Good for you, Elso. You might actually find more 3rd harmonic generated with the fet input, BUT it is a CLEAN 3rd, just like analog magnetic tape. The advantage of jfets is that they STAY QUIET over a range of source impedances, including both resistive and inductive. They don't require much input bias current either, SO you can leave out the input coupling cap. This is important! You can also use, with ease, MUCH more idle current on your input stage. This makes driving a comp cap easy, and potentially increases your slew rate. There is much more, but enough for now.
Re: noise of the 2SK389
In another thread, John Curl has said:
"Folks, the data sheets listed above are what the big boys use to make audio electronics. You can make either MC or MM phono inputs or even compatable phono stages for all MM or MC inputs. You usually have to adjust the input loading and the gain, however."
(http://www.diyvideo.com/forums/showthread.php?s=&postid=238209#post238209)
... refering to the 2SK389.
And on this thread:
"Neither device is 'better' than the other, for audio, yet a 2SK389 will beat both for low noise over a range of inputs."
I downloaded a copy of the Toshiba 2SK389 datasheet, and my reading of the two graphs illustrating noise (page 5) is that:
1. they are inconsistent: at the miminum quoted input resistance of 100 Ohms, the Noise figure differs on the two graphs by about 2 dB at 10 Hz and 1 kHz.
2. The noise figure of this device for typical MC loadings (<100 Ohms) is quite high. Much higher than a bipolar opamp such as AD797, which appears to indicate a ~1dB NF at 100 Ohms source resistance (I am guessing this figure is for frequencies above the 1/f corner, which is about 100 Hz for AD797).
John, on the basis of a choice by noise performance alone, would not the AD797 be a better choice for a MC input than the 2SK389?
What other factors do you consider?
Regards,
-- John
In another thread, John Curl has said:
"Folks, the data sheets listed above are what the big boys use to make audio electronics. You can make either MC or MM phono inputs or even compatable phono stages for all MM or MC inputs. You usually have to adjust the input loading and the gain, however."
(http://www.diyvideo.com/forums/showthread.php?s=&postid=238209#post238209)
... refering to the 2SK389.
And on this thread:
"Neither device is 'better' than the other, for audio, yet a 2SK389 will beat both for low noise over a range of inputs."
I downloaded a copy of the Toshiba 2SK389 datasheet, and my reading of the two graphs illustrating noise (page 5) is that:
1. they are inconsistent: at the miminum quoted input resistance of 100 Ohms, the Noise figure differs on the two graphs by about 2 dB at 10 Hz and 1 kHz.
2. The noise figure of this device for typical MC loadings (<100 Ohms) is quite high. Much higher than a bipolar opamp such as AD797, which appears to indicate a ~1dB NF at 100 Ohms source resistance (I am guessing this figure is for frequencies above the 1/f corner, which is about 100 Hz for AD797).
John, on the basis of a choice by noise performance alone, would not the AD797 be a better choice for a MC input than the 2SK389?
What other factors do you consider?
Regards,
-- John
FETs versus Bipolar Junky Transistors
John,
From the OPA604 datasheet:
"SOUND QUALITY
The sound quality of an op amp is often the crucial
selection criteria—even when a data sheet claims exceptional
distortion performance. By its nature, sound
quality is subjective. Furthermore, results of listening
tests can vary depending on application and circuit
configuration. Even experienced listeners in controlled
tests often reach different conclusions.
Many audio experts believe that the sound quality of a
high performance FET op amp is superior to that of
bipolar op amps. A possible reason for this is that
bipolar designs generate greater odd-order harmonics
than FETs. To the human ear, odd-order harmonics
have long been identified as sounding more unpleasant
than even-order harmonics. FETs, like vacuum tubes,
have a square-law I-V transfer function which is more
linear than the exponential transfer function of a bipolar
transistor. As a direct result of this square-law
characteristic, FETs produce predominantly even-order
harmonics. Figure 10 shows the transfer function
of a bipolar transistor and FET. Fourier transformation
of both transfer functions reveals the lower odd-order
harmonics of the FET amplifier stage."
So I guess you meant second order harmonics for the FET.
For the datasheet: www.ti.com
I agree on the advantage of omitting the input cap and the possibility to accomodate wildly varying source impedances with the FET. My moving coil cartridge headamp is direct coupled and balanced! I am also using direct input and output coupling in my line amp.
john curl said:
John,
From the OPA604 datasheet:
"SOUND QUALITY
The sound quality of an op amp is often the crucial
selection criteria—even when a data sheet claims exceptional
distortion performance. By its nature, sound
quality is subjective. Furthermore, results of listening
tests can vary depending on application and circuit
configuration. Even experienced listeners in controlled
tests often reach different conclusions.
Many audio experts believe that the sound quality of a
high performance FET op amp is superior to that of
bipolar op amps. A possible reason for this is that
bipolar designs generate greater odd-order harmonics
than FETs. To the human ear, odd-order harmonics
have long been identified as sounding more unpleasant
than even-order harmonics. FETs, like vacuum tubes,
have a square-law I-V transfer function which is more
linear than the exponential transfer function of a bipolar
transistor. As a direct result of this square-law
characteristic, FETs produce predominantly even-order
harmonics. Figure 10 shows the transfer function
of a bipolar transistor and FET. Fourier transformation
of both transfer functions reveals the lower odd-order
harmonics of the FET amplifier stage."
So I guess you meant second order harmonics for the FET.
For the datasheet: www.ti.com
I agree on the advantage of omitting the input cap and the possibility to accomodate wildly varying source impedances with the FET. My moving coil cartridge headamp is direct coupled and balanced! I am also using direct input and output coupling in my line amp.
I guess Mr. Curl was right..............
" I did NOT read this particular data sheet, but if I read such I statement I immediately become VERY suspicious."
Then go read the data sheets! There should be enough data to check the parameters important to your design. In designing for multiple parts sources, or for a redesign with a very similar part, a careful study of the data sheet and parameter spread should be a given. Testing with variations is even done in Spice simulation with Monte Carlo analysis. Is the part still going the work with minimum and maximum values for the parameters? Is it going to work over the temperature range in your product will be used? I spent many hours measuring circuits in an environmental chamber and watching things go to hell at very high or low temperatures. Sometimes part options are even designed in with stuffing options or alternate part values on the schematic and assembly drawings depending on the choice of the semiconductor availibility or pricing. It's the engineer's job function to be suspicious, it's your name on the document and" I didn't know" or "the data sheet didn't say" is not going to get much understanding from your boss. Been there, done that; it it was not fun.
The real gotcha is alternate parts that are supposed to be exact second sources but have some parameter poorly expressed or even undocumented in the data sheet. Sometimes "improvements" or process changes, that are not documented in the data sheet. can cause new problems not found with the older part. Ocasionally problems the manufacturer wasn't even aware of! The change from HC to AC CMOS Logic is a perfect example of this. The faster edge rates for the AC logic created signal integrity and RFI issues that led to whole classes of CMOS logic being developed to fix the "improvements" bought about by AC CMOS. This is more common than you think. When a parameter is left off of the data sheet it is often because the part is not as good in that particular area as the competition. Sometimes parameters are given but measured under different conditions. Like with contracts, it pays to read the fine print very closely. Data sheets are designed to be helpful, but are foremost designed to sell parts and present their virtues and hide the warts. Application notes and application engineers are the next stop when faced with things that are ambiguous or missing on the data sheet that you need to know for designing a part in.
Fred
" I did NOT read this particular data sheet, but if I read such I statement I immediately become VERY suspicious."
Then go read the data sheets! There should be enough data to check the parameters important to your design. In designing for multiple parts sources, or for a redesign with a very similar part, a careful study of the data sheet and parameter spread should be a given. Testing with variations is even done in Spice simulation with Monte Carlo analysis. Is the part still going the work with minimum and maximum values for the parameters? Is it going to work over the temperature range in your product will be used? I spent many hours measuring circuits in an environmental chamber and watching things go to hell at very high or low temperatures. Sometimes part options are even designed in with stuffing options or alternate part values on the schematic and assembly drawings depending on the choice of the semiconductor availibility or pricing. It's the engineer's job function to be suspicious, it's your name on the document and" I didn't know" or "the data sheet didn't say" is not going to get much understanding from your boss. Been there, done that; it it was not fun.
The real gotcha is alternate parts that are supposed to be exact second sources but have some parameter poorly expressed or even undocumented in the data sheet. Sometimes "improvements" or process changes, that are not documented in the data sheet. can cause new problems not found with the older part. Ocasionally problems the manufacturer wasn't even aware of! The change from HC to AC CMOS Logic is a perfect example of this. The faster edge rates for the AC logic created signal integrity and RFI issues that led to whole classes of CMOS logic being developed to fix the "improvements" bought about by AC CMOS. This is more common than you think. When a parameter is left off of the data sheet it is often because the part is not as good in that particular area as the competition. Sometimes parameters are given but measured under different conditions. Like with contracts, it pays to read the fine print very closely. Data sheets are designed to be helpful, but are foremost designed to sell parts and present their virtues and hide the warts. Application notes and application engineers are the next stop when faced with things that are ambiguous or missing on the data sheet that you need to know for designing a part in.
Fred
Elso, I meant FET's used in a diff pair input. The 2SK389 is a dual fet pair, and is mostly used for differential inputs. Differential input, by definition has NO second harmonic, if everything is matched. Please, everyone, think it through, before second guessing me on this topic. I have been designing at this level for more than 30 years.
I won't question the 2nd vs. 3rd harmonic issue for a diff pair, although "...if everything is matched" is perhaps the key here.
What I would like to question is your continued recommendation of the 2SK389 when said part has been dis-continued for years now TTBOMK. You might be able to get a limited number of them for a little while longer, but for how long is anybody's guess. Seems to me that designing them into anything new (even DIY projects I would go so far as to say) just isn't sensible. How about recommending an alternative? Some of the InterFET and Vishay parts look nice.
What I would like to question is your continued recommendation of the 2SK389 when said part has been dis-continued for years now TTBOMK. You might be able to get a limited number of them for a little while longer, but for how long is anybody's guess. Seems to me that designing them into anything new (even DIY projects I would go so far as to say) just isn't sensible. How about recommending an alternative? Some of the InterFET and Vishay parts look nice.
So much for thinking it through.....
"What I would like to question is your continued recommendation of the 2SK389"
I have even given sources on where to buy them in small quantities. Toshiba still list the K170 and J74 as current for the very nervous. The K389 is not difficult to find.
Interfet is a small company with a lot of focus on critical applications like instrumentation. The minimum order is $450 or $250 for a stocked standard part.
They are not geared for small quantity sales or the onslaught of free sample request from the many cheap skates who abuse the sample process elsewhere.
I don't believe Vishay or Interfet makes anything close to the Toshiba devices.
"What I would like to question is your continued recommendation of the 2SK389"
I have even given sources on where to buy them in small quantities. Toshiba still list the K170 and J74 as current for the very nervous. The K389 is not difficult to find.
Interfet is a small company with a lot of focus on critical applications like instrumentation. The minimum order is $450 or $250 for a stocked standard part.
They are not geared for small quantity sales or the onslaught of free sample request from the many cheap skates who abuse the sample process elsewhere.
I don't believe Vishay or Interfet makes anything close to the Toshiba devices.
So much for reading my post...
Don't chastise someone for not thinking it through until you've read and understood their post. I didn't say anything about not being able to find the parts. What I was getting at is that the availibility of the part is coasting on reserves of production, which ceased years ago. Finding them today is easy. How about tomorrow? Many good designs last a lot longer than the availability of the parts does. For production YMMV.
Fine parts if you want singles. And I do tend to be nervous, I've had more parts go obsolete on me on currently produced products than I care to count. After all, "it's the engineer's job function to be suspicious", and that includes part availability as well.
Maybe not close in every respect, but certainly some fine parts, and new, available, etc. Have you looked?
Don't chastise someone for not thinking it through until you've read and understood their post. I didn't say anything about not being able to find the parts. What I was getting at is that the availibility of the part is coasting on reserves of production, which ceased years ago. Finding them today is easy. How about tomorrow? Many good designs last a lot longer than the availability of the parts does. For production YMMV.
Fred Dieckmann said:
Fine parts if you want singles. And I do tend to be nervous, I've had more parts go obsolete on me on currently produced products than I care to count. After all, "it's the engineer's job function to be suspicious", and that includes part availability as well.
Fred Dieckmann said:
Maybe not close in every respect, but certainly some fine parts, and new, available, etc. Have you looked?
in all the right places
Have you looked?
Yes, I have, have you? Vishay/Siliconix highest transconductance device is about half of that the SK389. Interfet makes no direct equivalent and is not geared for small quantities. I have designed with jfets for about 20 years. I have heard the difference when Jeff Rowland (pre chip amp days) updated his amps with the Toshibas, it wasn't subtle. I can tell you that plenty of people design with the Toshibas and one would assume they plan to stay in business. The single devices (K170 and J74) are fairly inexpensive and jfets are easy to match if one chose to go that route.
In a hobby where people routinely design around 50 year old NOS vacuum tubes, I can't see the concerns about using one of the most popular jfets for audio. Have you had trouble finding them or you just trying have some fun with us?
Fred
Sorry for the focus but the pic is the inside of a fairly current Marantz home theater amp using a Toshiba jfet pair. One would assume Marantz buys fairy large numbers of them........
Have you looked?
Yes, I have, have you? Vishay/Siliconix highest transconductance device is about half of that the SK389. Interfet makes no direct equivalent and is not geared for small quantities. I have designed with jfets for about 20 years. I have heard the difference when Jeff Rowland (pre chip amp days) updated his amps with the Toshibas, it wasn't subtle. I can tell you that plenty of people design with the Toshibas and one would assume they plan to stay in business. The single devices (K170 and J74) are fairly inexpensive and jfets are easy to match if one chose to go that route.
In a hobby where people routinely design around 50 year old NOS vacuum tubes, I can't see the concerns about using one of the most popular jfets for audio. Have you had trouble finding them or you just trying have some fun with us?
Fred
Sorry for the focus but the pic is the inside of a fairly current Marantz home theater amp using a Toshiba jfet pair. One would assume Marantz buys fairy large numbers of them........
Attachments
Phred,
I have this vision of you as some audio equivalent of James Bond, surreptitiously sneaking around audio shows and hifi stores in the dead of night, taking internal pics with a tiny spy camera.
You do seem to have an unusually large and diverse collection of naked commercial electronics pics
Andy.
I have this vision of you as some audio equivalent of James Bond, surreptitiously sneaking around audio shows and hifi stores in the dead of night, taking internal pics with a tiny spy camera.
You do seem to have an unusually large and diverse collection of naked commercial electronics pics
Andy.
'Photographs, ay', he asked him knowlingly?
A friend let me help babysit his audio store for a week. It was slow and I got board. His sales guy in training ratted me out though. Jocko and I used to take our own screw drivers to stores and one of us would distract the salesman while the took the tops off. Open houses are better since the marketing guys show the stuff without the tops so Jocko and I can look at
the stuff and shake our heads no while snikering. Makes the marketing guys nervous as hell. Jocko looked in a preamp once and told the marketing guy " It looks like a capacitor farm!" I had to go out to the parking lot to laugh out loud so I didn't get thrown out.
Snap snap, grin grin, wink wink, nudge nudge, say no more?
Fred
A friend let me help babysit his audio store for a week. It was slow and I got board. His sales guy in training ratted me out though. Jocko and I used to take our own screw drivers to stores and one of us would distract the salesman while the took the tops off. Open houses are better since the marketing guys show the stuff without the tops so Jocko and I can look at
the stuff and shake our heads no while snikering. Makes the marketing guys nervous as hell. Jocko looked in a preamp once and told the marketing guy " It looks like a capacitor farm!" I had to go out to the parking lot to laugh out loud so I didn't get thrown out.
Snap snap, grin grin, wink wink, nudge nudge, say no more?
Fred
Attachments
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.