I saw a picture of you and your wife on the web, everyone should be so lucky. You could charge a lot more for that consulting than mere audio.
You're not listening. This is about all we are talking about. Using what we can get now.
Last edited:
2SJ74 devices are difficult to get. Yes, there are a few around here. However, WE find them almost impossible to get in quantity, and we would like to avoid them in less critical designs, JUST LIKE THE JC-3. Why do you think that we used IC's? However, there is a 'loophole' that I would like to explore. This is: What about JUST N channel Toshiba or equivalent devices? They are quiet, still available and not too expensive, YET! This could have made the JC-3 even better. That is why I am attempting to discuss topologies like you use in IC design, but in discrete.
Hard to get 2Sj74? Nope, easy. In DIY quantaties.
Here's my take on a "VendettaRevA" in Retin style. SNR is 75 dB rel 500uVrms 20-20kHz unweighted. Good enough, IMO. Gain is 64 dB. I remade the
DCservos a bit and I use OPA627 as DC servo OPamp.
I use Zetex ZVP/ZVN3310 at a failry high idle current instead of the original MOSFETs.
The 2nd stage in the first amp-stage is now made outof BJT:S running at around 20mA.
Here's my take on a "VendettaRevA" in Retin style. SNR is 75 dB rel 500uVrms 20-20kHz unweighted. Good enough, IMO. Gain is 64 dB. I remade the
DCservos a bit and I use OPA627 as DC servo OPamp.
I use Zetex ZVP/ZVN3310 at a failry high idle current instead of the original MOSFETs.
The 2nd stage in the first amp-stage is now made outof BJT:S running at around 20mA.
Attachments
Good point, John. I have advocated the use of N-channel-only phono input stages all along, for both MC and MM designs. The LSK389 is a good part.
Cheers,
Bob
It is not necessary to take the servo out and put in a coupling capacitor to achieve some LF rolloff to deal with warp. A properly-designed DC servo can be made to provide the desired first-order LF frequency response cut. Just remember that the DC servo is really in the audio path, and should be constructed with audio-grade components. It is part of the whole system. This is especially true when it is being used as a filtering function in the subsonic audio range.
Significantly, you can design the DC servo with an additional pole in its feedback path to achieve a second-order subsonic filter response if desired.
Cheers,
Bob
John,
For a slight shift in direction. In the past I have mentioned when using IC's I prefer to not use the same voltage for both rails. I take it in the JC3 both rails run at the same voltage?
My other new trick is in RC shunt filters. Instead of having the R go to signal common, I find you can use 2R x 2 with one going to each rail. Same filter action different resistor result. Of course it can be "voiced" with an additional wye resistor to signal common.
Now of course everyone knows there is no difference in those changes as long as the resistors sum to the same value, but of course there is a real difference!
ES
For a slight shift in direction. In the past I have mentioned when using IC's I prefer to not use the same voltage for both rails. I take it in the JC3 both rails run at the same voltage?
My other new trick is in RC shunt filters. Instead of having the R go to signal common, I find you can use 2R x 2 with one going to each rail. Same filter action different resistor result. Of course it can be "voiced" with an additional wye resistor to signal common.
Now of course everyone knows there is no difference in those changes as long as the resistors sum to the same value, but of course there is a real difference!
ES
How about voice in the desert.
Now the bypass caps are in the signal path as well as you have removed all the PSRR of your buffers/whatever, so what have you gained? Maybe an oppurtunity to spend more money.How about voice in the desert.
Yes, you do lose PSRR, but there is something you gain and Scott you got this correct before!
Yes, you can put in a say 10n cap before the 1000k resistor and get a 1st order roll-off at around 15 Hz.
No worries. Thinking back it would have been earlier than that, 02 maybe as I moved in 03. Mine was an RTP3. I'll likely be building another soon, so I'll go with his SSM suggestion.
OK, IngemarR, have it your way. I have plenty of P channel jfets, Charles Hansen has many more, Constellation has many thousands, and Parasound uses 100's every month. Sometimes I have to bail them out by selling them a few hundred. However, it is difficult to OEM P channel 2SJ74's, as I try virtually every week, and audiophiles tend to 'cry out' when something increases in price.
Personally, I already KNOW how to design a complementary jfet ultra low noise input in both single ended and differental topology. I work, at this time, on 3 different separate designs that will someday reach the marketplace. However, for the rest here, it can be difficult to make something comparable to a Vendetta with available products. Matching is necessary, and that takes a lot of parts. High current is necessary, because the folded cascode will not work properly without out high quiescent current. That is WHY I select only the top half of the V series for most designs, or parallel twice as many, and get into non-linear capacitance territory. NOW, where are your sources for P channel fets?
What Scott tends to quibble about, such as leakage, or leakage match with P's or nonlinear capacitance in general, I studied and experimented with, 38 years ago, when I designed the Levinson JC-2. I found, even then, that the advantage of the complementary jfet topology overcame some limitations in complementary characteristics for most applications. Bringing it up here, is just to muddle the water.
Personally, I already KNOW how to design a complementary jfet ultra low noise input in both single ended and differental topology. I work, at this time, on 3 different separate designs that will someday reach the marketplace. However, for the rest here, it can be difficult to make something comparable to a Vendetta with available products. Matching is necessary, and that takes a lot of parts. High current is necessary, because the folded cascode will not work properly without out high quiescent current. That is WHY I select only the top half of the V series for most designs, or parallel twice as many, and get into non-linear capacitance territory. NOW, where are your sources for P channel fets?
What Scott tends to quibble about, such as leakage, or leakage match with P's or nonlinear capacitance in general, I studied and experimented with, 38 years ago, when I designed the Levinson JC-2. I found, even then, that the advantage of the complementary jfet topology overcame some limitations in complementary characteristics for most applications. Bringing it up here, is just to muddle the water.
It would be a pity if your Vendetta went into the grave without some DIYers having the possibility to listen to it (their own Vendetta version)!
Finding V-grade is another ball game. Those I have no surce for.
If I show you my sources, I guess you will buy all the JFET:s...
Here you can find 400 pcs 2SJ74.
B&D Enterprises - Electronic Components Distributor
In Germany there is a source that sell 2Sj74 for 2 Euro each.
Now, I am ALSO interested in single ended input MC RIAA amplifiers. Please do not think otherwise. BF862 is another very interesting Nch JFET that has no Pch counterpart.
So, I will not show any more Vendetta versions. Let's focus on Nch only designs.
I am also very interesteed in PS which seems to have been forgotten all together...
Finding V-grade is another ball game. Those I have no surce for.
If I show you my sources, I guess you will buy all the JFET:s...
Here you can find 400 pcs 2SJ74.
B&D Enterprises - Electronic Components Distributor
In Germany there is a source that sell 2Sj74 for 2 Euro each.
Now, I am ALSO interested in single ended input MC RIAA amplifiers. Please do not think otherwise. BF862 is another very interesting Nch JFET that has no Pch counterpart.
So, I will not show any more Vendetta versions. Let's focus on Nch only designs.
I am also very interesteed in PS which seems to have been forgotten all together...
Ingemar. did you build your version of Vendetta phono preamp? Obviously, Mr. Curl does not wish to see any Vendetta clone. He wishes Vendetta to stay status symbol of selected few. He underestimates Mr. Borbely designs as compromised, but they are only with different design approaches compared to Mr. Curl. Kit designs are often consider inferior since are aimed for amateurs to be assembled. There are no reasons why they should be worse than factory made units. Mr. Borbely used folded cascode in the stage of his eb-804/419 ALL-jFET MC/MM phono preamp which I built myself ,including PCB with two ounce copper. It is very similar to Vendetta , so similar that I tried second stage Vendetta on the same PCB with only slight modification. Reading Mr. Curl posts between lines, I chose 15 ma fets in dual diff. stage. Instead of second stage Mosfets, i paralleled two 8ma fets ( X8 and X9 in your diagram)- it is C grade version. RIAA toplogy and values are the same. I put a small heatsink on output J-fets. Borbely circuit uses source follower at the output. Vendetta output changes open loop gain with the various line preamp impedances. All in all, original Borbely circuit is slightly more dynamic in sound reproduction.
There is no need to reinvent the wheel with SE phono stages using only N fets. Pass Aleph Xono is a almost perfect No-feedback MC SE gain stage. HPS 5.1 is also good circuit using BF862 JFets. Pass Lab latest phono preamp XP-25 is a complementary design. They obviously changed their design approach. At the same time they are constructed with SMD components.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.