Andrew , why did you not advise Quasi to do that ?? (post 254 , picture 2). As a project I am already at 2 boards per amp ... no external L//R. I can still do the NFB routing , heck I will , just to put this to sleep AND be better than the others.
This leaves me with a choice with the PB250 , which is worse taking NFB from halfway between 2 opposing outputs where I do now or running a real long trace right to the output pad. A way around this actually would be to run a grounded shielded wire right from the NFB input of the voltage board directly to the L//R pad , bypassing all the board contamination.
OS
Last edited:
See pix 1 below and tell me how the signal would be different from the present NFB point or the pad with this style layout. I won't argue the point .. as I will document AND add 1 more pad to run a shielded NFB from the present jumper to the L//R pad (method 2). 
Pix 2 is the PB 60 , absolute NFB takeoff , right at the pad.... shorter route.
OS
Pix 2 is the PB 60 , absolute NFB takeoff , right at the pad.... shorter route.
OS
Attachments
Last edited:
If you run the shielded wire just tack the ground shield to the voltage board ground connection on just 1 end of the wire . I did this on the first "frugalamp" (like the AX) no problems.
I am running the 220p on the "supersyms" now, no problems. On both the "supers" and these new ones , the input is shunted by a 100k resistor between the input and signal ground.
Do you see my point on the NFB , Kean. The middle of the op trace is the "fulcrum" for the opposing voltages coming through the emitter resistors .. each 40mm spacing between each p / n emitter pair is a separate amp in itself . To put the shield right to the L//R trace IS better , I agree ... but only by a VERY small degree. On a typical left - right layout (actual traces meeting at a center point) , going off-center by even a few millimeters can be seen as corruption and a bad choice for a takeoff point. I have no traces , almost like hard wiring (emitter to bus ). PB250 is a "wire with gain" Style , able to let even "beefy" MJL21193/4's "drop the load" into that center bus.
I am going to even use small gauge copper solderwick to augment the traces (V+ / V- / OP bus) Overkill in everyway.
OS
OS
Last edited:
Okay, it's been several posts, but... CCS DISTORTION!
Yes, solid state CCS's have distortion because of their nonlinear output impedance (very nonlinear). I like the thought about avoiding phase/response anomalies in the output impedance, but output impedance itself is important because it determines how much distortion is injected by the CCS.
I suppose the question is, at what point does CCS distortion exceed capacitor distortion? Does it?
Here are THD plots of the CCS's, at 34V pk-pk, 10mA, commonly found in the VAS.
Also, would it not be okay to connect the shield to the output and leave it unconnected at the other end?
- keantoken
Yes, solid state CCS's have distortion because of their nonlinear output impedance (very nonlinear). I like the thought about avoiding phase/response anomalies in the output impedance, but output impedance itself is important because it determines how much distortion is injected by the CCS.
I suppose the question is, at what point does CCS distortion exceed capacitor distortion? Does it?
Here are THD plots of the CCS's, at 34V pk-pk, 10mA, commonly found in the VAS.
Also, would it not be okay to connect the shield to the output and leave it unconnected at the other end?
- keantoken
Attachments
Last edited:
Concerning the baxendall CCS , I got it to sim in isolation no problem. when I used it to source a VAS or LTP , it "fell apart". At under 3ma or with a large modulated load (vas) it was quite "squirrelly "
Maybe if you could "plug it in" to an amp I could see it's benefits.
You should ground the shield , as this would negate any inductances from the OP bus or cap returns ...as you would have the wire under the board "sandwiched" between the FR-4 and the heatsink/chassis.
OS
That is what I said , used as a "strait" CCS , as you post them .. they work - all of them. But , I try to plug them into an amp
they "poop the bed". BTW , I tried the tweaked one first then the normal one , both worked in isolation perfectly but I had great difficulty with the more complex application.here, plug it into the AX (2 CCS's), see if you can get it to work.. (below)
OS
Last edited:
The question is rather what is the real importance of a CCS
as a THD source..
THD for a classical CCS implemented with a bjt and a voltage
reference is in the order of a few ppm, while the distorsion
due to the VAS itself is about 1 % and soaring with high
voltages swings...
indeed it is audible if designed this way :
http://www.diyaudio.com/forums/soli...iews-comments-here-please-46.html#post2267574
Ha , HA ...
"Designed" might be an improper term in this case. Carlos should of used 2 current sources like the AX.. 1/10th the distortion (I simmed his )OS
I have noticed that the LTP balance actually gets better up and beyond 10Khz with the 2 tranny CCS's .. very noticable on the AX and CX.
OS
Okay, I digress, the LTP of Carlos's amp uses a weak CCS.
For some reason the AC impedance of the voodoo CCS is the same as the two-Q version, in spite of the much lower DC impedance. This I do not understand.
Wait a sec - Ah! Ostripper, your KSC3503 model is botched I think... Try this one:
.MODEL SC3503 NPN IS=7.010E-13 BF=156.09 VAF=600 IKF=0.12325 ISE=1.2538E-14 NE=1.5 BR=0.64499 VAR=100 IKR=0.05102 ISC=6.4644E-09 NC=1.5 RE=0.108 RC=1.215 RB=12.134 RBM=0.034 IRB=3.0e-6 CJE=7.10E-12 CJC=8.20E-12 TF=7.025E-10 XTF=2 VTF=35 ITF=1 TR=1.0E-8 EG=0.76 XTB=1.5 FC=0.5
- keantoken
For some reason the AC impedance of the voodoo CCS is the same as the two-Q version, in spite of the much lower DC impedance. This I do not understand.
Wait a sec - Ah! Ostripper, your KSC3503 model is botched I think... Try this one:
.MODEL SC3503 NPN IS=7.010E-13 BF=156.09 VAF=600 IKF=0.12325 ISE=1.2538E-14 NE=1.5 BR=0.64499 VAR=100 IKR=0.05102 ISC=6.4644E-09 NC=1.5 RE=0.108 RC=1.215 RB=12.134 RBM=0.034 IRB=3.0e-6 CJE=7.10E-12 CJC=8.20E-12 TF=7.025E-10 XTF=2 VTF=35 ITF=1 TR=1.0E-8 EG=0.76 XTB=1.5 FC=0.5
- keantoken
"Botched"
it is the genuine fairchild model. I put yours in , no difference. Back to the topic of CCS's , it is alright to "share" the LED or diode referenced CCS's between LTP and VAS (my "supersym" has a RED LED CCS), but they are rather low performance to begin with. The 2 transistor ones work best alone since they are based on the gain of the transistors rather than an external referance.
OS
Last edited:
A question , Keen. Why do you use LT instead of multisim or microcap. And do you believe all models are created equal.
I have noticed that the newer models along with LT can simulate problems (oscillation , saturation) MUCH more reliably. I have multisim and it is too easy(optimistic). Where a circuit WILL work in multisim , sometimes it will not work or oscillate in LT. Could it be that LT's output is a more realistic representation of the real world ???
OS
I have noticed that the newer models along with LT can simulate problems (oscillation , saturation) MUCH more reliably. I have multisim and it is too easy(optimistic). Where a circuit WILL work in multisim , sometimes it will not work or oscillate in LT. Could it be that LT's output is a more realistic representation of the real world ???
OS
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.