GRollins said:
I posted something I called V2.2 (post #2905) which was simply a minor embellishment on my earlier posting. I doubt this is what NP was referring to though.
Ian.
I understand you not wanting to put an effort into someone you have just met, but thank you for replying because I just found out about this X100 project and I never knew that anyone existed that I could explore the current The State Of The Art amplification projects. I will be arround from now on and am studying the background material again now from Nelson Pass in order to bring my skills closer to the ones on this FORUMS, so you will get used to seeing me here and eventually I hope to build a design that has been built by others here, yours is the only clearly documented one I have been studying so far, I respect the clearness of your design following Pass's work, maybe I will build your design first.
Scatterbrain001
Scatterbrain001
Ian
is there a reason for not using something like a MRF151A MOSFET since we are building a state of the art amplifier, am I missing something here about how to choose an output transistor, at $56.00 each they are expensive though.
is there a reason for not using something like a MRF151A MOSFET since we are building a state of the art amplifier, am I missing something here about how to choose an output transistor, at $56.00 each they are expensive though.
scatterbrain001 said:
It's an RF transistor. If you can put up with a transconductance
of about 0.15 (that is, 7 gate volts for every amp output) it would
probably work, but the ordinary IRFP240 has a transconductance
more like 10.
😎
scatterbrain001 said:
1. Expensive
2. What does it do @ audio frequencies?
3.Are you trying to say Pass Labs is not state of the art?
scatterbrain001 said:
I think this question has already been answered but it might be useful to ask the same question of some of the other semiconductors. For example the ZVP2110 will be difficult to heatsink for the dissipation required and there may be better devices than ZTX450 and its complement for the cascodes.
Ian.
The original circuit could be built with most any reasonable vertical MOSFET. I believe it has also been built with lateral MOSFETs. In principle, it could be built with bipolars or even tubes, but I'm not aware of anyone actually doing so.
Are there better devices than the IRF MOSFETs? Arguably so, depending on what criteria you use. One useful thing to remember is that the problems with IRF MOSFETs are limited to the P-ch devices. The N-ch devices are still valid choices. With this in mind, and assuming that you want to bother changing things around, you could use something like the Fairchild P-ch devices for the front end and IRF or Fairchild devices for the outputs.
If you're going for a "UGS" front end, the choice of devices will depend on the exact topology you use.
It would be nice if someone had invented a radically new device in the past few years, but if they have I haven't heard of it. Mainly it's just the same candidates.
The Lovoltech LU1014D will pop into some peoples' minds, but it's a depletion mode device, which complicates biasing somewhat. That said, you can build an Aleph variant with the LU1014D...but you'll have to redesign the front end. Note that the LU1014D can draw Gate current and you'll have to allow for that in your front end design.
Grey
Are there better devices than the IRF MOSFETs? Arguably so, depending on what criteria you use. One useful thing to remember is that the problems with IRF MOSFETs are limited to the P-ch devices. The N-ch devices are still valid choices. With this in mind, and assuming that you want to bother changing things around, you could use something like the Fairchild P-ch devices for the front end and IRF or Fairchild devices for the outputs.
If you're going for a "UGS" front end, the choice of devices will depend on the exact topology you use.
It would be nice if someone had invented a radically new device in the past few years, but if they have I haven't heard of it. Mainly it's just the same candidates.
The Lovoltech LU1014D will pop into some peoples' minds, but it's a depletion mode device, which complicates biasing somewhat. That said, you can build an Aleph variant with the LU1014D...but you'll have to redesign the front end. Note that the LU1014D can draw Gate current and you'll have to allow for that in your front end design.
Grey
Nelson Pass thankyou for spicifically giving me information that I need tounderstand before choosing transistors, it is very helpfull as I build my understanding of why choices have been made while I look over all of the posible components aviable. I was just looking over power mosfet with high watts and amps, and now I will add Transconductance. The MRF151A has a Forward Transconductance of 7, you noted to me its transconductance was 0.15, are these two transconductance(s) not the same thing.
Transconductance as in the datasheet is stated for a certain working point which is usually not the one you're interested in. That's the reason why there are usually also transconducatance figures in the datasheet.
Well, if you're at it, you can also add gate capacitance to your list. And a lot more. It's not particularly easy to choose active elements so if you're unsure, stick with the recommendations!
All the best, Hannes
Well, if you're at it, you can also add gate capacitance to your list. And a lot more. It's not particularly easy to choose active elements so if you're unsure, stick with the recommendations!
All the best, Hannes
scatterbrain001 said:
Sorry, the spec sheet says 7 mhos. This is approximately half
of the IRFP240's
Bad math day...
😎
jacco vermeulen said:
Nelson's probably forgotten more math than I'll ever learn......sigh....maybe in my next life😉
Most of the necessary formulas are easy enough to remember. Two knowns and one unknown. Plug in the numbers, crank the handle, and out pops the answer. I have long resisted attempts by some of the EEs and like-minded folk to mystify the process. Fortunately, most of them have departed for other places where, presumably, they congratulate one another for being able to recite IR=E from memory. I have more confidence in the abilities of the rank and file members here and nothing to gain by building false barriers by insisting that the basic math is too hard.
Let it be known that you do not need calculus to build circuits. You need nothing more than rudimentary algebra to do some pretty sophisticated circuits. And if algebra sounds scary, just pick the formulas you need, make up a cheat sheet, and pin it to the wall.
Having learned multiple computer languages over the years, I used to write code to do various things. Fooey...these days I use an Excel spreadsheet that I've set up with all the usual suspects. It's no more accurate than a hand calculator, but it is a bit faster, and I value every second these days.
Grey
Let it be known that you do not need calculus to build circuits. You need nothing more than rudimentary algebra to do some pretty sophisticated circuits. And if algebra sounds scary, just pick the formulas you need, make up a cheat sheet, and pin it to the wall.
Having learned multiple computer languages over the years, I used to write code to do various things. Fooey...these days I use an Excel spreadsheet that I've set up with all the usual suspects. It's no more accurate than a hand calculator, but it is a bit faster, and I value every second these days.
Grey
It think it's more than this, Grey. It's not just the math. Some of you guys have been around long enough and experimented enough to know what will work and what will not. Like in what part of a circuit can you insert a wide range of values and still end up with the same results? Do you sit here and calculate the wattages of every resistor to make sure it's in a SAO? I doubt it. You know by just looking at a few facts and you just calculate it in your head.
Way before anyone here even brought up the Zen 9 or a JFet power amp, I researched it and found very little info on the net about them. I said, well maybe I can build one and see. I found the Lovoltech and thought there was enough current handling there to make a small amp with only a single JFet. You guys figured out in 2 seconds that the only solution was to cascode with a MosFet. How many Lovoltech JFets would I have to fry before I learned?
Answer me this- What do so many Pass DIY cirucuit use 1/4 watt 221 Ohm resistor? Why not 442 or 884 Ohms?
Face it, you guys know much more than simple Algebra. You can see a few steps a head. In Nelson's case it's probably like decades. You topologies you! 😉
I say this will all the respect in the world.
V~
Simple.
It's one of the building blocks of the universe, along with 220uf caps, Irfp240, and CL60 thermistors.
I'm with vdi_nenna, some of you guys do seem to see it a couple steps ahead. Thank goodness!
Vince,
No, I'm not offended. And yes, you're right, some of it is experience.
Rules of thumb and random points:
--Any resistor in the front end can be little, meaning 1/4W if you're Nelson, or 1/2W if it's me. Why the discrepancy? Because I still (rarely) do tube circuits and there are times when the extra heat dissipation comes in handy. And I don't have sufficient funds to double stock 1/4W and 1/2W of everything. Or the room. And the 1/2W only cost one or two cents more. I think of it as cheap insurance.
And yes, I popped a half watter a couple of months ago. Boy, was I peeved with myself. True, it's only a 16 cent part, but it's the principle of the thing. My ancestry shows a strong influx of blood from Scotland and I hate wasting money. Even 16 cents. But it happens. And I learned a thing from the poor resistor's demise, so it wasn't completely wasted money. But I still grumbled.
--Source/emitter resistors should be larger. The 3W Panasonics are a good choice. Okay, so you do the calculations and see that your dissipation is actually really small. Plan for peaks and failures. If an output goes, it frequently takes the resistor with it. I got tired of replacing resistors every time a 6550 (big octal output tube) flashed. So I went big. Now the tubes pop, but the resistor stays. Makes my life easier. Ditto for solid state.
The counterargument is that a burned resistor immediately tells you which outputs are suspect. You pays your money and you makes your choices.
A burned Gate stopper is a flashing neon light pointing at its output device.
--Regarding the Lovoltech JFETs. The case was the crux of the matter in my eyes. You just can't get that much heat out of the silly thing. How to cure heat? Reduce the current (which would come pretty close to being a deal killer because the whole point of the LU1014D is that it's high current) or reduce the voltage. Hmmm...lemme see...we can't do current, so that leaves voltage. Okay, so what are the options? Well, you can bridge the amp, which means that the output takes all the current, but only half the voltage. Good start (headphone amp, anyone?), but still not enough for a decent-sized power amp. Then there's this little devil on your shoulder whispering a single word, "Cascode!" That gives you control over the current and the voltage. Ah, now that sounds workable.
And it is.
--Gate stoppers. Experience. I've done everything from nothing (don't go there) to 10k (believe it or not, it worked okay--at least that one time). Be mindful of the fact that the resistor works against the capacitance of the Gate (which is why you're putting it there in the first place) and can have an influence on your frequency response. Nelson's tendency to use 221 Ohms is not a fundamental law of the universe. He buys them by the case. They're right next to him on the bench. It's a good compromise value between zero and infinity. Personally, I've been trending lower. Give 100 Ohms a try if you're fiddling a circuit. Be prepared to back up to 150 or 221 if your circuit gives you any backtalk.
--Do I see a step ahead? Yes, but through a clouded crystal ball. After all, I predicted that terrorists would take out a tall structure back in the early '90s (in the Darwin's Children series), but I pegged the Eiffel Tower, not the World Trade Center. I got it half right. But that's okay, I'm in good company. Orwell called his book 1984.
He was off by twenty years.
Grey
No, I'm not offended. And yes, you're right, some of it is experience.
Rules of thumb and random points:
--Any resistor in the front end can be little, meaning 1/4W if you're Nelson, or 1/2W if it's me. Why the discrepancy? Because I still (rarely) do tube circuits and there are times when the extra heat dissipation comes in handy. And I don't have sufficient funds to double stock 1/4W and 1/2W of everything. Or the room. And the 1/2W only cost one or two cents more. I think of it as cheap insurance.
And yes, I popped a half watter a couple of months ago. Boy, was I peeved with myself. True, it's only a 16 cent part, but it's the principle of the thing. My ancestry shows a strong influx of blood from Scotland and I hate wasting money. Even 16 cents. But it happens. And I learned a thing from the poor resistor's demise, so it wasn't completely wasted money. But I still grumbled.
--Source/emitter resistors should be larger. The 3W Panasonics are a good choice. Okay, so you do the calculations and see that your dissipation is actually really small. Plan for peaks and failures. If an output goes, it frequently takes the resistor with it. I got tired of replacing resistors every time a 6550 (big octal output tube) flashed. So I went big. Now the tubes pop, but the resistor stays. Makes my life easier. Ditto for solid state.
The counterargument is that a burned resistor immediately tells you which outputs are suspect. You pays your money and you makes your choices.
A burned Gate stopper is a flashing neon light pointing at its output device.
--Regarding the Lovoltech JFETs. The case was the crux of the matter in my eyes. You just can't get that much heat out of the silly thing. How to cure heat? Reduce the current (which would come pretty close to being a deal killer because the whole point of the LU1014D is that it's high current) or reduce the voltage. Hmmm...lemme see...we can't do current, so that leaves voltage. Okay, so what are the options? Well, you can bridge the amp, which means that the output takes all the current, but only half the voltage. Good start (headphone amp, anyone?), but still not enough for a decent-sized power amp. Then there's this little devil on your shoulder whispering a single word, "Cascode!" That gives you control over the current and the voltage. Ah, now that sounds workable.
And it is.
--Gate stoppers. Experience. I've done everything from nothing (don't go there) to 10k (believe it or not, it worked okay--at least that one time). Be mindful of the fact that the resistor works against the capacitance of the Gate (which is why you're putting it there in the first place) and can have an influence on your frequency response. Nelson's tendency to use 221 Ohms is not a fundamental law of the universe. He buys them by the case. They're right next to him on the bench. It's a good compromise value between zero and infinity. Personally, I've been trending lower. Give 100 Ohms a try if you're fiddling a circuit. Be prepared to back up to 150 or 221 if your circuit gives you any backtalk.
--Do I see a step ahead? Yes, but through a clouded crystal ball. After all, I predicted that terrorists would take out a tall structure back in the early '90s (in the Darwin's Children series), but I pegged the Eiffel Tower, not the World Trade Center. I got it half right. But that's okay, I'm in good company. Orwell called his book 1984.
He was off by twenty years.
Grey