I have assumed a fairly low source impedance, either a MC cartridge, or a step-up for a MC in my comments. If you are using a MM cartridge, the source resistance can exceed 1000 ohms depending on the model. So the cartridge’s self-generated noise can actually exceed that generated by a 6DJ8 and you wouldn’t notice as much of a difference between a preamp tubes. If you are using a MC and have only achieved the same noise performance with a 6DJ8 as a 12AX7, then there is probably something wrong with your circuit design. I have built preamps (microphone and RIAA) with many types of tubes, and if the circuit is done right, results will in fact match noise theory to a pretty close degree. Whenever theory and practice don’t match, by a long shot in this case, you should really ask why and figure it out. The most common problem I see with potentially low-noise high-gm tubes in the first RIAA slot is the use of resistors in either the cathode circuit or the grid circuit. If either resistor is more than a few tens of ohms, it can start to swamp the inherent noise of a quiet tube. Cathode resistors either must be bypassed or eliminated by using some form of grid bias. Resistors in series with the grid, often used in conjunction with the Miller capacitance to make a low-pass filter for RF, are not good for noise. It doesn’t matter whether the resistors are the best in the world, they will still make noise.
BTW, I too use an E810F (also called 7788) in my favorite design, but it is triode connected. Pentodes suffer from additional partition noise compared to triodes, although voltage gains can be higher with pentodes.
Erno Borbely wrote a very good explanation about phono preamp noise in the recent AudioXpress. His designs are JFET-based, not my cup of tea, but his explanations apply to tubes as well.
Konnichiwa,
I use MC including stepup.
Nope, my circuit design is fine, trust me.
I do not have any significant noise problems with either valve at MM (25mV maximum input, 5mV nominal @ 5cm/s). I can get even quieter, but my point is that you do not need very high transconductance for low noise in the audio band. The 12AY7 BTW, for whatever reasons is materially quieter than the ECC88.
All of that for a MC Cartridge with a Stepup Transformer, which of course for a fairly common Denon DL-103 results in a reflected source impedance between 4K (1:10 Stepup) and 16KOhm (1:20 Stepup). Even with a lower Impedance cartridge (Goldring Elite, 12R Coil Resistance) you get 1 - 4K source resistance.
By that time you are in a grey area where overall performance is controlled by complex issues.
HOWEVER, I still get about the same absolute noise levels (and yes, the noise is strictly Valve/Resistor noise, not PSU or elsewhere derived) for an ECC83 in fairly typhical operation (100K Anode Load) and an ECC88 in fairly typhical Operation (47k Anode load).
Agreed, but to sumarise, in a circuit with around 1-4K effective source impedance (eg a generic MC with generic stepup transformer) we are in a an area where noise performance is relative.
Also, I do have a number of reasons for using unbypassed cathode resistors in the first stage, noise perfornace does not in practical terms appear to suffer materially, in my particular applications.
Again, the partition noise is generally invoked, but it does not seem to prevent me from making very quiet Phonostages with pentodes operated as pentodes, nor did it prevent TAB from making very Quiet Microphone Preamplifiers that way.
I find a lot of what is claimed to "be good" to fall into the same category as the "High Gm reliably = low noise" hogwash. There is much more in how you do things than what you do, in my experience.
I tend to put fairly mosdest stock in theory, because usually theories are based on highly simplified and abstract models which are usefull to teaching students "how it works" but which are usually insufficiently complex and detailed o be of much use outside classrooms.
Sayonara
I use MC including stepup.
Nope, my circuit design is fine, trust me.
I do not have any significant noise problems with either valve at MM (25mV maximum input, 5mV nominal @ 5cm/s). I can get even quieter, but my point is that you do not need very high transconductance for low noise in the audio band. The 12AY7 BTW, for whatever reasons is materially quieter than the ECC88.
All of that for a MC Cartridge with a Stepup Transformer, which of course for a fairly common Denon DL-103 results in a reflected source impedance between 4K (1:10 Stepup) and 16KOhm (1:20 Stepup). Even with a lower Impedance cartridge (Goldring Elite, 12R Coil Resistance) you get 1 - 4K source resistance.
By that time you are in a grey area where overall performance is controlled by complex issues.
HOWEVER, I still get about the same absolute noise levels (and yes, the noise is strictly Valve/Resistor noise, not PSU or elsewhere derived) for an ECC83 in fairly typhical operation (100K Anode Load) and an ECC88 in fairly typhical Operation (47k Anode load).
Agreed, but to sumarise, in a circuit with around 1-4K effective source impedance (eg a generic MC with generic stepup transformer) we are in a an area where noise performance is relative.
Also, I do have a number of reasons for using unbypassed cathode resistors in the first stage, noise perfornace does not in practical terms appear to suffer materially, in my particular applications.
Again, the partition noise is generally invoked, but it does not seem to prevent me from making very quiet Phonostages with pentodes operated as pentodes, nor did it prevent TAB from making very Quiet Microphone Preamplifiers that way.
I find a lot of what is claimed to "be good" to fall into the same category as the "High Gm reliably = low noise" hogwash. There is much more in how you do things than what you do, in my experience.
I tend to put fairly mosdest stock in theory, because usually theories are based on highly simplified and abstract models which are usefull to teaching students "how it works" but which are usually insufficiently complex and detailed o be of much use outside classrooms.
Sayonara
KYW, You do have a high source resistance and it's therefore no wonder that your source noise will swamp tube noise and you wouldn't have noticed much of a difference. Here again, theory would have predicted that outcome! With my 4 ohm source feeding the input directly - tube choice makes a difference in noise! If I use a 1:10 transformer, I've got 400 ohms and there are still advantages to the hi-gm school at that point, but diminishing at 1:20.
Theory can certainly be abused, but it can be usefully applied for simple first-cut guidance, or it can be made as complex as you need to thoroughly understand something. I don't pretend that theory has adequately explained everything in audio, not by a long shot. But trial and error or just ear-tweaking can only take a designer so far.
Theory can certainly be abused, but it can be usefully applied for simple first-cut guidance, or it can be made as complex as you need to thoroughly understand something. I don't pretend that theory has adequately explained everything in audio, not by a long shot. But trial and error or just ear-tweaking can only take a designer so far.
Konnichiwa,
NOTE I MENTIONED MEASUREMENTS WITH THE INPUT SHORTED!
BTW, given that your 4 Ohm Cartridge likely has very low output and given that your 1:10 stepup does not stepup much you may have lower thermal noise from the resistance, BUT you end up with an overall poorer S/N ratio as I in my case with 40 Ohm and 1:20 Stepup where the Output from the cartridge is likely to be between 10 - 15db more than in your case and we gain added "level" with an extra 6db "passive" gain, so we have overall maybe 20db more signal on the input to the Phonostage overall (in my case 10mV @ 5cm/s).
We have an inherent noise level equivalent to 16KOhm or around 15nV|/Hz (can't be bothered to calculate it exactly) or around 2,000nV for a 20KHz bandwidth or indeed 2uV resistive noise, which with our 10mV signal translates into -74db @ 5cm/s and nearly 90db with reference to maximum level on LP at 25cm/s.
Comparing a case with an assumed Low Turns armature cartridge and 0.1mV @ 5cm/s and 4R we get 1mV @ 5cm/S after stepup with a resistive noise of 0.4uV which with our 1mV signal translates into -68db @ 5cm/s.
Given that a ECC83 has based on measurements has between 5-10nV|/Hz at 1KHz (and between 10-50nV|/Hz @ 100Hz) we have indeed a probelm with noise as the ECC83 produces almost 10db more noise than our cartridge. In my "high level" situation BTW the ECC83 has less noise than the source!
While we number numbers, ECC88's measure around 5nV|/Hz Ein @ 1KJHz, entierly comparable with a ECC83 and at low frequencies they are usually worse with 50 - 100nV|/Hz @ 100Hz being the "average". The quietest ECC83's and ECC88's i came across showed about the same level of Ein with 5nV|/Hz around 1Khz and 10-20nV|/Hz @ 100Hz. These are strictly empirical measurements in test circuits, not results of theoretical calculation or application circuits, before you ask.
Back to square one, what sort of S/N ratio you get and if your Phonostage ends up quiet or not depends on many things. Using High Gm valves does not do any harm, but neither is it of any particular merit over other approaches on grounds of noise at least.
Maybe, but that is not what I am using.
My point remains that in the real world, under real conditions you find no significant differences between a "low Gm" and a "high Gm" valve within the context of the application in the real world nor do I find it on the bench under idealised conditions.
Maybe I am looking wrong, I will readily admit, but that what it looks like from here...
Sayonara
Brian Beck said:
NOTE I MENTIONED MEASUREMENTS WITH THE INPUT SHORTED!
BTW, given that your 4 Ohm Cartridge likely has very low output and given that your 1:10 stepup does not stepup much you may have lower thermal noise from the resistance, BUT you end up with an overall poorer S/N ratio as I in my case with 40 Ohm and 1:20 Stepup where the Output from the cartridge is likely to be between 10 - 15db more than in your case and we gain added "level" with an extra 6db "passive" gain, so we have overall maybe 20db more signal on the input to the Phonostage overall (in my case 10mV @ 5cm/s).
We have an inherent noise level equivalent to 16KOhm or around 15nV|/Hz (can't be bothered to calculate it exactly) or around 2,000nV for a 20KHz bandwidth or indeed 2uV resistive noise, which with our 10mV signal translates into -74db @ 5cm/s and nearly 90db with reference to maximum level on LP at 25cm/s.
Comparing a case with an assumed Low Turns armature cartridge and 0.1mV @ 5cm/s and 4R we get 1mV @ 5cm/S after stepup with a resistive noise of 0.4uV which with our 1mV signal translates into -68db @ 5cm/s.
Given that a ECC83 has based on measurements has between 5-10nV|/Hz at 1KHz (and between 10-50nV|/Hz @ 100Hz) we have indeed a probelm with noise as the ECC83 produces almost 10db more noise than our cartridge. In my "high level" situation BTW the ECC83 has less noise than the source!
While we number numbers, ECC88's measure around 5nV|/Hz Ein @ 1KJHz, entierly comparable with a ECC83 and at low frequencies they are usually worse with 50 - 100nV|/Hz @ 100Hz being the "average". The quietest ECC83's and ECC88's i came across showed about the same level of Ein with 5nV|/Hz around 1Khz and 10-20nV|/Hz @ 100Hz. These are strictly empirical measurements in test circuits, not results of theoretical calculation or application circuits, before you ask.
Back to square one, what sort of S/N ratio you get and if your Phonostage ends up quiet or not depends on many things. Using High Gm valves does not do any harm, but neither is it of any particular merit over other approaches on grounds of noise at least.
Brian Beck said:
Maybe, but that is not what I am using.
My point remains that in the real world, under real conditions you find no significant differences between a "low Gm" and a "high Gm" valve within the context of the application in the real world nor do I find it on the bench under idealised conditions.
Maybe I am looking wrong, I will readily admit, but that what it looks like from here...
Sayonara
KYW,
Ok, I will have to dig up my lab notebooks and/or do some noise measurements again and share them here. The only other thing I can think of to explain your inability to detect a difference between hi-gm and low-gm tube noise is that flicker noise (1/f) must dominate in whatever tubes you're using. That is type, brand and individual tube variable.
Ok, I will have to dig up my lab notebooks and/or do some noise measurements again and share them here. The only other thing I can think of to explain your inability to detect a difference between hi-gm and low-gm tube noise is that flicker noise (1/f) must dominate in whatever tubes you're using. That is type, brand and individual tube variable.
Konnichiwa,
Actually, the strong rise towards LF and the great variability suggests that it is indeed flicker noise that dominates, however I referred to noise overall, without splitting it into different noise sources and excluding all but 1/Gm....
Hence my point of High Gm = Low Noise being hogwash, as especially for riaa applications we find flicker LF noise (in other words flicker noise) to dominate.
So, I repeat again, in the real world, using real valves there is no reliably observable noise advantage of the "high Gm" ECC88 over a "low Gm" ECC83, where overall noise is concerned and observational the 6072A/12AY7 is materially quieter than either of these.
Sayonara
Brian Beck said:
Actually, the strong rise towards LF and the great variability suggests that it is indeed flicker noise that dominates, however I referred to noise overall, without splitting it into different noise sources and excluding all but 1/Gm....
Hence my point of High Gm = Low Noise being hogwash, as especially for riaa applications we find flicker LF noise (in other words flicker noise) to dominate.
So, I repeat again, in the real world, using real valves there is no reliably observable noise advantage of the "high Gm" ECC88 over a "low Gm" ECC83, where overall noise is concerned and observational the 6072A/12AY7 is materially quieter than either of these.
Sayonara
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.