Why not use a dc coupled cathode follower with the other half of a 12AX7 run at 1mA ? Or use another higher transconductance, higher current tube like 5687 or 5842 as a follower. This will give you the low source impedance you crave and if you use a good ccs on the plate of the X7 you will get gain very close to the available mu of the triode. Even using a resistor as plate load if judiciously chosen will get you gains of up to 36dB in a single stage.
Cathode bypass cap is not necessary, note that omitting it will slightly improve the linearity, will significantly raise RP but unfortunately you will now also be amplifying the cathode resistor's johnson noise as well which will degrade your snr.
Just my opinion: Why over elaborate the design?? Tubes when correctly employed are pretty good amplifying devices and don't need or particularly benefit from the application of solid state devices in the signal path. Why bother with tubes at all if you are buffering everything with semiconductors? IMHO it never sounds as good as the tube by itself. And yes I have heard plenty of devices with ccs loads and use ccs myself sometimes in differential tails.. (Not that I currently use those either.)
Kevin
Cathode bypass cap is not necessary, note that omitting it will slightly improve the linearity, will significantly raise RP but unfortunately you will now also be amplifying the cathode resistor's johnson noise as well which will degrade your snr.
Just my opinion: Why over elaborate the design?? Tubes when correctly employed are pretty good amplifying devices and don't need or particularly benefit from the application of solid state devices in the signal path. Why bother with tubes at all if you are buffering everything with semiconductors? IMHO it never sounds as good as the tube by itself. And yes I have heard plenty of devices with ccs loads and use ccs myself sometimes in differential tails.. (Not that I currently use those either.)
Kevin
Triodes are the most linear "open-loop" voltage amplifiers available if loaded with high impedance, there is internal grid-plate feedback but it's part of the physics of the device. This linearity degrades with lower impedance loads especially for high-mu. Transistors are much less linear "open-loop" and have non-linear capacitances but their higher transconductance makes for better current amplifiers, and their small size and lack of filaments makes complexity easier to deal with. Combining triodes for voltage amplification and transistors for current amplification is a great combo if managed well, transistor junction capacitance must be watched, IMO Pimm is the master of this. His anode CCS designs with floating gate bias have much higher PSRR than the C4S design which is compromised by the bias resistor for the LED string, of course this resistor could be replaced by another CCS but at that point Pimm's CCS is less complex. To me Pimm is a fine example of someone who seeks the most satisfying solution without following dogma.
Kevinkr,
Right on. The original RCA is elegant in its modesty and simplicity, and I can appreciate someone wanting to build it for those reasons. But once we start to add in so many complications (including those FET followers), we’ve really deviated from the RCA spirit. Then we throw open our options, and we are free to consider many other topologies and design choices, including the use of tubes other than the 12AX7 in the first gain stage (say, a 6DJ8) which offer lower shot and flicker noise. And if a FET follower is contemplated, why not use a cathode follower instead? The passive EQ concept is certainly worthy. I think Morgan Jones, in his fine book “Valve Amplifiers”, offers some sensible RIAA stage options. His simplest 6DJ8 design seems a reasonable choice.
Right on. The original RCA is elegant in its modesty and simplicity, and I can appreciate someone wanting to build it for those reasons. But once we start to add in so many complications (including those FET followers), we’ve really deviated from the RCA spirit. Then we throw open our options, and we are free to consider many other topologies and design choices, including the use of tubes other than the 12AX7 in the first gain stage (say, a 6DJ8) which offer lower shot and flicker noise. And if a FET follower is contemplated, why not use a cathode follower instead? The passive EQ concept is certainly worthy. I think Morgan Jones, in his fine book “Valve Amplifiers”, offers some sensible RIAA stage options. His simplest 6DJ8 design seems a reasonable choice.
Konnichiwa,
Yup, I noticed that myself when I needed to design a simple passive 2-stage Phono.
Carefull. It does that mainly on paper. In real MM Applications the higher grid(noise)current and other stuff tends to end up with more noise than the ECC83 (even with unbypassed cathode R) in reality.
Also, it is a good idea to avoid daisychaining too many identical type valves.
Sayonara
Brian Beck said:
Yup, I noticed that myself when I needed to design a simple passive 2-stage Phono.
Brian Beck said:
Carefull. It does that mainly on paper. In real MM Applications the higher grid(noise)current and other stuff tends to end up with more noise than the ECC83 (even with unbypassed cathode R) in reality.
Also, it is a good idea to avoid daisychaining too many identical type valves.
Sayonara
hey, thanks for the advice!
does using a pair of 6922 triodes as input tubes make any sense? the thought of the high transconductance and the low noise and rp makes me smile, but is it feasable?
does using a pair of 6922 triodes as input tubes make any sense? the thought of the high transconductance and the low noise and rp makes me smile, but is it feasable?
hacknet said:
The Bottlehead Seduction uses a pair of 6922s and passive EQ. User feedback is favorable. Net gain with CCS loading is about 40 dB., which requires mating to a line stage with gain.
I'm a DODO when it comes to working RIAA EQ networks out. The RCA original is attractive to me, in no small part, because of the time proven EQ network.
My objectives are to shore up the weaknesses of the original and have enough net gain to function well in combination with a near unity gain line stage.
Konnichiwa,
The EQ Network may be old, but it is not proven, it has large deviations from the nominal RIAA Curve.
Unless your Amplifier has extremely low gain around 40db gain for a Phonostage is just right for MM.
Sayonara
Eli Duttman said:
The EQ Network may be old, but it is not proven, it has large deviations from the nominal RIAA Curve.
Eli Duttman said:
Unless your Amplifier has extremely low gain around 40db gain for a Phonostage is just right for MM.
Sayonara
Hi
I'd say the RCA RIAA is not bad if you like the vintage tube sound 😎
It was one of the first RIAA which I had build when I was studying the principles how a phonostage can be made and listened to different tube-sounds.
In the RCA Application Note (mentioned above) the 7025 tube is used which is very close or nearly identical to the 6SL7 which I used here.
Before I built the amp, I did some kind of reverse engineering, to verify the quality of the circuit.
When this circuit was simulated with PSPICE on a workstation of a friend who is a specialist of this Software the deviation was max. 0,6db of the RIAA-Curve. The Low-Frequency-Cutoff was at 18 Hz.
A mathematician from Berlin who was also engaged in the theoretical discussion of the circuit agreed with that and said to me:
"The RIAA 470k -- ( (22k || 3,3n) + 10n ) is already quite precise. In fact it is more precise than I would have expected. The time constants are 3054 (instead 3180), 289 (instead 318) and 70 (instead 75), all in µs.
The more exact values would be 22,814k instead of 22k, 3,559nF instead of 3,3nF and 10,379nF instead 10nF, assuming the 470k are remaining in place"
I have build different prototypes of this phonostage. One of it with LED-biasing. (Morgan Jones had done studies on LED-Biasing and he was so friendly to send me his favourite LED from Great Britain for testing)
Here is a picture of one example I have built with very nice vintage KENRAD 6SL7 Tubes.
Best regards, Jo
I'd say the RCA RIAA is not bad if you like the vintage tube sound 😎
It was one of the first RIAA which I had build when I was studying the principles how a phonostage can be made and listened to different tube-sounds.
In the RCA Application Note (mentioned above) the 7025 tube is used which is very close or nearly identical to the 6SL7 which I used here.
Before I built the amp, I did some kind of reverse engineering, to verify the quality of the circuit.
When this circuit was simulated with PSPICE on a workstation of a friend who is a specialist of this Software the deviation was max. 0,6db of the RIAA-Curve. The Low-Frequency-Cutoff was at 18 Hz.
A mathematician from Berlin who was also engaged in the theoretical discussion of the circuit agreed with that and said to me:
"The RIAA 470k -- ( (22k || 3,3n) + 10n ) is already quite precise. In fact it is more precise than I would have expected. The time constants are 3054 (instead 3180), 289 (instead 318) and 70 (instead 75), all in µs.
The more exact values would be 22,814k instead of 22k, 3,559nF instead of 3,3nF and 10,379nF instead 10nF, assuming the 470k are remaining in place"
I have build different prototypes of this phonostage. One of it with LED-biasing. (Morgan Jones had done studies on LED-Biasing and he was so friendly to send me his favourite LED from Great Britain for testing)
Here is a picture of one example I have built with very nice vintage KENRAD 6SL7 Tubes.
An externally hosted image should be here but it was not working when we last tested it.
Best regards, Jo
yes, instead of R/C.
The difference is not audible - to my ears. But because I did not recognize an improvement, I normally build my amps the conventional way.
Jo
The difference is not audible - to my ears. But because I did not recognize an improvement, I normally build my amps the conventional way.
Jo
Here's a link to the little Zetex MOSFET's data sheet.
FET Data Sheet
The b*gger has some toughness, being able to dissipate 700 mW. Given its 50 Ohm internal resistance, 10 mA. of drain current will not warm it up nearly enough to be concerned about. In addition to preventing parasitic oscillation, a Carbon comp. gate stopper resistor will isolate the FET thermally from the tube it buffers.
FET Data Sheet
The b*gger has some toughness, being able to dissipate 700 mW. Given its 50 Ohm internal resistance, 10 mA. of drain current will not warm it up nearly enough to be concerned about. In addition to preventing parasitic oscillation, a Carbon comp. gate stopper resistor will isolate the FET thermally from the tube it buffers.
Konnichiwa,
AT 25 Centigrade and then the chip inside will have very high temperature (near the 150 Centigrade Limit). The inside of your Valve Phonostage is likely to run closer to 40 Centigrade, hotter if external temperatures are up on the 20..25 centigrade level.
Other E-Line TO92 devices have 150C/W thermal resistance from the junction to ambient, in other words, if the ambient temperature is 0 Centigrade the junction will be heated to 150 Centigrade by 1 Watt dissipated.
As we know that 0.7W with 25 Centigrade Ambient temperature raise the junction temperature to 150 Centigrade we know the thermal resistance is 125C/0.7W or 180C/W.
If we want to avoid to exceed 125 Centigrade junction temperature (to avoid unduly stressing the device) and have an ambient temperature of 50 Centigrade allowed we arrive a temperature difference of 75 Centigrade between junction and ambient. This suggests a limit of no more than 400mW in operation for the FET.
I am not sure what the internal resitance has to do with it, but if you had (for arguments sake) 100V across the Fet and where using 10mA current you would dissipate 1000mW, which would in relatively short time release the black smoke kept inside the FET which is what makes it work. 😉
So, if you can reliably hold the air around the FET at 25 Centigrade you can allow no more than 70V across it with 10mA drain current. To be realistic it should be no more than around 40V @ 10mA to avoid pushing device limits.
I am not sure how a resistor isolates thermally.
Anyway, the FET has very low capacitance, but it's usefulness is severely limited by low dissipation.
Sayonara
Eli Duttman said:
AT 25 Centigrade and then the chip inside will have very high temperature (near the 150 Centigrade Limit). The inside of your Valve Phonostage is likely to run closer to 40 Centigrade, hotter if external temperatures are up on the 20..25 centigrade level.
Other E-Line TO92 devices have 150C/W thermal resistance from the junction to ambient, in other words, if the ambient temperature is 0 Centigrade the junction will be heated to 150 Centigrade by 1 Watt dissipated.
As we know that 0.7W with 25 Centigrade Ambient temperature raise the junction temperature to 150 Centigrade we know the thermal resistance is 125C/0.7W or 180C/W.
If we want to avoid to exceed 125 Centigrade junction temperature (to avoid unduly stressing the device) and have an ambient temperature of 50 Centigrade allowed we arrive a temperature difference of 75 Centigrade between junction and ambient. This suggests a limit of no more than 400mW in operation for the FET.
Eli Duttman said:
I am not sure what the internal resitance has to do with it, but if you had (for arguments sake) 100V across the Fet and where using 10mA current you would dissipate 1000mW, which would in relatively short time release the black smoke kept inside the FET which is what makes it work. 😉
So, if you can reliably hold the air around the FET at 25 Centigrade you can allow no more than 70V across it with 10mA drain current. To be realistic it should be no more than around 40V @ 10mA to avoid pushing device limits.
Eli Duttman said:
I am not sure how a resistor isolates thermally.
Anyway, the FET has very low capacitance, but it's usefulness is severely limited by low dissipation.
Sayonara
it looks like the devil and the deep blue sea.
larger fets seem to have much larger capacitances. i smell a jfet brewing...
larger fets seem to have much larger capacitances. i smell a jfet brewing...
Konnichiwa,
Electronic Design is always like that.
For example, I find that low impedance RIAA Networks sound better IF ALL IS EQUAL, however often the valves needed to get the anode impedance of the first low enough are in practice too noisy for use and once buffers are added to the mix it does no longer sound the same.
So, there very rarely is a free lunch, in electronics as in reality and good design consists in choosing the compromise between the devil or the blue sea in a way that suits you objective.
Sayonara
hacknet said:
Electronic Design is always like that.
For example, I find that low impedance RIAA Networks sound better IF ALL IS EQUAL, however often the valves needed to get the anode impedance of the first low enough are in practice too noisy for use and once buffers are added to the mix it does no longer sound the same.
So, there very rarely is a free lunch, in electronics as in reality and good design consists in choosing the compromise between the devil or the blue sea in a way that suits you objective.
Sayonara
Carbon is a poor conductor of heat. So, you don't carry heat into the FET along the wire. The resistor does nothing about the ambient air temp.
If the FET's gate is connected to 250 VDC. and the FET's source is connected to ground via a 25 KOhm resistor, the amount of heat generated internally is SMALL.
hacknet said:
Not if you're looking for 10mA and 100V Vds. You can cascode the jfet but then the jfet source (I assume njfet) has to drive the gate/base/grid of the cascode device and the jfet needs enough Vds to work. Look at Gary Pimm's website for principles and practice.
The FET power dissipation is Vds*Id, so what is Vds? With B+=250, Rp=100k and Ip=0.5mA then Vp=200V (250-100k*0.5mA). If the FET drain is +250 and gate is +200 then Vds=50V, I assume Vgs=0, actually 1 - 3V but close enough. FET power dissipation is then 50V*10mA=500mW. With KYW's estimate of 40C ambient and 180C/W thermal resistance from junction to ambient then Tj=40+180*0.5=130C, above the 125C generally accepted for good reliability, but hey this is DIY so let's live dangerously. In any case the internal temperature will be high but it will probably work OK for evaluation.
Is 10mA really needed? The Zetex FET has gm of 100mS at 100mA, probably has 5mS at 5mA or 200 ohms output resistance, 5mA would get Pd down to a comfortable level. If you need lower output resistance, consider cascading the FET with a bipolar transistor (Fetlington?), the BJT could be a CRT grid driver with low capacitance.
A FET drain voltage of as low as 210V might be another option that would allow the 10mA, perhaps by adding a resistor or zener diode between drain and B+, certainly simpler than cascoding.
The 5 mA./200 Ohm situtation is FINE. Being able to drive 10 KOhm loads and reasonable amounts of cable capacitance is all that is required. Looking good. 🙂
Konnichiwa,
While you can conduct heat that way, usually the effect is exponential with distance...
AGreed. But you will also have no current flowing and no use of the FET. You also need to attach the drain somewhere and have a voltage across the FET sufficient to supply the whole required swing. If you are econimic in not applying a seperate supply to each drain selected to rather well matched voltage things get a lot more toasty.
Look, you desperatly want to throw FET's, CCS's and other "High Tech" where it is not exactly needed, no grief from me. But I think you may wish to check your calculations, just to make sure the black smoke inside Solid state stuff stays inside.
Sayonara
Eli Duttman said:
While you can conduct heat that way, usually the effect is exponential with distance...
Eli Duttman said:
AGreed. But you will also have no current flowing and no use of the FET. You also need to attach the drain somewhere and have a voltage across the FET sufficient to supply the whole required swing. If you are econimic in not applying a seperate supply to each drain selected to rather well matched voltage things get a lot more toasty.
Look, you desperatly want to throw FET's, CCS's and other "High Tech" where it is not exactly needed, no grief from me. But I think you may wish to check your calculations, just to make sure the black smoke inside Solid state stuff stays inside.
Sayonara
- Status
- Not open for further replies.