The buffer is a source follower,
having the trans-conductance, “gm”
and the source resistor, “Rs”
I understand that the voltage gain at the source
is about Rs/(Rs+1/gm), where “Rs” is a fixed value
but, “gm” is variable depending on the Vds
variation (called modulation?) and the non-linear
trans-conductance curve
In this small signal handling,
I think that the Vds variation is minor concern
But, the non-linear trans-conductance might
be a real thing in connection to the output distortion
For me, these are uneasy monsters to handle
upon theory and measurement
I would try to find the spot using my ears
Drinking a cup of coffee for lunch . . .
having the trans-conductance, “gm”
and the source resistor, “Rs”
I understand that the voltage gain at the source
is about Rs/(Rs+1/gm), where “Rs” is a fixed value
but, “gm” is variable depending on the Vds
variation (called modulation?) and the non-linear
trans-conductance curve
In this small signal handling,
I think that the Vds variation is minor concern
But, the non-linear trans-conductance might
be a real thing in connection to the output distortion
For me, these are uneasy monsters to handle
upon theory and measurement
I would try to find the spot using my ears
Drinking a cup of coffee for lunch . . .
Thanks babowana,
I've got the little beast dialed in and sounding like a dream come true. I'm primarily interested in knowing why it works, and works so well. It's a great feeling to build something that works. But, I feel kind of dumb if I can't figure out why it works.
I think the addition of the buffer was the best trick I've learned this year! All the previously missing detail is now present. Perfect.
John🙂
I've got the little beast dialed in and sounding like a dream come true. I'm primarily interested in knowing why it works, and works so well. It's a great feeling to build something that works. But, I feel kind of dumb if I can't figure out why it works.
I think the addition of the buffer was the best trick I've learned this year! All the previously missing detail is now present. Perfect.
John🙂
carpenter said:But, I feel kind of dumb if I can't figure out why it works.
Me too
I also do
and
But, with my big $1.42 book on electronics, I'm learning so much! I had no idea that AC and DC operated so differently in an amplifier. I now understand how "holes and free electrons" operate. Now, if I could only find free transformers, heatsinks and electronic components... they're not so free and they leave a big hole in my wallet.
All well...
I'm certain that with time and studying, I'll be as smart as you, babowana😀 😀
John🙂
All well...
I'm certain that with time and studying, I'll be as smart as you, babowana😀
😀 John🙂
Basically I would say it works because of all the points N.P. mentioned in the ZV4 article and the ZV7 article😀 After seeing my ZV9-L square wave response and gain numbers. I'm thinking of going that rout with mine (input buffer).
I posted a dual JFET gain stage idea with a buffer, some time ago for someone who was thinking about a 100W ZV9. I'm not going there but I think I might just try a buffered dual JFET ZV9 even if I end up with a resistor load instead of the big L. Need to do a little refinement on the circuit first.
I posted a dual JFET gain stage idea with a buffer, some time ago for someone who was thinking about a 100W ZV9. I'm not going there but I think I might just try a buffered dual JFET ZV9 even if I end up with a resistor load instead of the big L. Need to do a little refinement on the circuit first.
You'll like that buffer, flg.
It just bugs me that I'm so far behind the curve with regards to being able to set up I/V graphs to understand the Q points of these amps. I'm close to understanding, though. I'm not the sharpest chisel in the box, so I have to dig and ponder a great deal to hammer down the details.
I did take apart an IRF transfer characteristics graph and convert it from logarithmic spacing to equal distance spacing per grid. It's amazing how vertically straight that curved line becomes.
What would be really cool, is to read an N.P. article that goes through the numbers and the theory (load lines, transfer graphs, etc.) and then demonstrates how all of this information creates such a superb piece of audio equipment. I'm so visual, that a graph really helps my ability to comprehend the outcome.
The buffer in ZV4 would have been an excellent opportunity to learn how the two stages combine in DC and AC configurations, i.e., how the resistors parallel through the coupling capacitor and so forth.
It probably isn't fair to request this, though. I have to pay my dues and study. Then I'll know firsthand why things work the way they do.
I'm am learning... and boy do I ramble on,
John🙂
It just bugs me that I'm so far behind the curve with regards to being able to set up I/V graphs to understand the Q points of these amps. I'm close to understanding, though. I'm not the sharpest chisel in the box, so I have to dig and ponder a great deal to hammer down the details.
I did take apart an IRF transfer characteristics graph and convert it from logarithmic spacing to equal distance spacing per grid. It's amazing how vertically straight that curved line becomes.
What would be really cool, is to read an N.P. article that goes through the numbers and the theory (load lines, transfer graphs, etc.) and then demonstrates how all of this information creates such a superb piece of audio equipment. I'm so visual, that a graph really helps my ability to comprehend the outcome.
The buffer in ZV4 would have been an excellent opportunity to learn how the two stages combine in DC and AC configurations, i.e., how the resistors parallel through the coupling capacitor and so forth.
It probably isn't fair to request this, though. I have to pay my dues and study. Then I'll know firsthand why things work the way they do.
I'm am learning... and boy do I ramble on,
John🙂
My current project reminds me alot of the ZV4 progression. If you had big enough input and feedback resistors, your high frequency started to loose it. It was a little short on gain too.
So, the buffer provided a low source impeadance and enough drive current to the gain transistor to improve the the high frquencies. It isolated the feeedback resistor so the gain transistor did not have to see the high R value. It has very high input impeadance itself so you could use a big enough input resistor without loading down your preamp to much. You can also insert a volume control in front of it if you needed to. Gain of the ZV4 might have still been a problem for some.
If I go to a dual JFET design , I beleive the buffer will be necessary to drive both and still maintain good high frequency response. I will get more gain, and be able to use at least a 10K input resistor so any preamp or music source with a few volts output can drive the amp.
All of this dosen't solve the low frequency square wave "tilt" and saturatin thing though. That I beleive is the inductors problem. If I had an aleph current source things would improve. A really high power resistor works good too, at least now that it is winter time 😀
So, the buffer provided a low source impeadance and enough drive current to the gain transistor to improve the the high frquencies. It isolated the feeedback resistor so the gain transistor did not have to see the high R value. It has very high input impeadance itself so you could use a big enough input resistor without loading down your preamp to much. You can also insert a volume control in front of it if you needed to. Gain of the ZV4 might have still been a problem for some.
If I go to a dual JFET design , I beleive the buffer will be necessary to drive both and still maintain good high frequency response. I will get more gain, and be able to use at least a 10K input resistor so any preamp or music source with a few volts output can drive the amp.
All of this dosen't solve the low frequency square wave "tilt" and saturatin thing though. That I beleive is the inductors problem. If I had an aleph current source things would improve. A really high power resistor works good too, at least now that it is winter time 😀
By "big enough", do you mean greater resistance? This is an important point, and I want to learn. Please elaborate.
John🙂
btw, the high frequency response sounds as if it's doubled with the buffer stage.
Yes, the capacitance seen at the gate of the gain FET reacts with the high value resistors to roll off high frequencies. It is also a variable capacitance in the ZV4 type circuit (and yours too) which causes distortion. If you lower the feedback resistor value it helps some, but that makes you lower the input resistor to maintain the same gain. As I did in my ZV9-L a few posts ago. With input resistors of lower values(less than 10K), your preamp starts to have difficulty driving it, espeacially the tube guys. A preamp with good output drive capability, or lower output impeadance, will do much better.
However, with the buffer driving the gain FET, it only sees a very low impeadance (source follower buffer output impeadance) and the high frequency performance is better. The feedback and input resistors have no trouble driving the buffer transistor with it's much lower capacitance (Very high input impeadance).
Also, in the ZV9, the cascoding of the JFET should help the capacitance problem. The varying Drain voltage of the ZEN causes a wild swing of Capacitance which the cascode of ZV9 tries to maintain at a constant. It does not seem to help enough in my circuit though. It might be that I am degrading the cascode benifits with the .2 ohm source Rs of the cascode transistors??? It reminds me of the Zen before the buffer was added. Great improvement over the direct input Zen so I'm thinking of going that route.
However, with the buffer driving the gain FET, it only sees a very low impeadance (source follower buffer output impeadance) and the high frequency performance is better. The feedback and input resistors have no trouble driving the buffer transistor with it's much lower capacitance (Very high input impeadance).
Also, in the ZV9, the cascoding of the JFET should help the capacitance problem. The varying Drain voltage of the ZEN causes a wild swing of Capacitance which the cascode of ZV9 tries to maintain at a constant. It does not seem to help enough in my circuit though. It might be that I am degrading the cascode benifits with the .2 ohm source Rs of the cascode transistors??? It reminds me of the Zen before the buffer was added. Great improvement over the direct input Zen so I'm thinking of going that route.
John, Take a look at this...
http://www.diyaudio.com/forums/showthread.php?s=&postid=1086535#post1086535
😀 😀 😀
http://www.diyaudio.com/forums/showthread.php?s=&postid=1086535#post1086535
😀 😀 😀
O.K. John, Here's my take(refrenced to post # 227...)
I could be wrong of coarse, I don't have a Physics degree or work at a well renowned amp comapany(N.P. Thought I was overqualified ) ... But...😀
First, your gain is, I think, 442kll442k(in paralell with)/221. Forget it😀 Thats 1000. Your actually running open loop. That's not so bad. Some of Nelson's amps do that just fine. No feedback, is what that means. I might take an off the wall guess and say you probably have a gain of 20 or so, due only to the gm of the gain transistors. In other words, you put in about 1 Volt Signal for full output???... But, like I said, let's take a look at the bias point of the follower and see what might be an option with your current topology...
If you have 23 Volts at your output(speaker,not V+,???). Then, 23V/2 (442ll442) = 11.5V at the buffer gate. Is that right? Maybe 15V (11.5+3.5 (Vgs)) at the buffer source??? Is that right??? So (V+)-15/(R9+R13) = Iq, of the buffer stage... 10mA -30mA would be good propbaly😀 Time for a Sailor Jerry's and Egg-Nog, just a minute... Are you calculating the Iq yet
O.K. Hey😕 What is on the source of your Q3??? looks like 11Vs across that guy??? And several amps??? I'l bet that puppy is a little hot??? Maybe we need to lower that bias point and cool him down, and, at the same time give the amp a little more voltage swing??? Are you with me??? You should probably only have 4-6Volts on that guy.... That means, you can't have the source of the follower at 15V - the Vgs of the gain transistor(4V or whatever) = 11V on the drian of Q3. Right...
O.K. Any questions
I could be wrong of coarse, I don't have a Physics degree or work at a well renowned amp comapany(N.P. Thought I was overqualified
) ... But...😀 First, your gain is, I think, 442kll442k(in paralell with)/221. Forget it😀 Thats 1000. Your actually running open loop. That's not so bad. Some of Nelson's amps do that just fine. No feedback, is what that means. I might take an off the wall guess and say you probably have a gain of 20 or so, due only to the gm of the gain transistors. In other words, you put in about 1 Volt Signal for full output???... But, like I said, let's take a look at the bias point of the follower and see what might be an option with your current topology...
If you have 23 Volts at your output(speaker,not V+,???). Then, 23V/2 (442ll442) = 11.5V at the buffer gate. Is that right? Maybe 15V (11.5+3.5 (Vgs)) at the buffer source??? Is that right??? So (V+)-15/(R9+R13) = Iq, of the buffer stage... 10mA -30mA would be good propbaly😀 Time for a Sailor Jerry's and Egg-Nog, just a minute... Are you calculating the Iq yet
O.K. Hey😕 What is on the source of your Q3??? looks like 11Vs across that guy??? And several amps??? I'l bet that puppy is a little hot??? Maybe we need to lower that bias point and cool him down, and, at the same time give the amp a little more voltage swing??? Are you with me??? You should probably only have 4-6Volts on that guy.... That means, you can't have the source of the follower at 15V - the Vgs of the gain transistor(4V or whatever) = 11V on the drian of Q3. Right...
O.K. Any questions
Boy
You're still sleeping
Wake up!
DC divider is known as 147K//442K (or very similar)
DC at the gate is about 6V
DC at the source is about 6+3.5
DC across R13 is about 28-9=18V
DC bias for the buffer is about 18/571x1000=30mA
Maybe somewhat lesser
Carpenter must have a correct answer
I think you need the fifth edition 😀
You're still sleeping
Wake up!
DC divider is known as 147K//442K (or very similar)
DC at the gate is about 6V
DC at the source is about 6+3.5
DC across R13 is about 28-9=18V
DC bias for the buffer is about 18/571x1000=30mA
Maybe somewhat lesser
Carpenter must have a correct answer
I think you need the fifth edition 😀
No, Not sleeping 😉 must be my medication 
No, just call me Grey...(hi Grey 
) I gotta go find a 147K 
How about my thinking??? If I had the right values was I helping😕 😕 😕
Ho Ho Ho
No, just call me Grey...(hi Grey ) I gotta go find a 147K How about my thinking??? If I had the right values was I helping😕 😕 😕
Ho Ho Ho
flg said:
One night
I was at a crowded drinking bar in Oslo
One Swedish guy was sitting by me
He offered a beer, smiling,
"ho ho ho . . ."
Trying to hold my hand
"ho ho ho . . ."
I was scared
It was a scary night, ho ho ho . . .
Babowana, sounds like he wanted to make you his little ho...
flg, the DC divider is 147ll142. R13 is 571 ohm.
After I wake up a bit, I'll try to digest what you've written.
You guys are great. Just ask that Swedish guy in Oslo.😀 😉 😀
Merry Holidays.
John🙂
flg, the DC divider is 147ll142. R13 is 571 ohm.
After I wake up a bit, I'll try to digest what you've written.
You guys are great. Just ask that Swedish guy in Oslo.😀 😉 😀
Merry Holidays.
John🙂
O.K. it's late for me too. But I'l try... R13 😕 147K O.K. Just a minute... Is 23 Volts right??? (442ll147)(just call me Grey, Hi Grey ) / R2, is AC gain. It's late(sorry Grey, jump in dude!) it's still well beyond OL Gain so I'll call it 20! Am I wrong??? Babowana, 30mA great bud thanks...
The other issue is what the follower output voltage does to the current source and the available voltage swing, there is a cap there so I'm sleeping... Babowana, Thanks Bud 😀
The gain transisitor gate bias is 23 Volts(I Think)/(R3+RP1)*RP1. But What is Bias R out???
I wont post for a while
O.K. Just a minute... Is 23 Volts right??? (442ll147)(just call me Grey, Hi Grey ) / R2, is AC gain. It's late(sorry Grey, jump in dude!) it's still well beyond OL Gain so I'll call it 20! Am I wrong??? Babowana, 30mA great bud thanks...The other issue is what the follower output voltage does to the current source and the available voltage swing, there is a cap there so I'm sleeping... Babowana, Thanks Bud 😀
The gain transisitor gate bias is 23 Volts(I Think)/(R3+RP1)*RP1. But What is Bias R out???
I wont post for a while
No you weren't. Oh boy... sorry guys.
I screwed up... must have been the nap.
DC divider is 147k ll 442K.
R13 is 571 ohm; at least I got that right.
John
I replaced the air-core inductor tonight; I used Surplus Sales of Nebraska's big I/E power transformer. Long story short, it works great, sounds wonderful. The piano sounds smooth in the upper register, as does the flute. Before I added the buffer stage, the highs were a bit grainy.
I remember reading Nelson's comment about symmetrical feedback having greater effect when there's more than one stage. He wasn't b.s.ing.
So, it could be the buffer stage offering additional feedback, or it could be that the tranny is a bit loose in the winding/coupling dept. Maybe all the above. Nelson also mentioned that I/E trannies don't couple as efficiently; and I presumed that this defect might be useful. Who knows? All I know is that it does the trick.
My air-core choke was beginning to melt it plastic bobbin--cheap plastic!😀 Couldn't be the heat?!?
Merry X-mas, Happy Hanukkah, bring on the reindeer and the Hanukia. Above all, peace on Earth. God, I wish...
John 🙂
I remember reading Nelson's comment about symmetrical feedback having greater effect when there's more than one stage. He wasn't b.s.ing.
So, it could be the buffer stage offering additional feedback, or it could be that the tranny is a bit loose in the winding/coupling dept. Maybe all the above. Nelson also mentioned that I/E trannies don't couple as efficiently; and I presumed that this defect might be useful. Who knows? All I know is that it does the trick.
My air-core choke was beginning to melt it plastic bobbin--cheap plastic!😀 Couldn't be the heat?!?
Merry X-mas, Happy Hanukkah, bring on the reindeer and the Hanukia. Above all, peace on Earth. God, I wish...
John 🙂
Attachments
flg--several post back, you mentioned that 147K ll 442K / R2 gives the gain. If this is so, then I should replace R2 (the 221 ohm resister) with something on the order of 10K? That would give a gain of 11; assuming I understood your formula correctly.
John🙂
John🙂
- Status
- Not open for further replies.