Folks, I would very much appreciate the benefit of everyone's expertise on a problem that puzzles me.
In another post someone referred to VSE's FVP5A preamp (with built-in phono section). So, naturally, I went to Mr. Wright's web pages to see for myself. I discovered alot of very interesting information and since I'm a simulation type of guy, I decided to simulate the line stage portion of the FVP5A. After I did this, and not noticing any particular special features of this topology, I decided to compare it to a plain old unbypassec grounded cathode coupled to a CF. I've shown my simulation diagram below. The GK and CF stages are biased identically for both amps.
Now, here is where I would really like your help to understand what I'm seeing.
But first, let me admit that simulations are just that. OTOH, I do believe that while they cannot always do a great job of predicting absolute values of certain things, like THD, they can do an excellent job of comparing two topologies.
I've compared these two designs for THD, frequency response, and phase shift (from 10Hz to 10MHz), and Zo. Folks, I cannot find any discernable difference between these two line stage topologies. Absolutely none, except that the plain CF has a slightly lower Zo and drops off a little faster above about 1MHz.
Since I know there are many things that I don't understand and not meaning to in any way disparage a highly respected supplier of tube equipment, I would very much like to know what I am missing. Thanks very much for the help.
In another post someone referred to VSE's FVP5A preamp (with built-in phono section). So, naturally, I went to Mr. Wright's web pages to see for myself. I discovered alot of very interesting information and since I'm a simulation type of guy, I decided to simulate the line stage portion of the FVP5A. After I did this, and not noticing any particular special features of this topology, I decided to compare it to a plain old unbypassec grounded cathode coupled to a CF. I've shown my simulation diagram below. The GK and CF stages are biased identically for both amps.
Now, here is where I would really like your help to understand what I'm seeing.
But first, let me admit that simulations are just that. OTOH, I do believe that while they cannot always do a great job of predicting absolute values of certain things, like THD, they can do an excellent job of comparing two topologies.
I've compared these two designs for THD, frequency response, and phase shift (from 10Hz to 10MHz), and Zo. Folks, I cannot find any discernable difference between these two line stage topologies. Absolutely none, except that the plain CF has a slightly lower Zo and drops off a little faster above about 1MHz.
Since I know there are many things that I don't understand and not meaning to in any way disparage a highly respected supplier of tube equipment, I would very much like to know what I am missing. Thanks very much for the help.
An externally hosted image should be here but it was not working when we last tested it.
I did basically the same experiment, but in hardware, ie, I built both and measured and listened a while back. I expected them to measure the same (modified AWA THD meter), and sound the same. They measured near enough to identical with a resistive load, but add a few metres of cable, and they sounded nothing alike with the SLCF being very good, much to my surprise. I wasn't expecting what I found, expecting they'd both be pretty poor.
FTR there's still a modified diff/bal version of this circuit in my pre chassis, but it isn't in use as I found something else I prefer more. I still use the phono section for my Decca.
So far, with all I've seen, I'm yet to share your confidence with modelling, so I'm not sure what you're asking. Lots of different circuits measure closely enough IME and sound completely different.
FTR there's still a modified diff/bal version of this circuit in my pre chassis, but it isn't in use as I found something else I prefer more. I still use the phono section for my Decca.
So far, with all I've seen, I'm yet to share your confidence with modelling, so I'm not sure what you're asking. Lots of different circuits measure closely enough IME and sound completely different.
Actually, you've begun to answer my question already! I was hoping that someone had done a hardware comparison.
I did wonder if the FPV would be better at handling cables, since there seems to be otherwise no real difference between it and a plain design. But, I don't really see how, looking at the circuit. Not that I don't see what the design is, but I don't see how it handles the characteristics of an audio cable better than an ordinary CF.
So, I would really like to understand what it is about this topology that makes it better able to handle "transmission lines".
Thanks for the help.
BTW, if the answer is on VSE's website, just point me there.
Thanks for the suggestion. I had already tried that, but I wasn't sure that I had reliable parameters for the cable. So I did some searching this morning on cable parameters and found some LRC for Belden shielded cable which were lower than the numbers I was using.
But, here's what I did. I modelled three feet of cable with:
200nH per foot
30mOhms per foot
100pf per foot to shield and other wire
100K termination load at the input to the next device
50pf capacitance at the input to the next device
I think most good audio cables are better than this.
In any case, I see no difference except that the FVP5A begins to show a resonance above 1MHz and so instead of tailing off after 1MHz the signal begins to rise a bit.
I'm happy to post these results if anyone wants to see them or to check my work.
But, here's what I did. I modelled three feet of cable with:
200nH per foot
30mOhms per foot
100pf per foot to shield and other wire
100K termination load at the input to the next device
50pf capacitance at the input to the next device
I think most good audio cables are better than this.
In any case, I see no difference except that the FVP5A begins to show a resonance above 1MHz and so instead of tailing off after 1MHz the signal begins to rise a bit.
I'm happy to post these results if anyone wants to see them or to check my work.
Fair enough. In my comparison, both output buffers are idling at over 10mA and the Zo for both is less than 100R. Either of them can work into fairly difficult loads, as witnessed by my simulations of transmission lines with higher than normal inductances and capacitances.
The FVP does have a CCS at the bottom reducing current swing in the tubes, but this hardly linearizes it much more than the ordinary CF and does not reduce distortion noticeably.
So, I am still puzzled as to why this topology is better than a regular CF.
The FVP does have a CCS at the bottom reducing current swing in the tubes, but this hardly linearizes it much more than the ordinary CF and does not reduce distortion noticeably.
So, I am still puzzled as to why this topology is better than a regular CF.
Code:
So, I am still puzzled as to why this topology is better than a regular CF.A regular CF with low value cathode resistor and grid leak tied to ground is not very linear but if the cathode resistor value is increased either by using negative voltage supply or by increasing grid potential linearity improves considerably.
There is a interesting comparison of CF linearity in Walley & Wallman, "Vacuum tube amplifiers" in the section of DC amplifiers where the linearity of an ordinary CF is compared to a CF using negative voltage supply with a large value cathode resistor and also compared to a CF with penthode CCS. The result speaks for itself, the ordinary CF is worst and the neagtive supply CF and the CCS CF is quite similar especially for relatively low voltage swing.
In your comparison you compared a CF with relatively high value cathode resistor with a CCS coupled one and the result is as expected I think.
I have built different kind of CF's using -100V cathode power supply and either resistor or CCS using pentodes or FET's and I have also compared this with an ordinary CF with grid tied to ground, my experience is that pentode CF or large cathode resistor CF measure almost the same and also give the best sound, sorry but although I tried I dont like the sound of FET based circuits.
My current line stage is based on a SRPP DC coupled to a cathode grounded stage followed by a CF using -100V for cathode resistor supply, I use DC and AC feedback to control gain, (I believe that feedback used correctly can give excellent results both sonically and in measurements)
Regards Hans
> I cannot find any discernable difference between these two line stage topologies.
One on the left has more tubes. It has to be better.
Myself, with a 6DJ8 and these values, I would not expect any significant difference in simulation or measurement. Voltages and currents are high, Mu is high. What the extra tubes do is waste supply voltage and reduce maximum undistorted signal swing, but there's tons of swing left. I happily agree that the extra tubes may clip more musically, though I doubt any normal audio level takes it near clipping on the worst transients.
Your models are of course imperfect. They probably only cover x^2 and x^3 order terms. This will perfectly model most large-scale behavior, but real devices have higher order bends. Maybe the stacking cancels (or enhances?) these high-order nonlinearities that can't be modeled at the current state of simulation (or tube production).
> I wasn't sure that I had reliable parameters for the cable.
Use 25-30pFd for cable capacitance.
What kind of fat-stuff has 100pFd/foot??? (Or was it pFd/Meter, and you mis-read?)
I doubt the inductance or resistance has any effect for audio inside the home.
6 feet is common.
You might find 200pFd parasitic on some inputs.
Round-up to 1,000 pFd. If the line may run across the room (preamp at chair, power amp at speakers), figure 2,000pFd.
I assume you have checked at high levels like 3V peak?
One on the left has more tubes. It has to be better.
Myself, with a 6DJ8 and these values, I would not expect any significant difference in simulation or measurement. Voltages and currents are high, Mu is high. What the extra tubes do is waste supply voltage and reduce maximum undistorted signal swing, but there's tons of swing left. I happily agree that the extra tubes may clip more musically, though I doubt any normal audio level takes it near clipping on the worst transients.
Your models are of course imperfect. They probably only cover x^2 and x^3 order terms. This will perfectly model most large-scale behavior, but real devices have higher order bends. Maybe the stacking cancels (or enhances?) these high-order nonlinearities that can't be modeled at the current state of simulation (or tube production).
> I wasn't sure that I had reliable parameters for the cable.
Use 25-30pFd for cable capacitance.
What kind of fat-stuff has 100pFd/foot??? (Or was it pFd/Meter, and you mis-read?)
I doubt the inductance or resistance has any effect for audio inside the home.
6 feet is common.
You might find 200pFd parasitic on some inputs.
Round-up to 1,000 pFd. If the line may run across the room (preamp at chair, power amp at speakers), figure 2,000pFd.
I assume you have checked at high levels like 3V peak?
Hans, thanks for the information. It is as I expected too. CFs don't do well into small loads, just like plate loaded topologies.
Federico, here is my modified model from Norman Koren's website:
.SUBCKT 6DJ8 1 2 3 ; P G C; NEW MODEL
+ PARAMS: MU=33 EX=1.3 KG1=220 KP=250 KVB=200 RGI=2000
+ CCG=2.3P CGP=2.1P CCP=.7P ; ADD .7PF TO ADJACENT PINS; .5 TO OTHERS.
E1 7 0 VALUE=
+{V(1,3)/KP*LOG(1+EXP(KP*(1/MU+V(2,3)/SQRT(KVB+V(1,3)*V(1,3)))))}
RE1 7 0 1G
G1 1 3 VALUE={(PWR(V(7),EX)+PWRS(V(7),EX))/KG1}
RCP 1 3 1G ; TO AVOID FLOATING NODES IN MU-FOLLOWER
C1 2 3 {CCG} ; CATHODE-GRID; WAS 1.6P
C2 2 1 {CGP} ; GRID-PLATE; WAS 1.5P
C3 1 3 {CCP} ; CATHODE-PLATE; WAS 0.5P
D3 5 3 DX ; FOR GRID CURRENT
R1 2 5 {RGI} ; FOR GRID CURRENT
.MODEL DX D(IS=1N RS=1 CJO=0 TT=1N)
.ENDS
PRR, my conclusion is the same as yours. The extra tubes are a waste of heater power. It would be awfully nice to do a strict AB blind comparison to see if any differences are audibly discernable.
The models are somewhat imperfect against the real imperfections of actual tubes and maybe these are audible, but I am doubting it at this time until I have further information.
Regarding the cable, I selected values that I thought would be much worse than any real audio cable so there wouldn't be any room for doubt. The values for the Belden cable that I did locate were well below the values I used.
I will do a few more tests with high load cap and higher signals to see if there are any noticeable changes.
Federico, here is my modified model from Norman Koren's website:
.SUBCKT 6DJ8 1 2 3 ; P G C; NEW MODEL
+ PARAMS: MU=33 EX=1.3 KG1=220 KP=250 KVB=200 RGI=2000
+ CCG=2.3P CGP=2.1P CCP=.7P ; ADD .7PF TO ADJACENT PINS; .5 TO OTHERS.
E1 7 0 VALUE=
+{V(1,3)/KP*LOG(1+EXP(KP*(1/MU+V(2,3)/SQRT(KVB+V(1,3)*V(1,3)))))}
RE1 7 0 1G
G1 1 3 VALUE={(PWR(V(7),EX)+PWRS(V(7),EX))/KG1}
RCP 1 3 1G ; TO AVOID FLOATING NODES IN MU-FOLLOWER
C1 2 3 {CCG} ; CATHODE-GRID; WAS 1.6P
C2 2 1 {CGP} ; GRID-PLATE; WAS 1.5P
C3 1 3 {CCP} ; CATHODE-PLATE; WAS 0.5P
D3 5 3 DX ; FOR GRID CURRENT
R1 2 5 {RGI} ; FOR GRID CURRENT
.MODEL DX D(IS=1N RS=1 CJO=0 TT=1N)
.ENDS
PRR, my conclusion is the same as yours. The extra tubes are a waste of heater power. It would be awfully nice to do a strict AB blind comparison to see if any differences are audibly discernable.
The models are somewhat imperfect against the real imperfections of actual tubes and maybe these are audible, but I am doubting it at this time until I have further information.
Regarding the cable, I selected values that I thought would be much worse than any real audio cable so there wouldn't be any room for doubt. The values for the Belden cable that I did locate were well below the values I used.
I will do a few more tests with high load cap and higher signals to see if there are any noticeable changes.
OK, I created five foot cables (and then I hit the PSpice demo parts limit).
200nH, 30p, 30mR per foot and 2000p, 100K load.
1Vp input, 7Vp output.
28mV second harmonic for the FVP
25mV second harmonic for the standard CF
Or, basically, no difference.
If there is something in the real world, other than the micro-features that PRR refers to, I sure would like to know what it is.
200nH, 30p, 30mR per foot and 2000p, 100K load.
1Vp input, 7Vp output.
28mV second harmonic for the FVP
25mV second harmonic for the standard CF
Or, basically, no difference.
If there is something in the real world, other than the micro-features that PRR refers to, I sure would like to know what it is.
fdegrove said:
If you look at the evolution of this pre-amp (the 5 stands for the 5th version) over the last 20 yrs or so you will see that AW has continued to stive for better power supplies & better downward dynamic range. Miss Piggy (his term for a straight cathode follower) has always been something he hasn't liked the sound of and this is many years of work to make a low impedance output stage that sounds good... Joe is using a very similar circuit in the front end of his tube buffered gain clone.
Give me a year or 2 and i'll reposrt back, the parts for an RTP are slowly gathering (and the skill to actually assemble it)
dave
Hi,
As it is, I don't feel all too comfortable with the idea of anyone needing twenty years to perfect an idea.
I may well be wrong but at the the time of writing I feel the WCF with its PS I presented before are doing what I expect from a buffer.
For all things audio, I can't find fault with it.
As for dynamic range, given the PS voltages, I'm darn right sure the circuit I presented wins hands down...
I invite anyone to think it through, build it or preferably both.
Cheers,
As it is, I don't feel all too comfortable with the idea of anyone needing twenty years to perfect an idea.
I may well be wrong but at the the time of writing I feel the WCF with its PS I presented before are doing what I expect from a buffer.
For all things audio, I can't find fault with it.
As for dynamic range, given the PS voltages, I'm darn right sure the circuit I presented wins hands down...
I invite anyone to think it through, build it or preferably both.
Cheers,
fdegrove said:
That is a pretty ill conceived statement Frank... many of the best designers in the world have been honing their craft for longer than this ... Nelson Pass' 1st Thresholds are now over 20 years old and i'm sure he was starting to work things out long before... Ted Jordan is still working on perfecting a design i 1st ran into 30 years ago, and the list goes on & on & on....
dave
Hi runeigth, thankyou
I see that your parameters are different
from those in the koren website.
Are they based on published
plate curves (which one?) or have you fitted
a real tube?
However, I think Koren model is
able to well describe tube behavior.
Surely in the negative grid region.
Federico
I see that your parameters are different
from those in the koren website.
Are they based on published
plate curves (which one?) or have you fitted
a real tube?
However, I think Koren model is
able to well describe tube behavior.
Surely in the negative grid region.
Federico
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.