GRollins
You have highlighted my point succinctly. You cannot build an Aleph-X with only one FET, you need at least two to create the symmetry. Now think about what happens when there is a 20mV mismatch between those two FETs and compare that to the same variance spread over a bank of eight FETs.
Nelson Pass
Thank you for illuminating the "darkness". About five years ago I posted my thoughts on current vs. voltage drive amplification schemes on this very forum. I had only my ears and physics to guide me but even then, I suspected that voltage only approximates the signal to the moving coil. Once accounting for time and deflection issues, the higher (ie: faster) the voltage, the better the approximation of the signal. The current on the other hand, IS the signal on the coil. Mauro Penesa and you with your F4 have provided working examples which give further credence to these notions. I am fairly well schooled in many things but I suspect my ignorance of electronics is still evident so please forgive me in advance if my post seems impertinent to you. As I stated, I built my Aleph-X on the hifizen boards. I don't suspect the Aleph-X circuit or part selection as much as my implementation. My power supply is very large, very quiet and in a separate enclosure so I suspect the board and or my connector layout is the reason I am disappointed in its performance. I see the potential for improvement on the input parts selection but implementing them on that board would not be a "drop-in" for me. There were nearly 1000 hifizen boards sold if I recall correctly. If you are feeling generous this holiday season, I suspect nothing would bring more happiness here than a few hints on how to optimize the design for that particular board.
You have highlighted my point succinctly. You cannot build an Aleph-X with only one FET, you need at least two to create the symmetry. Now think about what happens when there is a 20mV mismatch between those two FETs and compare that to the same variance spread over a bank of eight FETs.
Nelson Pass
Thank you for illuminating the "darkness". About five years ago I posted my thoughts on current vs. voltage drive amplification schemes on this very forum. I had only my ears and physics to guide me but even then, I suspected that voltage only approximates the signal to the moving coil. Once accounting for time and deflection issues, the higher (ie: faster) the voltage, the better the approximation of the signal. The current on the other hand, IS the signal on the coil. Mauro Penesa and you with your F4 have provided working examples which give further credence to these notions. I am fairly well schooled in many things but I suspect my ignorance of electronics is still evident so please forgive me in advance if my post seems impertinent to you. As I stated, I built my Aleph-X on the hifizen boards. I don't suspect the Aleph-X circuit or part selection as much as my implementation. My power supply is very large, very quiet and in a separate enclosure so I suspect the board and or my connector layout is the reason I am disappointed in its performance. I see the potential for improvement on the input parts selection but implementing them on that board would not be a "drop-in" for me. There were nearly 1000 hifizen boards sold if I recall correctly. If you are feeling generous this holiday season, I suspect nothing would bring more happiness here than a few hints on how to optimize the design for that particular board.
Nelson,
Agreed. I was assuming that he wasn't going to hack the PCB to retrofit JFETs, bipolars, or whatever. That would make it an apples-to-apples comparison if he were to try a different set of output devices. If he's willing to revamp the front end, then options abound.
nania,
The idea that a smaller number of output devices somehow leads to relaxed matching requirements is, well, let's just say it's unorthodox. It's always important. Given that the boys (I have two year-old twins) decided to be party animals all last night, everyone from my neck of the woods is a bit short on sleep and in danger of being grumpy, so I'll leave it at that. If you feel that you can sell others on your matching philosophy then by all means have at it, but I don't know how many takers you'll find.
Regarding the type of output devices, I'll relate a story that I've told before:
The ever-generous MikeW sent me an Aleph 3 to play with. (Thanks Mike.) I've got a quad-amped system and as Mike and I had discussed, I put the 3 on the tweeters and sat down to listen. A short while later, I started scratching my head. The right and left channels didn't sound the same. More particularly, the high frequencies didn't sound the same. After a bit more head scratching, I started investigating and found that the output devices on one channel were different from the other. I rummaged around and put some others in so as to have the same thing on both sides and everything came back together.
All the factors you mention can work either for or against the sound, but I think you underrate the importance of parts selection--MOSFETs, in this case--to the overall sound of the amp. I realize that you say you put a lot of time into selecting the MOSFETs you put into the amp, but I can't help feeling that there's a touch of Beranek's Law at work, here.
You might take a moment to compare the IRFP240 (which I believe is the part Nelson uses most in his output stages) to yours. He chose it, in part, for overall balance, although obviously the voltage and current ratings, heat dissipation and so forth all played a part. A "bigger" MOSFET tends, in the same circuit, to tilt the sound towards the bottom end. A "smaller" one tends to tilt it towards the top. You went for a big-ish one and the result you report is consistent with that choice.
It's your amp, your system, and your money. You're free to do what you wish. But I'm getting the impression that you've already closed off some options that you might benefit from reconsidering.
Grey
P.S.: The reason I can get away with choosing a heavyweight MOSFET like the '044 for the Aleph-X is that I'm not running the amp full range. As a result, I don't have to worry about the full range balance, only the sound from 5kHz up. In a full range application, the amp might very well require re-voicing.
Agreed. I was assuming that he wasn't going to hack the PCB to retrofit JFETs, bipolars, or whatever. That would make it an apples-to-apples comparison if he were to try a different set of output devices. If he's willing to revamp the front end, then options abound.
nania,
The idea that a smaller number of output devices somehow leads to relaxed matching requirements is, well, let's just say it's unorthodox. It's always important. Given that the boys (I have two year-old twins) decided to be party animals all last night, everyone from my neck of the woods is a bit short on sleep and in danger of being grumpy, so I'll leave it at that. If you feel that you can sell others on your matching philosophy then by all means have at it, but I don't know how many takers you'll find.
Regarding the type of output devices, I'll relate a story that I've told before:
The ever-generous MikeW sent me an Aleph 3 to play with. (Thanks Mike.) I've got a quad-amped system and as Mike and I had discussed, I put the 3 on the tweeters and sat down to listen. A short while later, I started scratching my head. The right and left channels didn't sound the same. More particularly, the high frequencies didn't sound the same. After a bit more head scratching, I started investigating and found that the output devices on one channel were different from the other. I rummaged around and put some others in so as to have the same thing on both sides and everything came back together.
All the factors you mention can work either for or against the sound, but I think you underrate the importance of parts selection--MOSFETs, in this case--to the overall sound of the amp. I realize that you say you put a lot of time into selecting the MOSFETs you put into the amp, but I can't help feeling that there's a touch of Beranek's Law at work, here.
You might take a moment to compare the IRFP240 (which I believe is the part Nelson uses most in his output stages) to yours. He chose it, in part, for overall balance, although obviously the voltage and current ratings, heat dissipation and so forth all played a part. A "bigger" MOSFET tends, in the same circuit, to tilt the sound towards the bottom end. A "smaller" one tends to tilt it towards the top. You went for a big-ish one and the result you report is consistent with that choice.
It's your amp, your system, and your money. You're free to do what you wish. But I'm getting the impression that you've already closed off some options that you might benefit from reconsidering.
Grey
P.S.: The reason I can get away with choosing a heavyweight MOSFET like the '044 for the Aleph-X is that I'm not running the amp full range. As a result, I don't have to worry about the full range balance, only the sound from 5kHz up. In a full range application, the amp might very well require re-voicing.
Hi Grey,
Interesting last couple of posts.
What Mosfets do you like to use in the upper range, the mids, the lower? I've, too, have noticed that fets have their own flavor.
Interesting last couple of posts.
What Mosfets do you like to use in the upper range, the mids, the lower? I've, too, have noticed that fets have their own flavor.
GRollins
Please point out how I said this? This seemed to be your assertion and my view was diammetrical. Be that as it may, the 044 part is a 50V part so running it at a lower rail (ie:15V) will still operate it within the linear part of its output profile (ie: less distortion). Running a higher volt rated part like the 240 that low, will not produce equivalent results and that is why the standard recommendation was for 25V rails. I could not economically find the lower volt rated FETs for my implementation but given a choice between the 240 part and the 150 for my 19V rails, the 150 seemed to be a better choice. Looking at the profile curves, I do not see any potential benefit to using the 240 FET but if I am wrong, please help me to understand the fault in my reasoning.The idea that a smaller number of output devices somehow leads to relaxed matching requirements is, well, let's just say it's unorthodox.
Funny you should ask...
The Aleph 3 thing took me totally by surprise. Once I figured out that the output devices were different on the right and left channels, I assumed that the lower current (hence lower capacitance) one would be the better of the two. I knew which channel I liked better and it was a simple matter to...what?...what's this?...the bigger one was the one that sounded better?
I repeat: This does not apply to amps used full range! (For some reason, people seem hell-bound to confuse the balance question with the quality of the highs taken on their own merits, or lack of same.)
This lead me to more head scratching. It wasn't a simple single variable matter. The general pattern is that the higher capacitance devices are that way because they pass more current; there's more Gate surface area on the die and it's just like you'd expect if if you made a capacitor out of Saran Wrap and aluminum foil...the more surface area, the more capacitance. More capacitance, all things being equal, means slew rate falls and distortion increases. As a consequence of controlling more current you have higher transconductance. More transconductance means more gain, which means more negative feedback assuming that the closed loop gain remains constant.
More feedback means...what?
More distortion means...what?
To what degree do they cancel each other out?
So was I falling prey to the dreaded first impression of "more information" that seduces so many people over to the Dark Side in a higher feedback amp, or was I being drawn in by distortion components that added things to the highs that shouldn't be there?
Or did it sound better because of some other factor?
Somebody fetch me a drink...like right now!
It's not as easy to settle questions like that in solid state as it is in tubes (for what it's worth, 6550s are better at bass and 6L6s are better at mids and highs). It's nothing to build a tube circuit that's got a knob on the front that allows you to crank the feedback up and down. A lot of solid state circuits go nuts if you try that. If nothing else, you can get some pretty serious DC offset going if you're not careful. I had a set of experiments in mind, but...
Complicate this with the fact that MOSFET manufacturing processes seem to be changing, which in turn is having a deleterious effect on their..."flatness," for want of a better term. This being the high frequency problem that Nelson identified a couple of years ago, such that even nominally same devices might not sound the same.
To escape this, I went over to the Fairchild devices (which Nelson says are now falling prey to the same changes). The devices I used for the GR-25 are more-or-less equivalent to the IRFP240/IRFP9240, but so far those are the only Fairchilds I've used, due to time constraints. That means I can't recommend specific Fairchild devices, either, at least until I get beyond a universe of one each, N and P parts.
The long and short of it is that I don't have a good answer for you, although I can pose some pretty good questions.
For full range use, which is what I gather nania is doing, you're in a bind because you want to voice the circuit such that the lows, mids, and highs are balanced. Yes, you can see some of this on meters and oscilloscopes, but not all of it. In the end it's an ear judgment sort of thing.
Grey
The Aleph 3 thing took me totally by surprise. Once I figured out that the output devices were different on the right and left channels, I assumed that the lower current (hence lower capacitance) one would be the better of the two. I knew which channel I liked better and it was a simple matter to...what?...what's this?...the bigger one was the one that sounded better?
I repeat: This does not apply to amps used full range! (For some reason, people seem hell-bound to confuse the balance question with the quality of the highs taken on their own merits, or lack of same.)
This lead me to more head scratching. It wasn't a simple single variable matter. The general pattern is that the higher capacitance devices are that way because they pass more current; there's more Gate surface area on the die and it's just like you'd expect if if you made a capacitor out of Saran Wrap and aluminum foil...the more surface area, the more capacitance. More capacitance, all things being equal, means slew rate falls and distortion increases. As a consequence of controlling more current you have higher transconductance. More transconductance means more gain, which means more negative feedback assuming that the closed loop gain remains constant.
More feedback means...what?
More distortion means...what?
To what degree do they cancel each other out?
So was I falling prey to the dreaded first impression of "more information" that seduces so many people over to the Dark Side in a higher feedback amp, or was I being drawn in by distortion components that added things to the highs that shouldn't be there?
Or did it sound better because of some other factor?
Somebody fetch me a drink...like right now!
It's not as easy to settle questions like that in solid state as it is in tubes (for what it's worth, 6550s are better at bass and 6L6s are better at mids and highs). It's nothing to build a tube circuit that's got a knob on the front that allows you to crank the feedback up and down. A lot of solid state circuits go nuts if you try that. If nothing else, you can get some pretty serious DC offset going if you're not careful. I had a set of experiments in mind, but...
Complicate this with the fact that MOSFET manufacturing processes seem to be changing, which in turn is having a deleterious effect on their..."flatness," for want of a better term. This being the high frequency problem that Nelson identified a couple of years ago, such that even nominally same devices might not sound the same.
To escape this, I went over to the Fairchild devices (which Nelson says are now falling prey to the same changes). The devices I used for the GR-25 are more-or-less equivalent to the IRFP240/IRFP9240, but so far those are the only Fairchilds I've used, due to time constraints. That means I can't recommend specific Fairchild devices, either, at least until I get beyond a universe of one each, N and P parts.
The long and short of it is that I don't have a good answer for you, although I can pose some pretty good questions.
For full range use, which is what I gather nania is doing, you're in a bind because you want to voice the circuit such that the lows, mids, and highs are balanced. Yes, you can see some of this on meters and oscilloscopes, but not all of it. In the end it's an ear judgment sort of thing.
Grey
nania said:GRollins Please point out how I said this? This seemed to be your assertion and my view was diammetrical. Be that as it may, the 044 part is a 50V part so running it at a lower rail (ie:15V) will still operate it within the linear part of its output profile (ie: less distortion). Running a higher volt rated part like the 240 that low, will not produce equivalent results and that is why the standard recommendation was for 25V rails. I could not economically find the lower volt rated FETs for my implementation but given a choice between the 240 part and the 150 for my 19V rails, the 150 seemed to be a better choice. Looking at the profile curves, I do not see any potential benefit to using the 240 FET but if I am wrong, please help me to understand the fault in my reasoning.
My fault, I said that backwards (I'm at work and I write things between doing stuff here--I get called away, lose my train of thought, and don't say what I meant to say). My objection still stands, though.
Think of a bank of output devices as a choir. The more matched they are, they more they will sing together. The more mismatched they are, the more dissonant they will be. Having a larger number of mismatched voices does not mean that they will somehow average out to being on pitch.
As for running an '044 at 15V being lower distortion than at 50V...it's quite the opposite, for any of a half-dozen reasons.
For instance, well, there are curves and then there are curves. The question isn't so much one of distortion (although distortion factors into it) as it is of slew rates and frequency and volume-dependent distortions. The closest you can come to this is the graph showing Gate capacitance against voltage, and even that kinda misses the point. The thing you have to remember is that IRF, et. al., don't make these parts to be used in audio equipment. They're marketed to people who use them as switches. They don't care about the needs, wants, and desires of the audio crowd because we're less than the proverbial drop in the bucket in terms of their sales volume. Yes, Toshiba and Hitachi make audio-specific MOSFETs, but they're expensive and hard to get here in the States. And no, they're not drop-in replacements in Nelson's amps, although you could rework some of the circuits to accept them; some more easily than others.
Anyway, back to MOSFETs...the charts and graphs aren't as useful as you might wish. Not for us, anyway. To obsess over one part vs. another because a ruler held against the curve shows a smidgen straighter line is a waste of time. The unit-to-unit variations are wide enough that you should only look at the graphs through squinted eyes, as a rough hint of what's going on.
Here's a particular case to consider. Start with the Gate capacitance/Drain voltage graph. Now, buy a couple hundred parts, select one that happens to have that precise curve, put it in a simple common Source circuit, turn the circuit on and...gee, it does exactly what you thought it would do. Whoopee! Declare feasting and celebration. Punch your time card and go home, patting yourself on the back for a job well done.
Come back the next morning, turn the amp on, and play music through it.
Ooops.
Big Oooooooops!
What's wrong?
Well...
Go back to the graph. It shows the variation of the Gate capacitance vs. the voltage at the Drain. When you selected an operating point, a 'rail' voltage, a load, and so forth, you were thinking DC. Once the music starts, things change and it ain't pretty. The voltage at the Drain changes. As a result, the capacitance at the Gate changes in real time with the music. With no signal, the Gate rests quietly and all is well. With a rapidly varying AC signal, the Gate capacitance becomes a real drag, because you have to charge and discharge varying (large) amounts of capacitance or face the consequences. And as if that wasn't bad enough, it's volume dependent!
Oh, fer cryin' out loud! Can't a guy get a break?
Well, yes...and no...
There are things you can do to speed up that charge/discharge cycle, but they have consequences of their own, sometimes obvious, sometimes subtle. The easiest thing to do is to increase the amount of current available to drive the Gate. (This gets back to why I was saying that JFETs [low current devices] aren't a full solution to the problem. It's also what Nelson was alluding to when he mentioned paralleling devices. [Which offsets the low current available from one JFET by paralleling two, three, five...whatever you think is needed to get the job done. But then--you guessed it--you need to match the JFETs...there's no end to it. When all is said and done, you chose a set of compromises and live with it.] Enough with the nested parenthetical stuff...) But in the classic Aleph or Aleph-X circuit, you're looking at a MOSFET driving a resistive load for the front end, which means that the current is actively driven one way, but only passively drained the other. Which means that the drive is asymmetrical to some extent. Which increases distortion. Which means that...
Oh drat...I've gone cross-eyed...
Don't get overly worried about that part. It's just an example of the myriad factors that you have to juggle. Scads and scads of solid state and tube pieces over the years have used precisely that sort of circuit to produce beautiful music with low distortion. There are tricks and techniques to use Judo-esque techniques to make things work out.
The take-away message is that you need to think in terms of charging and discharging that Gate capacitance so that you're not asking the feedback loop to fix the errors that tend to accumulate. The more you ask of your negative feedback, the more it will intrude on the music.
Out of time. Need to earn my keep.
Grey
GRollins
I said this? I think it may be better to not address you in the future. You're a creative fellow and that is evident in the way you have interpreted my posts.As for running an '044 at 15V being lower distortion than at 50V...it's quite the opposite, for any of a half-dozen reasons.
I think the simpler path will be to recommend that you read some books on electronics before you attempt to interpret any more graphs.
Grey
Grey
nania said:GRollins
I said this? I think it may be better to not address you in the future. You're a creative fellow and that is evident in the way you have interpreted my posts.
Be that as it may, the 044 part is a 50V part so running it at a lower rail (ie:15V) will still operate it within the linear part of its output profile (ie: less distortion).
Bane2871
If you follow the context of the previous posts and complete the entire quote, you will extract this:
You can run a lower rated part like the IRFP150N with lower voltage rails than a higher voltage rated part like the IRFP240 and still be in the linear operating part of the device. The 240 part gets wonkier below 25V rails than the 150 or 044 part. I never said that higher voltage rails would cause more distortion and that is why I feel the replying poster that presumed to instruct me should heed his own advice. The replying poser...er..., I mean poster 🙂 was looking for ways to assert himself in some sort of ridiculous pecking order. I'm not into that game, I'm just here to learn.
If you follow the context of the previous posts and complete the entire quote, you will extract this:
You can run a lower rated part like the IRFP150N with lower voltage rails than a higher voltage rated part like the IRFP240 and still be in the linear operating part of the device. The 240 part gets wonkier below 25V rails than the 150 or 044 part. I never said that higher voltage rails would cause more distortion and that is why I feel the replying poster that presumed to instruct me should heed his own advice. The replying poser...er..., I mean poster 🙂 was looking for ways to assert himself in some sort of ridiculous pecking order. I'm not into that game, I'm just here to learn.
nania, I'm glad you're here to learn.
BTW:
Grey's one of our forum favorites. He deserves a great deal of respect.
BTW:
Grey's one of our forum favorites. He deserves a great deal of respect.
carpenter
I am aware of the contribution made by GRollins. I probably would not have this Aleph-X amp if it were not for him. It was a brilliant idea to conceive it and it is probably the spark that inspired so many great audio designers to come to this forum and take notice but looking at it from my point of view, I had two successive posts misrepresented in direct replies filled with mire and obfuscation. If one looks at it cynically, you could determine that the replies were designed to reverse my stated positions and make me look like a jackass. I responded to the first reply cordially but I was annoyed at the second one because little apparent care of my meaning was displayed so I politely put forward the best way I could think of to avoid being annoyed in the future. I didn't think it warranted a public slap in return. As you may have noticed, I am not of the ilk that turns the other cheek.
I am aware of the contribution made by GRollins. I probably would not have this Aleph-X amp if it were not for him. It was a brilliant idea to conceive it and it is probably the spark that inspired so many great audio designers to come to this forum and take notice but looking at it from my point of view, I had two successive posts misrepresented in direct replies filled with mire and obfuscation. If one looks at it cynically, you could determine that the replies were designed to reverse my stated positions and make me look like a jackass. I responded to the first reply cordially but I was annoyed at the second one because little apparent care of my meaning was displayed so I politely put forward the best way I could think of to avoid being annoyed in the future. I didn't think it warranted a public slap in return. As you may have noticed, I am not of the ilk that turns the other cheek.
nania said:As you may have noticed, I am not of the ilk that turns the other cheek.
I've noticed.
The two Ian's........
What have I done now?
Have not spoken to the other Ian (McMillian) for a bit.....been busy for work and excursions to places like Japan.
Interesting discussion.
Nania,
I have read your posts and can see what you are saying.
Probably the best recommendation I can make is to read Nelson's excellent articles on testing and matching mosfets over at the Pass Diy web pages.
In the beautifully written article(s) Nelson discusses the various characteristics of the devices and the distortion mechanisms. The graphs in the articles are particularly telling of the relationships between bias current and Drain- Source voltage and distortion products.
There are up sides and down sides to using both high and low transconductance devices and their voltage limitations. More particularly it comes down to the power output and voltage rails of your chosen design and the heatsink real estate you have available.
Nelson makes a point of using at least 25 volts on the devices which as much overall bias as heatsink real estate allows and he mentioned this on numerous occassions. For diy where numbers and commerical specifications are not so important the operating conditions are a bit more relaxed.
Using parrellel Jfets on the front end is an interesting offshoot of the X Aleph. But more than anything I think setting up the current share of the CCS correctly and obtaining the right amount of NF has a bigger bearing on how these amps perform from the get go.
Ian
What have I done now?
Have not spoken to the other Ian (McMillian) for a bit.....been busy for work and excursions to places like Japan.
Interesting discussion.
Nania,
I have read your posts and can see what you are saying.
Probably the best recommendation I can make is to read Nelson's excellent articles on testing and matching mosfets over at the Pass Diy web pages.
In the beautifully written article(s) Nelson discusses the various characteristics of the devices and the distortion mechanisms. The graphs in the articles are particularly telling of the relationships between bias current and Drain- Source voltage and distortion products.
There are up sides and down sides to using both high and low transconductance devices and their voltage limitations. More particularly it comes down to the power output and voltage rails of your chosen design and the heatsink real estate you have available.
Nelson makes a point of using at least 25 volts on the devices which as much overall bias as heatsink real estate allows and he mentioned this on numerous occassions. For diy where numbers and commerical specifications are not so important the operating conditions are a bit more relaxed.
Using parrellel Jfets on the front end is an interesting offshoot of the X Aleph. But more than anything I think setting up the current share of the CCS correctly and obtaining the right amount of NF has a bigger bearing on how these amps perform from the get go.
Ian
macka
Good to read you in this thread again. I have read those articles several times and they have been instructive but new parts are made every day and new opportunities present themselves. These new parts might offer potential improvement.
To be more specific about what I am aiming for; my upper mids and highs are smeared (the soundstage sort of collapses in the middle) and there is a lack of 'juicyness'. Sorry but that is the only way I can describe it. My speakers are Vandersteen V3s and they dip down to 2.6 ohm in the low frequencies. My rails measure +/-19V so I think I'm delivering enough current. I have 4 amps of bias (250 milliohm source resistors) and my current share ratio is set high (60/40). Moving it to 50/50 doesn't make a notable change so I don't think my speakers are current starved. The 150's case measure within 2 degrees of each other. I have been looking around for alternatives that might work themselves into the hifizen Aleph-X board front. I would like to keep the 150's on it because they're matched fairly well but I think the 9610 pair might have better alternatives. Have you looked at the Philips BS108/208 family of small FETs. The noise numbers look pretty good to me and I was hoping they might be a more bang for the buck solution than the JFET. Do you have any suggestions on how to work these in?
Good to read you in this thread again. I have read those articles several times and they have been instructive but new parts are made every day and new opportunities present themselves. These new parts might offer potential improvement.
To be more specific about what I am aiming for; my upper mids and highs are smeared (the soundstage sort of collapses in the middle) and there is a lack of 'juicyness'. Sorry but that is the only way I can describe it. My speakers are Vandersteen V3s and they dip down to 2.6 ohm in the low frequencies. My rails measure +/-19V so I think I'm delivering enough current. I have 4 amps of bias (250 milliohm source resistors) and my current share ratio is set high (60/40). Moving it to 50/50 doesn't make a notable change so I don't think my speakers are current starved. The 150's case measure within 2 degrees of each other. I have been looking around for alternatives that might work themselves into the hifizen Aleph-X board front. I would like to keep the 150's on it because they're matched fairly well but I think the 9610 pair might have better alternatives. Have you looked at the Philips BS108/208 family of small FETs. The noise numbers look pretty good to me and I was hoping they might be a more bang for the buck solution than the JFET. Do you have any suggestions on how to work these in?
nania said:
...new parts are made every day and new opportunities present themselves. These new parts might offer potential improvement.
To be more specific about what I am aiming for; my upper mids and highs are smeared (the soundstage sort of collapses in the middle) and there is a lack of 'juicyness'.
The IRF parts are far from new.
I've suggested a course of action to remedy your problem. You have decided--based on less than stellar reasoning--that my suggestion is bogus. Apparently, you are determined to consider everything under the sun except my suggestion. That's cool. It's your amp and you can do what you want with it.
Reminds me of someone I once worked with. This person was pathologically suspicious of everyone, always convinced that people were trying to "get one over on her." Anything and everything that was said to her was immediately disregarded...with prejudice. Why? Because she was utterly, morally convinced that people were trying to take advantage of her in one way or another. In consequence, (and assuming that everyone was speaking truth...which we were...no one was trying to harm her in any way) the truthful things that were said to her were disregarded as lies and she then set about trying to find out what their motives really were. The problem is that, once you have discarded the truth, all that remains is un-truth. Sad to say, she then took this to mean that she had, in fact, "caught" people in lying to her. They weren't. It was an unavoidable consequence of the way she approached life. (If you find this confusing and frustrating, don't worry, you're not alone, so did we.)
The booger of it all was that she was really a pretty neat person if you could overlook that one personality flaw. Bummer. Tragic bummer.
Okay...
With the disclaimer that there are hundreds of things that influence the sound that you hear, and all of them contribute to the sound in greater or lesser ways, I think you have discarded the single greatest culprit in your high end problem.
Again...it's your choice, your amp, your system, and your ears.
The (word which is rendered as *s here, but which means female dog) of the situation is that even if you were to attempt my suggestion, I am getting the impression that you'd rather experience Death By Fire Ants (shudder!) than admit it might have helped.
Merry Christmas to you in spite of yourself, Mr. Scrooge.
Grey
nania said:macka
Good to read you in this thread again. I have read those articles several times and they have been instructive but new parts are made every day and new opportunities present themselves. These new parts might offer potential improvement.
To be more specific about what I am aiming for; my upper mids and highs are smeared (the soundstage sort of collapses in the middle) and there is a lack of 'juicyness'. Sorry but that is the only way I can describe it. My speakers are Vandersteen V3s and they dip down to 2.6 ohm in the low frequencies. My rails measure +/-19V so I think I'm delivering enough current. I have 4 amps of bias (250 milliohm source resistors) and my current share ratio is set high (60/40). Moving it to 50/50 doesn't make a notable change so I don't think my speakers are current starved. The 150's case measure within 2 degrees of each other. I have been looking around for alternatives that might work themselves into the hifizen Aleph-X board front. I would like to keep the 150's on it because they're matched fairly well but I think the 9610 pair might have better alternatives. Have you looked at the Philips BS108/208 family of small FETs. The noise numbers look pretty good to me and I was hoping they might be a more bang for the buck solution than the JFET. Do you have any suggestions on how to work these in?
Okay we need apic on the amp interior and your schematic details of the set up.
A few thinks are in order.
Wiring from the drive board to the fets is important and can add significant capacitance unless care is taken.
When setting the CCS share run the amp for 1 hour with 16 volts across 8 ohms ar 60 hertz using a balanced source.
Then disconnect the current share resister / series cap and measure ac millvolts on the lower active side and compare with the a/c millvolts on the outout sense resisters.
Calculate the a/c current and the proportion.
Re connect the resister/capacitor and measure again and adjust until the lower side is 50% of the current share of the sense resisters.
The X Aleph is not a lush and romantic sounding amp at the best of times. The traditional Aleph is and you may be better to look to the Aleph 2
The speakers could also be on the analytical side so bear that in mind
macka
I would like the accuracy if the soundstage was sharp and deep but it isn't and that is what is irritating. The lower to mid frequencies place well in the space. It just feels out of focus if you know what I mean.
I haven't been able to put pics up yet but I will provide an adequate physical description. I mentioned earlier that I ran the stock Aleph-X board, two IRFP150N per side soldered directly on the boards and mounted against the heatsink with no additional wire. The two heatsinks form the lateral sides of the box and the whole enclosure is aluminum except for the faceplate, which is birchwood backed with coppersheet. The enclosure interior holds only the 280mF capacitor banks standing on the bottom and the two populated hifizen boards directly above it. The amp is juiced by an external enclosure which holds a 60 amp knockdown EI home transformer, the rectification network and the CR part of the CRC supply. A nine foot grounded wire connects the two boxes and the amp is dead silent when idling unattenuated. The amp works well but for what I invested in time and money I expected more. One more thing, shouldn't this discussion be in the builders thread? I wanted to discuss the possibility of building the front end with the Philips BS208 part instead of the IRF9610 here. It seems tighter (less leakage) and has half the capacitance of the 9610 which should help bring better bandwidth if implemented properly.
I would like the accuracy if the soundstage was sharp and deep but it isn't and that is what is irritating. The lower to mid frequencies place well in the space. It just feels out of focus if you know what I mean.
I haven't been able to put pics up yet but I will provide an adequate physical description. I mentioned earlier that I ran the stock Aleph-X board, two IRFP150N per side soldered directly on the boards and mounted against the heatsink with no additional wire. The two heatsinks form the lateral sides of the box and the whole enclosure is aluminum except for the faceplate, which is birchwood backed with coppersheet. The enclosure interior holds only the 280mF capacitor banks standing on the bottom and the two populated hifizen boards directly above it. The amp is juiced by an external enclosure which holds a 60 amp knockdown EI home transformer, the rectification network and the CR part of the CRC supply. A nine foot grounded wire connects the two boxes and the amp is dead silent when idling unattenuated. The amp works well but for what I invested in time and money I expected more. One more thing, shouldn't this discussion be in the builders thread? I wanted to discuss the possibility of building the front end with the Philips BS208 part instead of the IRF9610 here. It seems tighter (less leakage) and has half the capacitance of the 9610 which should help bring better bandwidth if implemented properly.
- Home
- Amplifiers
- Pass Labs
- The Aleph-X