Hi Ken, it's been some time, hope you are doing well and surviving. Last month I got Covid after all and that was after four vaccinations/boosters (as I regard myself as a science person, I am definitely a believer). Life here has been the usual roller coaster.
The sound of ECC88/6922 vs ECC81/12AT7 is curious, all I can say is that when I have replaced a 6922 with a "T7" in any circuit where possible, I have preferred the T7 sound. When I have done it for others, they did not ask me to undo it. This included the input tube on power amps. Do I know the reason? Nope! But I have a theory, the 12AT7 is a regular down the middle signal tube and just does the job with ease, a word that describes the sound well.
The FVP-6 is right up there sound wise. You can still get in touch with Vacuum State and get a quote for the 6922 based SVP-2 and they will build it for you and it will take some months. The price will make your eyes water and it is good. Even then, I will prefer the FVP-6 - but this means getting the power supply right.
Hi, Joe! I’m doing fine, although, I haven’t yet hit the lottery yet. Although, I’m still looking forward to being one of those, seemingly more frequent, lottery billionaires 🤣. Sorry to hear that you had a bout with Covid. That thing is nasty. My wife caught it in early November, and she also had four vaccinations also. She was miserable, with flu like symptoms, including an elevated temp for four days that occasionally topped 103 degrees Fahrenheit. I, fortunately, never came down with it though.
Back to something more pleasant, audio 😀. Regarding getting the power supply right, do you have a suggested B+ circuit? I haven’t built a tube amplification stage before, and wondering what I may be subjectively missing out on. So, I’m seriously considering using your FVP-6A design as the output amplifier in a diy NOS DAC which I’m currently working out the basic circuit elements for.
One more observation about the FVP5A all tube version. While reading Morgan Jones "Valve Amplifiers" third edition book (page 96) I came across a phrase:
"An important point to note with cascodes is that the only general purpose valve that was designed to work well in cascode is the ECC88/6DJ8 or E88CC/6922 (special quality version). Try other valves, by all means, but do not expect the performance to be as good." Does it mean that the upper last two tubes are better to leave 6922 and change only first two halves with T7?
"An important point to note with cascodes is that the only general purpose valve that was designed to work well in cascode is the ECC88/6DJ8 or E88CC/6922 (special quality version). Try other valves, by all means, but do not expect the performance to be as good." Does it mean that the upper last two tubes are better to leave 6922 and change only first two halves with T7?
Does it mean that the upper last two tubes are better to leave 6922 and change only first two halves with T7?
*Morgan is spot on.
Well, there is a couple of ways to do it, You could use one side each of a T7 for the input stage, but that means one tube handling both channels. That has been done before but channel separation might take a hit. Then use another *6922/ECC88 for the SLCF, again half and half, and run 8-10mA for a maybe sub-100 Ohm output impedance? But of course, you could double of on the T7 and that would give you better channel balance, but now you are talking three tubes and not two. The way I am reading this, channel separation on the input tube is more critical than the SLCF buffer stage. There it sees fairly low impedances on all three electrodes, it is higher impedance signals that erode separation.
* The Sovtek 6922 can handle these voltages. That's the one I usually go for.
So, I’m seriously considering using your FVP-6A design as the output amplifier in a diy NOS DAC which I’m currently working out the basic circuit elements for.
I see, that could be interesting. Which NOS-DAC and what is its requirements? Are we talking TDA1541A? Or?
I have been eyeing a shunt reg based on using TL431 as a reference, below is borrowed from John Broskie's tubecad.com, but I have used many variations based on TL431, it's been very versatile to me:
Easy to set to 250V and then use a current source (can show what I would use), to make a simple but effective two-stage "SuperReg" and use about +300V or near it.
I was definitely thinking about posting something more fleshed out.
Unfortunately, I never had a TDA1541A chip. I’ve always been curious about that DAC because of its reputation for good sound, but those are priced higher than gold anymore. I do have a cache of AD1865 chips, along with a pair of AD1862, and also PCM1704. I suppose that I’ll utilize the AD1865 here. The DAC will operate NOS, utilize a T/H circuit to eliminate slew-rate limiting distortion via exponential-settling, and also feature an experimental 8-pole passive LC elliptical reconstruction filter which I’ve designed.
I’ve never actually built a tube stage before, so any tips you might have on implementing a good sounding B+ supply for the FVP-6 would be appreciated.
As far as that Broskie shunt reg. design, we think alike. I‘ve also been eyeing it. Broskie later published a slight variation of that circuit, attached below.
I’ve never actually built a tube stage before, so any tips you might have on implementing a good sounding B+ supply for the FVP-6 would be appreciated.
As far as that Broskie shunt reg. design, we think alike. I‘ve also been eyeing it. Broskie later published a slight variation of that circuit, attached below.
Last edited:
I have another project in mind that requires +100V and I changed values for +250V and it looks like below.
NOT YET BUILT - SUGGESTED ONLY:
A couple of explanations. The LM317L sets up 1mA and all the current goes through that LM431. This is better than using a resistor. It means that the added 1mA to the end result is also current sourced. The TIP32C redirects the current, the (red?) LED does mean that it is limited to 20mA in the shunt, but without the LED it can sink quite high currents, use appropriate heatsinks. The *47R needs to be adjust as per current required. Allow up to 30mA for the circuit and 20mA in the shunt.
I used 22K here instead of LM317L, but on the surface it does seem it might work. The 0.1uF I would use X2 class polypropylene. No values above are set in cement.
Any suggestions and critique is welcome.
.
NOT YET BUILT - SUGGESTED ONLY:
A couple of explanations. The LM317L sets up 1mA and all the current goes through that LM431. This is better than using a resistor. It means that the added 1mA to the end result is also current sourced. The TIP32C redirects the current, the (red?) LED does mean that it is limited to 20mA in the shunt, but without the LED it can sink quite high currents, use appropriate heatsinks. The *47R needs to be adjust as per current required. Allow up to 30mA for the circuit and 20mA in the shunt.
I used 22K here instead of LM317L, but on the surface it does seem it might work. The 0.1uF I would use X2 class polypropylene. No values above are set in cement.
Any suggestions and critique is welcome.
.
Thanks, Joe. Just for information purposes, here is a Broskie CCS circuit which is of similar complexity to his shunt-reg.
Joe, what did you make of Allen's "LM317 begone" modification for the SuperReg? This was in the same position where you have the TL431 in the CCS. I took the LM317 out of all four of the SuperRegs I built, but didn't check for oscillation before or after. Is there something intrinsic about the LM317 that makes it prone to instability, or is it just in this position?
Alex
Alex
Joe, what did you make of Allen's "LM317 begone" modification for the SuperReg? This was in the same position where you have the TL431 in the CCS. I took the LM317 out of all four of the SuperRegs I built, but didn't check for oscillation before or after. Is there something intrinsic about the LM317 that makes it prone to instability, or is it just in this position?
Hi Alex
(Please note that the LM317L in the above schematic performs an entirely different function. It only handles 1mA going through the LM431. A resistor could be used here, such as 22K = 1mA approx.)
Oh boy, it is such a long time ago that I was discussing this with Allen and this only now you brought it back to me. Yes, getting rid of the LM317L and replacing it with a resistor as you did, that solved a potential stability problem. Yes, it is the position that the LM317L is in, you are right about that.
I felt that a different solution was required and I spotted something in the LM431 datasheet that gave me the idea of coming up with a better solution than just replacing LM317L with a resistor (a resistor makes the circuit a lot less tight DC wise).
So what you see above is my solution to the problem. In the original circuit, the LM317L saw all of the current going through the HT HexFet. I identified that as the culprit. Using an LM431, the current now goes through a resistor and it does not go through the LM431. An elegant solution.
There is another thing I would add to the above schematic and so have amended it below:
If you have 2.5V voltage drop across *47R (=50mA approx), then another series *47R will drop another 2.5V. This value is not critical, just make sure that it drops a few volts. Note also that it is on the Current Source side.
Please note, some aspects of the above circuit needs to be tested by actually building it.
Thanks, Joe. Just for information purposes, here is a Broskie CCS circuit which is of similar complexity to his shunt-reg.
I see, looks like an interesting variation to study.
Joe, I don’t recall whether you’ve before mentioned it somewhere, but do you favor a particular volume attenuator solution?
Hi Alex
(Please note that the LM317L in the above schematic performs an entirely different function. It only handles 1mA going through the LM431. A resistor could be used here, such as 22K = 1mA approx.)
Oh boy, it is such a long time ago that I was discussing this with Allen and this only now you brought it back to me. Yes, getting rid of the LM317L and replacing it with a resistor as you did, that solved a potential stability problem. Yes, it is the position that the LM317L is in, you are right about that.
I felt that a different solution was required and I spotted something in the LM431 datasheet that gave me the idea of coming up with a better solution than just replacing LM317L with a resistor (a resistor makes the circuit a lot less tight DC wise).
So what you see above is my solution to the problem. In the original circuit, the LM317L saw all of the current going through the HT HexFet. I identified that as the culprit. Using an LM431, the current now goes through a resistor and it does not go through the LM431. An elegant solution.
Thanks, Joe. I see - the LM317 provides a constant current for the LM431, so stabilising the voltage reference, whereas Allen used it in series with the MOSFET to supply the gate voltage reference. I recall Emile Sprenger did something similar in his HPHV.
Alex
Thanks, Joe. I see - the LM317 provides a constant current for the LM431, so stabilising the voltage reference
Yep. But I admit that using the LM317L to provide a constant 1mA to has a drawback, the voltage must be under the max 40V rating. That should not be a problem with +275V input and +250V shunt regulated. If larger voltages are present, then you will need to use a resistor, so if for example you have +300V, then you will need around 50K to give you 1mA. That 50K degrades the performance (maybe not by a lot). I seem to recall that Allen went with very high resistance values because there is no Gate current on the MosFet.
Do you have the schematic there that shows what you did when you got rid of the LM317L? Did you increase the resistor value(s))?
whereas Allen used it in series with the MOSFET to supply the gate voltage reference.
Right. I believe the stability issue when the LM317L is related to the MosFet as they both see the same current. MosFets have capacitances issues. But my solution using the LM431 solves that. Alas, I am sure that Allen would have approved of this idea.
Since we are talking about Vacuum State and Allen's legacy still living on on here and on the RTP thread, I got an idea...
Last night I was looking for something on my hard drive and I came across a folder that contained pics of a car trip Allen and I took back in 2002 from Aarhus in Denmark to Munich. I met Emile Sprenger there in Aarhus for the first time. It was there and then we decided to rename Aarhus Triode Festival and to this day it is called the "European Triode Festival" and being going strong every year since then except for Covid.
We then drove to Munich where I stayed with Allen for some days and he showed me around Munich. Then on the morning when I took the train back to Copenhagen, I took this photo of him with his Opel car, just before heading towards Munich train station:
When we got to Munich train station, I passed my camera to Allen and he took this pic of me just before I got on the ICE train to Hamburg. Those were the days.
Thanks for the photos, Joe! I am still sorry I never met Allen in person, though he gave me plenty of advice on my SVP and RTP3 builds.
The Begone Mod consisted of measuring the voltage across the LM317, then replacing it, along with the 100R current sense resistor. with a resistor whose value would give the required current at that voltage.
Alex
Do you have the schematic there that shows what you did when you got rid of the LM317L? Did you increase the resistor value(s))?
The Begone Mod consisted of measuring the voltage across the LM317, then replacing it, along with the 100R current sense resistor. with a resistor whose value would give the required current at that voltage.
Alex
Yes, and if we want to not use the LM317L and we have something like +300V input, then we can't use it anyway, so in that case go something like this:
NOTE: NOBODY NOTICED THE ERROR IN THE SHUNT REG - IT WAS CORRECTED THIS TIME.
NOTE: NOBODY NOTICED THE ERROR IN THE SHUNT REG - IT WAS CORRECTED THIS TIME.
Last edited:
Here’s a interesting simple BJT cascode CCS circuit which I don’t believe I’ve ever seen before. I stumbled across it in the below linked thread. The biasing FETs are depletion-mode, so they could be either JFETs or small depletion MOSFETs. Raw supply voltage will be rather limited with most JFETs, though. The Zener on the far right is just there as the shunt voltage regulating element. Obviously, it could be replaced with something a bit more sophisticated.
https://www.superbestaudiofriends.org/index.php?threads/lets-build-a-phono-stage.2594/
https://www.superbestaudiofriends.org/index.php?threads/lets-build-a-phono-stage.2594/
Would there be any benefit not use any electrolytic capacitors but PP instead?Yes, and if we want to not use the LM317L and we have something like +300V input, then we can't use it anyway, so in that case go something like this:
View attachment 1264045
NOTE: NOBODY NOTICED THE ERROR IN THE SHUNT REG - IT WAS CORRECTED THIS TIME.
Yes, and if we want to not use the LM317L and we have something like +300V input, then we can't use it anyway, so in that case go something like this:
View attachment 1264045
NOTE: NOBODY NOTICED THE ERROR IN THE SHUNT REG - IT WAS CORRECTED THIS TIME.
Joe,
Clever, tying that 10uF cap. tied to the MOSFET’s source in the CCS.
I tried to build this version. Input voltage 236V to get 130V to plate pin 1 and 230V to pin 6. Pin 3 has 2.15V which is less than marked here. Whould that be a problem?
Now I have a problem. 😀 It is generating quite loud high frequency noise. Could be that the IRF820 which is connected to the plate pin 1 is off the pcb with wires? Resistors standing too high from the pcb? Using different tube is the same. Any other thoughts?
What is the voltage drop across the 560K?
If you are using a lower 236V (and not 250V), then the circuit will need some adjustment.
Only 6V drop from 236V. No, it needs to be lower than that.
That's not a problem here, it can vary. It is still in the ballpark and telling us that the current here is 2.15V/2K2 = 1mA. That's fine.
This seems to tell me that the input stage is OK (130V and 2.15V respectively looks OK). But since you have dropped 14V on the power supply, it may be the Darlington not having enough voltage drop across it.
Hence I come across the original question, the voltage drop across 560K. You may need to double it if you only have a 6V drop across the Darlington.
Please note that circuit layout, particular ground technique, can also be a problem. But let us see if we can get the voltages right.
If you are using a lower 236V (and not 250V), then the circuit will need some adjustment.
230V to pin 6
Only 6V drop from 236V. No, it needs to be lower than that.
in 3 has 2.15V which is less than marked here
That's not a problem here, it can vary. It is still in the ballpark and telling us that the current here is 2.15V/2K2 = 1mA. That's fine.
get 130V to plate pin 1
This seems to tell me that the input stage is OK (130V and 2.15V respectively looks OK). But since you have dropped 14V on the power supply, it may be the Darlington not having enough voltage drop across it.
Hence I come across the original question, the voltage drop across 560K. You may need to double it if you only have a 6V drop across the Darlington.
Please note that circuit layout, particular ground technique, can also be a problem. But let us see if we can get the voltages right.
The voltage drop across 560K is about 6.6 volts. If I raise the voltage to 250 V then pin 1 voltage raises to 143.8 V and pin 3 voltage to 2.35 V. Pin 6 has 243 V. So the voltage drop remains pretty much a same with 236 V and with 250 V.
What would be good HV regulator output mA for this FVP-6?
What would be good HV regulator output mA for this FVP-6?
- Home
- Amplifiers
- Tubes / Valves
- Upgrading Vacuum State FVP FVP-5 Tube Line Stage to FVP-6