Unfortunately over time I have become visually impaired due to glaucoma and I have only a 91 oscilloscope and a homemade signal generator left... I can't do any more tests, sorry.
I can only say that that simple adapted connection between the driver tube and the power amp performs a simple feedback that linearizes everything and even a less than perfect output transformer benefits from it. The sound becomes natural and engaging... I like it
I can only say that that simple adapted connection between the driver tube and the power amp performs a simple feedback that linearizes everything and even a less than perfect output transformer benefits from it. The sound becomes natural and engaging... I like it
This is the last version of circuit
I add a pot to trim the value of FB
I am working with Hammond mainly and after some tests, with the others
There is also a little changing on Vg2, just to find a better THD, 20 -30 Vdc less.
The 10% of FB declared by Schade I think is found varing the pot of FB ( in my case 220K + 100k, in case) from the value read in pentode mode measured at load.
This make a sense for me, there is a little loss for due the magnetic copupling and Rdc of secondary but it is negligible.
In my case I set as 0 dB ( reference) 3 volt on 8 ohm, it is a start point.
Another consideration is about the Zs on generator that is involved on circuit; in my case with Ecc88 the Zout is around 600 ohm, but the impact is very low; with AP from 20 ohm to 600 hom some little changes are visible.
I will check with the use of 6H30 and, last step, with a common cathode stage with higher Zout.
Walter
I add a pot to trim the value of FB
I am working with Hammond mainly and after some tests, with the others
There is also a little changing on Vg2, just to find a better THD, 20 -30 Vdc less.
Hi am playing with this; yesterday I have done dozen of test with different setting.
The 10% of FB declared by Schade I think is found varing the pot of FB ( in my case 220K + 100k, in case) from the value read in pentode mode measured at load.
This make a sense for me, there is a little loss for due the magnetic copupling and Rdc of secondary but it is negligible.
In my case I set as 0 dB ( reference) 3 volt on 8 ohm, it is a start point.
Another consideration is about the Zs on generator that is involved on circuit; in my case with Ecc88 the Zout is around 600 ohm, but the impact is very low; with AP from 20 ohm to 600 hom some little changes are visible.
I will check with the use of 6H30 and, last step, with a common cathode stage with higher Zout.
Walter
Some other test
Another circuit ( the previous one is in stand by), Hammond 1627SE used
This one uses a ECC83 to drive a KT88 with the use of Schade and UL
The circuit
The start point is to set the Vout at 1 watt on 8 ohm as 0 dB.
This was done in pentode mode
Vdc = 335 v
Ia = 80 mA
Rk = 390 ohm
Rg = 270 kohm
Ecc83
Vdc = 330 V
Ra = 270 kohm
Rk = 2k2
Ia = 0,65 mA
Vanode = 165 v
Pout 1 w
con 37 mVin
THD = 1,6 %
The FFT
The response ( poor)
The two resistor of FB in the test was changed for test
Schade with 6 dB of FB ( from 0 dB in pentode)
Vin = 85 mV for 1 w
THD = 1,25 %
Rfb = 1 Mohm (feedback)
The FFT
The response
12 dB of FB
Vin = 170 mV for 1 w
THD = 4,5%
Rfb = 348 kohm
The FFT
Last
Same circuit taking the Va for 83 from anode of 88
FB =10 dB
Vin = 105 mV for 1 w
Thd = 5%
Rfb = 800 kohm
The FFT
The response
-----------------------------------------------------------------------------------------------------------------
UL config
40%
no additional FB
The FFT
The response
The consideration will be done at the end, when other test will be done
I got some first impression, of course
Walter
Another circuit ( the previous one is in stand by), Hammond 1627SE used
This one uses a ECC83 to drive a KT88 with the use of Schade and UL
The circuit
The start point is to set the Vout at 1 watt on 8 ohm as 0 dB.
This was done in pentode mode
Vdc = 335 v
Ia = 80 mA
Rk = 390 ohm
Rg = 270 kohm
Ecc83
Vdc = 330 V
Ra = 270 kohm
Rk = 2k2
Ia = 0,65 mA
Vanode = 165 v
Pout 1 w
con 37 mVin
THD = 1,6 %
The FFT
The response ( poor)
The two resistor of FB in the test was changed for test
Schade with 6 dB of FB ( from 0 dB in pentode)
Vin = 85 mV for 1 w
THD = 1,25 %
Rfb = 1 Mohm (feedback)
The FFT
The response
12 dB of FB
Vin = 170 mV for 1 w
THD = 4,5%
Rfb = 348 kohm
The FFT
Last
Same circuit taking the Va for 83 from anode of 88
FB =10 dB
Vin = 105 mV for 1 w
Thd = 5%
Rfb = 800 kohm
The FFT
The response
-----------------------------------------------------------------------------------------------------------------
UL config
40%
no additional FB
The FFT
The response
The consideration will be done at the end, when other test will be done
I got some first impression, of course
Walter
Two test of the input stage with ECC83
Signal taken on grid of 88 ( not inserted) with a differential probe having a great Zin
Gain = 63
100 mVin = 6,3 Vout with 0,76%
The FFT
The response
There are two lines because I used two different signal cables with different value of capacitance
Walter
Signal taken on grid of 88 ( not inserted) with a differential probe having a great Zin
Gain = 63
100 mVin = 6,3 Vout with 0,76%
The FFT
The response
There are two lines because I used two different signal cables with different value of capacitance
Walter
As Robert Gribnau wrote in another thread:
See for an explanation of this effect pages 396 to 399 of this book:
Fundamentals of Radio Valve-Technique, Deketh, Philips, 1949
The test confirm as defined in the book.
Other test will come
Walter
See for an explanation of this effect pages 396 to 399 of this book:
Fundamentals of Radio Valve-Technique, Deketh, Philips, 1949
The test confirm as defined in the book.
Other test will come
Walter
Ciao Walter,
please do not be upset, but with that circuit you will never get good results.
It is a waste of time to perform measurements without knowing what works best in a specific scenario.
As I already wrote you, to get best performance you need a driver able to easily drive low loads.
A 12AX7 with 270k as plate load will collapse.
I got good results (no measurements, just set the amp to hear some music) with:
B+ around 400V, g2 at 200V, Ra 3k, the KT88 fixed biased at 35W to keep 220k from g1 to bias.
As a driver a 12AT7 with 33k as plate load and 150 Ohm as unbypassed cathode.
100k from kt88 plate to 12at7 plate.
I suggest you to try this configuration and play around it.
It will help you make reasonable measurements and get a better idea of how Schade limits can be reached.
please do not be upset, but with that circuit you will never get good results.
It is a waste of time to perform measurements without knowing what works best in a specific scenario.
As I already wrote you, to get best performance you need a driver able to easily drive low loads.
A 12AX7 with 270k as plate load will collapse.
I got good results (no measurements, just set the amp to hear some music) with:
B+ around 400V, g2 at 200V, Ra 3k, the KT88 fixed biased at 35W to keep 220k from g1 to bias.
As a driver a 12AT7 with 33k as plate load and 150 Ohm as unbypassed cathode.
100k from kt88 plate to 12at7 plate.
I suggest you to try this configuration and play around it.
It will help you make reasonable measurements and get a better idea of how Schade limits can be reached.
The test confirm the limit of this configuration
Also with Srpp the results are not very good
And why was never used by great brand
But I will follow with other check
I can’t accept technical consideration without a test lab
Walter
I understand that you put measures above everything, and I really appreciate that you started doing measures about this topic after I talked about it on the italian forum.
I’m sure that some more tests will be much appreciated by everyone.
But to have useful data, first of all it is needed to master the theory, otherwise you will only prove that doing things in a wrong way gives no results.
I started my journey in Hi-Fi with this kind of feedback, understanding its limits in PP and SE.
Then I switched from parallel to series applied voltage feedback, again understanding its limits and strenghts.
This lead me to patent an improved system with the purpose of removing drawbacks of both kind of feedbacks.
If you want to perform some reasonable tests, you must chose a low rp driver and an high gm output tube with limited plate swing.
12AT7 and EL84 in pentode are the combo I suggest you to start with using logic based on theory.
Parallel applied voltage feedback trades gm for rp on the output tube, at the expense of its input impedance.
Using a weak 12AX7 to drive a KT88 in SE you are doing exactly the opposite of what theory suggests to be the right way to make it work properly.
I’m sure that some more tests will be much appreciated by everyone.
But to have useful data, first of all it is needed to master the theory, otherwise you will only prove that doing things in a wrong way gives no results.
I started my journey in Hi-Fi with this kind of feedback, understanding its limits in PP and SE.
Then I switched from parallel to series applied voltage feedback, again understanding its limits and strenghts.
This lead me to patent an improved system with the purpose of removing drawbacks of both kind of feedbacks.
If you want to perform some reasonable tests, you must chose a low rp driver and an high gm output tube with limited plate swing.
12AT7 and EL84 in pentode are the combo I suggest you to start with using logic based on theory.
Parallel applied voltage feedback trades gm for rp on the output tube, at the expense of its input impedance.
Using a weak 12AX7 to drive a KT88 in SE you are doing exactly the opposite of what theory suggests to be the right way to make it work properly.
Last edited:
In your other thread you wrote about ax7 after the post with the link on Philips
I need to send you a copy?
In every case we are speaking of 2,8 vrns / 1 watt on 8 ohm not the max power so for the configuration I used it is safe
The main concept is : are you able to post the test results ( with same form of graphics) with same range of config ?
Just to confirm your point if you that is reasonable for me
If yes we continue to discuss
If not we are loosing time
This is not ‘impression’ that are personal and I can understand
I have seen that you copy and paste of different link on books, this tells something to me but it is limited until there aren’t real test
Last, I have seen lot of amo where the mix of tube and sand is terribile . With lot of FB
They can’t call tube amp, for me
But hibrid is much better and aren’t in my interest
The Baby Huey is one horror for the complexity of mixture with ss an FB everywhere
But it is only my opinion
Walter
Hi
The test are done on air, there is a photo.
But I use a beautiful KIKUSUI PAB 350, power regulated from 0 to 350 Vdc - 100mA max and the results are due the quality of this stuff
I have had a previous model that now is dead , thi one is te last level.
Walter
Walter, input impedance is independent from the power you are working at and I already suggested you an all tube amp to test.
I agree that BH is overcomplicated and I wrote it clearly in the other theead.
I also explained why in that specific case 12AX7 is fine.
I you want to go on measuring a circuit that cannot work without catching suggestions, you will only prove that you don’t understand how this kind of feedback work.
Sad to see your otherwise useful experience wasted by childish ego.
I agree that BH is overcomplicated and I wrote it clearly in the other theead.
I also explained why in that specific case 12AX7 is fine.
I you want to go on measuring a circuit that cannot work without catching suggestions, you will only prove that you don’t understand how this kind of feedback work.
Sad to see your otherwise useful experience wasted by childish ego.
Thank you! Interesting, that was going to be my next step, go to the garden table and try there with the laptop,
That's a bit harsh. Looking at the big picture, most of what we do here is, in the opinion of the 99% of the rest of the world, a waste of time, in the era of hand-held supercomputers. But it is fun for us, and sometimes the journey is as important as the result. As Soichiro Honda said once, "success is 99% failure" 🙂
Again, have you the test lab?
Otherwise are words.
And we italian are master!!!!! 🙂
Walter
Other test with ECC88 as input
Va = 300 Vdc
Ra = 19 kohm
Rk = 390 ohm
Ia = 7,7 mA
G = 21
Pentode config
Vin = 80 mV
THD = 1,7% 1 watt
Freq Resp
FFT
Schade 6db FB from Pentode
Vin =170 mV
THD = 0,64 % 1 w
Rfb = 111 kohm
Freq. Response
FFT
Schade 12dB FB
Vin = 320 mV
THD = 3,55 % 1 w
Rfb = 40 kohm
Freq. response
12dB = violet curve
6 dB = red curve
Other tests will come
Better info for me
Walter
Va = 300 Vdc
Ra = 19 kohm
Rk = 390 ohm
Ia = 7,7 mA
G = 21
Pentode config
Vin = 80 mV
THD = 1,7% 1 watt
Freq Resp
FFT
Schade 6db FB from Pentode
Vin =170 mV
THD = 0,64 % 1 w
Rfb = 111 kohm
Freq. Response
FFT
Schade 12dB FB
Vin = 320 mV
THD = 3,55 % 1 w
Rfb = 40 kohm
Freq. response
12dB = violet curve
6 dB = red curve
Other tests will come
Better info for me
Walter
I give up Walter.
As always, I’m trying to help you to the good path and you just attack me and hide behind measurements without understanding the theory behind.
I hope sooner or later you will follow my suggestion, at least to prove with measurements that I’m wrong. 😃
As always, I’m trying to help you to the good path and you just attack me and hide behind measurements without understanding the theory behind.
I hope sooner or later you will follow my suggestion, at least to prove with measurements that I’m wrong. 😃
We are loosing the manuality to build and test something.
The simualtion often doesn't help in this field, in my opinion
The capacity to discriminate the issues on a circuit becaming hard
Walter