There is another way ... since it seems your input pair is running "hotter"
(1.8-2ma).
R16/13 (1K) , determine bias of Q7/8. By reducing these to 820R ,
Q7/8 will conduct less .... allowing 120-150R for R17/21 (while keeping 7ma+).
Having a smaller R for R17/21 will give a greater global loop gain
for the IPS as a whole. 220R is not bad , but you lose 6-10db LF
open loop gain. THD goes up just a little (220R /7ma -I simulated).
Other combinations are 680R for R16/13 and 100R for R17/21 , you
can trim the final cascode I (ma) by trimming blue led I - R18 (47k/33k/27k).
On the VFA amps, (wolverine) this is easier ... just "trim the trimmer" for
the main LTP CCS. (changes VAS current)
OS
yes, you absolutely right CFA is more need housekeeping than VFAThere is another way ... since it seems your input pair is running "hotter"
(1.8-2ma).
R16/13 (1K) , determine bias of Q7/8. By reducing these to 820R ,
Q7/8 will conduct less .... allowing 120-150R for R17/21 (while keeping 7ma+).
Having a smaller R for R17/21 will give a greater global loop gain
for the IPS as a whole. 220R is not bad , but you lose 6-10db LF
open loop gain. THD goes up just a little (220R /7ma -I simulated).
Other combinations are 680R for R16/13 and 100R for R17/21 , you
can trim the final cascode I (ma) by trimming blue led I - R18 (47k/33k/27k).
On the VFA amps, (wolverine) this is easier ... just "trim the trimmer" for
the main LTP CCS. (changes VAS current)
OS
i'm still learning how to use more function of LTSpice.
Uncle where is Test Point location i should read when i have to trim R18... to get correct I (ma)?
Thank you
Last edited:
(below) "baby" is done...
You should wait ... because , for the baby ... I
will include modern choices for the EXACT H/K680R components.
The "baby" can be a heavy duty version EXACT clone of
this excellent OPS.
Harmon used ...
-toshiba to-220 drivers (2sc4793/sa1837)
-2sd667/sa647 for predriver
-2sc3907/sa1516 for OP
Sanken discontinued the OP , but newer sankens
are perfect subs.
With these ... .1-.3mv bias stabilty is the rule for this OPS.
OS
You should wait ... because , for the baby ... I
will include modern choices for the EXACT H/K680R components.
The "baby" can be a heavy duty version EXACT clone of
this excellent OPS.
Harmon used ...
-toshiba to-220 drivers (2sc4793/sa1837)
-2sd667/sa647 for predriver
-2sc3907/sa1516 for OP
Sanken discontinued the OP , but newer sankens
are perfect subs.
With these ... .1-.3mv bias stabilty is the rule for this OPS.
OS
Attachments
WOW
Do you consult for NASA or NSA?
All thats in another group photo.... to 6GHz ! TDR -- yes, have that also.
Thx-RNMarsh
So include the complementary spectrum analyzer too Please.
I was thinking maybe you could use a roll of Cat-6 and remotely do some evaluations of designs over the Internet at your place and charge bitcoins for it. Just the input to output THD and Slew rates for a flat fee.
It always helps to have the RIGHT equipment to do the job RIGHT
Shucks let me get off of here, my keyboard is starting short out from the drooling.
Ostripper stay away from difficult to find parts.
Looking for these predrivers and out. i can't see easy available.
Thats my opinion only
No , I said I will find 21'st century equivalents.
The old device documentation is good for
X-referencing new devices.
I used 2sc3263/sa1294 as OP replacements in the 680r.
Toshiba drivers are current.
Predrivers i'm still choosing.
I only discussed this as the "base" OPS was
Japanese semi based.
OS
Yes , "baby" is 60/100 8/4R (Very conservative). The equivalant HK680r is rated a straight 105w/8R
@52V rails.
"Baby" will be for 35-55V rails.
With more builds , I might choose the "best" VFA/CFA and integrate the OPS.
The modular style , as I have it ... adds no parasitic's or R to the design. It
essentially is integrated with IPS "swap-ability". Everything is low current/
high-Z at the interface.
OS
Ostripper,
With your current method of a separate ops and input section am I to assume that the global feedback circuit is done by an external wire and whatever type, TMC, TPC, etc. that would not be part of the board layout the way you are doing this now? I know that this is over my head at this point but I am learning by watching and reading what you and the other gurus are doing here and the CFA thread.
I actually attempted to install LTSpice today so I can start to see if I can follow some of what you are all doing but of course I got an error message that it would not install on my Win7 machine. I'll have to figure out what is stopping it from doing an install.
With your current method of a separate ops and input section am I to assume that the global feedback circuit is done by an external wire and whatever type, TMC, TPC, etc. that would not be part of the board layout the way you are doing this now? I know that this is over my head at this point but I am learning by watching and reading what you and the other gurus are doing here and the CFA thread.
I actually attempted to install LTSpice today so I can start to see if I can follow some of what you are all doing but of course I got an error message that it would not install on my Win7 machine. I'll have to figure out what is stopping it from doing an install.
It work fine in W7 although I had a slight issue at first.
Does this help,
http://www.diyaudio.com/forums/software-tools/213981-help-ltspice-w7-x64-professional-wont-run.html
I also have the LTspice shortcut to "run as admin". That's important. The other difference I found between running on W7 and Vista is that W7 objects to directly opening a simulation file just by clicking on it. It works OK but there is a message something like "the command couldn't be sent to the program". I always just open LTspice from the shortcut anyhow and browse for whatever I want from there.
Also... I always save simulations in the user areas (documents and so on) and not in the default folder which is buried in the operating system.
Thanks Molly. My problem is that it won't install, says I don't have enough disk space, that is not right or a permission is stopping that. I looked and I am allowing the program to make changes to my computer and still it gets to a certain point in the install and won't install all the components of the program. I'll have to figure out what is really happening with the install before I can do anything with it. And yes I do have Win7 Pro.
All the "fine stuff" (TMC/MIC/any other comp.) is done on The IPS.
The main FB is with an insulated wire. With the VFA's , one could
get really picky and drop the amplitude by splicing the larger value
NFB R at the "NFB" OP of the OPS PCB.
This would just allow a lower amplitude signal physically close to the IPS.
You could do this trick with the VSSA CFA as well.
NAF's CFA (with the servo) has 2 FB routes , one for the servo and
the main CF divider. Here I chose to run the full signal right to the IPS.
With many CM's of AIR as the insulator between NFB and sensitive
current gain/IP stages , any inductive FB is avoided.
With the OPS/IPS PCB's at a 90 degree angle (as designed) , the FB path
is 12Cm long and floating in the air...(a stiff, solid ,insulated 18ga wire).
IDEAL !
You must give LT full admin. on install. Also , once installed ... "take ownership"
of the LT folder (where the .asc's are) AND give Scad3.exe full admin.
privilege (by tweaking it's shortcut).
I had issues (W7-32) ... all fixed !
OS
