I thought I did something wrong, BUT... I reviewed PCB and
schema.. the 22r is R36, at the top of it ,the DC NFB cap and
input comp. are attached, giving an isolated ground reference,
(A in attached schema). Also both "B's" (also in attachment)
are connected by separate trace to star ground on PCB.(GRND)
Zobel ground connects back to connector at star ground(GRND) , too.
All grounding seems to be in order..
The input connector ground connects right at "A" in attachment,
this might be where a slight confusion could arise..will correct.
I return the speaker ground right back to the main star/supply
ground also.
OS
Attachments
yes, Signal Ground connects at A in FA3.
But I was referring to FA1 in my post641.
It would help if we could adopt different symbols for the different Grounds/Returns. Then there would be no confusion. Even better if we gave them all individual names. They all do a different job and deserve to be differentiated. But the software designers will never do that for us.
But I was referring to FA1 in my post641.
It would help if we could adopt different symbols for the different Grounds/Returns. Then there would be no confusion. Even better if we gave them all individual names. They all do a different job and deserve to be differentiated. But the software designers will never do that for us.
Good suggestion, Even as they are right on the PCB, a standard
convention must be observed , I alway did it right on the real
thing, never experiencing "hum" on any of the FA's (even with
various unbalanced inputs..soundcards/cd outs,etc.)
I will also change ground designations on both FA1/3..
OS
Hey, FA1 and 3 downloaders... download the ZIP files...
to download the original uncompressed bitmap of the PCB is
unneeded.. all are in the ZIP's (FA1_full.zip and FA3_full.zip), much
faster download...
OS
PS ..If I make any changes/improvements I will also update
the zipfiles for your convenience(speed).
to download the original uncompressed bitmap of the PCB is
unneeded.. all are in the ZIP's (FA1_full.zip and FA3_full.zip), much
faster download...
OS
PS ..If I make any changes/improvements I will also update
the zipfiles for your convenience(speed).
Trouble with triples
Already at that point, HM , have it hooked to my SUB!! bass
is incredible! I temporarily tried it on a fullrange and it sounded
as good or better than FA2.. but the bass is the surprise.
I want to put a "beast" OPS on it (8 devices)but I'm having
"trouble with triples". I can drive the 4 pairs well just from
my present drivers (mje15032/33) but for my 4R peerless
I would like to use a leach type OP stage.
When I simulate the triple, any crossover distortion or
clipping makes the whole thing breaK out in oscillation.
Is the triple just a unstable type of Op stage??Or is there
a trick to it??
My only other option for More power is a "quasi" type
6/8 device mosfet OP stage running higher rails.
Schema of "beast" is attached.. any triple experts welcome
to comment..
OS
Already at that point, HM , have it hooked to my SUB!! bass
is incredible! I temporarily tried it on a fullrange and it sounded
as good or better than FA2.. but the bass is the surprise.
I want to put a "beast" OPS on it (8 devices)but I'm having
"trouble with triples". I can drive the 4 pairs well just from
my present drivers (mje15032/33) but for my 4R peerless
I would like to use a leach type OP stage.
When I simulate the triple, any crossover distortion or
clipping makes the whole thing breaK out in oscillation.
Is the triple just a unstable type of Op stage??Or is there
a trick to it??
My only other option for More power is a "quasi" type
6/8 device mosfet OP stage running higher rails.
Schema of "beast" is attached.. any triple experts welcome
to comment..
OS
Attachments
Os why dont you use the cfp ef, works well, the triples are tricky sometimes, for 4 outputs try increase the cap between predriver bases a bit more, a sure way to stabilise it more is a small cap from the base of predriver to the rail. It helps when the predriver is a low cob device so 1381 would be good in that position, I dont think you running high current there anyway. I dont know if any will help because you talking of clipping here, this is not easy. It might not oscilate real time but if you wrong their might be some damage which is not good.
For me.. 300w for the hungry sub and 120w+ for the fullranges will do, but for the young un's, they crave POWER ,so triples or
mosfets (maybe both) to get to that magic 3-400 watts will
be neccessary.
My problem with the triple is "getting there" .I changed
devices, emitter resistor values, base stoppers..etc.
I need to create a better (more APT) T- comp (6 Vbe's now),
as the "Beast" will no longer be a frugalamp.
My problem with the mosfets is that I can't (don't know how
) to set up the .SUBCKT models in LT.
maybe that can be resolved..
I thought CPF OPS's were just as "squirrly" (proneto oscillate) as triples , doug self says so because of the local
feedback loop. Mr self also does not want to discuss triples
,so it seems. any answers to these problems would be
helpful.
OS
Andy beat me to it, othervise works no problem.
Andy I played a bit around with the input stage and and now must try real time for listening. Using the cascode and dropping the cfp in this case brings no improvement or reduction in THD but it seems it has to be the one or the other. Transient wise the cascode allows slightly extended bandwith.
Andy I played a bit around with the input stage and and now must try real time for listening. Using the cascode and dropping the cfp in this case brings no improvement or reduction in THD but it seems it has to be the one or the other. Transient wise the cascode allows slightly extended bandwith.
Os if you look on the lt scad3 file, near the end their is instruction on how to do this page 190 thereabouts, ive deleted my copy, but its near the end.
Is this coming from outputstage alone, have you tried the input stage with a ideal output. I am looking at it at 20khz but cant get rid of it.
Is this coming from outputstage alone, have you tried the input stage with a ideal output. I am looking at it at 20khz but cant get rid of it.
I did that Sim on the FA3 HP and ran it with 90V rails.
I tried it on the real (sub) , no warm zobel... on the sim at
20khz with both clipping and induced Xover distortion.. NO
problem. Only with the triple... but it DOES greatly reduce
with larger emitter "suckout cap" across the main drivers.
the VAS seems to be the problem , by changing loopgain and
lowering Ic (3-4mA) solves this.
That is wishful thinking. One would need 14 OP devices to
acheive that (SOA). I would even recommend Class D at that
point. UCD 700 amp is only 4X6 inches with NO heat..
you would only get 10-11 DB over the normal frugalamp..
for all those devices..
In simulation I clipped at 170V p-p(roughly 1500W) but that is way over rated..
I would like to go the "quasi" route and just use 6-8 NPN
IRF 250's to get my 300W.. it might even stay "frugal" with
that arrangement..
OS
OS
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.