Hi there Rongon. If you're using a DL110 wouldn't a 47K resistor be more appropriate at the input rather than 1K?
In my experience triode-connected E810Fs are more stable than triode-connected D3as. Interestingly, triode-connected they are interchangeable (as are the triode-connected E280F and E282F).
In my experience triode-connected E810Fs are more stable than triode-connected D3as. Interestingly, triode-connected they are interchangeable (as are the triode-connected E280F and E282F).
I think 40dB gain would be fine in my case. My line stage is a single 5687 triode (plate loaded voltage amp) per side.
I'm thinking the El Cheapo Valve preamp might be worth a try. I like its simplicity. The 12AX7 in the first stage gives me pause, but it would make it easy to build.
Other option is RJM 6DJ8, although I'm not crazy about the idea of running a 6DJ8 with only 5mA of plate current. I've found 6DJ8 at such low current sounds "transistory" -- somehow sizzly or hashy.
Are there other [simple] 6DJ8 RIAA preamp designs I should look at?
Re: Denon DL-110 -- My understanding is that the load impedance chosen should be between 5 and 10 times the source impedance. (DL-110 and DL-160 specs show output Z of 140 Ohms.) Perhaps a 1.2k or 1.5k resistor would work better?
The complaint I've seen is that the DL-110 run into 47k will sound "edgy" or overly bright.
--
I'm thinking the El Cheapo Valve preamp might be worth a try. I like its simplicity. The 12AX7 in the first stage gives me pause, but it would make it easy to build.
Other option is RJM 6DJ8, although I'm not crazy about the idea of running a 6DJ8 with only 5mA of plate current. I've found 6DJ8 at such low current sounds "transistory" -- somehow sizzly or hashy.
Are there other [simple] 6DJ8 RIAA preamp designs I should look at?
Re: Denon DL-110 -- My understanding is that the load impedance chosen should be between 5 and 10 times the source impedance. (DL-110 and DL-160 specs show output Z of 140 Ohms.) Perhaps a 1.2k or 1.5k resistor would work better?
The complaint I've seen is that the DL-110 run into 47k will sound "edgy" or overly bright.
--
Yeah, this is one of those things I really don't understand.
You want to load the source properly, but with the DL-110's Zout of 160 Ohms, a standard 47k Ohm load is almost 300 times the cartridge's Zout. That seems really excessive. (Incidentally, the low-output MC DL-103's Zout is 40 Ohms, and you'll usually see it run into a 100 Ohm load. Why would a high-output MC like the DL-110 work by different rules? They are both moving coils, so inductance is less an issue than with MM's, right?)
I read a lot of forum posts about the Denon DL-110/DL-160 carts and there was no shortage of debate over this. Most claimed the 1000 Ohm load was better sounding. That also seems to agree with common practice, where the load impedance is usually 4 to 10 times the source impedance. I don't understand the physical reasoning behind this, only that it's "common practice" as far as I can tell.
Does cartridge loading follow the same rules as in an amplifier stage? Any amplifier stage will have an output impedance. Let's take a 12AX7 with something like 70k Ohm internal (plate) resistance. That stage will usually get a plate load resistor of between 100k and 270k Ohms. The total Zout might be about 30k to 50k Ohms. (It can't be as simple as rp and Rp in parallel, can it? Where's that RDH4...)
The load resistor on the output of the stage combined with the Zin of the next stage will usually be roughly 80k to 200k Ohms. That's 3 to 4 times larger than the preceding stage Zout. A 10:1 ratio would be more "ideal."
Judging by that logic, it might be better to have the DL-110 running into 1500 Ohms, or 10 times the Zout of the cartridge. But 47k Ohms? I don't get it.
Maybe someone with more expertise than I have can shed some light on this...
Forum posts about this...
http://www.vinylengine.com/phpBB2/v...start=15&sid=8376f841e2c79c063f9402d090f9f0b7
http://www.pinkfishmedia.net/forum/showthread.php?t=42777
One of those has a post where it's suggested that 30X the source impedance is optimal, which would put the best input resistance for a DL-110 at 5k Ohms. Maybe I'll try that...
--
You want to load the source properly, but with the DL-110's Zout of 160 Ohms, a standard 47k Ohm load is almost 300 times the cartridge's Zout. That seems really excessive. (Incidentally, the low-output MC DL-103's Zout is 40 Ohms, and you'll usually see it run into a 100 Ohm load. Why would a high-output MC like the DL-110 work by different rules? They are both moving coils, so inductance is less an issue than with MM's, right?)
I read a lot of forum posts about the Denon DL-110/DL-160 carts and there was no shortage of debate over this. Most claimed the 1000 Ohm load was better sounding. That also seems to agree with common practice, where the load impedance is usually 4 to 10 times the source impedance. I don't understand the physical reasoning behind this, only that it's "common practice" as far as I can tell.
Does cartridge loading follow the same rules as in an amplifier stage? Any amplifier stage will have an output impedance. Let's take a 12AX7 with something like 70k Ohm internal (plate) resistance. That stage will usually get a plate load resistor of between 100k and 270k Ohms. The total Zout might be about 30k to 50k Ohms. (It can't be as simple as rp and Rp in parallel, can it? Where's that RDH4...)
The load resistor on the output of the stage combined with the Zin of the next stage will usually be roughly 80k to 200k Ohms. That's 3 to 4 times larger than the preceding stage Zout. A 10:1 ratio would be more "ideal."
Judging by that logic, it might be better to have the DL-110 running into 1500 Ohms, or 10 times the Zout of the cartridge. But 47k Ohms? I don't get it.
Maybe someone with more expertise than I have can shed some light on this...
Forum posts about this...
http://www.vinylengine.com/phpBB2/v...start=15&sid=8376f841e2c79c063f9402d090f9f0b7
http://www.pinkfishmedia.net/forum/showthread.php?t=42777
One of those has a post where it's suggested that 30X the source impedance is optimal, which would put the best input resistance for a DL-110 at 5k Ohms. Maybe I'll try that...
--
Tweaking, tweaking...
I checked out the TubeCAD Journal RIAA calculator. Using the separate 2122Hz and 50Hz filters, I saw that those 100k series resistors did indeed look wrong. However, looking in my old collection of Sound Practices mags from fifteen years ago (!), Arthur Loesch and JC Morrison were both using 100k Ohms in this spot in every RIAA schematic of theirs I could find. I think I see what they were doing. The pole frequency of 2122 Hz is reachable with RC of 100k and 750pF. 50Hz is obtained from 100k and .031uF. But that completely ignores Miller capacitance of the subsequent stage and the Zout of the previous stage.
There is a setting for R series before the series load resistor. Is this meant to be the Zout of the previous stage? I took it to mean this, and did some fiddling.
The 100k resistors are difficult to get to, and I have a boxful of polystyrene caps, so I tried changing cap values.
In the calculator, using the the previous stage's Zout for Rs, I arrived at the 2122Hz pole by changing to 900pF. (Actually, it's a little off, but by 2%, a lot less than before.)
I had to make a ballpark guess at the 50Hz pole, but .036uF looked to be closer, in the calculator at least.
My LCR meter is a cheapie, but I think it's in the ballpark.
According to the calculator, that should be closer to correct. But I have absolutely no idea. I tried to multiply RC to get the time constants (75uS, 318.3uS or 3183uS) and am not even close.
I went to http://www.muzique.com/schem/filter.htm and tried to figure out the pole frequencies.
I figure Zout of 7119 = 3k Ohms
and Zout of 7788 triode = 2k Ohms
For 2122Hz RC I got 102000 Ohms * .0009uF = 1734Hz. 750pF would be closer, at 2081.5Hz. For 2122Hz, C would need to be 735.7pF!
For 50Hz pole RC I got 103000 * .031uF = 49.9Hz.
Why doesn't the TubeCAD Journal calculator match up with the simple RC calculations?
Miller capacitance would be in series with the signal, no? Since the Miller capacitance is not known, is there any ballpark estimate people make for 7119/5687 family of tubes?
Help?
--
I checked out the TubeCAD Journal RIAA calculator. Using the separate 2122Hz and 50Hz filters, I saw that those 100k series resistors did indeed look wrong. However, looking in my old collection of Sound Practices mags from fifteen years ago (!), Arthur Loesch and JC Morrison were both using 100k Ohms in this spot in every RIAA schematic of theirs I could find. I think I see what they were doing. The pole frequency of 2122 Hz is reachable with RC of 100k and 750pF. 50Hz is obtained from 100k and .031uF. But that completely ignores Miller capacitance of the subsequent stage and the Zout of the previous stage.
There is a setting for R series before the series load resistor. Is this meant to be the Zout of the previous stage? I took it to mean this, and did some fiddling.
The 100k resistors are difficult to get to, and I have a boxful of polystyrene caps, so I tried changing cap values.
In the calculator, using the the previous stage's Zout for Rs, I arrived at the 2122Hz pole by changing to 900pF. (Actually, it's a little off, but by 2%, a lot less than before.)
I had to make a ballpark guess at the 50Hz pole, but .036uF looked to be closer, in the calculator at least.
My LCR meter is a cheapie, but I think it's in the ballpark.
According to the calculator, that should be closer to correct. But I have absolutely no idea. I tried to multiply RC to get the time constants (75uS, 318.3uS or 3183uS) and am not even close.
I went to http://www.muzique.com/schem/filter.htm and tried to figure out the pole frequencies.
I figure Zout of 7119 = 3k Ohms
and Zout of 7788 triode = 2k Ohms
For 2122Hz RC I got 102000 Ohms * .0009uF = 1734Hz. 750pF would be closer, at 2081.5Hz. For 2122Hz, C would need to be 735.7pF!
For 50Hz pole RC I got 103000 * .031uF = 49.9Hz.
Why doesn't the TubeCAD Journal calculator match up with the simple RC calculations?
Miller capacitance would be in series with the signal, no? Since the Miller capacitance is not known, is there any ballpark estimate people make for 7119/5687 family of tubes?
Help?
--
You also need to consider source resistance...
In any case, you can figure out the Miller capacitance pretty easily. Get a 7119 datasheet. Look up the mu and rp at the current you're running. You can then calculate the gain (A) by mu x R/(R + rp). With gain in hand, you can approximate the input capacitance by Cin = Cgk + A(Cgp). The Miller capacitance shunts the final capacitor to ground.
If there's cathode degeneration, replace rp by rp + (mu + 1)xRk.
In any case, you can figure out the Miller capacitance pretty easily. Get a 7119 datasheet. Look up the mu and rp at the current you're running. You can then calculate the gain (A) by mu x R/(R + rp). With gain in hand, you can approximate the input capacitance by Cin = Cgk + A(Cgp). The Miller capacitance shunts the final capacitor to ground.
If there's cathode degeneration, replace rp by rp + (mu + 1)xRk.
Thanks for that. So... If the 7119's Miller capacitance is say, 70pF, that would be in parallel with the cap forming the 75uS time constant, correct?
So if the 75uS time constant is derived from a 750pF cap shunt to ground and a 10k series resistor, I'd want to subtract the Miller capacitance of 70pF from that 750uF cap to get a final value of 680pF...
Right?
And how does the source resistance (is that the previous stage's Zout?) affect the 75uS time constant in this instance? In series with the actual Rseries (so added to it)?
--
So if the 75uS time constant is derived from a 750pF cap shunt to ground and a 10k series resistor, I'd want to subtract the Miller capacitance of 70pF from that 750uF cap to get a final value of 680pF...
Right?
And how does the source resistance (is that the previous stage's Zout?) affect the 75uS time constant in this instance? In series with the actual Rseries (so added to it)?
--
You find the value of the 75uS capacitor from the surrounding impedances rather than starting with a fixed value of capacitor. You add the output impedance of the 7788 (a parallel combination of the load resistor and the internal impedance of the 7788 at the operating point) to the series resistor (100K), then work out the parallel combination of this number with the grid resistor of the 7119 (presumably 1M). You then divide this figure into 75uS to get a capacitance and subtract the input capacitiance of the 7119 from it to get the final value of the capacitor you need. All this is derived from Morgan Jones' book, which has worked-out examples to make things clearer.
barretter said:
Agree with everything else, but would like to amend this a bit. The preceding tube section impedance is essentially a given value, once the operating conditions have been selected. But the added series resistance is not. You can vary this value as well. If it's several times higher than the tube section impedance, then as the tube ages, or tubes are changed, the effect on the RIAA curve will be minimized. On the other hand, if the resistance is too high, the noise, and miller capacitance (which is usually an estimate) of the following stage will be more significant. So somewhere between the two extremes is the place to be.
Sheldon
I need to get the M. Jones books. I gather the "Valve Amplifiers" book focus on design, while "Making Valve Amplifiers" focuses on layout, construction, etc. I need a really good discussion of grounding...
Re: larger value series R -- I see I goofed up again, and described the series resistor as 10k (which would yield a 7.5uS time constant!). Yes, 100k...
OK, if 100k makes the Miller capacitance of the following 7119 grid too influential, what if I changed the 100k resistance to 49.9k, then the cap would need to be doubled to 1500pF, correct? I could do that. 70pF or so wouldn't do much to that value. Not as much as it would to 750pF. Makes sense.
Since this is a split RIAA network, can I assume that the change in this RC combo's values won't substantially affect the RC values I choose for the 3180uS RC (in the next stage)?
SY -- A decade higher than the rp of the preceding stage? Would that be a convenient rule of thumb? In that case, I'd only really need 30k or in that ballpark...
Re: larger value series R -- I see I goofed up again, and described the series resistor as 10k (which would yield a 7.5uS time constant!). Yes, 100k...
OK, if 100k makes the Miller capacitance of the following 7119 grid too influential, what if I changed the 100k resistance to 49.9k, then the cap would need to be doubled to 1500pF, correct? I could do that. 70pF or so wouldn't do much to that value. Not as much as it would to 750pF. Makes sense.
Since this is a split RIAA network, can I assume that the change in this RC combo's values won't substantially affect the RC values I choose for the 3180uS RC (in the next stage)?
SY -- A decade higher than the rp of the preceding stage? Would that be a convenient rule of thumb? In that case, I'd only really need 30k or in that ballpark...
No it won't affect the next stage. By the way you have to implement two further time constants : 3180uS and 318uS, not just 3180. You still seem to think that you can implement the time constants by a simple RC combination, but it's more complicated than that. You have to take the resistances and capacitances of the preceding and following amplifying stages around your R and C otherwise your filter will be inaccurate. I recommend that you buy Morgan's book.
A decade higher than the source impedance of the driving stage. That's the effective rp in parallel with the following stage's grid leak.
I second, third, and fourth the STRONG recommendation that you get the Morgan Jones book and follow his analysis. 99% of what you've been asking is explained there in great detail (and in his very readable style).
75uS, 318us, 3180uS.... OK, a light just went on. That's the one on the sign that reads, "You Don't Know What You're Getting Into!"
This will be a prolonged process, I can tell.
In the meantime, I guess I do need to equip myself with a phono stage that won't drive me crazy. Hagtech or "El Cheapo Valve"?
Then a lot of reading and a re-wiring of the current project.
BTW, did you see that the Arthur Loesch preamp is available again?
http://www.tempoelectric.com/phono.htm ($5500)
--
This will be a prolonged process, I can tell.
In the meantime, I guess I do need to equip myself with a phono stage that won't drive me crazy. Hagtech or "El Cheapo Valve"?
Then a lot of reading and a re-wiring of the current project.
BTW, did you see that the Arthur Loesch preamp is available again?
http://www.tempoelectric.com/phono.htm ($5500)
--
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.