i want to make zv9 for more power,
i know i must increase rail to rail voltage of zv9, and parallelize power jfet lu1014d , but because zv9 use cascode circuit ,
i must parallelize mosfets and jfets both . so how to do it ?
and after parallel , how can i use feedback , because parallelized jfets have serial resistor to attach G pin of jfet.
so anyone draw a schematic to me ,thanks
i know i must increase rail to rail voltage of zv9, and parallelize power jfet lu1014d , but because zv9 use cascode circuit ,
i must parallelize mosfets and jfets both . so how to do it ?
and after parallel , how can i use feedback , because parallelized jfets have serial resistor to attach G pin of jfet.
so anyone draw a schematic to me ,thanks
john-china, O.K. I did a little more work This should be a good topic for the thread This should get you close to 50W I believe O.K. it's not 100+ Watts I could not imagine needing more than this anyway
more power than this and we would need to call it Monster Zen Version 9
Any comments?
more power than this and we would need to call it Monster Zen Version 9
Any comments?
Attachments
flg said:........
Any comments?
your circuit for cascode is as what i thought. i think whole circuit is like zv4 + zv9 because you use zvp3310 as input buffer. i dont like input buffer because i dont care high input capacitance. i can use diy pre amp to resolve this problem.
i saw you did not serialize any resistors to attach gate of jfet. you just attach gate of jfet togather straightly, is it feasible ?
your 50w circuit gave me much inspiration, thanks.
Zen Mod said:from top of my head-symetry or even susy is better way than paralleling in this case
at first do paralleling jfet, secondly do alephx+zv9(susy)
step by step . i can run whithout falling,
I have done some thinking with JFETs, SUSY and coupled inductor loading like ZV7-T http://www.diyaudio.com/forums/showthread.php?threadid=85246
You can change the load to the aleph CCS
Here is my latest schez
Only maybe 25Watts out
You can change the load to the aleph CCS
Here is my latest schez
Only maybe 25Watts out
Attachments
flg said:john-china, O.K. I did a little more work This should be a good topic for the thread This should get you close to 50W I believe O.K. it's not 100+ Watts I could not imagine needing more than this anyway
more power than this and we would need to call it Monster Zen Version 9
Any comments?
Since JFETs are high current devices, you can get away with just one while paralleling MOSFETs for CCS and Cascode circuits.
This was also mentioned by NP in another thread.
Allan
Well, regarding the Super ZV9 Not that I know better or anything, but let me explain my reasoning.
The JFETs, or one JFET as I think you are suggesting, carrying 4A Idle current, would mean a Vds of less than 2V in order to maintain a reasonable Pd and reliability. Also, I would be afraid that operating at such low voltages and high currents, you are likely to encounter voltage drops in wires, traces and connections that would become a substaintial percentage of the operating parameters. Basically, screwing things up, unless they are accounted for in the design, which is difficult Also, with a desire for a gain of about 20, this is more easily accomplished with 2 devices after the degenerating/biasing Source R's.
The addition of the ZV4 buffer is also possibly a compromising addition However, opperating these JFETs at very low voltages you are in the range of higher capacitances and slightly more capacitance variability with "modulated" Vds As with ZV4 I would expect improved performance over driving 2 of them dirrectly
I could be thinking a little over board here But, that is why I came up with this configuration
The JFETs, or one JFET as I think you are suggesting, carrying 4A Idle current, would mean a Vds of less than 2V in order to maintain a reasonable Pd and reliability. Also, I would be afraid that operating at such low voltages and high currents, you are likely to encounter voltage drops in wires, traces and connections that would become a substaintial percentage of the operating parameters. Basically, screwing things up, unless they are accounted for in the design, which is difficult Also, with a desire for a gain of about 20, this is more easily accomplished with 2 devices after the degenerating/biasing Source R's.
The addition of the ZV4 buffer is also possibly a compromising addition However, opperating these JFETs at very low voltages you are in the range of higher capacitances and slightly more capacitance variability with "modulated" Vds As with ZV4 I would expect improved performance over driving 2 of them dirrectly
I could be thinking a little over board here But, that is why I came up with this configuration
Nelson Pass said:a 100W ZV9 would be a simple matter of scaling the existing
design with parallel Mosfets. It is probably not necessary to
parallel JFETs, as the chip is capable of operating at the
necessary 3A bias with as much as 2.5 V across it (7.5W)
without undue stress.
risking to repeat my self ,repeating Papa,maybe is time for pat. No.5343166 ?
Well, I guess I should have done a little more thinking. Nelson is more correct than I was. I just replaced the light bulb load with the Aleph CS without any other specific modicifications, or thinking. As posted I guess my Supper ZV9 version might get close to 80 Watts into 4.5 ohms or so.
If the circuit had a little more voltage howerver, I could be taking more advantage of the Aleph circuit. It would take about 16 more volts and the required parralell MOSFETs but, something like 100 Watts into 6 ohms is possible or about 40 more volts for something like 140 Watts into 8
As Nelson said, maybe 3.35A Iq and the Aleph load at 50% I gain, with enough voltage of coarse, would net 100 Watts into 8 ohms. That is if my thinking is better today than over the weekend
I suppose you would still run into the gain issue or need plenty of drive voltage
I wonder though, could we still be in that sweet spot for ideal cascode modulation And, does the D-S JFET modulation hold true using some cascode source resistors in addition to the JFET source resistors
If the circuit had a little more voltage howerver, I could be taking more advantage of the Aleph circuit. It would take about 16 more volts and the required parralell MOSFETs but, something like 100 Watts into 6 ohms is possible or about 40 more volts for something like 140 Watts into 8
As Nelson said, maybe 3.35A Iq and the Aleph load at 50% I gain, with enough voltage of coarse, would net 100 Watts into 8 ohms. That is if my thinking is better today than over the weekend
I suppose you would still run into the gain issue or need plenty of drive voltage
I wonder though, could we still be in that sweet spot for ideal cascode modulation And, does the D-S JFET modulation hold true using some cascode source resistors in addition to the JFET source resistors
flg said:I wonder though, could we still be in that sweet spot for ideal cascode modulation
The Jfet sweet spot will still be at 2.5V across its cascoded D-S. At 4A bias (8% of the max 50A) that's about 12W (17% of the max dissipation) that needs (more) heatsinking.
Blues said:
...that's about 12W (17% of the max dissipation) that needs (more) heatsinking.
Sorry, that's 10W and 14.5%.
O.K. Lets say for the sake of understanding we can burn 8.25 Watts in 1 device at idle (2.5V x 3.3A) this would require a lower G-S voltage. In the ZV9 circuit this would mean a lower value Rsource. For the sake of understanding we'll say about .9V G-S gives us 3.3A Drain current. At .9V/3.3A that would be a .27 ohm source resistor
Now, an arbitrary signal is applied to the input causeing 1A higher drain current. This would cause a .27V increase in the voltage on the source resistor. Since the cascode FET is holding the Drain at a stable voltage, the .27V across the source resistor causes the D-S voltage to drop .27 volts. The trouble is, Nelson said the magic number for optimal modulation is about .6V per amp Hmmm In ZV9 we have about .6 ohms in the Source where a 1A change would cause a .6V D-S shift
I'm guessing this number changes for a different Drain current/Gate voltage And probably gets smaller as that "Triode" curve gets more strait (lower Gate voltage). I havent figured out these details exactly
But, there is another problem When paralleling all those Cascode devices. I believe there is a need to help them stay balanced or equal. Typically we use a source resistor to do this. The source resistor will degrade the costant cascode voltage on the Drain of the JFET just like the JFETs source resistor does. Will it not???
If we use a resistor(s), in the cascode transistor(s) source that makes up the other .33V over that 1 A change, we now have the magic .6V/A modulation
But, is that what we want
Or does the necessary modulation need to be lower with higher current
Now, an arbitrary signal is applied to the input causeing 1A higher drain current. This would cause a .27V increase in the voltage on the source resistor. Since the cascode FET is holding the Drain at a stable voltage, the .27V across the source resistor causes the D-S voltage to drop .27 volts. The trouble is, Nelson said the magic number for optimal modulation is about .6V per amp Hmmm In ZV9 we have about .6 ohms in the Source where a 1A change would cause a .6V D-S shift
I'm guessing this number changes for a different Drain current/Gate voltage And probably gets smaller as that "Triode" curve gets more strait (lower Gate voltage). I havent figured out these details exactly
But, there is another problem When paralleling all those Cascode devices. I believe there is a need to help them stay balanced or equal. Typically we use a source resistor to do this. The source resistor will degrade the costant cascode voltage on the Drain of the JFET just like the JFETs source resistor does. Will it not???
If we use a resistor(s), in the cascode transistor(s) source that makes up the other .33V over that 1 A change, we now have the magic .6V/A modulation
But, is that what we want
Or does the necessary modulation need to be lower with higher current
Zen Mod said:Papa,maybe is time for pat. No.5343166?
You certainly can apply that technique if you like, but the
biggest issue here is probably the dissipation of the JFET,
where that circuit won't help us.
flg said:maybe 3.35A Iq and the Aleph load at 50% I gain, with enough voltage of coarse, would net 100 Watts into 8 ohms.
I wonder though, could we still be in that sweet spot for ideal cascode modulation And, does the D-S JFET modulation hold true using some cascode source resistors in addition to the JFET source resistors
For beginners: An Aleph current source at 50% gain means that
for a given bias current, the peak output is twice the bias
current. This gives the peak output wattage as 4 times the
bias squared times the impedance. For 2 amp bias and an
8 ohm load this would be 4 * 2 * 2 * 8 = 128 watts peak and
64 watts rms. Of course you always have to bias a little higher
than this figure in real life due to the imperfections of the gain
devices.
There is a sweet spot for Source resistance and cascode
voltage/modulation, but you have to go looking for it. A
distortion analyzer is a big help.
Originally posted by flg When paralleling all those Cascode devices. I believe there is a need to help them stay balanced or equal. Typically we use a source resistor to do this.
Or does the necessary modulation need to be lower with higher current
Yes, you will want to have some Source resistance in addition
to some matching for Vgs. The sweet spot for modulation is
individual with a device and will vary with the DC current and
DC voltage.
Originally posted by flg "a 100W ZV9 would be a simple matter of scaling the existing design with parallel Mosfets." Is it actually this simple
More or less. You might have to adjust the values a bit.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
- Home
- Amplifiers
- Pass Labs
- how to make above 100w zv9