If you want to use N-JFETs in the input, you'll have to redo the output to use P-ch MOSFETs. Not really a problem. All you have to do is think upside-down.
The 2SJ109, being P, is the one I was trying to think of above. It should work well.
Putting JFETs in is no problem at all. All you've got to do is reduce the current source from 20 mA or so (where it is now) to your target current for your new differential.
The next step is to adjust the value of the load resistors in the differential. For the moment, assume that you want to keep the DC offset across those resistors in the 4V range (about like it is now). Take your current through your differential--let's say 8 mA total, and divide it in half, since half the current will be going up each side. That leaves us with 4 mA, seeking a 4V DC offset. From there it's a simple Ohm's Law question:
4V/.004A= 1000 ohms
<i>Ta da!</i>
Your new load resistor value will be 1k. Nothing to it.
If you use a JFET front end, bear in mind that the amp's gain will probably drop unless you start heading into active loads and such, at which point you'll have to reconsider the DC offset question.
Watch the pinouts, as there's no guarantee that a JEFT will have the right pins in the right places.
As a pass device for a current source, you might look into the 2N5462. Some mud-for-brains is doing a denial of service on us here at work and I'm having trouble getting to Motorola to get more specifics on the part; my connection is spotty at best.
Grey
The 2SJ109, being P, is the one I was trying to think of above. It should work well.
Putting JFETs in is no problem at all. All you've got to do is reduce the current source from 20 mA or so (where it is now) to your target current for your new differential.
The next step is to adjust the value of the load resistors in the differential. For the moment, assume that you want to keep the DC offset across those resistors in the 4V range (about like it is now). Take your current through your differential--let's say 8 mA total, and divide it in half, since half the current will be going up each side. That leaves us with 4 mA, seeking a 4V DC offset. From there it's a simple Ohm's Law question:
4V/.004A= 1000 ohms
<i>Ta da!</i>
Your new load resistor value will be 1k. Nothing to it.
If you use a JFET front end, bear in mind that the amp's gain will probably drop unless you start heading into active loads and such, at which point you'll have to reconsider the DC offset question.
Watch the pinouts, as there's no guarantee that a JEFT will have the right pins in the right places.
As a pass device for a current source, you might look into the 2N5462. Some mud-for-brains is doing a denial of service on us here at work and I'm having trouble getting to Motorola to get more specifics on the part; my connection is spotty at best.
Grey
Nice job, Grey
It helps your average carpenter a great deal when you explain your work with a mathematical example. Using your example, I can plug and play with different components and ponder ways to use this formula in related situations. Plus, I can see first hand the practicalities of using Ohm's law.
Thanks........
It helps your average carpenter a great deal when you explain your work with a mathematical example. Using your example, I can plug and play with different components and ponder ways to use this formula in related situations. Plus, I can see first hand the practicalities of using Ohm's law.
Thanks........
Well, first you take a 2x4" and cut off a piece 4 and 1/8" inches long...
Don't get me started on carpentry. I just got done installing an access hatch into a wall so as to be able to get to the backside of some plumbing. Should have been a trivial job, except for the fact that one of the studs had twisted and warped. Considering that I wanted a hinged hatch (instead of just a piece of plywood screwed into a hole in the wall), life got interesting. I got the job done, and it even looks tidy. Persistence is the key, but cussin' helps oil the wheels to keep things moving.
Yeah, have at it. Mix and match front ends to your heart's desire. Just watch that DC offset on the load resistors or you're not going to be able to get the DC at the output under control later.
Grey
Don't get me started on carpentry. I just got done installing an access hatch into a wall so as to be able to get to the backside of some plumbing. Should have been a trivial job, except for the fact that one of the studs had twisted and warped. Considering that I wanted a hinged hatch (instead of just a piece of plywood screwed into a hole in the wall), life got interesting. I got the job done, and it even looks tidy. Persistence is the key, but cussin' helps oil the wheels to keep things moving.
Yeah, have at it. Mix and match front ends to your heart's desire. Just watch that DC offset on the load resistors or you're not going to be able to get the DC at the output under control later.
Grey
And been out of a job when the economy went flooey...
Nah.
Besides, we need you here, Paul. At least you can rest assured that we're all paid according to the same scale.
I can't remember the quote exactly, but I think Edison said something about "1% inspiration and 99% perspiration." That about sums it up. You can analyse these things until you're so lost in the trees that you can't see the forest (examples of this approach are spread liberally throughout this thread), or you can start tossing parts around and see what happens.
Incidentally, that's how Ian Macmillan got the prize on the DC offset dingus. I had the same idea, but managed to argue myself into believing that it wouldn't work. Too much thinking, and not enough doing. Shame on me. Embarrassing.
In physics, there are theoretical and experimental physicists. The theoretical guys can argue until the cows come home, but sooner or later (if anything is to happen at all) they've got to go down the hall and talk to the hands-on guys. It's the experimental physicists who have all the fun...unless you just want to jaw.
Me? I like results. Arguing is fine for the first five minutes, but then I'm tired of it. Theory always takes a back seat to reality, and I'm much happier in the front seat--the view is much better.
Speaking of which...this Rushmore stuff has me thinking about speakers (which I shouldn't be doing right now), and I just had an oddball idea for a cabinet configuration.
No rest for the weary.
Grey
Nah.
Besides, we need you here, Paul. At least you can rest assured that we're all paid according to the same scale.
I can't remember the quote exactly, but I think Edison said something about "1% inspiration and 99% perspiration." That about sums it up. You can analyse these things until you're so lost in the trees that you can't see the forest (examples of this approach are spread liberally throughout this thread), or you can start tossing parts around and see what happens.
Incidentally, that's how Ian Macmillan got the prize on the DC offset dingus. I had the same idea, but managed to argue myself into believing that it wouldn't work. Too much thinking, and not enough doing. Shame on me. Embarrassing.
In physics, there are theoretical and experimental physicists. The theoretical guys can argue until the cows come home, but sooner or later (if anything is to happen at all) they've got to go down the hall and talk to the hands-on guys. It's the experimental physicists who have all the fun...unless you just want to jaw.
Me? I like results. Arguing is fine for the first five minutes, but then I'm tired of it. Theory always takes a back seat to reality, and I'm much happier in the front seat--the view is much better.
Speaking of which...this Rushmore stuff has me thinking about speakers (which I shouldn't be doing right now), and I just had an oddball idea for a cabinet configuration.
No rest for the weary.
Grey
Responding to an earlier posr, we use the 2SJ109's and
2SK389's whenever we need matched JFETs. The matching
on them is great, although we still end up trimming the
circuit manually.
If you need more current, you parallel them.
On another front:
"In physics, there are theoretical and experimental physicists. The theoretical guys can argue until the cows come home, but sooner or later (if anything is to happen at all) they've got to go down the hall and talk to the hands-on guys. It's the experimental physicists who have all the fun...unless you just want to jaw."
Statistically, the theoreticians have a better sense of humor,
and thus make better company. By all accounts Enrico Fermi
was the greatest combination of 20th century theoretician
and experimentalist.
2SK389's whenever we need matched JFETs. The matching
on them is great, although we still end up trimming the
circuit manually.
If you need more current, you parallel them.
On another front:
"In physics, there are theoretical and experimental physicists. The theoretical guys can argue until the cows come home, but sooner or later (if anything is to happen at all) they've got to go down the hall and talk to the hands-on guys. It's the experimental physicists who have all the fun...unless you just want to jaw."
Statistically, the theoreticians have a better sense of humor,
and thus make better company. By all accounts Enrico Fermi
was the greatest combination of 20th century theoretician
and experimentalist.
If you go with a dual part for the front end, you might want to consider making the two load resistors variable. Not because the devices won't be matched well enough...but because they're matched <i>too</i> well. The differential will be well-nigh perfect, but the output stages will be slightly different and you might find it easier to adjust the DC offset if you have individual control over the two sides.
To get around this, you could use individual JFETs and match them by hand--ideally in the circuit, even though that's a pain in the rump.
Still, there's no denying that the current source adjustment is more convenient.
Your choice.
I wanted to do a triple major--or failing that, a double major with a minor in physics. My advisor wouldn't let me do it. Bummer.
I wonder how one would go about gathering statistics on humor. I'd be afraid that the end result would be like the survey they did recently to find the world's funniest joke. The stuff that came out of that was barely enough to raise a smile, much less a good belly laugh. Seems that too many things are culturally based. We'd probably end up with theoretical humor and experimental humor; two distinct camps.
Grey
To get around this, you could use individual JFETs and match them by hand--ideally in the circuit, even though that's a pain in the rump.
Still, there's no denying that the current source adjustment is more convenient.
Your choice.
I wanted to do a triple major--or failing that, a double major with a minor in physics. My advisor wouldn't let me do it. Bummer.
I wonder how one would go about gathering statistics on humor. I'd be afraid that the end result would be like the survey they did recently to find the world's funniest joke. The stuff that came out of that was barely enough to raise a smile, much less a good belly laugh. Seems that too many things are culturally based. We'd probably end up with theoretical humor and experimental humor; two distinct camps.
Grey
because they're matched too well
I guess I am confused (or maybe someone else is). The better the front end devices are matched the lower the DC offset is. I will assume that we are talking about differential offset voltage which is the actual offset voltage seen by speaker. Mismatching the drain resistors will worsen the distortion figures and PSRR by degrading the balance in the open loop gain of the amp. Also, the negative feedback will try to adjust the amp's output offset to follow the front end's offset in proportion to the closed loop gain. This swamps differences in the open loop offset that unequal front end drain resistors would adjust. The ultimate perfomance of the amp depends close matching of each half from my understanding of balanced circuits from the theory I have read. Unbalancing the open loop gain and offset between two halves of the amp is is not an effective way to trim DC offset and increases distortion.
Adjustment of the current source for the front end adjust the common mode offset and not the differential offset. Common mode offset is not seen by the speaker.
Loren
I guess I am confused (or maybe someone else is). The better the front end devices are matched the lower the DC offset is. I will assume that we are talking about differential offset voltage which is the actual offset voltage seen by speaker. Mismatching the drain resistors will worsen the distortion figures and PSRR by degrading the balance in the open loop gain of the amp. Also, the negative feedback will try to adjust the amp's output offset to follow the front end's offset in proportion to the closed loop gain. This swamps differences in the open loop offset that unequal front end drain resistors would adjust. The ultimate perfomance of the amp depends close matching of each half from my understanding of balanced circuits from the theory I have read. Unbalancing the open loop gain and offset between two halves of the amp is is not an effective way to trim DC offset and increases distortion.
Adjustment of the current source for the front end adjust the common mode offset and not the differential offset. Common mode offset is not seen by the speaker.
Loren
Offset voltage
I understand as follows:
Even though the input voltage is zero, we see that an output voltage can exist. That is the output offset voltage and it is mainly due to a difference between the two Vgs values of the paired FETs. This difference acts like a small ac input and is amplified into an output offset voltage (open loop output offset voltage). However, this output offset voltage can be much smaller (i.e. closed-loop output offset voltage) with negative feedback.
If some of the output offset voltage is fed back to the inverting input (i.e. negative feedback), an out-of-phase voltage arrives at the output, canceling most of the original output offset voltage. When the loop gain is much greater than 1, the closed-loop output offset voltage is much smaller than the open-loop output offset voltage.
Anyhow, using the matching FETs with respect to Vgs values (with the maximum acceptable difference recommended by the One and Only, and also by other DIYers) is always desirable, in my opinion.
JH
I understand as follows:
Even though the input voltage is zero, we see that an output voltage can exist. That is the output offset voltage and it is mainly due to a difference between the two Vgs values of the paired FETs. This difference acts like a small ac input and is amplified into an output offset voltage (open loop output offset voltage). However, this output offset voltage can be much smaller (i.e. closed-loop output offset voltage) with negative feedback.
If some of the output offset voltage is fed back to the inverting input (i.e. negative feedback), an out-of-phase voltage arrives at the output, canceling most of the original output offset voltage. When the loop gain is much greater than 1, the closed-loop output offset voltage is much smaller than the open-loop output offset voltage.
Anyhow, using the matching FETs with respect to Vgs values (with the maximum acceptable difference recommended by the One and Only, and also by other DIYers) is always desirable, in my opinion.
JH
Nelson Pass said:
That was back in the days when they were playing with nukes and atom smashers. Nowadays all pure physicists do is spend a lot of money and get really frustrated, or they try to make a living by writing fiction on currently unprovable theories. It is so rare for them to succeed that they give the Nobel prize as soon as they do.
I feel sad as I see similar. Anyway, I like these words:grataku said:
Physics is just collection of theories—the right ones, the ones that have been found by experimental test to best describe physical phenomena.
The first concern of a physicist is whether the physics is “right,” but the next most important is that it be “simple.”
grataku,
It's hard to pull that stunt twice in a row. I'd already done it with psych. Nobody would have arrested me, I guess, but it's nice to have the high muckety-mucks on your side when it's time to graduate and you need a bunch of forms signed.
Loren,
If this amp were a 'normal' amp, a perfectly balanced differential would be a no-brainer.
Think of it this way--the front end differential is the fulcrum for a playground see-saw. It's perfectly balanced, and I mean <i>perfectly</i>. Then along come two kids who weigh exactly the same. Exactly. They sit on the see-saw. Just sit. They balance perfectly. Now, another Kid comes along who's got an EE in his back pocket. Otherwise he weighs exactly the same, but that EE makes him heavier by an ounce or so. He sits on the see-saw, and since he's got that EE, he's a heavy (or at least he thinks he is), so that end of the see-saw sinks to the ground.
The solution, of course, is to move the fulcrum point to an imperfect place; to offset it a bit. Then the see-saw comes back into balance, even though the fulcrum is no longer ideal when viewed on its own.
The fact that the outputs aren't going to match exactly means that the amp is going to 'tilt' in one direction or another. You could make the differential loads adjustable to give the user the ability to control each side individually, or you could adjust the fulcrum itself by using ever-so-slightly mismatched devices for the front end...or you could simply decide to live with the DC. Since it's absolute, the speaker won't see it. As long as it's not too great in magnitude, it won't really effect the operation of the amp.
Does that make it clearer?
grataku again,
I don't keep up with developments in physics as closely as I used to, just whatever they decide to toss into <i>Scientific American</i>. Unfortunately, the theoreticians seem to have run <i>way</i> ahead of the experimentalists. The sums of money necessary to check some of these concepts have gone beyond the cost of simple DIY cloud chambers. That leaves the experimental guys spending their time begging for money, which leaves them frustrated, too. I know I'd hate to have to cough up the money to build a neutrino or gravity wave detector in my back yard...not that I'd mind hosting one.
Not unlike us--ideas, but not enough resources to put them to the test.
Grey
It's hard to pull that stunt twice in a row. I'd already done it with psych. Nobody would have arrested me, I guess, but it's nice to have the high muckety-mucks on your side when it's time to graduate and you need a bunch of forms signed.
Loren,
If this amp were a 'normal' amp, a perfectly balanced differential would be a no-brainer.
Think of it this way--the front end differential is the fulcrum for a playground see-saw. It's perfectly balanced, and I mean <i>perfectly</i>. Then along come two kids who weigh exactly the same. Exactly. They sit on the see-saw. Just sit. They balance perfectly. Now, another Kid comes along who's got an EE in his back pocket. Otherwise he weighs exactly the same, but that EE makes him heavier by an ounce or so. He sits on the see-saw, and since he's got that EE, he's a heavy (or at least he thinks he is), so that end of the see-saw sinks to the ground.
The solution, of course, is to move the fulcrum point to an imperfect place; to offset it a bit. Then the see-saw comes back into balance, even though the fulcrum is no longer ideal when viewed on its own.
The fact that the outputs aren't going to match exactly means that the amp is going to 'tilt' in one direction or another. You could make the differential loads adjustable to give the user the ability to control each side individually, or you could adjust the fulcrum itself by using ever-so-slightly mismatched devices for the front end...or you could simply decide to live with the DC. Since it's absolute, the speaker won't see it. As long as it's not too great in magnitude, it won't really effect the operation of the amp.
Does that make it clearer?
grataku again,
I don't keep up with developments in physics as closely as I used to, just whatever they decide to toss into <i>Scientific American</i>. Unfortunately, the theoreticians seem to have run <i>way</i> ahead of the experimentalists. The sums of money necessary to check some of these concepts have gone beyond the cost of simple DIY cloud chambers. That leaves the experimental guys spending their time begging for money, which leaves them frustrated, too. I know I'd hate to have to cough up the money to build a neutrino or gravity wave detector in my back yard...not that I'd mind hosting one.
Not unlike us--ideas, but not enough resources to put them to the test.
Grey
If this amp were a 'normal' amp, a perfectly balanced differential would be a no-brai
I think it is a normal amp in that it obeys the laws of physics and is a simple enough design for some reasoned analysis.
"using ever-so-slightly mismatched devices for the front end...or you could simply decide to live with the DC. Since it's absolute, the speaker won't see it."
The DC offset that we concerned with is differential. The speaker see's it. What you are calling absolute is common mode offset and is not dependent on the DC imbalance of the front end diff pair.
An electrical engineering degree is not necessary to build or design amps nor is it a hinderance. I believe Mr. Pass has a degree in Physics. I have met many more EEs that are not analog designers than are.
When troubleshooting and adressing design issues one does need a good understanding of how an amplifier works. Familiarity
with the concepts such as closed loop gain, open loop gain, common mode, differential mode, and device characteristics such as transconductance and saturated drain current are essential to design amplifiers. I have seen plenty of good advice from those with a strong grasp of these concepts and have learned a lot on this forum from Mr. Pass and others here. Good audio design is never a no-brainer.
And some good advice for posters that even I try to keep in mind when I post:
Giving design advice, without the fundamentals to know that it is usefull and correct information, is a diservice to the trusting and less experienced members of the forum. Bluff on the forum about tangible concepts, and you will be caught. You may not be confronted, but your readers will know when you are going out a limb with a saw in your hand. Your credibility will diminish each time till it is all gone at some point with little chance of regaining it. Have fun but be responsible, friends don't let friends type bunk.
Loren
I think it is a normal amp in that it obeys the laws of physics and is a simple enough design for some reasoned analysis.
"using ever-so-slightly mismatched devices for the front end...or you could simply decide to live with the DC. Since it's absolute, the speaker won't see it."
The DC offset that we concerned with is differential. The speaker see's it. What you are calling absolute is common mode offset and is not dependent on the DC imbalance of the front end diff pair.
An electrical engineering degree is not necessary to build or design amps nor is it a hinderance. I believe Mr. Pass has a degree in Physics. I have met many more EEs that are not analog designers than are.
When troubleshooting and adressing design issues one does need a good understanding of how an amplifier works. Familiarity
with the concepts such as closed loop gain, open loop gain, common mode, differential mode, and device characteristics such as transconductance and saturated drain current are essential to design amplifiers. I have seen plenty of good advice from those with a strong grasp of these concepts and have learned a lot on this forum from Mr. Pass and others here. Good audio design is never a no-brainer.
And some good advice for posters that even I try to keep in mind when I post:
Giving design advice, without the fundamentals to know that it is usefull and correct information, is a diservice to the trusting and less experienced members of the forum. Bluff on the forum about tangible concepts, and you will be caught. You may not be confronted, but your readers will know when you are going out a limb with a saw in your hand. Your credibility will diminish each time till it is all gone at some point with little chance of regaining it. Have fun but be responsible, friends don't let friends type bunk.
Loren
When I was at school, one of the students in the physicsGRollins said:
department conducted the following experiment:
The 4th floor was a theoreticians floor, and the 3rd was
experimentalists. He posted signs in the lavatories on both
floors, labeling urinals, toilets, and sinks "Experimentalists"
and "Theoreticians".
After a couple of days, all the signs on the experimentalist
floor had been torn down and thrown away. On the
theoretician's floor, the signs had been embellished upon
and one "Experimentalist" graced the trash can.
The student in question was the now famous game designer
Chris Crawford.
