nelsonvandel,
I have a suggestion for you to try. Get rid of the buffer and replace it with a complementary output pair biased with a Vbe multiplier (or diode string) attached to the output side of the folded cascode.
Sometimes simple is better.
Jam
I have a suggestion for you to try. Get rid of the buffer and replace it with a complementary output pair biased with a Vbe multiplier (or diode string) attached to the output side of the folded cascode.
Sometimes simple is better.
Jam
Since I'm a neutrality freak, the brightness of this amp has been bugging me. Since I've found SA970/SC2240 to be on the bright side of neutral in other amps, I gave BC559 from Philips/NXP a shot again as cascode transistors.
The difference is remarkable. It's far more neutral now - even sligthly on the soft side. The transistors are just out of the box, but I don't really expect any major difference after burning in. It just sounds good now, very clean and neutral, but I'd like it a bit more aggressive.
Careful selection of active components seems to be as important as anything else.
I wonder what active current sources on the LTP would do to the sound. Can I use a simple current mirror there? It simulates OK.
The difference is remarkable. It's far more neutral now - even sligthly on the soft side. The transistors are just out of the box, but I don't really expect any major difference after burning in. It just sounds good now, very clean and neutral, but I'd like it a bit more aggressive.
Careful selection of active components seems to be as important as anything else.
I wonder what active current sources on the LTP would do to the sound. Can I use a simple current mirror there? It simulates OK.
Hi Nelson
You seem very hard to please, can take ages before you get the exact sound you want. Earlier i gave you a little advice on compensation with this amp of yours, you should try it.
For the jfets, until last year I had a problem using jfets, Ill discribe it as similar to yours, the sound that is. John Curl gave some good advice, jfets like to be biased hard, up close to Idss, watch the power. This will take away the graininess, but dont think it will be good for batteries.
You seem very hard to please, can take ages before you get the exact sound you want. Earlier i gave you a little advice on compensation with this amp of yours, you should try it.
For the jfets, until last year I had a problem using jfets, Ill discribe it as similar to yours, the sound that is. John Curl gave some good advice, jfets like to be biased hard, up close to Idss, watch the power. This will take away the graininess, but dont think it will be good for batteries.
homemodder said:
OK, thanks for the advice again. Sure, I'm hard to please. Aren't we all? I'll try to crank up the currents to about 4 mA and see what it does to the sound. I use BL grade, and as I remember it, their Idss is something like 6 or 7 mA. But after changing SA970 for BC559 most of the grain is gone, and there's very little of left "the JFET sound". I'm not really sure I could point it out in a blind test.
I just changed from SA970 to BC559, used as VAS in my JLH like amp, resulting in a smoother and more natural sound in this amp as well. I don't think I like those Toshiba transistors, or maybe I'm using them in a suboptimal way.
> or maybe I'm using them in a suboptimal way.
Yes, sir.
And if you want the most out of Toshiba JFETs, I would consider 4mA bias absolute minimum. Under 3mA they also get very noisy, as stated in the datasheet.
Patrick
Yes, sir.
And if you want the most out of Toshiba JFETs, I would consider 4mA bias absolute minimum. Under 3mA they also get very noisy, as stated in the datasheet.
Patrick
Most bjt, bipolar, lownoise transistors
like BC549, BC550. BC559, BC560, 2SC2240, 2SA970
works as 'a best compromise' at around 1 mA.
-----------------------------------------------------------------
for lowest noise we can use 0.15 - 0.30 mA
allround compromise is often 0.6 - 1.5 mA range.
for more gain and speed we can use 2.5 - 5.0 mA
-----------------------------------------------------------------
whatever choice, we can not have it all
lower noise --- less gain & not as high speed
higher gain --- higher noise
compromise --- not lowest noise, not best gain/speed
do not use BC549/550/559/560 above 5 mA
if you need more than 5 mA,
then 2SC2240/2SA970 or 2N5551/2N5401 is better choice
2N5551 / 2N5401, if a bit higher current + higher voltage supply
( because even if 2SC2240 / 2SA970 can take 120 Volt
they do not take much Power = Current x Voltage )
> 2N5551/5401 are often rated at least 0.625 Watt
> while 2SC2240, 2SA970 are only 0.2 - 0.3 Watt
> BC549,550,559,560 are in the middle, with 0.5 Watt,
which is also the most normal for one TO-92 plastic capsule
---------------------------------------------------------------------
About JFET bias level.
As said, most cases use them rather high, by John Curl recommendation.
For one 6 mA IDSS 2SK170BL
I would go for 4.5 - 5.0 mA = 75-80%
Nelson Pass in some special amplifiers, use them as buffers
at 100 % .... 6mA for one 6 mA IDSS !!!!
But one LTP differential JFET input is NOT one Buffer follower.
like BC549, BC550. BC559, BC560, 2SC2240, 2SA970
works as 'a best compromise' at around 1 mA.
-----------------------------------------------------------------
for lowest noise we can use 0.15 - 0.30 mA
allround compromise is often 0.6 - 1.5 mA range.
for more gain and speed we can use 2.5 - 5.0 mA
-----------------------------------------------------------------
whatever choice, we can not have it all
lower noise --- less gain & not as high speed
higher gain --- higher noise
compromise --- not lowest noise, not best gain/speed
do not use BC549/550/559/560 above 5 mA
if you need more than 5 mA,
then 2SC2240/2SA970 or 2N5551/2N5401 is better choice
2N5551 / 2N5401, if a bit higher current + higher voltage supply
( because even if 2SC2240 / 2SA970 can take 120 Volt
they do not take much Power = Current x Voltage )
> 2N5551/5401 are often rated at least 0.625 Watt
> while 2SC2240, 2SA970 are only 0.2 - 0.3 Watt
> BC549,550,559,560 are in the middle, with 0.5 Watt,
which is also the most normal for one TO-92 plastic capsule
---------------------------------------------------------------------
About JFET bias level.
As said, most cases use them rather high, by John Curl recommendation.
For one 6 mA IDSS 2SK170BL
I would go for 4.5 - 5.0 mA = 75-80%
Nelson Pass in some special amplifiers, use them as buffers
at 100 % .... 6mA for one 6 mA IDSS !!!!
But one LTP differential JFET input is NOT one Buffer follower.
I tend to disagree. 2mA is fine for the SK170/SJ74.
Looking at the datasheets, TOS even specifies many noise graphs with Id = 1mA and I cannot understand why they would use a current that puts their JFETs in a bad view.
Looking at the datasheets also show that 2mA is a fine curretn noise wise.
Sigurd
Looking at the datasheets, TOS even specifies many noise graphs with Id = 1mA and I cannot understand why they would use a current that puts their JFETs in a bad view.
Looking at the datasheets also show that 2mA is a fine curretn noise wise.
Sigurd
EUVL said:
You might want to use another pot if your sonical goals are very high. I think you use an ALPS pot and I feel them grainy.
How about trying a stepped pot instead with some nice resistors?
Regarding ressitors, what types are you using?
You can try Sanyo 2SC2362K/2SA1016K instead of 2SC2270/SA970 - they are excellent BJTs.
Sigurd
How about trying a stepped pot instead with some nice resistors?
Regarding ressitors, what types are you using?
You can try Sanyo 2SC2362K/2SA1016K instead of 2SC2270/SA970 - they are excellent BJTs.
Sigurd
nelsonvandal said:
Thanks for all the replies.
I use SK170 as diff stage in my JLH like amp at 1.23 mA with 82R degeneration resistors. It has the lowest noise level I've ever experienced. Far less than any monolithic opamp I've tried. The folded cascode is noisier. I think part of it is due to the lowish voltage drop for the cascode bias. This is a compromise because of battery operation/portability. I'm going to try to bypass the degeneration resistors and see if lowers the noise. I do my noise tests with 16 Ohm in ear monitors, not by measurement.
Lineup, what will happen to a "too high" biased bipolar transistor? More distortion? Is the result audible, and in what way?
The "JLH amp" is a bit grainy, so I'll try your suggestions regarding currents in this amp as well.
Where do I find those Sanyo transistors?
I'm using Firstronics 0.6 W metal film resistors. RN55 specified. That's the only one I can get at ELFA.
I use SK170 as diff stage in my JLH like amp at 1.23 mA with 82R degeneration resistors. It has the lowest noise level I've ever experienced. Far less than any monolithic opamp I've tried. The folded cascode is noisier. I think part of it is due to the lowish voltage drop for the cascode bias. This is a compromise because of battery operation/portability. I'm going to try to bypass the degeneration resistors and see if lowers the noise. I do my noise tests with 16 Ohm in ear monitors, not by measurement.
Lineup, what will happen to a "too high" biased bipolar transistor? More distortion? Is the result audible, and in what way?
The "JLH amp" is a bit grainy, so I'll try your suggestions regarding currents in this amp as well.
Where do I find those Sanyo transistors?
I'm using Firstronics 0.6 W metal film resistors. RN55 specified. That's the only one I can get at ELFA.
Sigurd Ruschkow said:
With 2mA one will get something like En=0,9 nV/sqrtHz, operate them at Idss and you will get En=0,85 nV/sqrtHz...so we have more then 90% of the noise performance at 2mA, right?
Hi Sigurd
Are the 2sa1016k trannies the ones you purchased from Profusion?? So you got to try them. I told you they were great, out perform even the 2sa970 and sound better too, rather more like Jfets, softer sound. Main reason I prefer to use all Sanyo trannies. The 2sa872 from Hitachi also very good.
With regards to the Jfets I used Curls advice and he is spot on, the Jfets do sound better when highly biased whether in buffer or LTP, I dont even take into account the noise, try it. I am waiting for a order of Sanyo Jfets 2sk222 to see how these perform.
Nelson, looking at the datasheets I would say the 2sa970 would be better at higher voltage and at around 2 to 3ma. Its about using the transistors in their most Linear reagion. What happens with smaller degeneration resistors?. What earphones do you use, must be good ones to hear every minute change in sound. I also use headphones HD600s to evaluate the sound.
Lineup those 2n5551 tranies must be the worst Ive ever used, even the BC types are better, dont just take my word for it ask PMA who did some reallife testing. they dont sound good to me.
Are the 2sa1016k trannies the ones you purchased from Profusion?? So you got to try them. I told you they were great, out perform even the 2sa970 and sound better too, rather more like Jfets, softer sound. Main reason I prefer to use all Sanyo trannies. The 2sa872 from Hitachi also very good.
With regards to the Jfets I used Curls advice and he is spot on, the Jfets do sound better when highly biased whether in buffer or LTP, I dont even take into account the noise, try it. I am waiting for a order of Sanyo Jfets 2sk222 to see how these perform.
Nelson, looking at the datasheets I would say the 2sa970 would be better at higher voltage and at around 2 to 3ma. Its about using the transistors in their most Linear reagion. What happens with smaller degeneration resistors?. What earphones do you use, must be good ones to hear every minute change in sound. I also use headphones HD600s to evaluate the sound.
Lineup those 2n5551 tranies must be the worst Ive ever used, even the BC types are better, dont just take my word for it ask PMA who did some reallife testing. they dont sound good to me.
You can get the Sanyos from http://www.profusionplc.com/
Send me a PM, and I'll send you a few.
Regarding resistors, I highly recommend you to change the ones you are using into something like the red PRP:s which you cna get
for little money at PartsConnexxion.
Sigurd
Send me a PM, and I'll send you a few.
Regarding resistors, I highly recommend you to change the ones you are using into something like the red PRP:s which you cna get
for little money at PartsConnexxion.
Sigurd
nelsonvandal said:
PM sent.
Sigurd, are those Sanyo's good as VAS?
I use Sennheiser HD650. They are quite colored with a dip in the 3k region and a slight peak just after this dip making them sound like a horn speaker. The sound is very clean and fast and it's not too hard to hear colorations from amplifiers.
I never thought transistor would have that much impact, but maybe I should give them a try.
Sigurd, are those Sanyo's good as VAS?
I use Sennheiser HD650. They are quite colored with a dip in the 3k region and a slight peak just after this dip making them sound like a horn speaker. The sound is very clean and fast and it's not too hard to hear colorations from amplifiers.
I never thought transistor would have that much impact, but maybe I should give them a try.
Well, the 2SA1016/SC2362 are maybe not the best VAS BJT around. Depending on current needed, Sanyo makes some superb VAS transistors but are hard to get these days as most of them are made for CRT:s, and CRT:s are scarce these days...
2SA1407/2sc3601 or 2SA1406/2SC3600 are some of the best and in this league the 2SA1360/2SC3423 also is.
More on VAS at
http://www.diyaudio.com/forums/showthread.php?s=&threadid=124142&perpage=25&highlight=&pagenumber=2
I also have a pair of the HD650 and they are made to be kinda audiophile style with a falling freq curve.
You might try one of the 3rd party cables for it.
Sigurd
2SA1407/2sc3601 or 2SA1406/2SC3600 are some of the best and in this league the 2SA1360/2SC3423 also is.
More on VAS at
http://www.diyaudio.com/forums/showthread.php?s=&threadid=124142&perpage=25&highlight=&pagenumber=2
I also have a pair of the HD650 and they are made to be kinda audiophile style with a falling freq curve.
You might try one of the 3rd party cables for it.
Sigurd
nelsonvandal said:
Sigurd Ruschkow said:
you disagree with mr Curl, sigge 😉
good to know .. i wont be too far off
if using 2SK170GR at 1-2 mA
this was also what I got from datasheets .. until I adopted Juhn Curl thinking
nice we have different persons' experiences to lean on
and I know for a fact you sigge have used many JFET
and I trust your high fidelity ambitions & audio findings
You mention Noise
I do not often design my headphone / pre amplifiers regading noise
... just because I think it is one very overrated parameter
higher gain & pusg & drive for each transistor is for me more important
as i use as few stages, few transistors as possible .. a bit of Curl philosophy
in this matter, you can say
... and also Nelson Pass type of amplifiers
He exactly is not the one that bothers about a bit of MOSFET or JFET npise, you know
but more about linearity and gain
he uses current a lot .. in JFET and in his HOT Class A HEXFETs
--------------------------
bottomline:
if you design for low noise .. use lower current
if you design for more gain, less devices/less stages .. use as much current you can
without getting too hot and and overly stress your transistors
Lineup Audio - noisy gain audio lab
😎
I am not sure what John Curl has said about the Sk170/SJ74 and their noise vs Id performance,
but my comment was on what EUVL said about the noise performance in the datasheets. As aparatusonitus points out, running the JFETs at 2mA gives 90% of the eq short circuit rms nose, and I would personally worry more about the noise spread within JFETs than 10%. Espcially for a headphone amp....
If I recall correctly, John Curl's super low noise friend only uses noise-selected GR grade Sk389 JFETs.
For optimal sonics, there are different views on what Id to use. JC likes - as far as I know - the V grade, and runs them hard.
Borbely runs his input stage JFETs often at 2mA. Other stages need more current. Ouput stage maybe 10mA.
A german guy found that 5mA was the optimal Id for the SK389 in a RIAA inputstage.
Personally, I run them at about 2-8mA for input stage work.
10-15mA for output or driver stages.
Sigurd
but my comment was on what EUVL said about the noise performance in the datasheets. As aparatusonitus points out, running the JFETs at 2mA gives 90% of the eq short circuit rms nose, and I would personally worry more about the noise spread within JFETs than 10%. Espcially for a headphone amp....
If I recall correctly, John Curl's super low noise friend only uses noise-selected GR grade Sk389 JFETs.
For optimal sonics, there are different views on what Id to use. JC likes - as far as I know - the V grade, and runs them hard.
Borbely runs his input stage JFETs often at 2mA. Other stages need more current. Ouput stage maybe 10mA.
A german guy found that 5mA was the optimal Id for the SK389 in a RIAA inputstage.
Personally, I run them at about 2-8mA for input stage work.
10-15mA for output or driver stages.
Sigurd
lineup said:
Hi homemodder,
yep, they are from Profusion. Thanks for the tip! I bought 500 of each type. They will actually replace the 2SA872AE/2SC1775AE which I have a hard time to source. I do have some but need something else which is still in production. This time, the 2SA1016K will be used as cascodes, but I look forward to use them in a LTP.
When you say biased harder, I guess you mean higher % of Idss?
What kind of difference in sonics did you find when biasing the SK170/SJ74 harder?
The 2sk222 do looks good - I just wish there were a Pch for it.
I would be very interested in your performance evaluation.
Sigurd
yep, they are from Profusion. Thanks for the tip! I bought 500 of each type. They will actually replace the 2SA872AE/2SC1775AE which I have a hard time to source. I do have some but need something else which is still in production. This time, the 2SA1016K will be used as cascodes, but I look forward to use them in a LTP.
When you say biased harder, I guess you mean higher % of Idss?
What kind of difference in sonics did you find when biasing the SK170/SJ74 harder?
The 2sk222 do looks good - I just wish there were a Pch for it.
I would be very interested in your performance evaluation.
Sigurd
homemodder said:
Sigurd,
I found that a higher bias gets rid of graininess and also in a particular amp that i tried this the voices became more real life, total absence of siblence. This could be for other reasons too, but Ill take Curls judgement about this, he has a lot of experience with Jfets.
It was a battle to get just 10 pcs 2sk222, like 15 years ago when I started using 2sc1016k, I had to pull a lot of strings. Im very new to using Jfets in amps, I think I will try send some to Scott Wurcer to test as he has equipment and better understanding of parameters for Jfets. I will only comment on their sound compaired to Toshiba. Parts should arrive in a week or two. Real Pity about P parts.
Dont know if you have heard that here in Portugal succesful Jfets have been produced using paper. I have sent email to the electrical department of the univeristy asking for more details and whether N and P parts can be manufactured. I like the way Jfets sounds, I think a lot of us are hoping for some new matched P and N parts to replace Toshiba.
I found that a higher bias gets rid of graininess and also in a particular amp that i tried this the voices became more real life, total absence of siblence. This could be for other reasons too, but Ill take Curls judgement about this, he has a lot of experience with Jfets.
It was a battle to get just 10 pcs 2sk222, like 15 years ago when I started using 2sc1016k, I had to pull a lot of strings. Im very new to using Jfets in amps, I think I will try send some to Scott Wurcer to test as he has equipment and better understanding of parameters for Jfets. I will only comment on their sound compaired to Toshiba. Parts should arrive in a week or two. Real Pity about P parts.
Dont know if you have heard that here in Portugal succesful Jfets have been produced using paper. I have sent email to the electrical department of the univeristy asking for more details and whether N and P parts can be manufactured. I like the way Jfets sounds, I think a lot of us are hoping for some new matched P and N parts to replace Toshiba.
Thank you again for all replies. I'll see what those Sanyo transistors can do to the sound, and maybe try some better resistors.
Homemodder,
yes, I read about the portugesian paper JFETs. Not sure they are made for audio, but interesting anyway!
Scott W measured some other very interesting JFETs and those were the SMD parts BF861 (or BF862). I have a bunch of those but no Pch for them either.
The sonics of the 2SK222 vs Sk170 would be very interesting - and that not only for me.
Sigurd
yes, I read about the portugesian paper JFETs. Not sure they are made for audio, but interesting anyway!
Scott W measured some other very interesting JFETs and those were the SMD parts BF861 (or BF862). I have a bunch of those but no Pch for them either.
The sonics of the 2SK222 vs Sk170 would be very interesting - and that not only for me.
Sigurd
homemodder said:
- Status
- Not open for further replies.