Last night I changed the R8 to a 820R resistor. This seems a good compromise between the overbloom of 1K and the slight hissness of 660R.
I repeat - try a pot to set this to your own taste.
I'am now toying with the idea of doing an ultimate version. Up the voltage to 250V. Substitute a 12B4. Replace the top Mosfet with another PCC88. Add reed relays for switching. Might be good.
Shoog
I repeat - try a pot to set this to your own taste.
I'am now toying with the idea of doing an ultimate version. Up the voltage to 250V. Substitute a 12B4. Replace the top Mosfet with another PCC88. Add reed relays for switching. Might be good.
Shoog
Hi Shoog
Sounds interesting, that ultimate version. I am curious
Just one more question, also to andypairo!!
What is the dissipation for the IRF in this design? Lets take the original one...I assume it is about half the rail voltage (~80VDC) multiplied with the current through the tube (and that is...10mA?). that would make for a dissipation of about 0,8W. The BS107 (costs 6 cents, and is easy to get) would be to small, as its maximum dissipation is 350mW...even at 5mA of current ot would still be too small. The IRFD220 is rated at 1W, that would be brely sufficient for 80V/10mA...and probably I will be using 100V of rail supply. I will contact the guy from who I always buy and see what he has got in his assortiment.
Many thanks for reading
Erik
Sounds interesting, that ultimate version. I am curious
Just one more question, also to andypairo!!
What is the dissipation for the IRF in this design? Lets take the original one...I assume it is about half the rail voltage (~80VDC) multiplied with the current through the tube (and that is...10mA?). that would make for a dissipation of about 0,8W. The BS107 (costs 6 cents, and is easy to get) would be to small, as its maximum dissipation is 350mW...even at 5mA of current ot would still be too small. The IRFD220 is rated at 1W, that would be brely sufficient for 80V/10mA...and probably I will be using 100V of rail supply. I will contact the guy from who I always buy and see what he has got in his assortiment.
Many thanks for reading
Erik
Hi Shoog
Thanks, many thanks... And of course I was wrong. Let me, at least, try to explain...there must always be a drop across the plate, with the use of a resistor that is half the B+. WIthout thinking much I applied this logic to your circuit...but off course this is wrong, with an active load this drop can be quiet small...
The BS107 is rated 350mW, 7X the 50mW quiescent current. I think I will order some...
Erik
Thanks, many thanks... And of course I was wrong. Let me, at least, try to explain...there must always be a drop across the plate, with the use of a resistor that is half the B+. WIthout thinking much I applied this logic to your circuit...but off course this is wrong, with an active load this drop can be quiet small...
The BS107 is rated 350mW, 7X the 50mW quiescent current. I think I will order some...
Erik
The function of the Mosfet is to test the output signal, invert it and apply the opposite voltage to the plate of the output valve. In this way the voltage across the output tube is kept constant in relation to the fixed current supplied by the CCS. Hence it is a Constant Voltage Constant Current buffer. This makes the output tube behave in a more linear way hence the name Super Linear Cathode Follower.
That is my understanding of the principle on which this circuit works.
If you want to know how to substitute a valve for the Mosfet, I suggested you visit the Vacuum State website and look at the original FVP5 circuit(s).
Shoog
That is my understanding of the principle on which this circuit works.
If you want to know how to substitute a valve for the Mosfet, I suggested you visit the Vacuum State website and look at the original FVP5 circuit(s).
Shoog
Important newsflash.*****
I just realised there is a mistake in the original schematic. The final filter stage of the schematic is shared in my preamp, whereas in the schematic it is seperated. This means that R8 is half of its required value. ie, 820R should be 1K6. Alternatively you can share the final stage between both channels.
Hope that saves some hassle.
Shoog
I just realised there is a mistake in the original schematic. The final filter stage of the schematic is shared in my preamp, whereas in the schematic it is seperated. This means that R8 is half of its required value. ie, 820R should be 1K6. Alternatively you can share the final stage between both channels.
Hope that saves some hassle.
Shoog
I took note of peoples comments about there possably been a problem with the Mosfets gate capacitance. With the Mosfet we are using, this is about 100pf. This will effectively mean that the buffer will pass slightly more high frequencies than low.
I have done a little experiement to try to compensate. I placed a 100pf mica cap in parallel with the resistor to ground between the two stages. This should roll off the top end by a similar amount to the boost recieved by the gate capacitance. Hopefully this will flatten the response slightly. Can't say that I can hear much difference though.
Shoog
I have done a little experiement to try to compensate. I placed a 100pf mica cap in parallel with the resistor to ground between the two stages. This should roll off the top end by a similar amount to the boost recieved by the gate capacitance. Hopefully this will flatten the response slightly. Can't say that I can hear much difference though.
Shoog
Lowering the gain without changing the tube
Hi Shoog and all,
I was thinking about a way to have moderate gain from this amp without changing tubes and it came to my mind that a fellow Italian diyer, Giaime, had made use of some feedback from the plate to lower gain in a E88cc based preamp.
His circuits can be found here:
http://giaime.altervista.org/plate_follower.html
What do you think about such approach?
Cheers
Andrea
Hi Shoog and all,
I was thinking about a way to have moderate gain from this amp without changing tubes and it came to my mind that a fellow Italian diyer, Giaime, had made use of some feedback from the plate to lower gain in a E88cc based preamp.
His circuits can be found here:
http://giaime.altervista.org/plate_follower.html
What do you think about such approach?
Cheers
Andrea
I don't see why it wouldn't work but I don't think its the best approach.
12B4 are very linear and extremely cheap compared to ECC88's.
This circuit doesn't need feedback. Infact the whole point is to eliminate feedback. So 12B4 = cheaper and better.
I put the circuit to the scope today. Putting a simple sine wave through it gave flat response all the way to about 50Khz, only after that does it start to roll off a tiny amount. There was no rise in response- which I half expected to find. It just goes to show that BBC broadcast quality is sibilant and not very good quality.
Shoog
12B4 are very linear and extremely cheap compared to ECC88's.
This circuit doesn't need feedback. Infact the whole point is to eliminate feedback. So 12B4 = cheaper and better.
I put the circuit to the scope today. Putting a simple sine wave through it gave flat response all the way to about 50Khz, only after that does it start to roll off a tiny amount. There was no rise in response- which I half expected to find. It just goes to show that BBC broadcast quality is sibilant and not very good quality.
Shoog
Bring up this old thread just to give credit to what it is due.
I built this FVP5 with the help of Shoog, of course. I used what I had in my parts bin as much as possible with the exception of the irf510a, no exotic parts. Chassis and PT are from an Eico tuner. Refurnish the chassis and prep the chassis took longer than the wiring. The preamp started the first note about one month ago and had a very noticeable hum. To make a long story short, the hum is 99% gone, this after tons of advise from Shoog via email. A few experience to share:
= same as Shoog's, there is no regulation, simple 6x4 -> RCRCRCLC
filtering.
= I had DC heater supply but didn't help reducing hum and had noticeable improvement over AC, so they are gone.
= B+ is usually between 144 and 146, depending on the hours of the day.
= I have my first stage set at 84V vs Shoog's at 80.
= Both channels draw 28ma. First stage draws approx 5ma, 2nd stage
draws 9ma each channel. Seem to be a lot higher than Shoog's.
Now the most important part, how it sounds. One word, great. To put
it to perspective, I compared to a factory wired Transcendent Sound ground grid preamp and a p2p version that I built. Transparency/natural is the GG's long suit, but the FVP5 clone is just as good, but when it comes to imaging and vocal, it beats the GG hands down. The interesting part is that it wasn't obvious until I switched back to the GG for A/B comparison. The first I noticed when switched back to the GG was the fussiness of the 3D image. There seemed to be a layer of silk in between, kind of like, you can tell where is the singer and instruments but not laser sharp in focus.
In summary, the result is very positive. While I can't say how it compares to the real FVP5, it is well worth the money and the time to build it. It is simply a very good sounding preamp. I can see this to be my main preamp in the forseeable future until .........
Good listening!!
I built this FVP5 with the help of Shoog, of course. I used what I had in my parts bin as much as possible with the exception of the irf510a, no exotic parts. Chassis and PT are from an Eico tuner. Refurnish the chassis and prep the chassis took longer than the wiring. The preamp started the first note about one month ago and had a very noticeable hum. To make a long story short, the hum is 99% gone, this after tons of advise from Shoog via email. A few experience to share:
= same as Shoog's, there is no regulation, simple 6x4 -> RCRCRCLC
filtering.
= I had DC heater supply but didn't help reducing hum and had noticeable improvement over AC, so they are gone.
= B+ is usually between 144 and 146, depending on the hours of the day.
= I have my first stage set at 84V vs Shoog's at 80.
= Both channels draw 28ma. First stage draws approx 5ma, 2nd stage
draws 9ma each channel. Seem to be a lot higher than Shoog's.
Now the most important part, how it sounds. One word, great. To put
it to perspective, I compared to a factory wired Transcendent Sound ground grid preamp and a p2p version that I built. Transparency/natural is the GG's long suit, but the FVP5 clone is just as good, but when it comes to imaging and vocal, it beats the GG hands down. The interesting part is that it wasn't obvious until I switched back to the GG for A/B comparison. The first I noticed when switched back to the GG was the fussiness of the 3D image. There seemed to be a layer of silk in between, kind of like, you can tell where is the singer and instruments but not laser sharp in focus.
In summary, the result is very positive. While I can't say how it compares to the real FVP5, it is well worth the money and the time to build it. It is simply a very good sounding preamp. I can see this to be my main preamp in the forseeable future until .........
Good listening!!
pchw said:
Hi Guys,
Thanks, it's nice to know one's work is appreciated. BUT as you guys know, the FVP was a complete Phono and Line Stage preamp. The Phono was designed by Allen Wright and the Line Stage by yours truly. Indeed I coined the term Super Linear Cathode Follower in the latter half of the 90's.
Re the constant voltage element that sits atop the CF tube, it is important to keep its input capacitance low as it can be quite non-linear. This means that if a Fet is chosen - choose high voltage that is equal to or greater than your rail HT and low capacitance. We originally used HexFets because they were easy to get and high voltage/rugged etc. - ended up choosing IRF710. But now I would tend to favour a bipolar device. I have used a Darlington pair made from TIP48 (or TIP48/50). I also have tried to use a single TIP50 and used the current derived from its Hfe. That way you can typically use a single 470K to 1M resistor to the HT (second R not required to form a voltage divider) and only a cap, say 0.47uF, to the output, cathode of the tube.
So there are a variety of options, and the parts are not exotic and yet work very well.
You can of course do what the new SVP does, and make the constant voltage follower another tube still. This requires that HT needs to be high enough.
Joe R.
Hi Joe,
I owed you a from the bottom of my heart, "thank you" especially my clone is only an one input line stage, no phono
I really like how it sounds and am trying to tweak it one parts at a time. Your comments came in the best timing. In fact, I am in the process of building an outboard power supply for the next iteration. While it is not super reg, it will be regulated supply all around. I will keep the current version around so that I can compare.
Thanks again!!
I owed you a from the bottom of my heart, "thank you" especially my clone is only an one input line stage, no phono
I really like how it sounds and am trying to tweak it one parts at a time. Your comments came in the best timing. In fact, I am in the process of building an outboard power supply for the next iteration. While it is not super reg, it will be regulated supply all around. I will keep the current version around so that I can compare. Thanks again!!
Hi Joe,
Thanks for the input. Your ideas are much appreciated.
Since pchw copied my copy I have made a few subtle changes. I replaced the LED's in the buffer stage with simple resistors. After readjusting the operating point of the first stage I got a subtle but significant tightening of the bass. The sound is slightly less relaxed overall - all changes have to be carefully judged.
If I was doing it again I would drop the current boosting network in favour of a simple resistor from the HV rail. I don't think it would have any sonic impact, but it would be simpler to implement.
My next version will have a EF86 front end running at 1mA. I will take the screen from the output node of the buffer stage and use a voltage divider to DC bias the grid of the buffer stage off of the first stage anode. This should function as a sort of UL stage with the anode of the EF86 held about 80V above the screen, such that the screen will draw tiny current. It should also help to moderate the gain. I will also put another triode on in place of the MOSFET. I'm hoping this will all produce a significant step up. The only thing I haven't decided on is how to further stiffen the Constant Current Sink, and if there is much sonic gain in the effort.
Be careful, my experience is that anything but the best regulators hurt the sound.
Good luck.
Shoog
Thanks for the input. Your ideas are much appreciated.
Since pchw copied my copy I have made a few subtle changes. I replaced the LED's in the buffer stage with simple resistors. After readjusting the operating point of the first stage I got a subtle but significant tightening of the bass. The sound is slightly less relaxed overall - all changes have to be carefully judged.
If I was doing it again I would drop the current boosting network in favour of a simple resistor from the HV rail. I don't think it would have any sonic impact, but it would be simpler to implement.
My next version will have a EF86 front end running at 1mA. I will take the screen from the output node of the buffer stage and use a voltage divider to DC bias the grid of the buffer stage off of the first stage anode. This should function as a sort of UL stage with the anode of the EF86 held about 80V above the screen, such that the screen will draw tiny current. It should also help to moderate the gain. I will also put another triode on in place of the MOSFET. I'm hoping this will all produce a significant step up. The only thing I haven't decided on is how to further stiffen the Constant Current Sink, and if there is much sonic gain in the effort.
Be careful, my experience is that anything but the best regulators hurt the sound.
Good luck.
Shoog
Hi Shoog
Looking at what you have done, I am inclined to use the LED as a voltage reference with, say 1mA, on the grid of V1B - the CS. You have to draw that from a + source via a suitable resistor to give you 1mA. It could be the filament supply or, if you have to, from the HT. Then use a resistor on the cathode of CS (V1B) to adjust current.
BUT, to get the tightest CS, use a small suitable Fet instead of the resistor, adjust as a current source. Put a 1K resistor on the gate and then 100R (adjust) on the Source to give you the desired current. I am not sure IF I am painting a clear word picture here?
Re the the voltage gain stage, I am aware that it has a lot of gain for a Line stage. But I cannot stress the simplest and yet utmost importance of the "unloaded Triode".
Many look at the SLCF and make wonderful remarks about it. Yes, it is what it is! That is why it makes such a good UG (Unity Gain) buffer. BUT as far as gain is concerned, the key to understanding the sonic beauty of a Triode gain stage is: it sees virtually no load on the anode because of the high input Z of the SLCF buffer. I am not sure, but I recall another notable designer for Audio Note who understood this. In effect, when a Triode is unburdened of any load, it is effectively in a similar mode to that of constant current. I know this takes a bit of left field thinking, but it is fairly true as a Triode stage dealing with line level signals and very high voltage overheads. So as the Triode swings, there is very little/minimal change in current. And this is when they sound best. After that, the SLCF is good enough to hear that real Triode quality.
Re EF86, I am yet to be convinced. The problem remains, the high gain?
Joe R.
Looking at what you have done, I am inclined to use the LED as a voltage reference with, say 1mA, on the grid of V1B - the CS. You have to draw that from a + source via a suitable resistor to give you 1mA. It could be the filament supply or, if you have to, from the HT. Then use a resistor on the cathode of CS (V1B) to adjust current.
BUT, to get the tightest CS, use a small suitable Fet instead of the resistor, adjust as a current source. Put a 1K resistor on the gate and then 100R (adjust) on the Source to give you the desired current. I am not sure IF I am painting a clear word picture here?
Re the the voltage gain stage, I am aware that it has a lot of gain for a Line stage. But I cannot stress the simplest and yet utmost importance of the "unloaded Triode".
Many look at the SLCF and make wonderful remarks about it. Yes, it is what it is! That is why it makes such a good UG (Unity Gain) buffer. BUT as far as gain is concerned, the key to understanding the sonic beauty of a Triode gain stage is: it sees virtually no load on the anode because of the high input Z of the SLCF buffer. I am not sure, but I recall another notable designer for Audio Note who understood this. In effect, when a Triode is unburdened of any load, it is effectively in a similar mode to that of constant current. I know this takes a bit of left field thinking, but it is fairly true as a Triode stage dealing with line level signals and very high voltage overheads. So as the Triode swings, there is very little/minimal change in current. And this is when they sound best. After that, the SLCF is good enough to hear that real Triode quality.
Re EF86, I am yet to be convinced. The problem remains, the high gain?
Joe R.
Since the screen constitutes 100% feedback, the gain should be as if it was triode strapped, and further reduced by the voltage divider to the buffer grids. I know what you mean about the triode probably been the best option overall - but I want to try this anyway. If it doesn't work out it should be a simple matter of triode strapping the EF86, which I understand to make an excellent triode in itself.
I will probably put a simple little FET in the CS cathode as you suggest.
At the moment I am planning a power amp build so this project will have to go on the back burner.
Shoog
Shoog said:
There is a way to lower the gain stage of a Triode stage. Increase the local feedback by increasing the value of the cathode resistor. Of course, then you lose the current through the tube and up goes your anode voltage and you've lost your operation point. But what if you now make up the missing current by inserting a small parallel fet across the increased value cathode resistor. The anode voltage will return to normal, but the overall gain is reduced. This trick only works in this situation - it is not recommended as a cure-all. It just happens to work in this case where the anode sees effectively no load.
Joe R.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.