Found these in my library, gain not that high though
.MODEL SUB734 PNP IS =2E-13 NF =1 BF =330 IKF=0.8
+VAF=44 ISE=1E-13 NE =1.4 RCO=4.5 GAMMA=5E-9
+NR =1 BR =20 IKR=0.2 VAR=10 ISC=2e-13 NC =1.25
+RB =0.15 RE =0.15 RC =0.2 QUASIMOD=1
+CJC=36E-12 MJC=0.45 VJC=0.75 CJE=110E-12
+TF =0.8E-9 TR =10e-9 XTB=1.4
.ENDS FMMT734
.MODEL SUB634 NPN IS=4.5E-14 ISE=1E-14 NF =1 NE =1.45 BF =250 IKF=.16
+RE=.09 RB=1 BR =12 IKR=.15 VAF=240 CJE=65E-12 CJC=7E-12
+VJC=.85 MJC=.45 TF =.95E-9 TR =300E-9
.ENDS FMMT634
.MODEL SUB734 PNP IS =2E-13 NF =1 BF =330 IKF=0.8
+VAF=44 ISE=1E-13 NE =1.4 RCO=4.5 GAMMA=5E-9
+NR =1 BR =20 IKR=0.2 VAR=10 ISC=2e-13 NC =1.25
+RB =0.15 RE =0.15 RC =0.2 QUASIMOD=1
+CJC=36E-12 MJC=0.45 VJC=0.75 CJE=110E-12
+TF =0.8E-9 TR =10e-9 XTB=1.4
.ENDS FMMT734
.MODEL SUB634 NPN IS=4.5E-14 ISE=1E-14 NF =1 NE =1.45 BF =250 IKF=.16
+RE=.09 RB=1 BR =12 IKR=.15 VAF=240 CJE=65E-12 CJC=7E-12
+VJC=.85 MJC=.45 TF =.95E-9 TR =300E-9
.ENDS FMMT634
homemodder said:
These aren't valid. A Darlington "model" will actually be a subcircuit (that is, specified with a .SUBCKT statement instead of .MODEL). What you've posted here looks like fragments of a SUBCKT (there is a .ENDS at the end but no corresponding .SUBCKT at the beginning).
Once OS gets a subcircuit model of the device, he'll need to figure out how to associate the SUBCKT with a symbol, which is not transparent in LTspice as it is with .MODEL. It's too late at the moment to go into that, but I can help with that tomorrow if need be.
All done modelled KSA 700/800 Hfe -3500-10k..
here is the damage..
Gain is 5db more ,phase is the same...
but the balance is "SWAG"ed as B.cordell puts it..
By the way, most Ksa 1381's have a Hfe of 500+
(almost a darlington) I played with different Re degenerations
What is amazing is that the loop gain plot barely deviated
even with this gross modification.
here is the damage..
Gain is 5db more ,phase is the same...
but the balance is "SWAG"ed as B.cordell puts it..
By the way, most Ksa 1381's have a Hfe of 500+
(almost a darlington) I played with different Re degenerations
What is amazing is that the loop gain plot barely deviated
even with this gross modification.
Attachments
But this plot is the bummer.. the balance on the LTP is destroyed,
going from a few hundreth's of a mA to tenths..must be this
holman topology as the reduced load should not of done that..
I'll have to play around more.
unbalanced plot below...
OS
going from a few hundreth's of a mA to tenths..must be this
holman topology as the reduced load should not of done that..
I'll have to play around more.
unbalanced plot below...
OS
I had good luck with the model (it worked)but Symbol did not update and read as "Model D"
Attachments
ostripper said:
In line with previous comments from you I think it’s clear whom this presumptuous whine is directed at.
Just for my own enjoyment (but to the obvious distress of some) I’ve spent considerable time developing esoteric designs with performance in target areas to match the component count.
I’ve optimised these designs in sim, tested them on the bench and shared key aspects of the circuit operation here numerous times at length.
Documenting the construction of such designs here to any degree mostly has the effect of inciting presumptuous and resentful BS from pseudo-technical hacks and subjectivist whacko’s, so I don’t bother much with that any more.
BTW, I can assure you that there are plenty of PCB layouts and completed assemblies. In fact you can see some of the current production run in the pic below.
Attachments
This is a extreme case of paranoia...
I "steal" useful techniques from even 50 transistor
discussions. To assume I refer to your designs is indeed
presumptuous.
My point is that the average constructor will "pass over"
the 40 device project and embrace the simple.
No offense intended but your designs are not for everybody, as
They will migrate to the "quasi", blameless" DX...
OS
I "steal" useful techniques from even 50 transistor
discussions. To assume I refer to your designs is indeed
presumptuous.
My point is that the average constructor will "pass over"
the 40 device project and embrace the simple.
No offense intended but your designs are not for everybody, as
They will migrate to the "quasi", blameless" DX...
OS
ostripper said:
OS,
Have you measured this (hfe ~500)? The data sheet spec is way lower
and spice models even lower again.
I find generally, spice models are very conservative, ie on the low
side wrt hfe. It depends on the cct but in many instances the high
current gain is desireable.
cheers
Terry
no, not even a good indication.Terry Demol said:
This is mainly due to the fact that most of these DMM transistor testers operate with a near constant Ib and measure the change in Ic.
This change in Ic and unknown Vce ruins the temperature and Vce conditions that are essential to do any comparison.
If this were simply to batch the unmarked devices before measuring the hFE properly, then it is an acceptable indication, but you will find that some are wrongly batched due to these inherent errors, if the near constant Ib method is used.
Hi Os
Your response got me a bit worried so I quickly put the idea to test on a sim with just the front end. Some work may be required on your operating currents.
The theory is spot on, a simple test is to just replace the ksc1381 and ksc3503 with ztx696 and ztx796 which are higher gain devices. These arent very complementary but will do. Every transistor is more suitable for a intended purpose, here you have no cob or voltage issue, although it will help, but gain is a important issue for linearity. The higher gain might just overide some other issues.
Andrew has a good point about the hfe, the ksc1381 is no wonder transistor and I use some devices that will put them to shame, probably on hfe too. I have some original sanyos but not the fairchild second source ones. No model is going to help you here, must physically test. The sanyo 1381 came short against other devices I did a proper test on some years back.
I think you have a mistake somewhere, mod works for me, there are some issues, its a gain beast but Ill tame it and fine tune. Have just sent a japanese manufacturer a request for suitable complementary devices and models, sometimes helpfull to have contacts for these things.
A cfp will work here too, something I will try, but the initial proposition was to have no more added devices. A darlington is going to require some fine tuning here.
I think I might be just be able to turn the holman with the vas configuration as is into another -120db THD amp with just maybe 6 trannies addition. Ill try keep below 40 transistors but sometimes this is required to get extreme objective performance but I stop if it gets too much. The subjective performance is debatable but this new year 2009 Ill refrain from giving any.
Your response got me a bit worried so I quickly put the idea to test on a sim with just the front end. Some work may be required on your operating currents.
The theory is spot on, a simple test is to just replace the ksc1381 and ksc3503 with ztx696 and ztx796 which are higher gain devices. These arent very complementary but will do. Every transistor is more suitable for a intended purpose, here you have no cob or voltage issue, although it will help, but gain is a important issue for linearity. The higher gain might just overide some other issues.
Andrew has a good point about the hfe, the ksc1381 is no wonder transistor and I use some devices that will put them to shame, probably on hfe too. I have some original sanyos but not the fairchild second source ones. No model is going to help you here, must physically test. The sanyo 1381 came short against other devices I did a proper test on some years back.
I think you have a mistake somewhere, mod works for me, there are some issues, its a gain beast but Ill tame it and fine tune. Have just sent a japanese manufacturer a request for suitable complementary devices and models, sometimes helpfull to have contacts for these things.
A cfp will work here too, something I will try, but the initial proposition was to have no more added devices. A darlington is going to require some fine tuning here.
I think I might be just be able to turn the holman with the vas configuration as is into another -120db THD amp with just maybe 6 trannies addition. Ill try keep below 40 transistors
but sometimes this is required to get extreme objective performance but I stop if it gets too much. The subjective performance is debatable but this new year 2009 Ill refrain from giving any.- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.