Joe,
Attached is a picture of my idea for the next build.
I tried it with a 6AU6 and the screen and plate at the same voltage, it sounded grainy and the top end suffered. Smoking Amps then pointed out my error (to much screen current), so i rejigged it to the new scheme. The screen current should be well below 1mA most of the time so the buffer shouldn't brake a sweat supplying it.
I have no illusion that it will be an improvement - but it just might be !!
Shoog
Attached is a picture of my idea for the next build.
I tried it with a 6AU6 and the screen and plate at the same voltage, it sounded grainy and the top end suffered. Smoking Amps then pointed out my error (to much screen current), so i rejigged it to the new scheme. The screen current should be well below 1mA most of the time so the buffer shouldn't brake a sweat supplying it.
I have no illusion that it will be an improvement - but it just might be !!
Shoog
Attachments
Hi Shoog
OK, hope you don't mind if I critique your proposed schematic.
The voltage divider formed by R12 & R11 is a two-edged sword. The good: It will lower the noise floor. The bad: It will load the anode of the gain stage - bad for sound. I know which choice is more palatable - some noise and put the voltage divider on the input. One tube that could be tried for the gain stage is the U7/ECC82. Do a Manley trick and even parallel them if you want a single tube per channel (but I know some purists who don't like doing it - I'm pragmatic as well as purist). That will give you lower gain - have not tried it. But I did a U7 input stage on a power amp using similar conditions and it sounded very nice.
I do repeat: The unseen key to the excellence of the FVP5 line stage is not just the SLCF, but also the anode of the Triode gain stage sees a near infinite high impedance. In your instance R12 & R11 would need to be extreme high values - not practicable.
I violated your copyright and amended the schematic. The added Fet allows the parallel resistor to be much higher value and hence lower gain. With ECC88 there should just enough voltage for it to work.
If using EF86 and adding R2, you have introduced negative (loop) feedback in to a "Zero-Feedback" design. Sorry, but not impressed. If you are thinking 'Ultra-Linear" - then I have to admit NOT to be a fan. The best Triode (or Pentode) is where the input grid is totally in charge - does not happen with any form U-L. In transformer (power) outputs it is even worse if Push-Pull amps - now you have some real symmetry problems (think smearing). No, we are no U-L fans! Seriously thinking about starting a new trend... the anti-fan club.
My other suggestion - see my violated (your) schematic - in the SLCF I would use the LED as shown. You can even use TWO Leds. Feel like three? In theory this is a good thing as the resistor setting the current will be higher value. Of course a Fet current source here would convert that resistor to, like, huge equivalent R!
But, by all means build what you have in mind - but it takes away some of the things that makes the FVP5 so simple and yet so good.
Joe R.
OK, hope you don't mind if I critique your proposed schematic.
The voltage divider formed by R12 & R11 is a two-edged sword. The good: It will lower the noise floor. The bad: It will load the anode of the gain stage - bad for sound. I know which choice is more palatable - some noise and put the voltage divider on the input. One tube that could be tried for the gain stage is the U7/ECC82. Do a Manley trick and even parallel them if you want a single tube per channel (but I know some purists who don't like doing it - I'm pragmatic as well as purist). That will give you lower gain - have not tried it. But I did a U7 input stage on a power amp using similar conditions and it sounded very nice.
I do repeat: The unseen key to the excellence of the FVP5 line stage is not just the SLCF, but also the anode of the Triode gain stage sees a near infinite high impedance. In your instance R12 & R11 would need to be extreme high values - not practicable.
I violated your copyright and amended the schematic. The added Fet allows the parallel resistor to be much higher value and hence lower gain. With ECC88 there should just enough voltage for it to work.
If using EF86 and adding R2, you have introduced negative (loop) feedback in to a "Zero-Feedback" design. Sorry, but not impressed. If you are thinking 'Ultra-Linear" - then I have to admit NOT to be a fan. The best Triode (or Pentode) is where the input grid is totally in charge - does not happen with any form U-L. In transformer (power) outputs it is even worse if Push-Pull amps - now you have some real symmetry problems (think smearing). No, we are no U-L fans! Seriously thinking about starting a new trend... the anti-fan club.
My other suggestion - see my violated (your) schematic - in the SLCF I would use the LED as shown. You can even use TWO Leds. Feel like three? In theory this is a good thing as the resistor setting the current will be higher value. Of course a Fet current source here would convert that resistor to, like, huge equivalent R!
But, by all means build what you have in mind - but it takes away some of the things that makes the FVP5 so simple and yet so good.
Joe R.
Attachments
Joe,
I didn't quite understand the idea with the LED to stiffen the CCS - brilliant !!! Thats definately a keeper and will go into the next build. A string of 3 Green LED's would stiffen things up very nicely. It should also cut overall gain usefully. Is it concievable that this configuration could introduce a bit of PS cancellation ala the Aikido.
I appreciate where your coming from with the UL setup and i'm inclined to think your right. I have a particular fondness for the 5687 and this would make a good front end with a slightly lower gain, however it would run at a wasteful current of 10mA or so. Do you know of any other triode with moderate gain that sounds good at 4mA or there abouts.
Shoog
I didn't quite understand the idea with the LED to stiffen the CCS - brilliant !!! Thats definately a keeper and will go into the next build. A string of 3 Green LED's would stiffen things up very nicely. It should also cut overall gain usefully. Is it concievable that this configuration could introduce a bit of PS cancellation ala the Aikido.
I appreciate where your coming from with the UL setup and i'm inclined to think your right. I have a particular fondness for the 5687 and this would make a good front end with a slightly lower gain, however it would run at a wasteful current of 10mA or so. Do you know of any other triode with moderate gain that sounds good at 4mA or there abouts.
Shoog
Ahah! You get the idea now re the LEDs.
I think there are plenty of triodes that sound good in this configuration, which is best will be subject to taste. But I admit that I have mostly been using 6922 in FVPs - but in power amp front-ends where the load has been typically 2M2 (not near infinite Z as in the FVP but still very high) I have used a variety of triodes and they have all sounded pretty good, which leads me to think that the configuration is more important than the choice of triode.
The following does have an advantage, use a triode that gives you a higher voltage across the cathode resistor for the Fet to work with. So ECC82/U7 has moderate gain and fits the bill. I have not tried it in an FVP but it did work a treat in a power amp front-end.
Looking at the curves of 5687 published www.diyparadise.com/tubeloadline/tubeloadlines.html, with a plate voltage around 150V and 5ma it indicates -9V, that should be great for the Fet. Make anode load 20-25K, try cathode resistor 2K2 - these should be starter values.
Divide anode/cathode values is around 10, so AC voltage gain cannot be higher than around 10 - it ought to be a little lower than that, rule of thumb. Try it and see. If you then want to further decrease gain, increase cathode resistor to 4K7 and then add Fet CS to take up the slack current - and get the same total current the same and the anode voltage will settle back to where it was with the 2K2 resistor.
Take the above values only as a guide, OK.
Joe R.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.