A first timer's response to Mr. Pass's steps.....
Originally posted by Nelson Pass
My list of the most effective techniques for getting the best sound out of an Aleph:
1) measure the linearity of each output transistor in addition to matching and use the most linear
I may be able to figure this one out after some reading.
2) bias the hell out of it, with overkill heat sinking
If this refers to the R19 mod I'd like to try it.
3) substitute in combinations of diff pair devices, picking the best performing ones in terms of distortion, noise, and sound
Alittle bit puzzled but on the list for more reading.
4) a good pi filter on the supply
Should be within my grasp.
5) power transformer two feet away
Too impractical for me.
6) filter cap on reference Zener
Hmmmmm....
7) a glass of fine Cabernet
A definite must do!
8) better caps
I'll try and ask more questions here.
9) better connectors and wire
Cardas connectors and pure silver wire?
10) better resistors
1% metal film.
Kind of neat having an "expert" respond to my novice level query.
Regards,
Dan
Originally posted by Nelson Pass
My list of the most effective techniques for getting the best sound out of an Aleph:
1) measure the linearity of each output transistor in addition to matching and use the most linear
I may be able to figure this one out after some reading.
2) bias the hell out of it, with overkill heat sinking
If this refers to the R19 mod I'd like to try it.
3) substitute in combinations of diff pair devices, picking the best performing ones in terms of distortion, noise, and sound
Alittle bit puzzled but on the list for more reading.
4) a good pi filter on the supply
Should be within my grasp.
5) power transformer two feet away
Too impractical for me.
6) filter cap on reference Zener
Hmmmmm....
7) a glass of fine Cabernet
A definite must do!
8) better caps
I'll try and ask more questions here.
9) better connectors and wire
Cardas connectors and pure silver wire?
10) better resistors
1% metal film.
Kind of neat having an "expert" respond to my novice level query.
Regards,
Dan
<B>Questions…</B>
1. How much worse is a ferrite-cored inductor than the air-cored one?
2. Small built-in EMI filters are applicable for high current Aleph by its specification. Are they really so? (I have tried once in a different circuit, and seemed they are not)
3. How much better is putting such filter after the transformer from doing so before the transformer? (Not the same component values of course)
4. With advanced measurement apparatus, do you see any signal problems by running a 20cm cables from MOSFET to PCB? (Cannot with my ears)
5. With advanced measurement apparatus, do you find a problem using wire-wound 5W Source resistors? (Even so, I don’t believe this would be a bottleneck)
6. Has anybody tried or noticed the effect of offsetting front-end amplification (including feedback) ground level with a resistor?
7. Why is a good transformer not in Mr. Pass list? With that effort to separate the transformers and that oversize is not a problem, I think “E” transformer is preferable than the toroid.
8. What is a good manufacturer for the MOSFET? I used a trial-and-error (beside Vgs matching) to find the “magical” pairs. In a single setup, IR is the best, in paralleled MOSFET, I found that IR is not the best. What I observed was that these MOSFET have <B>lower</B> Vgs than IR MOSFET. Any comment?
9. With Cabernet, did Mr. Pass really talk about wine? Number 7, hmmm….
<B>Thanks…</B>
1. How much worse is a ferrite-cored inductor than the air-cored one?
2. Small built-in EMI filters are applicable for high current Aleph by its specification. Are they really so? (I have tried once in a different circuit, and seemed they are not)
3. How much better is putting such filter after the transformer from doing so before the transformer? (Not the same component values of course)
4. With advanced measurement apparatus, do you see any signal problems by running a 20cm cables from MOSFET to PCB? (Cannot with my ears)
5. With advanced measurement apparatus, do you find a problem using wire-wound 5W Source resistors? (Even so, I don’t believe this would be a bottleneck)
6. Has anybody tried or noticed the effect of offsetting front-end amplification (including feedback) ground level with a resistor?
7. Why is a good transformer not in Mr. Pass list? With that effort to separate the transformers and that oversize is not a problem, I think “E” transformer is preferable than the toroid.
8. What is a good manufacturer for the MOSFET? I used a trial-and-error (beside Vgs matching) to find the “magical” pairs. In a single setup, IR is the best, in paralleled MOSFET, I found that IR is not the best. What I observed was that these MOSFET have <B>lower</B> Vgs than IR MOSFET. Any comment?
9. With Cabernet, did Mr. Pass really talk about wine? Number 7, hmmm….
<B>Thanks…</B>
Re: How do you measure linearity of mosfets ?
Measuring linearity: Measure mosfets at several different currents and choose the pairs that match the best after collecting that data. For example, some may match well at low currents but not at high currents. If substituting another FET makes a better match at both points, you have a pair!
Sure placing the supply transformer farther away would help. It would reduce noise radiated into the low voltage circuitry.
Maybe NP or someone can comment on the likely effectiveness of DIY shielding around the transformer?
vinay said:
Measuring linearity: Measure mosfets at several different currents and choose the pairs that match the best after collecting that data. For example, some may match well at low currents but not at high currents. If substituting another FET makes a better match at both points, you have a pair!
Sure placing the supply transformer farther away would help. It would reduce noise radiated into the low voltage circuitry.
Maybe NP or someone can comment on the likely effectiveness of DIY shielding around the transformer?
Jay said:
IMO the best inductor is an EI inductor though you will need a big one for 3 Amps say a 10 mH. this is way better than other types, provided is well design.
And yes I would also go for EI PSU trafo, less induced noise and more tolerant of DC present on the line.
If you want to prize Nelson for his excellent designs send him a bottle or two of fine Cabernet wine.
As you can see I'm an EI fan!!!
Sandy H.
Regarding trafo in a seperate box, readhttp://www.diyaudio.com/forums/showthread.php?threadid=7903&perpage=15&pagenumber=2 .
Regards
Paul
Regarding trafo in a seperate box, readhttp://www.diyaudio.com/forums/showthread.php?threadid=7903&perpage=15&pagenumber=2 .
Regards
Paul
Hello All, this is by far one of the most intruiging forums in a while on DIYAUDIO.
I have been matching FET's for days with this box below based on the popular test method of Mr. Pass and his Good friend Dr. Thogard (did I get the spelling right?). I log the Data at different times and after different on and off periods. I have noted that strange similarities appear across different groups of FET's.
Groups of FET's will change in test results over periods of time sometimes up and sometimes down. Most of the time it will be 2 or 3 doing the same thing, and this from FET's all the same date code.
So now I have groups of FET's that are matched and sorted into sub groups of like "Linearity". My concern is how do I know which readings to rate them as in terms of volts. It would almost seem better to classify them using A,B,C. Where you do not mix B family with A Family.
To put into numbers what I mean here is one example.
a FET measure 3.163 volts initially, after 30 minutes it measured 3.143 volts, after being off for an hour it came back on at 3.202 volts and gradually dropped to 3.170 volts. When I came back to lable it the next day and measured it for "one last time" it read 2.995 volts!!!
I thought it an irregularity and put it aside and built a Device to test many FET's at one time. Well wouldn't you know it they all do the same thing regardless of type of FET or datecode!
It was shortly after that I began to document predictable behaviour and could even tell what the FET was going to do over time (sometimes)
I checked my circuits for errors and my component values, everything is working within specified parameters.
Any feedback? Is this usefull or useless?
Anthony
I have been matching FET's for days with this box below based on the popular test method of Mr. Pass and his Good friend Dr. Thogard (did I get the spelling right?). I log the Data at different times and after different on and off periods. I have noted that strange similarities appear across different groups of FET's.
Groups of FET's will change in test results over periods of time sometimes up and sometimes down. Most of the time it will be 2 or 3 doing the same thing, and this from FET's all the same date code.
So now I have groups of FET's that are matched and sorted into sub groups of like "Linearity". My concern is how do I know which readings to rate them as in terms of volts. It would almost seem better to classify them using A,B,C. Where you do not mix B family with A Family.
To put into numbers what I mean here is one example.
a FET measure 3.163 volts initially, after 30 minutes it measured 3.143 volts, after being off for an hour it came back on at 3.202 volts and gradually dropped to 3.170 volts. When I came back to lable it the next day and measured it for "one last time" it read 2.995 volts!!!
I thought it an irregularity and put it aside and built a Device to test many FET's at one time. Well wouldn't you know it they all do the same thing regardless of type of FET or datecode!
It was shortly after that I began to document predictable behaviour and could even tell what the FET was going to do over time (sometimes)
I checked my circuits for errors and my component values, everything is working within specified parameters.
Any feedback? Is this usefull or useless?
Anthony
An externally hosted image should be here but it was not working when we last tested it.
Regarding the linearity test, a good way is to hook it into a Zen type circuit using a resistor for the bias from the + supply and no load. (the resistor forms the load). Give it a sine wave at 100, 1k, and 10 KHz and test the distortion at a consistent level halfway between 0 and clipping. Select out the lowest distortion pieces.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.