TLDR;
a lot has to do with the output transformer and the hysteresis / B-H curve. Basically if you are a power hungry disciple with insentive small speakers then you often like PP, if you have efficient speakers and like fine detal then you often (eventually) end up with SE.
I've had the full range of amplifiers from small to big, solid state and tube. The ones that stayed in the end was a simple SE EL84 with UL and a SE EL506 (a premium 7868 / 7591, half the distortion) with UL and CFB.
a lot has to do with the output transformer and the hysteresis / B-H curve. Basically if you are a power hungry disciple with insentive small speakers then you often like PP, if you have efficient speakers and like fine detal then you often (eventually) end up with SE.
I've had the full range of amplifiers from small to big, solid state and tube. The ones that stayed in the end was a simple SE EL84 with UL and a SE EL506 (a premium 7868 / 7591, half the distortion) with UL and CFB.
The N Fdbk to the driver cathode in the Dutch schematic is a fairly standard local N Fdbk scheme, although possibly one of the earlier implementations.
It puts plenty of loop gain into the local N Fdbk (versus the more common lately, "shunt Schade" scheme), so is quite effective. Enough so that some have said it sounds "flat", ie., no 2nd H distortion left in the signal.
Merlin Blencowe posted a rather unknown patent scheme ( # US4614914A ) last Spring, from 1986 for Audio Precision, that removes the magnetizing current (and hysteresis) from OT type outputs, SE or P-P.
It just requires a low value current sense resistor in the output tube(s) cathode, with a high value resistor back to the driver cathode(s) ( above a small driver cathode R). It is adjusted to inject just enough drive signal (positive current Fdbk) to the driver to supply just enough output tube drive to overcome the OT primary resistive loss (+ magnetizing current loss) for the current sensed. Effectively removing the OT primary resistance and the magnetizing current from the signal transferred to the OT secondary. Removing the primary resistance should improve the damping factor for SE as well. (this is an application of negative resistance, on a limited scale)
For SE and all the hoopla of fixing the OT crossover magnetization, (which is actually greater for SE), this is a MUST read. Patent # US4614914A
IMO no tube amplifier with an OT can be considered a "serious" design effort without this easy to implement scheme. Effectively, it makes for an OTL like design, without the heat or expense of parallel tubes, but still using an impedance matching OT.
https://patents.google.com/patent/US4614914A/en?oq=us4614914
It puts plenty of loop gain into the local N Fdbk (versus the more common lately, "shunt Schade" scheme), so is quite effective. Enough so that some have said it sounds "flat", ie., no 2nd H distortion left in the signal.
Merlin Blencowe posted a rather unknown patent scheme ( # US4614914A ) last Spring, from 1986 for Audio Precision, that removes the magnetizing current (and hysteresis) from OT type outputs, SE or P-P.
It just requires a low value current sense resistor in the output tube(s) cathode, with a high value resistor back to the driver cathode(s) ( above a small driver cathode R). It is adjusted to inject just enough drive signal (positive current Fdbk) to the driver to supply just enough output tube drive to overcome the OT primary resistive loss (+ magnetizing current loss) for the current sensed. Effectively removing the OT primary resistance and the magnetizing current from the signal transferred to the OT secondary. Removing the primary resistance should improve the damping factor for SE as well. (this is an application of negative resistance, on a limited scale)
For SE and all the hoopla of fixing the OT crossover magnetization, (which is actually greater for SE), this is a MUST read. Patent # US4614914A
IMO no tube amplifier with an OT can be considered a "serious" design effort without this easy to implement scheme. Effectively, it makes for an OTL like design, without the heat or expense of parallel tubes, but still using an impedance matching OT.
https://patents.google.com/patent/US4614914A/en?oq=us4614914
Last edited:
There is no crossover distortion in SE since the current does not change direction, it only goes from more current to less current and visa versa. Unlike PP.
SE has FAR more magnetizing current than a P-P OT due to the air gap and low permeability. And the notion of "no crossover" is wrong as well. The DC causes a large portion of the "easy" magnetic domains to permanently align in the DC direction. The remaining "harder" (anisotropic magnetic) domains then flip back and forth with the AC component, just like in any P-P OT. (except it is using the "harder" anisotropic magnetic domains to do so, so very reduced permeability. (besides the air gap effect)
The patent fixes this.
Although using a high permeability, high inductance P-P OT, with a DC compensating current source on one side, has been available for a long time. ( or keeping the DC component out of the OT using an isolation cap, but that just transfers the magnetizing current problem to the additional parafeed load inductor )
SE gives you low level euphonic 2nd H dist. from the tube, and 3rd harmonic from the SE OT. The level depending on the output Z of the tube. The 3rd H will disappear with the patent, or the DC compensated P-P OT.
The patent fixes this.
Although using a high permeability, high inductance P-P OT, with a DC compensating current source on one side, has been available for a long time. ( or keeping the DC component out of the OT using an isolation cap, but that just transfers the magnetizing current problem to the additional parafeed load inductor )
SE gives you low level euphonic 2nd H dist. from the tube, and 3rd harmonic from the SE OT. The level depending on the output Z of the tube. The 3rd H will disappear with the patent, or the DC compensated P-P OT.
Last edited:
I have difficulty with forementioned patent since it simply ignores the hysteris and b-h loop. Furthermore the fact remains that SE transformers reproduce more fine nuances than PP. Perhaps it gets time to study vacuum tube technology at varsity rather than read rubbish on the web?
Hysteresis and the B-H loop DO manifest in the magnetizing current. The patent scheme monitors the current and effectively deletes the effect by putting in the additional drive to handle it. And then the OT removes it again, leaving just the original signal for the secondary. Besides removing the OT primary resistance, which is helpful for the damping factor. (almost always an issue with SE)
Varsity level? MIT Magnetics, Chih-Wen Chen Ferromagnetics
https://mitpress.mit.edu/books/magnetic-circuits-and-transformers
https://www.thriftbooks.com/w/magne...7lBoCmBQQAvD_BwE#idiq=5743871&edition=3307553
Varsity level? MIT Magnetics, Chih-Wen Chen Ferromagnetics
https://mitpress.mit.edu/books/magnetic-circuits-and-transformers
https://www.thriftbooks.com/w/magne...7lBoCmBQQAvD_BwE#idiq=5743871&edition=3307553
Last edited:
The distortion of Lundahl transformers is usually specified with such a negative resistance driving circuit - it gives them much better figures than a 0 ohm or positive impedance driving circuit. I never knew it was an old Audio Precision patent.
Hmmm, I would call that sneaky cheating on specs. Probably not noticed by many customers.
I do think Audio Precision knew what they were doing when they came up with this patent. Brilliant.
Last edited by a moderator:
HiHere are their curves for comparison.
Sylvania 6A5G (IDHT)
It would be nice to have -Ug step values?
-Ug starts from 0V or some other value?
Thanks
Cheers 🙂
Hi I checked out:
in Sylvania PDF for 6A5G noted:
"The ratings and characteristics are identical to those of type 6B4G
except for the class A power rating which is 3.75 watts for type 6A5G"
.
I compared measurements of 6A5G to 6B4G (General Electric datasheet)
The Ug step is -10V
The measurements are very well corresponds to 6B4G graph.
.
.
I found other measurements by Thomas Mayer
http://vinylsavor.blogspot.com/2019/12/tube-of-month-6a5.html
Ofcourse without any markings just plain graph... ?
Seems that is the horizontal X voltage axis step is 50V, vertical Y axes current is 20mA, -Ud=10V
That is most probable...
.
in Sylvania PDF for 6A5G noted:
"The ratings and characteristics are identical to those of type 6B4G
except for the class A power rating which is 3.75 watts for type 6A5G"
.
I compared measurements of 6A5G to 6B4G (General Electric datasheet)
The Ug step is -10V
The measurements are very well corresponds to 6B4G graph.
.
.
I found other measurements by Thomas Mayer
http://vinylsavor.blogspot.com/2019/12/tube-of-month-6a5.html
Ofcourse without any markings just plain graph... ?
Seems that is the horizontal X voltage axis step is 50V, vertical Y axes current is 20mA, -Ud=10V
That is most probable...
.
Well this is from 4 years ago. But yes, the curves start at zero and are in 10 volt steps. The point of that post was to compare the linearity of two tubes, one that is directly heated and one that is indirectly heated with otherwise the same internal construction.
Hi
Yes, but when the graphs are marked properly, clear, then we can have more useful informations about the tube?
When the effort was already done, by taking a measurements. (With special equipment not available to wider range of diy members...)
For instace we can make an Spice model...
Thanks 🙂
Code:
**** 6A5G_sylvania ******************************************
* Created on 03/30/2025 16:08 using paint_kit.jar 3.1
* www.dmitrynizh.com/tubeparams_image.htm
* Plate Curves image file:
* Data source link: 6A5G Sylvania measurements DIY Audio
*----------------------------------------------------------------------------------
.SUBCKT 6A5G_sylvania 1 2 3 ; Plate Grid Cathode
+ PARAMS: CCG=7P CGP=16P CCP=5P RGI=2000
+ MU=4.469 KG1=1546 KP=43.29 KVB=600 VCT=0 EX=1.484
* Vp_MAX=600 Ip_MAX=120 Vg_step=10 Vg_start=10 Vg_count=15
* Rp=5000 Vg_ac=50 P_max=15 Vg_qui=-50 Vp_qui=270
* X_MIN=113 Y_MIN=47 X_SIZE=661 Y_SIZE=646 FSZ_X=1578 FSZ_Y=808 XYGrid=true
* showLoadLine=y showIp=y isDHT=n isPP=n isAsymPP=n showDissipLimit=y
* showIg1=n gridLevel2=n isInputSnapped=n
* XYProjections=y harmonicPlot=y dissipPlot=y
*----------------------------------------------------------------------------------
E1 7 0 VALUE={V(1,3)/KP*LOG(1+EXP(KP*(1/MU+(VCT+V(2,3))/SQRT(KVB+V(1,3)*V(1,3)))))}
RE1 7 0 1G ; TO AVOID FLOATING NODES
G1 1 3 VALUE={(PWR(V(7),EX)+PWRS(V(7),EX))/KG1}
RCP 1 3 1G ; TO AVOID FLOATING NODES
C1 2 3 {CCG} ; CATHODE-GRID
C2 2 1 {CGP} ; GRID=PLATE
C3 1 3 {CCP} ; CATHODE-PLATE
D3 5 3 DX ; POSITIVE GRID CURRENT
R1 2 5 {RGI} ; POSITIVE GRID CURRENT
.MODEL DX D(IS=1N RS=1 CJO=10PF TT=1N)
.ENDS 6A5G_sylvania
*$I have properly marked graphs, but making spice models wasn't the point of this thread. Regardless, you can use a 2A3 spice model, they are identical as far as LTSpice simulations are concerned.
This a bit more accurate models based on measurements
for 6A5G Sylvania and Visseaux.
Models are very close to each other only 2 parameters are slight different
KP and VCT.
.
for 6A5G Sylvania and Visseaux.
Models are very close to each other only 2 parameters are slight different
KP and VCT.
.
Code:
***** 6A5G_sylvania ******************************************
* Created on 03/30/2025 20:22 using paint_kit.jar 3.1
* www.dmitrynizh.com/tubeparams_image.htm
* Plate Curves image file: Measurements by L0rdGwyn DIY Audio member
* Data source link: https:
* //www.diyaudio.com/community/threads/why-has-single-ended-output-become-popular.368322/page-18
*----------------------------------------------------------------------------------
.SUBCKT 6A5G_sylvania 1 2 3 ; Plate Grid Cathode
+ PARAMS: CCG=7P CGP=16P CCP=5P RGI=2000
+ MU=4.335 KG1=1546 KP=47.7 KVB=600 VCT=-1.4 EX=1.484
* Vp_MAX=600 Ip_MAX=120 Vg_step=10 Vg_start=10 Vg_count=17
* Rp=5000 Vg_ac=50 P_max=15 Vg_qui=-50 Vp_qui=270
* X_MIN=113 Y_MIN=45 X_SIZE=661 Y_SIZE=648 FSZ_X=1578 FSZ_Y=808 XYGrid=true
* showLoadLine=y showIp=y isDHT=n isPP=n isAsymPP=n showDissipLimit=y
* showIg1=n gridLevel2=n isInputSnapped=n
* XYProjections=y harmonicPlot=y dissipPlot=y
*----------------------------------------------------------------------------------
E1 7 0 VALUE={V(1,3)/KP*LOG(1+EXP(KP*(1/MU+(VCT+V(2,3))/SQRT(KVB+V(1,3)*V(1,3)))))}
RE1 7 0 1G ; TO AVOID FLOATING NODES
G1 1 3 VALUE={(PWR(V(7),EX)+PWRS(V(7),EX))/KG1}
RCP 1 3 1G ; TO AVOID FLOATING NODES
C1 2 3 {CCG} ; CATHODE-GRID
C2 2 1 {CGP} ; GRID=PLATE
C3 1 3 {CCP} ; CATHODE-PLATE
D3 5 3 DX ; POSITIVE GRID CURRENT
R1 2 5 {RGI} ; POSITIVE GRID CURRENT
.MODEL DX D(IS=1N RS=1 CJO=10PF TT=1N)
.ENDS 6A5G_sylvania
*$
*
*
**** 6A5G_visseaux ******************************************
* Created on 03/30/2025 20:19 using paint_kit.jar 3.1
* www.dmitrynizh.com/tubeparams_image.htm
* Plate Curves image file: Measurements by L0rdGwyn DIY Audio member
* Data source link:
* //www.diyaudio.com/community/threads/why-has-single-ended-output-become-popular.368322/page-18
*----------------------------------------------------------------------------------
.SUBCKT 6A5G_visseaux 1 2 3 ; Plate Grid Cathode
+ PARAMS: CCG=7P CGP=16P CCP=5P RGI=2000
+ MU=4.335 KG1=1546 KP=32.9 KVB=600 VCT=-2 EX=1.484
* Vp_MAX=600 Ip_MAX=120 Vg_step=10 Vg_start=10 Vg_count=17
* Rp=5000 Vg_ac=50 P_max=15 Vg_qui=-50 Vp_qui=270
* X_MIN=113 Y_MIN=47 X_SIZE=661 Y_SIZE=646 FSZ_X=1578 FSZ_Y=808 XYGrid=true
* showLoadLine=y showIp=y isDHT=n isPP=n isAsymPP=n showDissipLimit=y
* showIg1=n gridLevel2=n isInputSnapped=n
* XYProjections=y harmonicPlot=y dissipPlot=y
*----------------------------------------------------------------------------------
E1 7 0 VALUE={V(1,3)/KP*LOG(1+EXP(KP*(1/MU+(VCT+V(2,3))/SQRT(KVB+V(1,3)*V(1,3)))))}
RE1 7 0 1G ; TO AVOID FLOATING NODES
G1 1 3 VALUE={(PWR(V(7),EX)+PWRS(V(7),EX))/KG1}
RCP 1 3 1G ; TO AVOID FLOATING NODES
C1 2 3 {CCG} ; CATHODE-GRID
C2 2 1 {CGP} ; GRID=PLATE
C3 1 3 {CCP} ; CATHODE-PLATE
D3 5 3 DX ; POSITIVE GRID CURRENT
R1 2 5 {RGI} ; POSITIVE GRID CURRENT
.MODEL DX D(IS=1N RS=1 CJO=10PF TT=1N)
.ENDS 6A5G_visseaux
*$
*
Last edited:
Measurements of 6A5G by T. Mayer are a bit away from 6B4G. Other two measurement are much closer to target curves...
.
.
.
.
Code:
*** 6A5G_TMayer ******************************************
* Created on 03/30/2025 21:06 using paint_kit.jar 3.1
* www.dmitrynizh.com/tubeparams_image.htm
* Plate Curves image file: Measurements by T. Mayer
* Data source link:
*----------------------------------------------------------------------------------
.SUBCKT 6A5G_TMayer 1 2 3 ; Plate Grid Cathode
+ PARAMS: CCG=7P CGP=16P CCP=5P RGI=2000
+ MU=4.162 KG1=1546 KP=25.28 KVB=600 VCT=-3 EX=1.484
* Vp_MAX=500 Ip_MAX=120 Vg_step=10 Vg_start=10 Vg_count=14
* Rp=5000 Vg_ac=50 P_max=15 Vg_qui=-50 Vp_qui=255
* X_MIN=50 Y_MIN=113 X_SIZE=900 Y_SIZE=723 FSZ_X=1771 FSZ_Y=939 XYGrid=true
* showLoadLine=y showIp=y isDHT=n isPP=n isAsymPP=n showDissipLimit=y
* showIg1=n gridLevel2=n isInputSnapped=n
* XYProjections=y harmonicPlot=y dissipPlot=y
*----------------------------------------------------------------------------------
E1 7 0 VALUE={V(1,3)/KP*LOG(1+EXP(KP*(1/MU+(VCT+V(2,3))/SQRT(KVB+V(1,3)*V(1,3)))))}
RE1 7 0 1G ; TO AVOID FLOATING NODES
G1 1 3 VALUE={(PWR(V(7),EX)+PWRS(V(7),EX))/KG1}
RCP 1 3 1G ; TO AVOID FLOATING NODES
C1 2 3 {CCG} ; CATHODE-GRID
C2 2 1 {CGP} ; GRID=PLATE
C3 1 3 {CCP} ; CATHODE-PLATE
D3 5 3 DX ; POSITIVE GRID CURRENT
R1 2 5 {RGI} ; POSITIVE GRID CURRENT
.MODEL DX D(IS=1N RS=1 CJO=10PF TT=1N)
.ENDS 6A5G_TMayer
*$