp builder!
I don't have the courage to mess with an amp that I consider just about perfect! I can't wait to read about your results.
No amp is perfect. J2 (DIY) is good but not perfect. Great overall but I miss transparency.
Tokin VFET with above mentioned input sounds better, IMHO.
The problem I have is that my mechanical construction, while sturdy, is hard to dismantle to mod...
bad planning. Apart, I am so busy with my experimental amp, the Amnesis, that I have postponed other amp modifications. Cheers,
M.
Thanks for the link, Lazybutt.
So Choky's fork analogy is the bias spreading. He also mentions the LED tracking the output current, which means the optocoupler must also be performing some sort of compensation. For thermal tracking I assume?
In the J2 there's no push/pull and so no bias spread, leaving just the compensation part. So is it also for thermal tracking in the J2, or is it so the current source tracks the power needed (a la the Aleph current source)?
So Choky's fork analogy is the bias spreading. He also mentions the LED tracking the output current, which means the optocoupler must also be performing some sort of compensation. For thermal tracking I assume?
In the J2 there's no push/pull and so no bias spread, leaving just the compensation part. So is it also for thermal tracking in the J2, or is it so the current source tracks the power needed (a la the Aleph current source)?
no Bubba
no special need for thermal tracking of any sort ..... simple as that : diode will lit according to voltage sag across source resistors ... if tempco of mosfet is resulting in increasing current , diode will open bjt part more , squeezing gates closer
catch is to set values of resistors from rail to gates vs. optocoupler current transfer and voltage sag across source resistors , to have best behaving in voltage domain ( rails fluctuation - mains instability in time) and thermal domain ( mosfet TempCo)
though - as myriad times before - Papa did all the work ..... besides getting idea , in first place
so , we know that 47K-ish is good from gate to rail for M2
it's same principle in J2 .... even if you're guiding (in DC domain) just one gate with optocoupler thingie
lower gate is controlled from input , certainly even unaware of optothingie on top
no special need for thermal tracking of any sort ..... simple as that : diode will lit according to voltage sag across source resistors ... if tempco of mosfet is resulting in increasing current , diode will open bjt part more , squeezing gates closer
catch is to set values of resistors from rail to gates vs. optocoupler current transfer and voltage sag across source resistors , to have best behaving in voltage domain ( rails fluctuation - mains instability in time) and thermal domain ( mosfet TempCo)
though - as myriad times before - Papa did all the work ..... besides getting idea , in first place
so , we know that 47K-ish is good from gate to rail for M2
it's same principle in J2 .... even if you're guiding (in DC domain) just one gate with optocoupler thingie
lower gate is controlled from input , certainly even unaware of optothingie on top
Cool. I think we're partly saying the same thing: that the optocoupler smooths the MOSFETs output in the thermal domain. I was calling that "thermal tracking" but maybe that's the wrong term?
But I still think it's doing something Aleph-y in the J2. If the MOSFET were just a standard CCS, then its gate voltage would be constant (in the absence of any rail and/or tempco fluctuation), right?
Oh ****. Light bulb time. The MOSFET is "upside down". For a CCS the gate voltage needs to be "constant" relative to the source (aka the sinusoidal output), not relative to the drain (aka the more-or-less-linear V+ rail). So it is sort of like 1/2 a bias spreader.
Am I getting warm?
But I still think it's doing something Aleph-y in the J2. If the MOSFET were just a standard CCS, then its gate voltage would be constant (in the absence of any rail and/or tempco fluctuation), right?
Oh ****. Light bulb time. The MOSFET is "upside down". For a CCS the gate voltage needs to be "constant" relative to the source (aka the sinusoidal output), not relative to the drain (aka the more-or-less-linear V+ rail). So it is sort of like 1/2 a bias spreader.
Am I getting warm?
I found this helpful: Foolin' with FETs.
Now that I know what a mu-follower is, it's certainly simpler, although I'm not sure I'm quite at "simple" yet.
Anyway, thanks for your help, ZM.
Now that I know what a mu-follower is, it's certainly simpler, although I'm not sure I'm quite at "simple" yet.
Anyway, thanks for your help, ZM.