Same output, 5k Raa load:
Code:
Harmonic Frequency Fourier Normalized Phase Normalized
Number [Hz] Component Component [degree] Phase [deg]
1 1.000e+03 1.855e+01 1.000e+00 -0.08° 0.00°
2 2.000e+03 1.756e-02 9.466e-04 91.22° 91.30°
3 3.000e+03 1.420e-01 7.658e-03 -0.67° -0.59°
4 4.000e+03 1.929e-02 1.040e-03 90.12° 90.20°
5 5.000e+03 5.508e-03 2.970e-04 -0.91° -0.83°
6 6.000e+03 1.642e-03 8.851e-05 -90.82° -90.74°
7 7.000e+03 1.635e-03 8.814e-05 1.45° 1.53°
8 8.000e+03 1.815e-03 9.786e-05 -89.53° -89.45°
9 9.000e+03 1.192e-03 6.428e-05 179.62° 179.70°
Total Harmonic Distortion: 0.779358%(0.779367%)Hi, I post here the last simulation I've done this afternoon.
This is the harmonic distortion at 88 Wrms with a quad of KT77 per channel at 450V 3k3 Raa (6k6 per pair):
This is the harmonic distortion at 88 Wrms with a quad of KT77 per channel at 450V 3k3 Raa (6k6 per pair):
Code:
Harmonic Frequency Fourier Normalized Phase Normalized
Number [Hz] Component Component [degree] Phase [deg]
1 1.000e+03 3.757e+01 1.000e+00 -0.03° 0.00°
2 2.000e+03 5.793e-02 1.542e-03 90.04° 90.07°
3 3.000e+03 2.676e-01 7.122e-03 0.09° 0.12°
4 4.000e+03 1.808e-03 4.811e-05 90.68° 90.71°
5 5.000e+03 1.150e-02 3.061e-04 -0.11° -0.08°
6 6.000e+03 8.504e-04 2.264e-05 -89.25° -89.22°
7 7.000e+03 5.915e-03 1.574e-04 3.53° 3.56°
8 8.000e+03 3.976e-04 1.058e-05 92.39° 92.42°
9 9.000e+03 1.386e-02 3.688e-04 -179.08° -179.05°
Total Harmonic Distortion: 0.730497%(0.730585%)Had a look at this is it your intension the build something or just for simulation.
Don't mean to be critical but here's some things.
1) You currently have a perfect transformer. If you intend to implement NFB you will need to have something closer to the real thing. This will need to include coupling factor series resistance and primary capacitance.
2) Although your you current sources are just fine something much simpler would work just as well. A simple 6v8 reference which can be shared to several 2n5550's and just a single 2n5550 would be fine. You can also share the grid reference for the 12AX7. One to try in LT spice.
3) You only have about 5dB of NFB, typically one would aim for 17dB or so. However you need to add a gain stage at the front end to achieve this.
4) Maybe a ECC88 or 12BH7 may a better follower than a 12AT7 less likely to draw grid current.
5) JFET cascade is interesting idea. A J113 would give a bit more gain. If you wanted all valve you could use a EF86 matched pair.
Don't mean to be critical but here's some things.
1) You currently have a perfect transformer. If you intend to implement NFB you will need to have something closer to the real thing. This will need to include coupling factor series resistance and primary capacitance.
2) Although your you current sources are just fine something much simpler would work just as well. A simple 6v8 reference which can be shared to several 2n5550's and just a single 2n5550 would be fine. You can also share the grid reference for the 12AX7. One to try in LT spice.
3) You only have about 5dB of NFB, typically one would aim for 17dB or so. However you need to add a gain stage at the front end to achieve this.
4) Maybe a ECC88 or 12BH7 may a better follower than a 12AT7 less likely to draw grid current.
5) JFET cascade is interesting idea. A J113 would give a bit more gain. If you wanted all valve you could use a EF86 matched pair.
Thanks for your interest baudouin0,
Actually the hybrid cascode has a LSK170 on bottom and 12AX7 on top, but U440 seems more linear. It has one single pair of EL34 in UL
I would like to reuse the actual EL34 and transformers for some guitar projects I have, and buy new transformers for this KT77 project.
I've bought two monoblocks from a user of an italian forum, and then modified them to see how different nfb will sound like.
Actually the hybrid cascode has a LSK170 on bottom and 12AX7 on top, but U440 seems more linear. It has one single pair of EL34 in UL
I would like to reuse the actual EL34 and transformers for some guitar projects I have, and buy new transformers for this KT77 project.
It will be a custom transformer, so I don't have information about those aspects at the moment. It will be a Toroidy transformer. Do you have any suggestion as an estimation of those data?
The grids of the 12ax7 are indeed shared in reality, it was just simpler graphically to have twice the regulators. I will simulate with one single voltage reference. The pcb already has different references for the CCS, but it's a good idea to have just one if the PCB will be redone.
Do you think it would be better to reduce shunt feedback and increase gnfb? In the current design I've no gnfb at all.
Ideally it would be better to apply the biasing voltage to the grid of the driver and dc couple the driver to the grid of the KT77s. I will try your suggestion as well, trying to find a tube with good linearity without too much current needed through it.
I know a pensotde would be even better, but I've found the cascode being nice sounding and quite simple.
Have a look at post.
https://www.diyaudio.com/forums/tubes-valves/355686-weve-heartbeat-6x-6550-alive-new-post.html
I just making suggestions - having a look at your simulation and a simulation of 6550 amp. The problem you will have is getting enough closed loop gain without an additional input stage.
https://www.diyaudio.com/forums/tubes-valves/355686-weve-heartbeat-6x-6550-alive-new-post.html
I just making suggestions - having a look at your simulation and a simulation of 6550 amp. The problem you will have is getting enough closed loop gain without an additional input stage.
I have a model of the toroidal used in the K4040 amp however that transformer is a bit rubbish at HF.
TRANSFORMER LIBRARY
.SUBCKT DYNA_OUTPUT_XFRMR 1 2 3 4 5 6 7 8 9 ; PARAMETERS FOR MARK 3:
+PARAMS: LPRIM=60 LLKG=.040 RPRIM=125 CPRIM=1.04NF LRATIO={4/4300}
* ERIC BARBOUR ARTICLE: ~233H TOTAL PRIMARY L FOR MARK 3.
* MARK 3: LPRIM=60 LLKG=.040 RPRIM=125 CPRIM=1.04NF LRATIO={4/4300}
* LPRIM IS THE TOTAL PRIMARY L (VARIES WITH MEASUREMENT).
* LLKG IS THE LEAKAGE L (MEASURABLE: CONSISTENT).
* RPRIM IS THE TOTAL PRIMARY R.
* CPRIM IS THE MEASURED PRIMARY CAPACITANCE.
* LRATIO IS THE INDUCTANCE RATIO: (4 OHMS)/(PRIMARY Z).
.PARAM QFCTR={LPRIM/LLKG} ; Q-FACTOR.
CS1 1 5 {CPRIM} ; PRIMARY CAPACITANCE
RS1 1 5 300K ; SHUNT R FOR HIGH FREQUENCY EFFECTS.
LP1 1 12 {LPRIM*.09} ; .7164H ; PRIMARY
RP1 12 2 {RPRIM*.5}
LP2 2 3 {LPRIM*.04} ; .3184H
LP3 3 4 {LPRIM*.04}
LP4 4 45 {LPRIM*.09}
RP4 45 5 {RPRIM*.5}
LP5 7 6 {.34315*LPRIM*LRATIO} ; 8-16 OHM WINDING: (2-SQRT(2))^2
LP6 8 7 {.17157*LPRIM*LRATIO} ; 4-8 OHM WINDING: (SQRT(2)-1)^2
LP7 9 8 {LPRIM*LRATIO} ; COM-4 OHM WINDING
KALL LP1 LP2 LP3 LP4 LP5 LP6 LP7 {1-1/(2*QFCTR)} ; COUPLING
.ENDS
.SUBCKT SOWTER_UA21 1 2 3 4 5 6 7 8 9 10 11 12 13
+PARAMS: LPRIM=145 RPRIM=192 RSEC=.15 CPRIM=2.5n LRATIO={1/3600} LLKG=.05
* LPRIM IS THE TOTAL PRIMARY L (VARIES WITH MEASUREMENT).
* LLKG IS THE LEAKAGE L (MEASURABLE: CONSISTENT).
* RPRIM IS THE TOTAL PRIMARY R.
* CPRIM IS THE MEASURED PRIMARY CAPACITANCE.
* LRATIO IS THE INDUCTANCE RATIO: (4 OHMS)/(PRIMARY Z).#.PARAM
.PARAM QFCTR={LPRIM/LLKG} ; Q-FACTOR.
CPR 1 5 {CPRIM}
LP1 1 21 {LPRIM*.09} ; .7164H ; PRIMARY
RP1 21 2 {RPRIM*.3}
LP2 2 22 {LPRIM*.04} ; .3184H
RP2 22 3 {RPRIM*.2}
LP3 3 23 {LPRIM*.04}
RP3 23 4 {RPRIM*.2}
LP4 4 24 {LPRIM*.09}
RP4 24 5 {RPRIM*.3}
LP5 7 31 {LPRIM*LRATIO}
RP5 31 6 {RSEC}
LP6 9 32 {LPRIM*LRATIO}
RP6 32 8 {RSEC}
LP7 11 33 {LPRIM*LRATIO}
RP7 33 10 {RSEC}
LP8 13 34 {LPRIM*LRATIO}
RP8 34 12 {RSEC}
KALL LP1 LP2 LP3 LP4 LP5 LP6 LP7 LP8 {1-1/(2*QFCTR)} ; COUPLING
.ENDS
.SUBCKT PAT-4006-CFB 1 2 3 4 5 6 7 8 9 10 11
* PLITRON PAT-4006-CFB OUTPUT TRANSFORMER 2KOHM UL PRIMARY
* OL NUMBERS CORRESPOND TO TRANSFORMER SCHEMATIC.
.PARAM PRIML=392.5 ; TOTAL PRIMARY L (FROM SPECS).
.PARAM LRATIO={5/2000} ; INDUCTANCE RATIO: (5 OHMS)/(PRIMARY).
.PARAM QFCTR=400000 ; Q-FACTOR: PRIMARY SHUNT L/LEAKAGE L.
LP1 1 2 {PRIML*.09} ; PRIMARY
LP2 2 3 {PRIML*.04}
LP3 3 4 {PRIML*.04}
LP4 4 5 {PRIML*.09}
CP1 1 5 .342NF ; CAPACITANCE FROM SPECS
LP5 8 7 {PRIML*LRATIO/4} ; 1/2 SPEAKER SECONDARY
LP6 7 6 {PRIML*LRATIO/4} ; " "
LP7 11 10 {PRIML*LRATIO} ; 1/2 FBK WINDING
LP8 10 9 {PRIML*LRATIO} ; " "
KALL LP1 LP2 LP3 LP4 LP5 LP6 LP7 LP8 .9999987 ; 1-1/(2*403600) AWESOME!
.ENDS
* Model of 1650R transformer
.SUBCKT 1650R 100 102 104 106 108 204 202 200 302 300 ; Blu Blu/Yel Red Brn/Yel Brn Yel Grn/Yel Blk/Yel Grn Blk Case
.PARAM Lp=800 ; TOTAL PRIMARY L (FROM SPECS).
.PARAM Rp=100 ; TOTAL PRIMARY R (FROM SPECS).
.PARAM Rs=.3 ; SECONDARY R (FROM SPECS).
.PARAM MM=.999947
.PARAM Lrat={4/4300} ; INDUCTANCE RATIO: (4 OHMS)/(PRIMARY).
.PARAM Cp=300p
L10 100 101 {Lp*.09}
R10 101 102 {Rp*.3}
L11 102 103 {Lp*.04}
R11 103 104 {Rp*.2}
L12 104 105 {Lp*.04}
R12 105 106 {Rp*.2}
L13 106 107 {Lp*.09}
R13 107 108 {Rp*.3}
C10 100 108 {Cp}
L20 200 201 {Lp*Lrat}
R20 201 202 {Rs}
L21 202 203 {Lp*Lrat*.1716}
R21 203 204 {Rs*.4142}
L30 300 301 {Lp*Lrat}
R30 301 302 {Rs}
K1 L10 L11 L12 L13 L20 L21 L30 {MM}
.ENDS
* Model of 1650T transformer
.SUBCKT 1650T 100 102 104 106 108 204 202 200 302 300 ; Blu Blu/Yel Red Brn/Yel Brn Yel Grn/Yel Blk/Yel Grn Blk Case
+ PARAMS: Lp=.625H Rp=2R Cp=200p Ls=10.7mH Rs=0.025R MM=.999947
L10 100 101 {36*Lp}
R10 101 102 {6*Rp}
L11 102 103 {16*Lp}
R11 103 104 {4*Rp}
L12 104 105 {16*Lp}
R12 105 106 {4*Rp}
L13 106 107 {36*Lp}
R13 107 108 {6*Rp}
C10 100 108 {Cp}
*
L20 200 201 {49*Ls}
R20 201 202 {7*Rs}
L21 202 203 {9*Ls}
R21 203 204 {3*Rs}
*
L30 300 301 {49*Ls}
R30 301 302 {7*Rs}
*
K1 L10 L11 L12 L13 L20 L21 L30 {MM}
.ENDS
.SUBCKT ZD043 1 2 3 4 5 6 7 8
* ZD043 VELLEMAN OUTPUT TRANSFORMER 2KOHM UL PRIMARY
* OL NUMBERS CORRESPOND TO TRANSFORMER SCHEMATIC.
.PARAM CPL=1
.PARAM PRIML=10 ; TOTAL PRIMARY L (FROM SPECS).
.PARAM LRATIO={8/2500} ; INDUCTANCE RATIO: (5 OHMS)/(PRIMARY).
.PARAM STAP=.333333 ; Screen tap position @ 33%
.PARAM PRIMR=1
L1 1 22 {PRIML*(1-STAP)*(1-STAP)/4}
R1 22 2 {PRIMR*(1-STAP)/2}
L2 2 33 {PRIML*STAP*STAP/4}
R2 33 3 {PRIMR*STAP/2}
L3 3 44 {PRIML*STAP*STAP/4}
R3 44 4 {PRIMR*STAP/2}
L4 4 55 {PRIML*(1-STAP)*(1-STAP)/4}
R4 55 5 {PRIMR*(1-STAP)/2}
*C1 1 5 5NF ; CAPACITANCE FROM SPECS
L5 8 7 {4*PRIML*LRATIO/9} ; 1/2 SPEAKER SECONDARY
L6 7 6 {1*PRIML*LRATIO/9} ; " "
K1 L1 L2 L3 L4 L5 L6 {CPL}
.ENDS
TRANSFORMER LIBRARY
.SUBCKT DYNA_OUTPUT_XFRMR 1 2 3 4 5 6 7 8 9 ; PARAMETERS FOR MARK 3:
+PARAMS: LPRIM=60 LLKG=.040 RPRIM=125 CPRIM=1.04NF LRATIO={4/4300}
* ERIC BARBOUR ARTICLE: ~233H TOTAL PRIMARY L FOR MARK 3.
* MARK 3: LPRIM=60 LLKG=.040 RPRIM=125 CPRIM=1.04NF LRATIO={4/4300}
* LPRIM IS THE TOTAL PRIMARY L (VARIES WITH MEASUREMENT).
* LLKG IS THE LEAKAGE L (MEASURABLE: CONSISTENT).
* RPRIM IS THE TOTAL PRIMARY R.
* CPRIM IS THE MEASURED PRIMARY CAPACITANCE.
* LRATIO IS THE INDUCTANCE RATIO: (4 OHMS)/(PRIMARY Z).
.PARAM QFCTR={LPRIM/LLKG} ; Q-FACTOR.
CS1 1 5 {CPRIM} ; PRIMARY CAPACITANCE
RS1 1 5 300K ; SHUNT R FOR HIGH FREQUENCY EFFECTS.
LP1 1 12 {LPRIM*.09} ; .7164H ; PRIMARY
RP1 12 2 {RPRIM*.5}
LP2 2 3 {LPRIM*.04} ; .3184H
LP3 3 4 {LPRIM*.04}
LP4 4 45 {LPRIM*.09}
RP4 45 5 {RPRIM*.5}
LP5 7 6 {.34315*LPRIM*LRATIO} ; 8-16 OHM WINDING: (2-SQRT(2))^2
LP6 8 7 {.17157*LPRIM*LRATIO} ; 4-8 OHM WINDING: (SQRT(2)-1)^2
LP7 9 8 {LPRIM*LRATIO} ; COM-4 OHM WINDING
KALL LP1 LP2 LP3 LP4 LP5 LP6 LP7 {1-1/(2*QFCTR)} ; COUPLING
.ENDS
.SUBCKT SOWTER_UA21 1 2 3 4 5 6 7 8 9 10 11 12 13
+PARAMS: LPRIM=145 RPRIM=192 RSEC=.15 CPRIM=2.5n LRATIO={1/3600} LLKG=.05
* LPRIM IS THE TOTAL PRIMARY L (VARIES WITH MEASUREMENT).
* LLKG IS THE LEAKAGE L (MEASURABLE: CONSISTENT).
* RPRIM IS THE TOTAL PRIMARY R.
* CPRIM IS THE MEASURED PRIMARY CAPACITANCE.
* LRATIO IS THE INDUCTANCE RATIO: (4 OHMS)/(PRIMARY Z).#.PARAM
.PARAM QFCTR={LPRIM/LLKG} ; Q-FACTOR.
CPR 1 5 {CPRIM}
LP1 1 21 {LPRIM*.09} ; .7164H ; PRIMARY
RP1 21 2 {RPRIM*.3}
LP2 2 22 {LPRIM*.04} ; .3184H
RP2 22 3 {RPRIM*.2}
LP3 3 23 {LPRIM*.04}
RP3 23 4 {RPRIM*.2}
LP4 4 24 {LPRIM*.09}
RP4 24 5 {RPRIM*.3}
LP5 7 31 {LPRIM*LRATIO}
RP5 31 6 {RSEC}
LP6 9 32 {LPRIM*LRATIO}
RP6 32 8 {RSEC}
LP7 11 33 {LPRIM*LRATIO}
RP7 33 10 {RSEC}
LP8 13 34 {LPRIM*LRATIO}
RP8 34 12 {RSEC}
KALL LP1 LP2 LP3 LP4 LP5 LP6 LP7 LP8 {1-1/(2*QFCTR)} ; COUPLING
.ENDS
.SUBCKT PAT-4006-CFB 1 2 3 4 5 6 7 8 9 10 11
* PLITRON PAT-4006-CFB OUTPUT TRANSFORMER 2KOHM UL PRIMARY
* OL NUMBERS CORRESPOND TO TRANSFORMER SCHEMATIC.
.PARAM PRIML=392.5 ; TOTAL PRIMARY L (FROM SPECS).
.PARAM LRATIO={5/2000} ; INDUCTANCE RATIO: (5 OHMS)/(PRIMARY).
.PARAM QFCTR=400000 ; Q-FACTOR: PRIMARY SHUNT L/LEAKAGE L.
LP1 1 2 {PRIML*.09} ; PRIMARY
LP2 2 3 {PRIML*.04}
LP3 3 4 {PRIML*.04}
LP4 4 5 {PRIML*.09}
CP1 1 5 .342NF ; CAPACITANCE FROM SPECS
LP5 8 7 {PRIML*LRATIO/4} ; 1/2 SPEAKER SECONDARY
LP6 7 6 {PRIML*LRATIO/4} ; " "
LP7 11 10 {PRIML*LRATIO} ; 1/2 FBK WINDING
LP8 10 9 {PRIML*LRATIO} ; " "
KALL LP1 LP2 LP3 LP4 LP5 LP6 LP7 LP8 .9999987 ; 1-1/(2*403600) AWESOME!
.ENDS
* Model of 1650R transformer
.SUBCKT 1650R 100 102 104 106 108 204 202 200 302 300 ; Blu Blu/Yel Red Brn/Yel Brn Yel Grn/Yel Blk/Yel Grn Blk Case
.PARAM Lp=800 ; TOTAL PRIMARY L (FROM SPECS).
.PARAM Rp=100 ; TOTAL PRIMARY R (FROM SPECS).
.PARAM Rs=.3 ; SECONDARY R (FROM SPECS).
.PARAM MM=.999947
.PARAM Lrat={4/4300} ; INDUCTANCE RATIO: (4 OHMS)/(PRIMARY).
.PARAM Cp=300p
L10 100 101 {Lp*.09}
R10 101 102 {Rp*.3}
L11 102 103 {Lp*.04}
R11 103 104 {Rp*.2}
L12 104 105 {Lp*.04}
R12 105 106 {Rp*.2}
L13 106 107 {Lp*.09}
R13 107 108 {Rp*.3}
C10 100 108 {Cp}
L20 200 201 {Lp*Lrat}
R20 201 202 {Rs}
L21 202 203 {Lp*Lrat*.1716}
R21 203 204 {Rs*.4142}
L30 300 301 {Lp*Lrat}
R30 301 302 {Rs}
K1 L10 L11 L12 L13 L20 L21 L30 {MM}
.ENDS
* Model of 1650T transformer
.SUBCKT 1650T 100 102 104 106 108 204 202 200 302 300 ; Blu Blu/Yel Red Brn/Yel Brn Yel Grn/Yel Blk/Yel Grn Blk Case
+ PARAMS: Lp=.625H Rp=2R Cp=200p Ls=10.7mH Rs=0.025R MM=.999947
L10 100 101 {36*Lp}
R10 101 102 {6*Rp}
L11 102 103 {16*Lp}
R11 103 104 {4*Rp}
L12 104 105 {16*Lp}
R12 105 106 {4*Rp}
L13 106 107 {36*Lp}
R13 107 108 {6*Rp}
C10 100 108 {Cp}
*
L20 200 201 {49*Ls}
R20 201 202 {7*Rs}
L21 202 203 {9*Ls}
R21 203 204 {3*Rs}
*
L30 300 301 {49*Ls}
R30 301 302 {7*Rs}
*
K1 L10 L11 L12 L13 L20 L21 L30 {MM}
.ENDS
.SUBCKT ZD043 1 2 3 4 5 6 7 8
* ZD043 VELLEMAN OUTPUT TRANSFORMER 2KOHM UL PRIMARY
* OL NUMBERS CORRESPOND TO TRANSFORMER SCHEMATIC.
.PARAM CPL=1
.PARAM PRIML=10 ; TOTAL PRIMARY L (FROM SPECS).
.PARAM LRATIO={8/2500} ; INDUCTANCE RATIO: (5 OHMS)/(PRIMARY).
.PARAM STAP=.333333 ; Screen tap position @ 33%
.PARAM PRIMR=1
L1 1 22 {PRIML*(1-STAP)*(1-STAP)/4}
R1 22 2 {PRIMR*(1-STAP)/2}
L2 2 33 {PRIML*STAP*STAP/4}
R2 33 3 {PRIMR*STAP/2}
L3 3 44 {PRIML*STAP*STAP/4}
R3 44 4 {PRIMR*STAP/2}
L4 4 55 {PRIML*(1-STAP)*(1-STAP)/4}
R4 55 5 {PRIMR*(1-STAP)/2}
*C1 1 5 5NF ; CAPACITANCE FROM SPECS
L5 8 7 {4*PRIML*LRATIO/9} ; 1/2 SPEAKER SECONDARY
L6 7 6 {1*PRIML*LRATIO/9} ; " "
K1 L1 L2 L3 L4 L5 L6 {CPL}
.ENDS
Thanks! I’ve chosen KT77 for their linearity around that loadline with 43% UL, and their high gm (I need only 70Vpp instead of 106Vpp to reach full power). One more thing: I have 267 kOhm on 12ax7s vs 220k. Those two things help to reach the ballpark (by now I’m using it eith EL34 and a 40% UL, but the actual OT is inappropriate.
Yes you might get there with the lower drive of the KT77, slightly bigger plate resistors and a good choice of JFET. Try and get as much gain as possible so you can have as much negative feedback. I forgot to mention I would increase the grid resistors to the KT77 to reduce blocking distortion. Say 2k2 or 4k7 rather than 470R.
I am sure you understand blocking distortion. Making R bigger reduces the peak positive grid current and therefore reduces the bias shift. The aim is to reduce this shift and the duration as much as possible. There's plenty of stuff on blocking distortion - I just noticed that on simulation that with the cathode follower able to drive hard positive this amp is quite prone to it.
Are I see what you are saying. Not quite correct you need to drive the bias current down to zero in the final stage after the overload to get crossover distortion. So you have a window between the fixed bias current and zero to work in. So even if the final stage has only 2ma left in it this is still much better in terms of distortion than being driven out of conduction. The NFB also has a much more trouble correcting something with no gain (where it will go unstable) than even say half the gain. A complicated answer to a simple question.
Are I see what you are saying. Not quite correct you need to drive the bias current down to zero in the final stage after the overload to get crossover distortion. So you have a window between the fixed bias current and zero to work in. So even if the final stage has only 2ma left in it this is still much better in terms of distortion than being driven out of conduction. The NFB also has a much more trouble correcting something with no gain (where it will go unstable) than even say half the gain. A complicated answer to a simple question.
Last edited:
What a larger series grid stopper resistor mostly does is change the time constant.
Given enough time, the change of the voltage across the capacitor is almost the same.
That means with time, the bias shift is essentially the same.
A long low frequency transient will shift the bias very radically (such as the signal from the 6Hz canon on the Telarc recording of the 1812 overture).
Sorry that I used such an extreme, just to illustrate the point.
As I have always have said, if your amplifier distorts, blocks or both, then . . .
1. Turn the volume down.
2. Get a more powerful amplifier.
3. Live with the distortion and or blocking.
4. Change the recording that you are listening to.
At some volume, all amplifiers will clip, distort, etc.
Given enough time, the change of the voltage across the capacitor is almost the same.
That means with time, the bias shift is essentially the same.
A long low frequency transient will shift the bias very radically (such as the signal from the 6Hz canon on the Telarc recording of the 1812 overture).
Sorry that I used such an extreme, just to illustrate the point.
As I have always have said, if your amplifier distorts, blocks or both, then . . .
1. Turn the volume down.
2. Get a more powerful amplifier.
3. Live with the distortion and or blocking.
4. Change the recording that you are listening to.
At some volume, all amplifiers will clip, distort, etc.
I will send you an accurate simulation. It does change where the bias ends up as there is less current flow. If you can shift the bias point a little out of cut off it can make a big difference. It just that peak clipping often goes unnoticed whereas the ear will find blocking distortion objectionable.
Last edited:
A typical small to medium grid current will make the grid to cathode resistance be on the order of 1k Ohms.
A grid stopper resistor of:
A. 470 Ohms,
B. 4700 Ohms,
in series with a grid to cathode resistance of 1k
Using a 0.22uF coupling cap for example . . .
Time constant of 1.47k and 0.22uF = 323uS
Time constant of 5.70k and 0.22uF = 1254uS
The time of one alternation of 1kHz is 500uS (fairly close to the 2 time constants above).
Suppose the grid bias is -30V.
The driver swings to +30V peak. No problem.
The driver swings 1dB more than that, +33V, that is a problem. There will be bias shift.
Using a 1kHz note, the time of the excess 3V is quite short.
But instead, use a 40Hz bass note, the time of the excess 3V is much longer than 722uS.
So 20Hz, a "Golden" Hi Fi number, is 1444us (compare that to 1254us above)
IRMC (I hate acronyms, so I will write it out I Rest My Case).
A grid stopper resistor of:
A. 470 Ohms,
B. 4700 Ohms,
in series with a grid to cathode resistance of 1k
Using a 0.22uF coupling cap for example . . .
Time constant of 1.47k and 0.22uF = 323uS
Time constant of 5.70k and 0.22uF = 1254uS
The time of one alternation of 1kHz is 500uS (fairly close to the 2 time constants above).
Suppose the grid bias is -30V.
The driver swings to +30V peak. No problem.
The driver swings 1dB more than that, +33V, that is a problem. There will be bias shift.
Using a 1kHz note, the time of the excess 3V is quite short.
But instead, use a 40Hz bass note, the time of the excess 3V is much longer than 722uS.
So 20Hz, a "Golden" Hi Fi number, is 1444us (compare that to 1254us above)
IRMC (I hate acronyms, so I will write it out I Rest My Case).
Last edited:
I took Zintolo schematic and overloaded the input. I can see very quickly that blocking distortion develops. I changed the 470 grids to 4k7 and this made little difference. It was not until I got to 22k that things really improved. However at 22k this may well effect the stability of the NFB would have to check. So unlike guitar stages with 12ax7 which are using 100k upwards anything below 10k makes little difference. So there you are. It may be using a follower makes matters worse here as often you will already have say 33k plate resistors.
Oddly enough the 6550 schematic I throw together is much better with blocking distortion and again the grid resistor makes little difference. Maybe what's more important is the drive circuit should not go too much positive.
So YRYC
Oddly enough the 6550 schematic I throw together is much better with blocking distortion and again the grid resistor makes little difference. Maybe what's more important is the drive circuit should not go too much positive.
So YRYC
Last edited:
I've asked to Toroidy, they can build a custom transformer for a quad of KT77 per channel, adding a custom 5,5 Ohm tap on secondary, at a very good price. I'm towards that choice at the moment.
But I'll keep you suggestion to open a dedicated thread about it.
Thanks!
Do you think the SS section of the forum is a better place to ask for a better Jfet?
Thanks, I will!