Clarkcr said:
Hi C,
as this topology is a push-pull design, engineers tend to
make it as symmetric as possible. Using all N-Fets would
lead to nonsymmetry in the driverstage thus leading to
different behaviour at positive or negative swing.
As Aleph is of single ended nature it does not matter at all
what kind of fet you use in the CCS.
The X is a push-pull design, as simple as possible.
Using N-fets throughout the outputstage would lead to
a much more complicated circuit because you have to change
phase in one half of the gate driver. Second issue is that one
half is implemented as source follower the other not. This
again leads to more THD.
Uli
Use of both n and p types is the most used for push-pull amps. However I would like to see an all N-Channel X amp to. Anthony Holton (do a search for his name) has made an all N-Channel amp with very good results both sonic and measured. The amp only has two stages so it’s a good candidate for Xing. This type of topology is called quasi complementary.
Got it! Thanks. Makes perfect sense. But, since the x amps are Class A to a point....then push/pull.....would the Class A be all 'N' type to that point, then go 'P' and 'N' in a balanced fashion? OR does it just make good all around design since to use an equal amount of Ns and Ps??
Another question.....the Driver Stage...this is the stage with the large banks of MOSFETs right? In looking at a circuit diagram in the Pass Service Manuals, would it be necessary to include all the MOSFETS that are in the amplifier, or just a certain number? In other words, if I'm building a circuit diagram using some software program, do I actually draw out 80 MOSFETs???? How are they laid out in the X1000? Is there a PCB with 40 of them all in a row?? Is the PCB 2 feet long? OR do they simply mount directly to the heatsink? Are they in stages or arrays? 20, 20, 20, 20?
How is that thing laid out?
C
Another question.....the Driver Stage...this is the stage with the large banks of MOSFETs right? In looking at a circuit diagram in the Pass Service Manuals, would it be necessary to include all the MOSFETS that are in the amplifier, or just a certain number? In other words, if I'm building a circuit diagram using some software program, do I actually draw out 80 MOSFETs???? How are they laid out in the X1000? Is there a PCB with 40 of them all in a row?? Is the PCB 2 feet long? OR do they simply mount directly to the heatsink? Are they in stages or arrays? 20, 20, 20, 20?
How is that thing laid out?
C
Clarkcr said:
From what I understand, each P and N type will amplify the entire sinosudial waveform and then those two (complete) waves will be added out of phase to double the voltage delivered to the load. i.e. The "P" type would not be a positive waveform (as in a class B amp) but a full sinewave same for the N type but be out of phase with the P....
K-
K-amps,
adding two out of phase anything gets you a big distorted signal unless it's a very specific phase shift (180 deg).
Then again if you add two such signals you would get 0 and 2x if you subtract them but that is a differential not a PP, I think. 😉
adding two out of phase anything gets you a big distorted signal unless it's a very specific phase shift (180 deg).
Then again if you add two such signals you would get 0 and 2x if you subtract them but that is a differential not a PP, I think. 😉
source follower
As both P and N outputdevices are commonly used as
source followers in outputstages, they amplify the signal
IN PHASE!
To do quasi-complementary design you need to invert
phase of one of the 2 gate signals as the output from
the drain is inverted too.
Like in tube circuits you therefore need some kind of
phase splitter circuit.
Using both output nodes of a fet you then drive the
sourcefollower part of the outputstage with the drain node
thus controlling the gain with the gain of this particular fet.
The source node then drives the other part of the output
which in turn is controlled by the gain of the outputfet.
I see no need to construct a fragile circuit like this since
modern Mosfets are available in P-versions too with
good matching to N-types.
Uli
PS:Anthony Holton
As both P and N outputdevices are commonly used as
source followers in outputstages, they amplify the signal
IN PHASE!
To do quasi-complementary design you need to invert
phase of one of the 2 gate signals as the output from
the drain is inverted too.
Like in tube circuits you therefore need some kind of
phase splitter circuit.
Using both output nodes of a fet you then drive the
sourcefollower part of the outputstage with the drain node
thus controlling the gain with the gain of this particular fet.
The source node then drives the other part of the output
which in turn is controlled by the gain of the outputfet.
I see no need to construct a fragile circuit like this since
modern Mosfets are available in P-versions too with
good matching to N-types.
Uli
PS:Anthony Holton
Got it thanks! I'm going to practice building an amp by building two Aleph 1.2s. I'm ordering the parts today. Everything but the power supply caps and the transformer. Digikey doesn't seem to have 25000 uF caps in the catalog or such a large Toroid transformer. Maybe if I call them? special order?
Anyway, this will be fun and relatively easy. I will learn a thing or two about chassis design and construction.
In the meantime, I'm going to get the circuit builder software and pay around with the X1000 circuit and board layout. I'll run my versions by everyone on DIY to see what you think.
Thanks for the help😉
Anyway, this will be fun and relatively easy. I will learn a thing or two about chassis design and construction.
In the meantime, I'm going to get the circuit builder software and pay around with the X1000 circuit and board layout. I'll run my versions by everyone on DIY to see what you think.
Thanks for the help😉
I wouldn't call an A1.2 a "practice" amp.
I think your plan to start cold turkey to built something so complex with your level of experience is a recipe for failure. But hey, what do I know? I just got a ton of decommissioned crap in my attic which I built and didn't work. 😉
Only two people built the straight X amp that I know of, Petter and Uli. I think the latter had better luck.
How about something small and cheap first like an aleph3 (which is not all that small) or better yet a mini-A? MikeW may even have an extra board for it.
Do you have some simple instrumentation to test your result?
The miniA would pretty much behave like the A1.2 so at least you could get use to the circuit. You can always turn the minia into a HDPhone amp later.
By the way the A1.2 is not Nelson's fav. amp. For something big the A2 maybe better.
Good luck,
I think your plan to start cold turkey to built something so complex with your level of experience is a recipe for failure. But hey, what do I know? I just got a ton of decommissioned crap in my attic which I built and didn't work. 😉
Only two people built the straight X amp that I know of, Petter and Uli. I think the latter had better luck.
How about something small and cheap first like an aleph3 (which is not all that small) or better yet a mini-A? MikeW may even have an extra board for it.
Do you have some simple instrumentation to test your result?
The miniA would pretty much behave like the A1.2 so at least you could get use to the circuit. You can always turn the minia into a HDPhone amp later.
By the way the A1.2 is not Nelson's fav. amp. For something big the A2 maybe better.
Good luck,
What can be so difficult about it? I have built amps before, what I lack is the topography design knowledge. The physical building and soldering I've got down.
Where will I go wrong here? Tell me what I don't know that will lead my Aleph 1.2 to become a boat anchor?🙂
Now, the X1000 project is a challenge. Agreed. I want to have much more knowledge and understanding under my belt before that one. I believe there has been enough published though about the circuit to back engineer the monster.
Why does Nelson favor one amp over another? I have heard this from other builders..."Nelson's favorite is the Aleph 2". Well, why? I want monoblocks. What is Nelson's favorite monoblock?
😀
Thanks for all the advice!!
C
Where will I go wrong here? Tell me what I don't know that will lead my Aleph 1.2 to become a boat anchor?🙂
Now, the X1000 project is a challenge. Agreed. I want to have much more knowledge and understanding under my belt before that one. I believe there has been enough published though about the circuit to back engineer the monster.
Why does Nelson favor one amp over another? I have heard this from other builders..."Nelson's favorite is the Aleph 2". Well, why? I want monoblocks. What is Nelson's favorite monoblock?
😀
Thanks for all the advice!!
C
Clarkcr said:
Hi C,
IMHO Grataku just wants to tell that an A3 is basically the same
as A1.2 BUT scaled down. -> catastrophic failure is scaled down
too!
The reason why NP is believed to like A2 most is simply the best
sound! BTW A2 are monoblocks.
I believe X1000 is backengineered already but I for my part
won´t post a circuit on that, it is Mr. Pass´intellectual AND
commercial property.
Uli
Clark,
I am simply voicing a word of caution, that's all. Time and time again I have seen big plans like yours (and some of mine) evaporate like farts in the wind leaving a bad smell in your nose and alot less money in your pocket.
A priori I don't see the logic of building something huge as the A1.2 for testing and then build something even huge-er like the X1000. Again it's just me. 😉
The A1.2 sounds dark for the simple reason it uses 12 parallel mosfets (u need a total of 48 very closely matched mosfets) and the overall input capacitance gets to be ridiculously high even for a very low impedance driver like the 9610.
I think I made my point and I'll stay out of it while eagerly waiting for pictures of the finished amps.
I am simply voicing a word of caution, that's all. Time and time again I have seen big plans like yours (and some of mine) evaporate like farts in the wind leaving a bad smell in your nose and alot less money in your pocket.
A priori I don't see the logic of building something huge as the A1.2 for testing and then build something even huge-er like the X1000. Again it's just me. 😉
The A1.2 sounds dark for the simple reason it uses 12 parallel mosfets (u need a total of 48 very closely matched mosfets) and the overall input capacitance gets to be ridiculously high even for a very low impedance driver like the 9610.
I think I made my point and I'll stay out of it while eagerly waiting for pictures of the finished amps.
Hmmmm...less heat.......marginaly less power....less money.....what are the sonic differences? I'm not going to read the Stereophile review....but in Nelson's words, more or less, what difference sonically does the amp have. Does the 1.2 do anything better/worse than the A2?
I'm not worried about heat....the amplifiers will be in a seperate room that can be air conditioned or not. The room can reach the 60 C....doesn't matter. They won't be on for that long a period anyway....not like I have time to listen to music 4 hours a day 😡
In my ideal system....these Alephs will drive a pair of sub woofers. Until the X1000 are finished to drive my electrostatics, the Alephs will run the Monoliths. So, if there is more low end power and punchy bass in the 1.2, then those are for me. If it's imaging and delicate highs....I'm not that concerned about it.
I don't know what do you think? I feel like I'm going in circles. I have noticed that no one has really built a 1.2. So maybe I SHOULD go with the collective wisdom and experience and build A2s.
In regards to the X1000 back engineered...are you referring to Petters schematic? Do you think Petter's schematic is what's in the X1000 Service Manual?....I assume there is such a thing as an X1000 service manual.
Thanks again
Chris 😀
I'm not worried about heat....the amplifiers will be in a seperate room that can be air conditioned or not. The room can reach the 60 C....doesn't matter. They won't be on for that long a period anyway....not like I have time to listen to music 4 hours a day 😡
In my ideal system....these Alephs will drive a pair of sub woofers. Until the X1000 are finished to drive my electrostatics, the Alephs will run the Monoliths. So, if there is more low end power and punchy bass in the 1.2, then those are for me. If it's imaging and delicate highs....I'm not that concerned about it.
I don't know what do you think? I feel like I'm going in circles. I have noticed that no one has really built a 1.2. So maybe I SHOULD go with the collective wisdom and experience and build A2s.
In regards to the X1000 back engineered...are you referring to Petters schematic? Do you think Petter's schematic is what's in the X1000 Service Manual?....I assume there is such a thing as an X1000 service manual.
Thanks again
Chris 😀
IIRC, NP said that Petter's design was similar to the lower powered X amps, but the X-1000 and X-600 use a cascoded input stage, as shown in the patent. I seem to remember something about Jfets, too.
a 60C room plus the rise across the heat sinks sounds like your output stage wouldn't live long. Since you have a separate room anyway, why not use fan cooling with outside air? Even 35C+ Texas summers are better than keeping that heat in the house and operating at 60C ambient. Liquid cooling would be another way to get the heat outside.
That ought to get you started. keep us posted on your progress.
a 60C room plus the rise across the heat sinks sounds like your output stage wouldn't live long. Since you have a separate room anyway, why not use fan cooling with outside air? Even 35C+ Texas summers are better than keeping that heat in the house and operating at 60C ambient. Liquid cooling would be another way to get the heat outside.
That ought to get you started. keep us posted on your progress.
Clark
"Hmmmm...less heat.......marginaly less power....less money.....what are the sonic differences? I'm not going to read the Stereophile review....but in Nelson's words, more or less, what difference sonically does the amp have. Does the 1.2 do anything better/worse than the A2?"
As I said, high input cap means a highs are going to be attenuated ---> darker sound, or less highs, or tonally balanced toward lows. How else can I put it?
If a sub amp is what you need there are MUCH better alternatives for you than either the X or the A1.2 at a much lower cost, too.
"Hmmmm...less heat.......marginaly less power....less money.....what are the sonic differences? I'm not going to read the Stereophile review....but in Nelson's words, more or less, what difference sonically does the amp have. Does the 1.2 do anything better/worse than the A2?"
As I said, high input cap means a highs are going to be attenuated ---> darker sound, or less highs, or tonally balanced toward lows. How else can I put it?
If a sub amp is what you need there are MUCH better alternatives for you than either the X or the A1.2 at a much lower cost, too.
Aleph subwoofer amp? That would be a waste. The nice thing about class A is the lack of crossover and switching disorder. I doubt that you could hear either in a driver limited to 150 hz.
If you really wasnt to be ballsie, do something a little bigger than a 2 for the electrostatics. As Nelson says, Alephs laugh at reactive loads (electrostatic speakers are a step up transformer driving a capacitor. Quite ugly for your run of the mill class AB). After listening to an Aleph, you'll forget about building anything else.
For the sub, I would build a large class ab with huge damping factor and tons of nfb. If you stay below 500 hz, and bandwith limit somwhere, you avoid all of the ills of global feedback.
If you really wasnt to be ballsie, do something a little bigger than a 2 for the electrostatics. As Nelson says, Alephs laugh at reactive loads (electrostatic speakers are a step up transformer driving a capacitor. Quite ugly for your run of the mill class AB). After listening to an Aleph, you'll forget about building anything else.
For the sub, I would build a large class ab with huge damping factor and tons of nfb. If you stay below 500 hz, and bandwith limit somwhere, you avoid all of the ills of global feedback.
Uli said earlier that you'd need a phase splitter to run a quasi output.
You might try forking the output from each half of the input differential to the proper bank of output devices. Differentials make dandy phase splitters.
Note that I have not built this variation yet. I've got notes and a rough schematic. The only problem I can see is that the feedback might get wonky since each side of the differential effects both sides of the output. Unfortunately, I've got complications out here in the hinterlands that will prevent me from getting much audio done in the near future so it will be a while before I can try the circuit.
Grey
You might try forking the output from each half of the input differential to the proper bank of output devices. Differentials make dandy phase splitters.
Note that I have not built this variation yet. I've got notes and a rough schematic. The only problem I can see is that the feedback might get wonky since each side of the differential effects both sides of the output. Unfortunately, I've got complications out here in the hinterlands that will prevent me from getting much audio done in the near future so it will be a while before I can try the circuit.
Grey
Somebody can give a hint about how many mA running on differential and folded cascode on monster size amp like X1000?
If the output pairs is 20 pair (40 mosfets total), voltage is as high as +/-75V, I dont think figure as low as 10mA is the answer. Is it possible that the differential and folded cascode running at about 200mA or more?
If the output pairs is 20 pair (40 mosfets total), voltage is as high as +/-75V, I dont think figure as low as 10mA is the answer. Is it possible that the differential and folded cascode running at about 200mA or more?
20mA
Mine runs on ~20mA , don´t know howmuch is it on X1000,
but from photos and device choice I can imagine it running
on 40-60 mA.
Uli
Mine runs on ~20mA , don´t know howmuch is it on X1000,
but from photos and device choice I can imagine it running
on 40-60 mA.
Uli
X1000 uses TO-247 mosfet for output device. If each has 8nF input capacitance, 40 of them will have 320nF that will have to be charge-filled to achieve at least 100khz.
In folded cascode, the driving swing is equal to the swing in the differential. How to calculate how many mA needed to drive gate capacitance of final mosfets to certain output voltage at certain Khz?
I remember someone give me this formula
I=2xPixFxCxV,
where
I=current needed in Ampere
Pi=3.14
F=Frequency Hz
C=capacitance Farad
V= voltage
I calculate, if I wanted to charge all the mosfets for their Vgs (I take V=6volt), F=100khz, C=320nF, the result is 1.2A
Am I making a mistake in this calculation? Why the result is over 1 Ampere?
In folded cascode, the driving swing is equal to the swing in the differential. How to calculate how many mA needed to drive gate capacitance of final mosfets to certain output voltage at certain Khz?
I remember someone give me this formula
I=2xPixFxCxV,
where
I=current needed in Ampere
Pi=3.14
F=Frequency Hz
C=capacitance Farad
V= voltage
I calculate, if I wanted to charge all the mosfets for their Vgs (I take V=6volt), F=100khz, C=320nF, the result is 1.2A
Am I making a mistake in this calculation? Why the result is over 1 Ampere?
- Status
- Not open for further replies.