Pip
Well-known member
I always thought that negative feedback NFB effected THD Total Harmonic Distortion and circuit linearity the trade off was loss in gain from a given amplifier and not the noise of the circuit?
EF86
- Thread starter johnheath
- Start date
Help Support GroupDIY Audio Forum:
ruffrecords
Well-known member
emrr said:
No problem Doug. It is good that we can disagree or talk at cross purposes or whatever as gentlemen.
I totally agree with PRR. When I started designing tube mic pres I was very concerned about measured noise performance. All the SS preamps were quoting noise levels close the the theoretical Johnson noise from the source resistance but I found it hard to get within 10dB of that. The first half decent tube mic pre I built I connected to a Rode NT1A which is one of the quietest mics on the market. I turned the gain up to 70dB expecting to hear hiss but instead I could hear nothing but acoustic background noise.
Cheers
Ian
Pip said:
My implication is apparently lost on all: The noise will be less apparent with lower gain. It may be obnoxiously unbearable with higher gain. All these building blocks can be used to assemble systems of wildly varying gain. Some will not work out so well. That which you might use for a 40dB system might not work well stretched to a 70 dB system. A junior designer will not know this. NFB may or may not be part of the toolkit, it usually is but not always. Historically when building 40dB systems, pentodes are generally used in mic preamps to give excess gain which can be funneled into NFB, thereby eliminating use of an additional triode that would be required with large NFB.
Doug: I clearly stated that I had the idea to use the pentode as the first preamp tube feeding the second… a12AU7 that would be SRPP coupled as the output tube… no output transformer in this case.
Values of components are not yet decided… that will be an elaboration =)
Sincerely
/John
Values of components are not yet decided… that will be an elaboration =)
Sincerely
/John
CJ
Well-known member
there was a strange tax scheme back in the day which penalized radio makers by way of how many tubes they had on a chassis,
so the pentode was designed to take the place of 2 triodes, thus lowering the tax,
EF86 is close to EF804S which is used in the Telefunken preamps,
you can do things to cut down on microphonics and noise, mount the tube socket with rubber grommets, drop the B+ voltage, and operate it in triode, or in between triode and pentode which involves playing with screen circuit, change resistor and cap values, lower screen voltage, etc.
you can also lower noise by using NFB as your volume control, here is the front end of the V76 which is a very good circuit,
note the low 67 volts on the first plate for less noise,
note how the 50 uf cap in the FB circuit gets closer to ground as gain is changed so as to compensate for tone ,
as gain goes up, so does the bass,
why is the cathode biased up? to allow application of the NFB.
so the pentode was designed to take the place of 2 triodes, thus lowering the tax,
EF86 is close to EF804S which is used in the Telefunken preamps,
you can do things to cut down on microphonics and noise, mount the tube socket with rubber grommets, drop the B+ voltage, and operate it in triode, or in between triode and pentode which involves playing with screen circuit, change resistor and cap values, lower screen voltage, etc.
you can also lower noise by using NFB as your volume control, here is the front end of the V76 which is a very good circuit,
note the low 67 volts on the first plate for less noise,
note how the 50 uf cap in the FB circuit gets closer to ground as gain is changed so as to compensate for tone ,
as gain goes up, so does the bass,
why is the cathode biased up? to allow application of the NFB.
Attachments
ruffrecords
Well-known member
CJ said:
Oh dear, I am going to have to disagree again. Sorry CJ!!
It is unfortunate that the circuit is drawn in a way the makes it difficult to see what is happening. One big problem with NFB is stability and stability depends on open loop gain and the amount of NFB. To ensure stability it is a good idea to keep the difference between open loop gain and the gain after NFB costant. This means that as tou change the gain after NFB you also need to change the open loop gain. And THAT is what the neat circuit with the 50uF cap does. As you get to the higher after NFB gains you bypass the cathode resistors with lower values of resistor via the 50uF which automatically increases the gain of the pentode stage and keeps the difference nearly constant and ensures stability at all gain settings. I
Cheers
Ian
> tutorial by Great Guru Baxandall
> In dem days, mike input transformers were (relatively) even more expensive
> Actually he DOES recommend input transformers in...
Context:
In dem days, mike transformers were often IN the microphone. Your mixer had naked-grid inputs (possibly with jumpered sockets for optional MTs). It was generally cheaper to buy mikes with transformers one-by-one than to load-up the mixer with more transformers than you needed just yet. (There was also some fad for crystal mikes which work well direct to naked-grid.)
I believe Baxandall was assuming a Hi/Lo-Z microphone on Hi-Z.
Background: the impedance of a long microphone line should be 50-500 Ohms. Lower gets copper-loss. Higher gets high-cut and induced buzz. The naked mike likes to be 0.1 Ohms (ribbon) or 10 Ohms (dynamic). So to some degree, there had to be a transformer at the mike, to couple 0.1-10r to a hundred Ohm line. (Later dynamics were wound directly to 200 Ohms, no transformer, but the SM58 comes from a line of hi/lo-Z models and retains a 10 Ohm coil with a transformer.)
"Short" lines can be run at higher Z. If the mayor gives a speech, and you set up just off stage, and you don't want to hear anything over 3KC out of his mouth, 25 or 50 feet at 10K-20K source impedance is fine. In tape recording they sometimes put the recorder on the stage apron, 10' cords. Hi-Z mikes were widely used for decades.
The source impedance of dynamic and ribbon is nearly the (transformed) capsule impedance. On low-Z we expect 200 Ohms. Ideally the mike amp has an OSI near here.
The OSI of a tube is "super high". It has Noise Voltage, but the Noise Current is essentially too small to matter for audio involving lines and transformers. (Matters in the condenser head.) But take the Noise Voltage as an equivalent resistance and call that a OSI.
In a triode this may be 2,000 Ohms, 500 Ohms for heroic cases, though at low current perhaps 3,000 or more. (And we want low current because B+ ripple is a different kind of "noise" and B+ filtering is expensive, but less-so if the current is small.)
In a pentode part of the precious signal from grid-cathode action is diverted to Screen. Rather more than you might expect. The OSI is liable to be 10K-20K.
When OSI matches we have a 3dB Noise Figure. Good, not the best, but very close.
So 150r mike, triode of 3K OSI, we want at least 1:4.5 transformer ratio.
150r mike, pentode of 20K OSI, we want at least 1:11.5 transformer ratio.
So we often like a high ratio for best hiss.
As the step-up and secondary impedance rises, we lose bandwidth (and also must use hyper-fine wire which is a costly pain). A 600:600 transformer can be VERY flat over a wide range. A 200r:45K (1:15) transformer has to be very carefully partitioned and very gently loaded to beat 15KHz.
So we often like a low ratio for best highs per buck.
1:15 was used in a fair number of PA and broadcast consoles.
1:7 is more likely to give wide-range response at a fair price.
Don't focus on one number. Is it intended to be wide-band? Is it convenient to mount? Is it well shielded? We see many preamps here where the highest "noise" component is power transformer hum, making hiss-noise somewhat moot.
> In dem days, mike input transformers were (relatively) even more expensive
> Actually he DOES recommend input transformers in...
Context:
In dem days, mike transformers were often IN the microphone. Your mixer had naked-grid inputs (possibly with jumpered sockets for optional MTs). It was generally cheaper to buy mikes with transformers one-by-one than to load-up the mixer with more transformers than you needed just yet. (There was also some fad for crystal mikes which work well direct to naked-grid.)
I believe Baxandall was assuming a Hi/Lo-Z microphone on Hi-Z.
Background: the impedance of a long microphone line should be 50-500 Ohms. Lower gets copper-loss. Higher gets high-cut and induced buzz. The naked mike likes to be 0.1 Ohms (ribbon) or 10 Ohms (dynamic). So to some degree, there had to be a transformer at the mike, to couple 0.1-10r to a hundred Ohm line. (Later dynamics were wound directly to 200 Ohms, no transformer, but the SM58 comes from a line of hi/lo-Z models and retains a 10 Ohm coil with a transformer.)
"Short" lines can be run at higher Z. If the mayor gives a speech, and you set up just off stage, and you don't want to hear anything over 3KC out of his mouth, 25 or 50 feet at 10K-20K source impedance is fine. In tape recording they sometimes put the recorder on the stage apron, 10' cords. Hi-Z mikes were widely used for decades.
The source impedance of dynamic and ribbon is nearly the (transformed) capsule impedance. On low-Z we expect 200 Ohms. Ideally the mike amp has an OSI near here.
The OSI of a tube is "super high". It has Noise Voltage, but the Noise Current is essentially too small to matter for audio involving lines and transformers. (Matters in the condenser head.) But take the Noise Voltage as an equivalent resistance and call that a OSI.
In a triode this may be 2,000 Ohms, 500 Ohms for heroic cases, though at low current perhaps 3,000 or more. (And we want low current because B+ ripple is a different kind of "noise" and B+ filtering is expensive, but less-so if the current is small.)
In a pentode part of the precious signal from grid-cathode action is diverted to Screen. Rather more than you might expect. The OSI is liable to be 10K-20K.
When OSI matches we have a 3dB Noise Figure. Good, not the best, but very close.
So 150r mike, triode of 3K OSI, we want at least 1:4.5 transformer ratio.
150r mike, pentode of 20K OSI, we want at least 1:11.5 transformer ratio.
So we often like a high ratio for best hiss.
As the step-up and secondary impedance rises, we lose bandwidth (and also must use hyper-fine wire which is a costly pain). A 600:600 transformer can be VERY flat over a wide range. A 200r:45K (1:15) transformer has to be very carefully partitioned and very gently loaded to beat 15KHz.
So we often like a low ratio for best highs per buck.
1:15 was used in a fair number of PA and broadcast consoles.
1:7 is more likely to give wide-range response at a fair price.
Don't focus on one number. Is it intended to be wide-band? Is it convenient to mount? Is it well shielded? We see many preamps here where the highest "noise" component is power transformer hum, making hiss-noise somewhat moot.
12AU7 is often not a first-choice for a low-hiss input. The Gm (inversely related to OSI) is good (at high currents), but the voltage gain is only 15 or 16. You could get a Noise Figure of 1dB in the first stage. However with such low gain, into a similar second stage, you pick up almost another 1dB of hiss in the second stage.
12AX7 can give gain of 50, making second-stage hiss negligible. The 12AX7 will have more input capacitance, making it tougher to get >20KHz response on a higher-ratio transformer. However for many reasons the 12AX7 remains our go-to for simple effective music preamps.
The 12AU7 was never marketed for "low hiss". Gain is low. (The 12AY7, with more than twice the gain, was developed and marketed as "low hiss", though many of its improvements were brought into later 7025 and 12AX7 and the AY kinda got lost.)
12AX7 was only supposed to be a computer tube (analog computers, particularly anti-aircraft aiming computers). Low hiss was not the goal. However more designers grabbed it to simplify their audio (more gain in fewer stages, more NFB, hoo-ray!). Today the main market is guitar amps, and the first and second stages must not be real hissy/crackly. Yes, it is a low-bid world and there's crap 12AX7 on the market, but you can find (sort) some fine samples.
Triode-strapped EF86 has amplification very similar to 12AY7, and was always promoted as "low hiss". Again they have been made in many places and sometimes at too-low prices, so there will be crap EF86 around.
Many-many-MANY tubes can be fine mike preamps. Especially if you feel that BIG CLEAN B+ power is cheap. TV Tuner tubes were aimed for low snow (RF hiss) when working up around 10mA. They have enuff gain to overwhelm 2n-stage hiss. Video amplifiers may be suitable. You might also consider EL84 working up near 40mA in Triode. Huge Gm, don't need high-ratio transformer; probably don't want hi-ratio because in triode the input capacitance may be high.
Anyway if you MUST run 1dB NF and 30KHz response, what you really want is the $10 chip. If you really are lusting for tube preamps, you are wanting/wishing for things other than extended response and lowest possible hiss.
I'll never go back to tube preamps for practical reasons. They always crackle in their sockets. I don't like "damaging" a signal right-off... save the bending for post production.
12AX7 can give gain of 50, making second-stage hiss negligible. The 12AX7 will have more input capacitance, making it tougher to get >20KHz response on a higher-ratio transformer. However for many reasons the 12AX7 remains our go-to for simple effective music preamps.
The 12AU7 was never marketed for "low hiss". Gain is low. (The 12AY7, with more than twice the gain, was developed and marketed as "low hiss", though many of its improvements were brought into later 7025 and 12AX7 and the AY kinda got lost.)
12AX7 was only supposed to be a computer tube (analog computers, particularly anti-aircraft aiming computers). Low hiss was not the goal. However more designers grabbed it to simplify their audio (more gain in fewer stages, more NFB, hoo-ray!). Today the main market is guitar amps, and the first and second stages must not be real hissy/crackly. Yes, it is a low-bid world and there's crap 12AX7 on the market, but you can find (sort) some fine samples.
Triode-strapped EF86 has amplification very similar to 12AY7, and was always promoted as "low hiss". Again they have been made in many places and sometimes at too-low prices, so there will be crap EF86 around.
Many-many-MANY tubes can be fine mike preamps. Especially if you feel that BIG CLEAN B+ power is cheap. TV Tuner tubes were aimed for low snow (RF hiss) when working up around 10mA. They have enuff gain to overwhelm 2n-stage hiss. Video amplifiers may be suitable. You might also consider EL84 working up near 40mA in Triode. Huge Gm, don't need high-ratio transformer; probably don't want hi-ratio because in triode the input capacitance may be high.
Anyway if you MUST run 1dB NF and 30KHz response, what you really want is the $10 chip. If you really are lusting for tube preamps, you are wanting/wishing for things other than extended response and lowest possible hiss.
I'll never go back to tube preamps for practical reasons. They always crackle in their sockets. I don't like "damaging" a signal right-off... save the bending for post production.
Wow… first I must say that it lis very interesting to follow the discussions between the big elephants and I… (a very small elephant) can just sit down and try to suck it all in.
Anyway… yes, I am aiming for other benefits from a tube preamp than zero noise.
Actually I am already thinking about leaving the EF86 and stick to triodes… and maybe change the idea of using the 12AU7 as the first preamp tube =)
PRR: As I am from sweden and english is not my native language it is sometimes a bit difficult to grasp the whole picture… but If I understand you correctly your advise is to use a 12AX7/ 7025 as the first preamp tube based on its relationship between gain and noise…. high gain which will be enough before the noise floor will be disturbing? … rather than using the 12AU7 with its lower gain compared to its own noise floor?
If that would be the case I also believe that it would demand a "cooler" bias than the 12AU7 compared to the 12AX7?
Thank you all for a very informative discussion =)
Sincerely
/John
Anyway… yes, I am aiming for other benefits from a tube preamp than zero noise.
Actually I am already thinking about leaving the EF86 and stick to triodes… and maybe change the idea of using the 12AU7 as the first preamp tube =)
PRR: As I am from sweden and english is not my native language it is sometimes a bit difficult to grasp the whole picture… but If I understand you correctly your advise is to use a 12AX7/ 7025 as the first preamp tube based on its relationship between gain and noise…. high gain which will be enough before the noise floor will be disturbing? … rather than using the 12AU7 with its lower gain compared to its own noise floor?
If that would be the case I also believe that it would demand a "cooler" bias than the 12AU7 compared to the 12AX7?
Thank you all for a very informative discussion =)
Sincerely
/John
A practical problem in all of this is the amount of gain produced by the 'correct' input transformer plus the first stage. If you typically follow that with a standard voltage divider volume control, in any loud case you've already overdriven the input stage unless you pad the mic output. A lower mu input stage will buy you some headroom, and might allow you to avoid that pad. I speak from the position of using all sorts of vintage tube preamps in a rock and roll studio; most classic designs by nature do not have the input headroom required to take anything more than average dynamic mic output levels from loud sources without using an input pad, which is to defeat some of the purpose of the preamp in the first place. You might not get away with a modern high output condenser on a loud source with a pad even, they virtually produce line level signals, which call for a SS based preamp which can run minimum gains down close to unity; more of a buffer amp approach. If you go with feedback based volume control, you are defining a certain type of preamp which will also change in sonic quality based on the gain setting. An EF86 in triode is more or less equivalent to the classic octal 6J7 used as a triode, which may be about the most common input stage amongst professional equipment for a long long period of time. In other words, it's a good choice. I personally rarely like the sound of 12AX7's. When I do like them they are hobbled by smaller than spec plate resistors which reduces total gain; total not-by-the-book approach that is nonetheless fairly common in certain eras.
CJ
Well-known member
every 12AX7a that sounded good seemed to be a little microphonic, figure that out,
Once in history when I was young and played the guitar and worshipped Stevie Ray Vaughan and Jimi Hendrix to mention two of the top players I tried to get a decent amp and guitar to match their sound and a very famous swedish guitar player said to me…" It is more important how you play than what gear you use". I guess that is implied here as well… I guess we are talking details more or less beyond what we can grasp or hear… the music itself must be more important than these details.
I guess we all have good experience from playing it dirty and raw and also from playing clean and soft… ?
Since PRR's last post I have made a very nice preamp with a 12AY7 as the first preamp tube and a 5687 as the output tube… fiddling with resistors and NFB and not and it too sounds nice =)
I guess it is all down to a matter of taste in the end as long as it is sounding good?
Regards
/John
I guess we all have good experience from playing it dirty and raw and also from playing clean and soft… ?
Since PRR's last post I have made a very nice preamp with a 12AY7 as the first preamp tube and a 5687 as the output tube… fiddling with resistors and NFB and not and it too sounds nice =)
I guess it is all down to a matter of taste in the end as long as it is sounding good?
Regards
/John
CJ
Well-known member
JW
Well-known member
Hey folks,
I thought maybe I'd attach this question to this thread. I just finished up building my first tube preamp directly from a schematic. It's sort of an offshoot of people using the bones from Akai and Roberts tube reel to reel blocks and modifying them. But this guy redesigned the circuit a lot. He posted the schematic here: http://xfmr.blogspot.com/2012/06/brute-force-of-jekyll-hyde.html It's the first one, called the Brute Force. I haven't done any noise measurements, but so far it seems very quiet.
Anyhow, my question is concerning pin 8 of the EF86. I've seen in some designs (including the original Akai amp fwiw) pin 8 is grounded, and in this schematic he has it connected to the cathode (pin3). I wonder if somebody could talk about the difference? Having two channels, I've tried both ways, but I can't say I hear a difference.
I thought maybe I'd attach this question to this thread. I just finished up building my first tube preamp directly from a schematic. It's sort of an offshoot of people using the bones from Akai and Roberts tube reel to reel blocks and modifying them. But this guy redesigned the circuit a lot. He posted the schematic here: http://xfmr.blogspot.com/2012/06/brute-force-of-jekyll-hyde.html It's the first one, called the Brute Force. I haven't done any noise measurements, but so far it seems very quiet.
Anyhow, my question is concerning pin 8 of the EF86. I've seen in some designs (including the original Akai amp fwiw) pin 8 is grounded, and in this schematic he has it connected to the cathode (pin3). I wonder if somebody could talk about the difference? Having two channels, I've tried both ways, but I can't say I hear a difference.
> "It is more important how you play than what gear you use".
A brilliant trumpeter told a master class "It's just air!"
Meaning: don't be whining and lusting for yet another fancy-alloy mouthpiece. (Knowing his history, he musta rose from the wrong side of the tracks into a major metro orchestra with pawn-shop cast-off instruments.)
Anyway: if you "must" use tubes, "must" design it yourself, this is a good read:
http://www.americanradiohistory.com/Audio-Magazine.htm
1955 - Oct (10MB PDF file)
page 25 "Cascode Preamp...."
I've said elsewhere that I am not keen on cascodes generally. Here they are not wrong (but there are other ways to skin this cat).
Regardless how you feel about the "flavor" of 12AX7, you always find opinions that it is a computer tube not designed for low hiss or low hum. This was true, in the old days. Turn to page 8 in the rag above, Tung-Sol ad.
TUNG-SOL HI-FI TUBES
Premium performance to satisfy the most critical Hi-Fi enthusiast is engineered into these popular Tung-Sol Tubes. Their ratings, uniformity and dependability demonstrate that Tung-Sol quality control methods can achieve in volume production the performance levels required for highest quality equipment.
12AX7 TWIN TRIODE VOLTAGE AMPLIFIER
picks up low level signals without introducing hum.
This may be why the 12AY7 faded-away, the 7025 and 6EU7 are found in the fringes. While 6EU7's heater pins are maybe in better places, and the dealers would be pleased to sell a 7025 for a few bucks more than a 12AX7, the "hi-fi type" 12AX7 really became good-enuff for almost any use, and the changes were adopted into almost all 12AX7 production.
Using Kiebert's numbers, 12AY7's effective hiss resistance is 3.4K, 6BQ7's is 1.5K. Using his 2X to 4X criteria, the optimum source resistance runs 3K to 14K. A 7K source will usually be fine.
A 150r:7K transformer is around 1:7 voltage ratio. That is a fair amount of gain. Good if you are short on gain. Bad if, like Doug, you do Rock (or rocky C/W) with hot mikes at short range. Of course then hiss may be totally moot.
A brilliant trumpeter told a master class "It's just air!"
Meaning: don't be whining and lusting for yet another fancy-alloy mouthpiece. (Knowing his history, he musta rose from the wrong side of the tracks into a major metro orchestra with pawn-shop cast-off instruments.)
Anyway: if you "must" use tubes, "must" design it yourself, this is a good read:
http://www.americanradiohistory.com/Audio-Magazine.htm
1955 - Oct (10MB PDF file)
page 25 "Cascode Preamp...."
I've said elsewhere that I am not keen on cascodes generally. Here they are not wrong (but there are other ways to skin this cat).
Regardless how you feel about the "flavor" of 12AX7, you always find opinions that it is a computer tube not designed for low hiss or low hum. This was true, in the old days. Turn to page 8 in the rag above, Tung-Sol ad.
TUNG-SOL HI-FI TUBES
Premium performance to satisfy the most critical Hi-Fi enthusiast is engineered into these popular Tung-Sol Tubes. Their ratings, uniformity and dependability demonstrate that Tung-Sol quality control methods can achieve in volume production the performance levels required for highest quality equipment.
12AX7 TWIN TRIODE VOLTAGE AMPLIFIER
picks up low level signals without introducing hum.
This may be why the 12AY7 faded-away, the 7025 and 6EU7 are found in the fringes. While 6EU7's heater pins are maybe in better places, and the dealers would be pleased to sell a 7025 for a few bucks more than a 12AX7, the "hi-fi type" 12AX7 really became good-enuff for almost any use, and the changes were adopted into almost all 12AX7 production.
Using Kiebert's numbers, 12AY7's effective hiss resistance is 3.4K, 6BQ7's is 1.5K. Using his 2X to 4X criteria, the optimum source resistance runs 3K to 14K. A 7K source will usually be fine.
A 150r:7K transformer is around 1:7 voltage ratio. That is a fair amount of gain. Good if you are short on gain. Bad if, like Doug, you do Rock (or rocky C/W) with hot mikes at short range. Of course then hiss may be totally moot.
If you grew-up around NYC, see also page 32 (p34 in PDF) of Audio Oct 1955. Author is some guy who claims to sell more phonograph records than anybody else in the country. Name is Sam Goody.
FYI:
https://en.wikipedia.org/wiki/Sam_Goody
http://nodepression.com/article/story-sam-gutowitz
FYI:
https://en.wikipedia.org/wiki/Sam_Goody
http://nodepression.com/article/story-sam-gutowitz
DAMN YOU PRR!!
I'm never going to build another valve device especially a preamp. ( maybe a 20W EL34 UL amp ...
shut up! Just shut up! )
But I'm collecting all this info you keep throwing at us as though I am.
My biggest take so far is that triode connected EF86 is sorta 12AY7 / 7025 which explains why 7025 was so popular with sensible condensor mike makers.
Today, it may be a question of which valve is available with reliable performance. In da old days, you'd just specify Mullard or one of the many brands under the Philips empire.
Now, the brand is just a screen print which can hide any number of sins. A less well know valve might be better cos there will be less cheapo yucky fakes.
I'm never going to build another valve device especially a preamp. ( maybe a 20W EL34 UL amp ...
shut up! Just shut up! )But I'm collecting all this info you keep throwing at us as though I am.
My biggest take so far is that triode connected EF86 is sorta 12AY7 / 7025 which explains why 7025 was so popular with sensible condensor mike makers.
Today, it may be a question of which valve is available with reliable performance. In da old days, you'd just specify Mullard or one of the many brands under the Philips empire.
Now, the brand is just a screen print which can hide any number of sins. A less well know valve might be better cos there will be less cheapo yucky fakes.
Similar threads
Latest posts
-
-
-
-
-
-
Modifying the Collins 356A Mic Pre for variable Negative Feedback Control
- Latest: abbey road d enfer
-