Here is all the article:
http://www.giaime.altervista.org/SuperRIAA.html
I think the whole section on tube and resistor noise should be read from anyone serious about electronics.
The same question of all my recent threads: what do you think? 😀
http://www.giaime.altervista.org/SuperRIAA.html
I think the whole section on tube and resistor noise should be read from anyone serious about electronics.
The same question of all my recent threads: what do you think? 😀
I like the simple desing, the ability to switch between a variety of cartidges, and the lack of capacitors. I am looking to make a phono preamp, and this might be exactly what I am looking for
mmm... lack of capacitors? 
I'm not sure what schematic are you referring to, however in this design low parts count hasn't been an objective, like DC coupling.
Better good polarized stages than the absence of a capacitor 😉
I'm not sure what schematic are you referring to, however in this design low parts count hasn't been an objective, like DC coupling. Better good polarized stages than the absence of a capacitor 😉
Hi Giaime,
This circuit looks interesting. I use a hybrid cascode myself. I like what I hear, but don't fully understand why. I have some questions.
What is the reason for choosing a cascoded second and third MM stage? I thought they were best as the first stage. I remember when I mentioned the word cascode on the forum, the heavies questioned why I would want such a stage, I could only say I liked the sound.
If I wanted to try something other than the ECC88, would I just need to alter the plate resistors, cathode biasing, and adjust the output impedances for the RIAA EQ to work. Do you think it would be worth trying with 6SN7's?
This circuit looks interesting. I use a hybrid cascode myself. I like what I hear, but don't fully understand why. I have some questions.
What is the reason for choosing a cascoded second and third MM stage? I thought they were best as the first stage. I remember when I mentioned the word cascode on the forum, the heavies questioned why I would want such a stage, I could only say I liked the sound.
If I wanted to try something other than the ECC88, would I just need to alter the plate resistors, cathode biasing, and adjust the output impedances for the RIAA EQ to work. Do you think it would be worth trying with 6SN7's?
Hello Indm,
I admit I haven't tried yet the circuit. Like many of those designs, it's only a "paper" project, which can be tweaked by the brave fellow who tries. However in the future I'm already planning to build a new preamp so...
Why Cascode? This particular "cascode with nfb" as mentioned in Broskie's articles has particulary low distortion (and noise, very important), lower than what I would expect by just putting nfb. Spice has confirmed that.
Cascode for the second stage because I needed that gain, but obviously one could do in other ways, such a CCS loaded grounded cathode stage. They are like an ice cream: cascode is chocolate, grounded cathode is vanilla, etc etc, choose your favourite one. When PROPERLY IMPLEMENTED good gainstages should be indistinguishable 😉
And no, I wouldn't try 6SN7. This dual triode, bear in mind, was "invented" long before low noise and microphonics were key objective. This isn't to say that 6SN7 is noisy, because many good expensive NOS ones are very quiet. But why I would make a project suitable only for pricy NOS tubes?
Also there's a reason about transconductance: the higher the better, as stated in the article. Possible other tubes (in current production) to be considered are 6H30 and 6S45, all pricier than ECC88/6922, and the good 6S45 isn't dual.
However please note that the output stage (the 6CG7 - I love this tube - totem pole) is in developing stage, and I surely will change it, probably dc-coupling it with the previous stage (there's only 76V on the grid if I remove the coupling cap). I was thinking about the Plate Follower stage I use in my preamp.
Thank you all for the interest!
I admit I haven't tried yet the circuit. Like many of those designs, it's only a "paper" project, which can be tweaked by the brave fellow who tries. However in the future I'm already planning to build a new preamp so...
Why Cascode? This particular "cascode with nfb" as mentioned in Broskie's articles has particulary low distortion (and noise, very important), lower than what I would expect by just putting nfb. Spice has confirmed that.
Cascode for the second stage because I needed that gain, but obviously one could do in other ways, such a CCS loaded grounded cathode stage. They are like an ice cream: cascode is chocolate, grounded cathode is vanilla, etc etc, choose your favourite one. When PROPERLY IMPLEMENTED good gainstages should be indistinguishable 😉
And no, I wouldn't try 6SN7. This dual triode, bear in mind, was "invented" long before low noise and microphonics were key objective. This isn't to say that 6SN7 is noisy, because many good expensive NOS ones are very quiet. But why I would make a project suitable only for pricy NOS tubes?
Also there's a reason about transconductance: the higher the better, as stated in the article. Possible other tubes (in current production) to be considered are 6H30 and 6S45, all pricier than ECC88/6922, and the good 6S45 isn't dual.
However please note that the output stage (the 6CG7 - I love this tube - totem pole) is in developing stage, and I surely will change it, probably dc-coupling it with the previous stage (there's only 76V on the grid if I remove the coupling cap). I was thinking about the Plate Follower stage I use in my preamp.
Thank you all for the interest!
I used and liked the made-in-CCCP 6N6p. I won those lil double triodes on ebay for cheap (and I mean a couple of euro each). You can probably consider them
Ciao
Ciao
Giaime,
Jim McShane who KNOWS about these things tells me that the ElectroHarmonix labeled 6922 variant out of Saratov is very good in the noise dept. The EH tube should be a good affordable choice in the MC gain blocks.
A way to improve the CRITICAL noise performance in the MC gain block is to use paralleled 6922 sections in each leg of the difference amp. You'd be taking advantage of 2X gm and 1/2 Rp. I know the idea requires 2 bottles per channel for the MC gain blocks alone, but this project is an all out assault on performance.
Jim McShane who KNOWS about these things tells me that the ElectroHarmonix labeled 6922 variant out of Saratov is very good in the noise dept. The EH tube should be a good affordable choice in the MC gain blocks.
A way to improve the CRITICAL noise performance in the MC gain block is to use paralleled 6922 sections in each leg of the difference amp. You'd be taking advantage of 2X gm and 1/2 Rp. I know the idea requires 2 bottles per channel for the MC gain blocks alone, but this project is an all out assault on performance.
...never seen a cascode feeding another cascode.🙂
You may want to look at Steve Bench's cascode phono designs, as well as Alan Wrights hybrid cascode with a j-fet on the bottom... if you haven't already.
You may want to look at Steve Bench's cascode phono designs, as well as Alan Wrights hybrid cascode with a j-fet on the bottom... if you haven't already.
Oh yes I thougt about paralleling sections. SNR ratio should increase by 3dB if I'm not wrong.
Since EH 6922 are cheap (not for me, but for the one that needs those kind of phono pre in his rig 😀 ) it could be done.
Thank you pedroskova, I read all that I could on Steve's site, however I didn't pay attention too much to what Wrights said, I have to look better.
I still have to learn more about SS design, I strongly think that a FET MC stage can be a big improvement over tubes.
Since EH 6922 are cheap (not for me, but for the one that needs those kind of phono pre in his rig 😀 ) it could be done.
Thank you pedroskova, I read all that I could on Steve's site, however I didn't pay attention too much to what Wrights said, I have to look better.
I still have to learn more about SS design, I strongly think that a FET MC stage can be a big improvement over tubes.
This is the article I started with. http://www.vacuumstate.com/various/SP-15_Article.pdf
I used only the first stage, and didn't use the regulator.
I used only the first stage, and didn't use the regulator.
Sorry to say, with all the respect for this Allen Wright, but sorry, this guy can't write an article, I have difficulties in following it. And since I can follow Koren I don't think that's my fault...
Also, much much much much much internet-guru-300B-nonfb BS there. Many wrong techincal statements.
One of the interesting points is a thing I already did and I'm doing at the moment, carefully studing Tektronix scopes gain blocks. 😉
Also, much much much much much internet-guru-300B-nonfb BS there. Many wrong techincal statements.
One of the interesting points is a thing I already did and I'm doing at the moment, carefully studing Tektronix scopes gain blocks. 😉
Hi Eli,
-Chris
I can confirm that as well. They sound pretty good for this type also. Well worth a try. Sovtek they are not.
-Chris
Some news.
A new circuit, look for it at the end of the page.
http://www.giaime.altervista.org/SuperRIAA.html
I've done this to simplify things quite a bit, but to keep performance. Useful for "first serious" tube project 😉
A new circuit, look for it at the end of the page.
http://www.giaime.altervista.org/SuperRIAA.html
I've done this to simplify things quite a bit, but to keep performance. Useful for "first serious" tube project 😉
Attachments
Konnichiwa,
Sadly these sections are mainly misleading.
For example, correctly applied the ECC83 has much less noise than the ECC88, despite the latter having according to theory a lower noise.
Here the noise levels in a selection of ECC83 at 1mA anode current:
Compare this to the ECC88 at 3mA anode current:
This BTW from this Article:
http://www.jogis-roehrenbude.de/Leserbriefe/Rabus-VV/inhalt/Inhalt.html
We can see that at 100Hz the majority of the ECC83's have a noise of less than 20nV|/Hz, with the bulk of the ECC88 over a wider range of 25 - 75nV|/Hz, which is between a few db and nearly 12db noisier!!!!
My "Valve El Cheapo" Phonostage may appear to be "too simple" to offer "no compromise", however you will find within the design brief limits (which excluded such niceties as LCR RIAA Modules, expensive MC Stepup Transformers and such) it actually offers exactly that.
On top of this BTW, a J-Fet MC Pre-Pre such as I recommend for the El Cheapo has an average noise levels in the "a few nV|/Hz region at 100Hz, which is around 20db less noise than the ECC83 on average has and as much as 30db+ less noise than the ECC88.
I find also that pentodes in moving magnet level input sgates can be surpremely quiet. My quite extreme LCR equalised Phonostage uses a E810F operated as pentode on the input. It is actually the quietest of all my valve based phonostages!
You might find some of my previous Phonostage Projects of interrest as well:
http://thunderstoneaudio.nav.to
Sorry for the various popups....
Ciao T
Giaime said:
Sadly these sections are mainly misleading.
For example, correctly applied the ECC83 has much less noise than the ECC88, despite the latter having according to theory a lower noise.
Here the noise levels in a selection of ECC83 at 1mA anode current:
Compare this to the ECC88 at 3mA anode current:
This BTW from this Article:
http://www.jogis-roehrenbude.de/Leserbriefe/Rabus-VV/inhalt/Inhalt.html
We can see that at 100Hz the majority of the ECC83's have a noise of less than 20nV|/Hz, with the bulk of the ECC88 over a wider range of 25 - 75nV|/Hz, which is between a few db and nearly 12db noisier!!!!
My "Valve El Cheapo" Phonostage may appear to be "too simple" to offer "no compromise", however you will find within the design brief limits (which excluded such niceties as LCR RIAA Modules, expensive MC Stepup Transformers and such) it actually offers exactly that.
On top of this BTW, a J-Fet MC Pre-Pre such as I recommend for the El Cheapo has an average noise levels in the "a few nV|/Hz region at 100Hz, which is around 20db less noise than the ECC83 on average has and as much as 30db+ less noise than the ECC88.
I find also that pentodes in moving magnet level input sgates can be surpremely quiet. My quite extreme LCR equalised Phonostage uses a E810F operated as pentode on the input. It is actually the quietest of all my valve based phonostages!
You might find some of my previous Phonostage Projects of interrest as well:
http://thunderstoneaudio.nav.to
Sorry for the various popups....
Ciao T
Thank you for the contribution Thoersten. I find those graphs useful and maybe I will add a note in the article.
Sorry but the site you linked has too much popups to be read. Sometimes you select a link and it goes to ebay or other things like this. Some articles can't be read at all.
And I'm extremely sorry I can't understand German because the article at Jogi's site appears to be interesting...
(ps I didn't like too much your irony about "it might appear to you too simple". Instead of always pointing us to your excellent design, you could be more costructive, in my opinion.)
Sorry but the site you linked has too much popups to be read. Sometimes you select a link and it goes to ebay or other things like this. Some articles can't be read at all.
And I'm extremely sorry I can't understand German because the article at Jogi's site appears to be interesting...
(ps I didn't like too much your irony about "it might appear to you too simple". Instead of always pointing us to your excellent design, you could be more costructive, in my opinion.)
Hi Thoersten,
Man, those pop-ups are killing the browsing experience on your website! Otherwise it looks interesting.
-Chris
Man, those pop-ups are killing the browsing experience on your website! Otherwise it looks interesting.
-Chris
Hi Giaime,
Your second stage looks like a subtle mu stage. I thought I'd share with you my current push for a pentode loaded mu stage. The idea is floating around a couple of other threads right now. The good stuff is here http://www7.taosnet.com/f10/mustage.html
Your second stage looks like a subtle mu stage. I thought I'd share with you my current push for a pentode loaded mu stage. The idea is floating around a couple of other threads right now. The good stuff is here http://www7.taosnet.com/f10/mustage.html
Thorsten, Interesting data. That must have taken some time! Thanks for publishing it.
This looks like flicker noise at work (as well as other 1/f^n effects). These are quite device variable and even somewhat variable in a single device as it ages. Notice how the noise curves converge at higher frequencies as shot noise begins to dominate and things become more predictable.
Now I have to ask: Have you tried running the 6DJ8 at anything higher than a mere 3mA? Transconductance will only be around 4mS at 3mA of anode current. If you run it up to 15mA, you’ll get closer to the specified gm of 12.5mS. That alone will reduce the noise in the region above flicker noise by about 5 dB. You may find that some components of flicker noise will reduce too, as their current contributions take on a lower percentage of the larger anode current. I think that current is too often spared in the front ends of phono and mike preamps.
Next, since noise voltage is proportional to SQR(Hz), we must realize that the decade from 1KHz to 10KHz (where flicker noise is declining in effect) has 10dB more bandwidth significance (weight in the total noise voltage) than the decade from 100 to 1KHz. The logarithmic frequency axis can be misleading when the dependent variable is defined as units per SQR(Hz).
Finally, since the ear is not as sensitive to all frequencies, often a weighting such as the A-weighting is applied to crudely account for hearing sensitivity. A-weighting cuts the significance of 100Hz by 20dB relative to 1KHz, and cuts it by a whopping 50dB at 20Hz. These low frequencies are where the flicker effects are at their worst, thankfully. However, in an RIAA-equalized stage there is 13.2dB more gain to 100Hz than at 1KHz (and 19.4 dB more at 20Hz than at 1KHz). So, depending on the gain and equalization distribution, a portion, but not all, of the A-weighting would be reversed by the RIAA equalization.
So what do you have when you throw all these considerations into the hat? It gets pretty complicated. It’s hard to say anything definitive from the data as provided. My own experience (and that of many others) is that audible noise is generally lower when using higher transconductance triodes (richly biased) in RIAA and mike preamps, but it is not guaranteed since it is design dependent.
This looks like flicker noise at work (as well as other 1/f^n effects). These are quite device variable and even somewhat variable in a single device as it ages. Notice how the noise curves converge at higher frequencies as shot noise begins to dominate and things become more predictable.
Now I have to ask: Have you tried running the 6DJ8 at anything higher than a mere 3mA? Transconductance will only be around 4mS at 3mA of anode current. If you run it up to 15mA, you’ll get closer to the specified gm of 12.5mS. That alone will reduce the noise in the region above flicker noise by about 5 dB. You may find that some components of flicker noise will reduce too, as their current contributions take on a lower percentage of the larger anode current. I think that current is too often spared in the front ends of phono and mike preamps.
Next, since noise voltage is proportional to SQR(Hz), we must realize that the decade from 1KHz to 10KHz (where flicker noise is declining in effect) has 10dB more bandwidth significance (weight in the total noise voltage) than the decade from 100 to 1KHz. The logarithmic frequency axis can be misleading when the dependent variable is defined as units per SQR(Hz).
Finally, since the ear is not as sensitive to all frequencies, often a weighting such as the A-weighting is applied to crudely account for hearing sensitivity. A-weighting cuts the significance of 100Hz by 20dB relative to 1KHz, and cuts it by a whopping 50dB at 20Hz. These low frequencies are where the flicker effects are at their worst, thankfully. However, in an RIAA-equalized stage there is 13.2dB more gain to 100Hz than at 1KHz (and 19.4 dB more at 20Hz than at 1KHz). So, depending on the gain and equalization distribution, a portion, but not all, of the A-weighting would be reversed by the RIAA equalization.
So what do you have when you throw all these considerations into the hat? It gets pretty complicated. It’s hard to say anything definitive from the data as provided. My own experience (and that of many others) is that audible noise is generally lower when using higher transconductance triodes (richly biased) in RIAA and mike preamps, but it is not guaranteed since it is design dependent.
- Status
- Not open for further replies.