So, getting back to phone stages, indeed:
Looking at three likely scenarios:
1.) MC cartridge > step-up transformer > RIAA stage (40dB gain nominal)
A typical situation (mine actually) might be a 4 ohm MC source resistance driving a 20dB step-up transformer (1:10 turns ratio, or so). The source impedance reflected to the secondary is then about 400 ohms. The transformer will add some internal resistance, but the source resistance presented to the RIAA stage should still be but a few hundred ohms, if using a good transformer like the Jensen JT-346-AX. Right now, I’m toying with a toroid with just tens of ohms of added resistance in the secondary. So the source resistance presented to the RIAA stage can/should be just above the noise Req of a well-biased 6DJ8. It is surely a low enough source impedance to justify efforts to reduce any unnecessary RIAA stage noise. As I said earlier, resistors in and around the first tube’s grid and cathode should be at or below a few hundred ohms to keep noise as low as possible.
2.) MC cartridge > RIAA stage (60dB nominal, “headamp” included)
If using tubes to directly amplify the tiny source voltage of an MC as I have done in the past, then an all-out noise assault is not only justifiable, it’s absolutely necessary. Here’s where we break out the fancy tubes, put them in parallel, and make powers supplies with excruciatingly low ripple, etc. We keep resistances low, low (“Resistance is futile”). A 12AX7 need not apply here; sorry old friend. This is where solid state temptations arise, either by way of a decent opamp, or as SY suggested above, using a FET like a 2SK170 in a cascode with a high gm tube. The 2SK170 has an extremely low noise floor and a very low flicker noise turnover frequency as well. In this one use and location, I can excuse the concerns with using FETs. The signals are tiny, so that basic linearity isn’t a concern. The non-linear and large input capacitances are driven by the low MC source impedance (1.5 to 20 ohms) so the resultant poles and phase shifts are moved well above the audio band. The cascode tube’s cathode presents a moderately low drain load resistance to the FET so it can’t move much, keeping Miller effect low. Still, I’ve used an all-tube design with success, but it was complicated and ran hot. The effort made me respect the step-up transformer even more.
3.) MM cartridge > RIAA stage (40 dB nominal)
Are people still using MMs? 😉 Seriously, here is where a 12AX7 can still strut its stuff and in fact it has been doing so for decades. (But so can a 6DJ8 and other tubes as well). MM source impedances might be 5K in the midrange and around 30K on top due to the large source inductance. The 12AX7 can hold its ground against these values, but my point about being careful to not overextend front-end resistors values still applies. For example, in some RIAA designs, I’ve seen a large value resistor in series with the first grid (usually a 12AX7 in this case), in the range of 10K to 33K, presumably to create an RF filter with the Miller capacitance. I would attempt to address RF, if is a problem, by other means, and reduce or remove this noise-making resistor, nevermind whether it’s metal film or carbon comp.
If you plan to build a 40dB standard RIAA stage with enough flexibility to use either a MM or a MC with step-up, you’ll need to worry about noise (no, not to “lose sleep” over it). I would advise using a high gm tube as the more versatile approach. I think Morgan Jones covered RIAA design trade-offs well in his book. Worth a read if you haven’t already. Enjoy...
Kuei Yang Wang said:
Looking at three likely scenarios:
1.) MC cartridge > step-up transformer > RIAA stage (40dB gain nominal)
A typical situation (mine actually) might be a 4 ohm MC source resistance driving a 20dB step-up transformer (1:10 turns ratio, or so). The source impedance reflected to the secondary is then about 400 ohms. The transformer will add some internal resistance, but the source resistance presented to the RIAA stage should still be but a few hundred ohms, if using a good transformer like the Jensen JT-346-AX. Right now, I’m toying with a toroid with just tens of ohms of added resistance in the secondary. So the source resistance presented to the RIAA stage can/should be just above the noise Req of a well-biased 6DJ8. It is surely a low enough source impedance to justify efforts to reduce any unnecessary RIAA stage noise. As I said earlier, resistors in and around the first tube’s grid and cathode should be at or below a few hundred ohms to keep noise as low as possible.
2.) MC cartridge > RIAA stage (60dB nominal, “headamp” included)
If using tubes to directly amplify the tiny source voltage of an MC as I have done in the past, then an all-out noise assault is not only justifiable, it’s absolutely necessary. Here’s where we break out the fancy tubes, put them in parallel, and make powers supplies with excruciatingly low ripple, etc. We keep resistances low, low (“Resistance is futile”). A 12AX7 need not apply here; sorry old friend. This is where solid state temptations arise, either by way of a decent opamp, or as SY suggested above, using a FET like a 2SK170 in a cascode with a high gm tube. The 2SK170 has an extremely low noise floor and a very low flicker noise turnover frequency as well. In this one use and location, I can excuse the concerns with using FETs. The signals are tiny, so that basic linearity isn’t a concern. The non-linear and large input capacitances are driven by the low MC source impedance (1.5 to 20 ohms) so the resultant poles and phase shifts are moved well above the audio band. The cascode tube’s cathode presents a moderately low drain load resistance to the FET so it can’t move much, keeping Miller effect low. Still, I’ve used an all-tube design with success, but it was complicated and ran hot. The effort made me respect the step-up transformer even more.
3.) MM cartridge > RIAA stage (40 dB nominal)
Are people still using MMs? 😉 Seriously, here is where a 12AX7 can still strut its stuff and in fact it has been doing so for decades. (But so can a 6DJ8 and other tubes as well). MM source impedances might be 5K in the midrange and around 30K on top due to the large source inductance. The 12AX7 can hold its ground against these values, but my point about being careful to not overextend front-end resistors values still applies. For example, in some RIAA designs, I’ve seen a large value resistor in series with the first grid (usually a 12AX7 in this case), in the range of 10K to 33K, presumably to create an RF filter with the Miller capacitance. I would attempt to address RF, if is a problem, by other means, and reduce or remove this noise-making resistor, nevermind whether it’s metal film or carbon comp.
If you plan to build a 40dB standard RIAA stage with enough flexibility to use either a MM or a MC with step-up, you’ll need to worry about noise (no, not to “lose sleep” over it). I would advise using a high gm tube as the more versatile approach. I think Morgan Jones covered RIAA design trade-offs well in his book. Worth a read if you haven’t already. Enjoy...
Giaime,
Looking at your RIAA phonostage:
Are you pleased with the sound?
What cartridge are you using as you listen to it?
What kind of music do you listen to with it?
What are the other components in your system? Turntable and tonearm? Head amp? Linestage? Amplifier? Speakers? I would like to know these things before I attempt to build one myself.
Thank you,
John
Looking at your RIAA phonostage:
Are you pleased with the sound?
What cartridge are you using as you listen to it?
What kind of music do you listen to with it?
What are the other components in your system? Turntable and tonearm? Head amp? Linestage? Amplifier? Speakers? I would like to know these things before I attempt to build one myself.
Thank you,
John
Giaime, (Off thread alert). Let me also offer my congratulations on Italy's soccer victory! By the way, you live in what is probably my favorite city in the world. You and I share at least one environmental condition that can affect audio - humidity!
Yes Brian, it's extremely humid here, and if that isn't good for critical audio listening, it surely isn't good for studying 
To John: as you probably noticed from the title of this thread, those two phono stages that we're talking about, this
http://www.giaime.altervista.org/SuperRIAA.html
and this
http://www.giaime.altervista.org/NCPSsimpler.html
are in the developing stage. Nothing has been built or tested yet.
I'm now using a 2 stage phono amp, I've built it some time ago:
http://www.giaime.altervista.org/DefinitivePhonostage.gif
(note: the link might not work. If it is so, go here http://www.giaime.altervista.org/mypreamp.html and look for the picture under the "This is the definitive phono stage" text).
If you want to try prototyping one of them, and let us know how they sound, I will give all the support I can.
To John: as you probably noticed from the title of this thread, those two phono stages that we're talking about, this
http://www.giaime.altervista.org/SuperRIAA.html
and this
http://www.giaime.altervista.org/NCPSsimpler.html
are in the developing stage. Nothing has been built or tested yet.
I'm now using a 2 stage phono amp, I've built it some time ago:
http://www.giaime.altervista.org/DefinitivePhonostage.gif
(note: the link might not work. If it is so, go here http://www.giaime.altervista.org/mypreamp.html and look for the picture under the "This is the definitive phono stage" text).
If you want to try prototyping one of them, and let us know how they sound, I will give all the support I can.
Konnichiwa,
As I have been playing around with "current input" MC stages I did a survey of Cartridge sensitivities vs resistance across a wide range of MC Cartridges the mean was around 0.03mV/Ohm @ 5cm/S and an average output of 0.3mV for low output MC's. Significant outliers could be observed into either direction which suggested to me the need for a three-stepped "current gain" based on 0.01mV/Ohm, 0.03mV/Ohm and 0.1mV/Ohm.
For my evaluation of noisebehaviour in MC stepped up to MM level I tend towards using the mean values, see below.
If we assume a 40db gain Moving Magnet Phono Stage (which is about right as 5mV@5cm/S 0db and 25mV@25cm/S for "analogue full scale" give around 1.75V RMS out for "analogue full scale" which is about parity with CD if the CD was recorded conservatively) we need thus on average a 1:20 stepup.
Hence we will have on average a cartridge impedance of 10 Ohm, which translates into an output impedance from the MC Stepup Transformer as 4KOhm plus any of the transformers winding resistance (primary * 400 plus secondary), in the case of the S&B TX-103 around 450 Ohm.
So, your example seems to be from the extreme end of the range of MC cartridges.
So, shall we agree that depending on the MC Pickup and transformer used source impedances anywhere from a few hundert ohm (very rare) over a few kilohm (average) to as much as 20K (some low output Koetso's and the Denon DL-103 come to mind) will be experienced by an MM Phono tage used with MC Pickups and Stepup transformers?
So, for the average we are in the region of the equvalent noise impedance of the ECC83, not that this much in Phonostages where other effects dominate and most other valves tend to be noiser than the ECC83.
Yet both Aesthetix and Corad Johnson make Phonostages in exactly THAT way (each with 4 ECC83 sections in parallel) and get rather reasonably low noise, based on measurements by Stereophile and others. So, what gives?
Maybe it is less WHAT you use, but the how?
And maybe you should have another look at the bit's in Giame's article on noise. While making several claims that do not really apply to audio it makes a very crucial point!
Namely that while shot noise decreases with increasing anode current, the other noise sources which in our typhical valves INCREASE and these noise sources outweigh other sources by a very substantial degree below a few KHz and which are NOT significantly attenuated by the RIAA (unlike the frequencies at which shot noise becomes dominant).
So, running your ECC88 at anything near nominal Anode current will actually make the LF noise WORSE than the graphs I cited earlier show....
At the same time the ECC83 will have even lower LF noise than shown in the graph if operated at LESS current (and less anode voltage).
Where does this leaves us?
At a simple point that as with so many things what is noisy and what is quiet is much more in the skill of implementation and of course at the point that very little beats a 2SK147 in noise, except several 2SK147's in parallel.
Sayonara
Brian Beck said:
As I have been playing around with "current input" MC stages I did a survey of Cartridge sensitivities vs resistance across a wide range of MC Cartridges the mean was around 0.03mV/Ohm @ 5cm/S and an average output of 0.3mV for low output MC's. Significant outliers could be observed into either direction which suggested to me the need for a three-stepped "current gain" based on 0.01mV/Ohm, 0.03mV/Ohm and 0.1mV/Ohm.
For my evaluation of noisebehaviour in MC stepped up to MM level I tend towards using the mean values, see below.
If we assume a 40db gain Moving Magnet Phono Stage (which is about right as 5mV@5cm/S 0db and 25mV@25cm/S for "analogue full scale" give around 1.75V RMS out for "analogue full scale" which is about parity with CD if the CD was recorded conservatively) we need thus on average a 1:20 stepup.
Hence we will have on average a cartridge impedance of 10 Ohm, which translates into an output impedance from the MC Stepup Transformer as 4KOhm plus any of the transformers winding resistance (primary * 400 plus secondary), in the case of the S&B TX-103 around 450 Ohm.
So, your example seems to be from the extreme end of the range of MC cartridges.
So, shall we agree that depending on the MC Pickup and transformer used source impedances anywhere from a few hundert ohm (very rare) over a few kilohm (average) to as much as 20K (some low output Koetso's and the Denon DL-103 come to mind) will be experienced by an MM Phono tage used with MC Pickups and Stepup transformers?
So, for the average we are in the region of the equvalent noise impedance of the ECC83, not that this much in Phonostages where other effects dominate and most other valves tend to be noiser than the ECC83.
Brian Beck said:
Yet both Aesthetix and Corad Johnson make Phonostages in exactly THAT way (each with 4 ECC83 sections in parallel) and get rather reasonably low noise, based on measurements by Stereophile and others. So, what gives?
Maybe it is less WHAT you use, but the how?
And maybe you should have another look at the bit's in Giame's article on noise. While making several claims that do not really apply to audio it makes a very crucial point!
Namely that while shot noise decreases with increasing anode current, the other noise sources which in our typhical valves INCREASE and these noise sources outweigh other sources by a very substantial degree below a few KHz and which are NOT significantly attenuated by the RIAA (unlike the frequencies at which shot noise becomes dominant).
So, running your ECC88 at anything near nominal Anode current will actually make the LF noise WORSE than the graphs I cited earlier show....
At the same time the ECC83 will have even lower LF noise than shown in the graph if operated at LESS current (and less anode voltage).
Where does this leaves us?
At a simple point that as with so many things what is noisy and what is quiet is much more in the skill of implementation and of course at the point that very little beats a 2SK147 in noise, except several 2SK147's in parallel.
Sayonara
Kuei Yang Wang said:
On those notes, let's leave this in some agreement.
Chacun à son goût, as they say.
Konnichiwa,
I usually prefer to build and test things, in a variety of ways, prior to publishing them.
Your website seems to be down right, but I suppose the schematics are the same as posted in this thread?:
http://www.diyaudio.com/forums/showthread.php?s=&threadid=77512
If so, may I ask if you have actually verified the accuracy of the RIAA equalisation and measured distortion behaviour?
Past that, last time I proposed a linestage like the one you show "they" all jumped up and cried foul and went on to debate how awfull such a thing is. Nice to see I'm not alone with that approach.
Ciao/Sayonara
Giaime said:
I usually prefer to build and test things, in a variety of ways, prior to publishing them.
Giaime said:
Your website seems to be down right, but I suppose the schematics are the same as posted in this thread?:
http://www.diyaudio.com/forums/showthread.php?s=&threadid=77512
If so, may I ask if you have actually verified the accuracy of the RIAA equalisation and measured distortion behaviour?
Past that, last time I proposed a linestage like the one you show "they" all jumped up and cried foul and went on to debate how awfull such a thing is. Nice to see I'm not alone with that approach.
Ciao/Sayonara
Hi Thoersten,
You know, we all started somewhere. Many of us have learned through trial and error. I'll bet there are commercial offerings out there that are not as good as what Giaime is planning.
In the spirit of DIY, getting something built and running is the primary goal. Learning from that is even better. We can't just start by building the best whatever and getting it right the first time.
Anyone who has past this stage and doesn't feel like encouraging other DIYer's incrementally is missing out on the best part of human interactions. I am part way in my learning and wouldn't be this far if it weren't for the kind assistance of others. Returning the kindness to other people is the best way to return the favour.
So suggest better practices, write what you have found to be better, but allow some experimentation and personal discovery. Sometimes we need to make mistakes to learn. People will disagree from time to time.
-Chris
You know, we all started somewhere. Many of us have learned through trial and error. I'll bet there are commercial offerings out there that are not as good as what Giaime is planning.
In the spirit of DIY, getting something built and running is the primary goal. Learning from that is even better. We can't just start by building the best whatever and getting it right the first time.
Anyone who has past this stage and doesn't feel like encouraging other DIYer's incrementally is missing out on the best part of human interactions. I am part way in my learning and wouldn't be this far if it weren't for the kind assistance of others. Returning the kindness to other people is the best way to return the favour.
So suggest better practices, write what you have found to be better, but allow some experimentation and personal discovery. Sometimes we need to make mistakes to learn. People will disagree from time to time.
-Chris
Konnichiwa,
Absolutely. I learned more from the designs no-one ever hears about because they where "NDFG". I learned a lot for example from the time when I tried to convert the Arthur Loesch preamp schematic to use different valves without changing much else just eyeballing some minimal changes.
The RIAA EQ was miles off and it sounded horrible. So I build several more versions of it untill I understood the circuit and principles and had achieved a variation that I felt was worth mentioning.
Sayonara
anatech said:
Absolutely. I learned more from the designs no-one ever hears about because they where "NDFG". I learned a lot for example from the time when I tried to convert the Arthur Loesch preamp schematic to use different valves without changing much else just eyeballing some minimal changes.
The RIAA EQ was miles off and it sounded horrible. So I build several more versions of it untill I understood the circuit and principles and had achieved a variation that I felt was worth mentioning.
Sayonara
Hi Thoersten,
Well, I won't mention the horrors I created. But the point was, I created them! 😀
I did learn a great deal that way. Mind you, I still create some awful sounding stuff from time to time. 😉
-Chris
Well, I won't mention the horrors I created. But the point was, I created them! 😀
I did learn a great deal that way. Mind you, I still create some awful sounding stuff from time to time. 😉
-Chris
Very interesting points Chris and Thoersten,
but I'd like to make a little step back.
So, talking about my lastest design, this schematic attached here, as has been pointed out there are such points that I should improve, feel free to add more:
1) back to the drawing board defining operating points for the stages - since the general rule that I applied (high gm but low current) is a semplification.
2) R10 isn't needed 😉
3) re-thinking of the RIAA network to scale down resistors, R7 is going to make much noise.
4) probably C10 isn't needed due to the input capacitance of the following ECC81 stage.
Any comments?
but I'd like to make a little step back.
So, talking about my lastest design, this schematic attached here, as has been pointed out there are such points that I should improve, feel free to add more:
1) back to the drawing board defining operating points for the stages - since the general rule that I applied (high gm but low current) is a semplification.
2) R10 isn't needed 😉
3) re-thinking of the RIAA network to scale down resistors, R7 is going to make much noise.
4) probably C10 isn't needed due to the input capacitance of the following ECC81 stage.
Any comments?
Attachments
Konnichiwa,
First, consider ditching the cathode bypass capaictors. Electrolytic capacitors are a very poor choice here, as they are quite non-linear (nonlinear even for Douglas "All Sounds Same" Self to recommend against their use in sonically sensitive positions.
As in the Cathode they see the full output stage AC current, but the error voltgae is developed in the devices input loop the distortion of the capacitor is amplified by the devices gain.
Second, if you leave the first stage cathode unbypassed the full cascode will have a very stable and high output impedance. This way you can emply a direct full shunt RIAA and operate the whole frontend in pure transconductance mode. It also makes for lower impedances, which is a good thing, if there is no other penalty.
Changing your output Valve to something meatier than an ECC81 would allow you to take the output from the Anode (the high load impedance from the Mu-Follower is always good, the sound from it's cathode much less so, sadly). Get rid (again) of the cathode bypass and use fixed gridbias if you find the use of gridleak bias too "voodoo".
In fact, I'd be tempted to invert the configuration and to use the ECC81 (or better the 6072A) in the first stage and the ECC88 in the second.
Ciao/Sayonara
Giaime said:
First, consider ditching the cathode bypass capaictors. Electrolytic capacitors are a very poor choice here, as they are quite non-linear (nonlinear even for Douglas "All Sounds Same" Self to recommend against their use in sonically sensitive positions.
As in the Cathode they see the full output stage AC current, but the error voltgae is developed in the devices input loop the distortion of the capacitor is amplified by the devices gain.
Second, if you leave the first stage cathode unbypassed the full cascode will have a very stable and high output impedance. This way you can emply a direct full shunt RIAA and operate the whole frontend in pure transconductance mode. It also makes for lower impedances, which is a good thing, if there is no other penalty.
Changing your output Valve to something meatier than an ECC81 would allow you to take the output from the Anode (the high load impedance from the Mu-Follower is always good, the sound from it's cathode much less so, sadly). Get rid (again) of the cathode bypass and use fixed gridbias if you find the use of gridleak bias too "voodoo".
In fact, I'd be tempted to invert the configuration and to use the ECC81 (or better the 6072A) in the first stage and the ECC88 in the second.
Ciao/Sayonara
Thank you very much. In fact I'm thinking about a mu-follower frontend for the ECC81, I will use another type of RIAA network that will give less insertion loss at 1kHz, and will have smaller value resistors (so less noise).
But I was also thinking of a simple SS CCS on top of a grounded cathode ECC81.
What could be the best option? Obviously the second one looks interesting, only one triode section for the first stage.
But I was also thinking of a simple SS CCS on top of a grounded cathode ECC81.
What could be the best option? Obviously the second one looks interesting, only one triode section for the first stage.
Konnichiwa,
As noted before, stop worrying about the resistors, start worrying about the valves....
I would keep the cascode as frontend. In fact, that was the MAIN compromise in my "Valve El Cheapo", in the interrest of simplicity and ease of building, no cascode.
As the cascode is basically a very linear voltage controlled current source you can realise the EQ purely as frequency dependent shunt device, with the lowest timeconstant set by the cascodes anode load.
This way headroom (which is I believe what Bill Clinton called the Oval Office? I wonder what Shrub calls it?) is maximised.
Ciao/Sayonara
Giaime said:
As noted before, stop worrying about the resistors, start worrying about the valves....
I would keep the cascode as frontend. In fact, that was the MAIN compromise in my "Valve El Cheapo", in the interrest of simplicity and ease of building, no cascode.
As the cascode is basically a very linear voltage controlled current source you can realise the EQ purely as frequency dependent shunt device, with the lowest timeconstant set by the cascodes anode load.
This way headroom (which is I believe what Bill Clinton called the Oval Office? I wonder what Shrub calls it?) is maximised.
Ciao/Sayonara
The gain of the cascode would be higher than the mu stage. For noise it may be better to do RIAA on the higher level signal. Also, wouldn't a bypassed Rk or other constant voltage on the input stage be better for noise?
Konnichiwa,
Yup.
As remarked elsewhere, Valves are so noisy that at usual, optimised cathode resistor values their noise dominates.
Also, I found that local degeneration seems to reduce flicker noise and other LF noise, which is much higher than the devices shot (1/gm) noise.
What you gain from unbypassed cathode resistors is a stabilisation against parameter changes from different valves and aging, which is especially relevant here if you realise the the Valves anode impedance of the preceeding stage is always part of the RIAA EQ Circuit.
Sayonara
lndm said:
Yup.
lndm said:
As remarked elsewhere, Valves are so noisy that at usual, optimised cathode resistor values their noise dominates.
Also, I found that local degeneration seems to reduce flicker noise and other LF noise, which is much higher than the devices shot (1/gm) noise.
What you gain from unbypassed cathode resistors is a stabilisation against parameter changes from different valves and aging, which is especially relevant here if you realise the the Valves anode impedance of the preceeding stage is always part of the RIAA EQ Circuit.
Sayonara
Kuei Yang Wang said:
Good point. The problem is, with unbypassed resistors 6DJ8 have very low gain. Too low, I can't design a suitable RIAA network without too many losses.
One could employ 6S45 in the cascode, but that will double the tube count, and in general I wouldn't use difficult to obtain tubes, or maybe tubes that tomorrow can be unobtaniable.
I am thinking an ECC83 SRPP for the front-end...
Konnichiwa,
Well, ARE YOU SURE?
Lets run the lower valve in the cascode at 60V/10mA which needs around 0.75V Bias, so our cathode R is 75 Ohm. We use a +B of 250V and 100V acoss the upper triode we have 100V across the anode load resistor and thus 10K Anode load.
I'd estimate the transconductance of the lower valve at close to 12mA/V making the valves internal cathode impedance broadly comparable to the external resistor. So, in this case our gain is determined by the approximatly 200 Ohm cathode impedance and the 10K Anode Load as around 50.
If we now connect a 670 Ohm resistor in parallel with 330nF and both these in series with 100nF across the 10K anode load we will have a pretty good approximation of the RIAA EQ (this obviously needs finetuning, values are quick approximations) with a midband gain of around 14db at the anode of the upper ECC88.
The ECC81 in Mu follower or such will have a gain of around 32db, so in two stages we have managed 46db gain, with a minimal number of components.
Alternatively, we could make a ECC81 Cascode. I we again use 50V across the lower valve and run a reasonable deal of current (say 3mA) we have a cathode resistor (unbypassed) of 100 Ohm, the Anode load becomes 33K, transconductance eyballs somewhere around 3.3mA/V so our effective cathode impedance is around 400 Ohm and maximum gain is over 80.
Our RIAA components scale to 100n//2k2 in series with 33n and our midband gain is around 18db. If we now add the ECC88 as simple common cathode stage with fixed (semi-fixed) bias and use the Mu-follower part merely as high impedance anode load we can use my 12mA/70V operating condition and get a gain of around 30db into an "instrument" load of 100K, so we have around 48db gain with the ECC81 cascode frontend.
So, there is no need to bypass any cathode, the Anode load becomes the series resistor in our RIAA EQ and the whole thing is as simole as they come. If you can find nice 33nF Styroflex capaitors in 250V or higher, you could just buy 10pcs and use them for the RIAA and as coupling capacitors between stages.
Ciao/Sayonara
Giaime said:
Well, ARE YOU SURE?
Lets run the lower valve in the cascode at 60V/10mA which needs around 0.75V Bias, so our cathode R is 75 Ohm. We use a +B of 250V and 100V acoss the upper triode we have 100V across the anode load resistor and thus 10K Anode load.
I'd estimate the transconductance of the lower valve at close to 12mA/V making the valves internal cathode impedance broadly comparable to the external resistor. So, in this case our gain is determined by the approximatly 200 Ohm cathode impedance and the 10K Anode Load as around 50.
If we now connect a 670 Ohm resistor in parallel with 330nF and both these in series with 100nF across the 10K anode load we will have a pretty good approximation of the RIAA EQ (this obviously needs finetuning, values are quick approximations) with a midband gain of around 14db at the anode of the upper ECC88.
The ECC81 in Mu follower or such will have a gain of around 32db, so in two stages we have managed 46db gain, with a minimal number of components.
Alternatively, we could make a ECC81 Cascode. I we again use 50V across the lower valve and run a reasonable deal of current (say 3mA) we have a cathode resistor (unbypassed) of 100 Ohm, the Anode load becomes 33K, transconductance eyballs somewhere around 3.3mA/V so our effective cathode impedance is around 400 Ohm and maximum gain is over 80.
Our RIAA components scale to 100n//2k2 in series with 33n and our midband gain is around 18db. If we now add the ECC88 as simple common cathode stage with fixed (semi-fixed) bias and use the Mu-follower part merely as high impedance anode load we can use my 12mA/70V operating condition and get a gain of around 30db into an "instrument" load of 100K, so we have around 48db gain with the ECC81 cascode frontend.
So, there is no need to bypass any cathode, the Anode load becomes the series resistor in our RIAA EQ and the whole thing is as simole as they come. If you can find nice 33nF Styroflex capaitors in 250V or higher, you could just buy 10pcs and use them for the RIAA and as coupling capacitors between stages.
Ciao/Sayonara
Konnichiwa,
Wrong sequence of connection. Anyway, it's easier to see what's going on with a schematic. Here is what I would do with that kind of circuit, note, not build, just simulated in P-Spice.
With 1% capacitors the RIAA should be +/- 0.2db or so, gain 48db or therabouts, depending on valves.
Overly complex for my taste, but hey....
Ciao T
Kuei Yang Wang said:
Wrong sequence of connection. Anyway, it's easier to see what's going on with a schematic. Here is what I would do with that kind of circuit, note, not build, just simulated in P-Spice.
An externally hosted image should be here but it was not working when we last tested it.
With 1% capacitors the RIAA should be +/- 0.2db or so, gain 48db or therabouts, depending on valves.
Overly complex for my taste, but hey....
Ciao T
Very very very interesting, thank you. In fact I had doubts about your previous post, so I took a little time for thinking. This schematic, as you say, is better than 1000 words.
Thank you very much 😉 I have to study this.
Thank you very much 😉 I have to study this.
- Status
- Not open for further replies.