Hi RDF,
Thanks for taking trouble to check this out. I haven't yet started simulating with SE transformers due to the worry that I might get it all wrong and end up no where and waste a lot of time!
The grid stopper isn't what I like and is what's on the original circuit. I would have used a smaller value but you suggest taking it out all together.
Let me work on this . Your suggestion sounds interesting. I always like the idea of dumping caps !
I have just got all the components for this amp. I had bought the tubes 6 years ago ! Just waiting to get the chassis ready . Meantime I wanted to see if I can improve on it's original JELabs circuit.
Thanks,
Ashok.
PS. Do you use LT Spice for simulations or some other package ?
I have LT Spice and Beige Bag .
Thanks for taking trouble to check this out. I haven't yet started simulating with SE transformers due to the worry that I might get it all wrong and end up no where and waste a lot of time!
The grid stopper isn't what I like and is what's on the original circuit. I would have used a smaller value but you suggest taking it out all together.
Let me work on this . Your suggestion sounds interesting. I always like the idea of dumping caps !
I have just got all the components for this amp. I had bought the tubes 6 years ago ! Just waiting to get the chassis ready . Meantime I wanted to see if I can improve on it's original JELabs circuit.
Thanks,
Ashok.
PS. Do you use LT Spice for simulations or some other package ?
I have LT Spice and Beige Bag .
When the grid gets close to 0V or indeed goes positive, it becomes essentially a forward biassed vacuum diode, so grid current begins flowing. If you have a coupling cap to the grid, the grid side becomes effectively 'clamped' to 0V or thereabouts by the diode, and the diode current charges the cap to a positive voltage, referred to the grid end - in other words, you get rectification effects of the drive voltage. This emans that 'overdriving' actually makes the grid bias more negative, by 'adding' extra bias voltage over the coupling cap - so you get a operating point shift and distortion - the size of the cap really does not matter in principle, it only changes how long it takes for the operating point shift to occur and return to normal once overdrive is removed, but of course, the value will makedly change the sound of the amp when overdriven.
You could also look at the grid as a nonlinear impedance, which fairly abruptly drops several orders of magnitude near zero crossing. This means that you need something with a much lower output impedance to drive the grid near zero or positive, far lower than the impedance of the grid in this mode of operating, and that would be some sort of a follower, be that vacuum or solid state - the point being, it should be DC coupled. The bias voltage is then applied to the input of the follower and it is AC coupled to the driver. Of course, sufficient power rail headroom needs to be provided so that the same rectification echanism does not happen at the input of the follower.
You could also look at the grid as a nonlinear impedance, which fairly abruptly drops several orders of magnitude near zero crossing. This means that you need something with a much lower output impedance to drive the grid near zero or positive, far lower than the impedance of the grid in this mode of operating, and that would be some sort of a follower, be that vacuum or solid state - the point being, it should be DC coupled. The bias voltage is then applied to the input of the follower and it is AC coupled to the driver. Of course, sufficient power rail headroom needs to be provided so that the same rectification echanism does not happen at the input of the follower.
Just LTSpice. I'm at work and don't have access to the files but the circuit is simple. An ideal AC signal source drives a resistor (R1) and cap (C1) in series. R1 represents the driver's output impedance and C1 the coupler. The load is a diode, resistor (R2) and ideal DC voltage source in series. The diode is of course the grid under positive drive, R2 the grid impedance under positive drive (estimate from data sheets) and the DC voltage source set to the output tube bias voltage.
The resultant distortion spectrum when the AC source is set higher than the DC bias voltage is amazingly similar to what I see when using tubes. Comparing distortion profiles when playing around with R1 and C1 is highly educational.
The resultant distortion spectrum when the AC source is set higher than the DC bias voltage is amazingly similar to what I see when using tubes. Comparing distortion profiles when playing around with R1 and C1 is highly educational.
Re: MOSFET driver for 300B
Yeah, it will work. The driver will certainly provide enough current to drive the 300B's grid capacitance. However, the capacitor coupling defeats one of the advantages of using a grid driver. This won't help the clipping misbehaviour known as "blocking" when the inevitable transient comes along and clips, charging the coupling capacitor negative. Even if you don't hear an obvious clip, shifting the operating point down the loadline can cause sonic degradation.
Also, it looks like you're going to have to pull more source current. 2mA isn't much, and you'll be running that MOSFET at nearly cutoff. 20 - 30mA would be more like it.
ashok said:
Yeah, it will work. The driver will certainly provide enough current to drive the 300B's grid capacitance. However, the capacitor coupling defeats one of the advantages of using a grid driver. This won't help the clipping misbehaviour known as "blocking" when the inevitable transient comes along and clips, charging the coupling capacitor negative. Even if you don't hear an obvious clip, shifting the operating point down the loadline can cause sonic degradation.
Also, it looks like you're going to have to pull more source current. 2mA isn't much, and you'll be running that MOSFET at nearly cutoff. 20 - 30mA would be more like it.
I have a Yaqin MC-100B, with is 60 watts Ultralinear or 30 watts triode. The difference between Triode and UL is very subtle. UL doesn't sound harsh at all, and tends to have a little more bass. Triode sounds better when listening to Pink Floyd... but even then I can't reliably determine which mode I'm on in a blind test.
I leave it on triode mostly, because I've never actually needed the extra power from UL. 30 watts triode is a lot of power.
Oh... note, I'm not running stock tubes. The Yaqin is a little bright stock... Electro Harmonix 12AX7's fixed that completely. Sylvania chrome domes (circa 1948) really tightened the bass up too. KT88 tubes are stock though. Note that this amp *does* sound better biased correctly. Noticably better. Tube life may be compromised though.
Charles.
I leave it on triode mostly, because I've never actually needed the extra power from UL. 30 watts triode is a lot of power.
Oh... note, I'm not running stock tubes. The Yaqin is a little bright stock... Electro Harmonix 12AX7's fixed that completely. Sylvania chrome domes (circa 1948) really tightened the bass up too. KT88 tubes are stock though. Note that this amp *does* sound better biased correctly. Noticably better. Tube life may be compromised though.
Charles.
ilimzn said:
I thought so too but it doesn't appear to be so. The coupling cap is an impedance in series with the Zout of the driver. If the driver is a typical R loaded plate with a Zout in the Kohms, the cap isn't nearly as critical. If however the driver is a low impedance follower of some sort, the cap becomes a major factor because it's the primary impedance facing the output grid. In part this has nothing to do with shifting operating points since the phenomenon is clearly visible in the circuit below which contains no tubes. It's entirely due to the physics of a non-zero impedance voltage source driving a diode junction. Changing the cap from the shown 0.22u to 47u drops THD almost 40 dB for a constant Vac drive.
Attachments
rdf said:
Agreed, of course - there is simply not enough current to change the charge state of the cap. In most cases any blocking distortion that arises in such designs is simply cured by increasing the capacitance, as clipping is generally short term. Besides, the coupling cap already forms a high pass filter with the grid leak, so should be dimensioned for it's turnover frequency to be well into the infrasound range.
Of course, due to it's impedance becoming lower. The charging of the cap is a rectification thing, so 'nearly DC', not harmonically related, not steady state and will not show in a THD plot. However, if we were only monitoring AC grid drive, which is what your circuit demonstrates, what you say is certainly true - the AC voltage of course gets majorly distorted by the ratio of the follower + cap impedance to the non-linear grid impedance in A2.
That being said, the actual DC at the grid obviously impacts how an actual tube will behave, meaning, the operating point of the tube shifts towards cut-off. The thing is, because music is a complex and very 'peaky' signal, by increasing the value of the cap you ot only reduce it's impedance compared to the grid impedance (which you need to do anyway), but you also make the time needed for it to charge and discharge possibly long enough so that operating point shifts become small enough under real input signal conditions, to make the design perform adequatly. In other words, the 'evil day' is postponed till a condition of more long term or harder overdrive happens. Because we are not talking about a single stage in a real amp, further trouble may well occur in the other stages which makes the point of shifting bias obscure - as always, it's a question of balance.
While I don't disagree with what you're saying I have to stress again the distortion in A2 sims is strongly correlated to the composite impedance driving the grid. Getting rid of the cap and increasing the drive 'resistance' has the same result. And while blocking is a consideration it's one which to an extent can be tamed via grid chokes (not without their own issues.) Ultimately though there's little distortion benefit to a very low Zout follower drive stage when coupled with a small value cap, even with a time constant in the infrasonic. The cap's impedance apparently becomes the dominant factor determining distortion under A2 conditions.
This has been an interesting discussion and I'm looking at factors that I didn't consider earlier.
It all started because I saw a post suggesting that a low Zout using a MOSFET follower will help driving the 300B better. So I assumed that adding a MOSFET follower after the second stage on the JE Labs circuit will work better ! Now it looks like there is a lot of disagreement about this.
Mainly because of the coupling capacitor which is there in many circuits .
The consensus seems to be that with a higher Zout in the driver stage and using a R load on it , the coupling cap has a lesser influence in the performance. Right ? If so then a CCS on the anode of the driver should be a shade better still.
I can't see a clear conclusion from the discussion . It appears that the standard R loaded driver would give a better performance (?).
It all started because I saw a post suggesting that a low Zout using a MOSFET follower will help driving the 300B better. So I assumed that adding a MOSFET follower after the second stage on the JE Labs circuit will work better ! Now it looks like there is a lot of disagreement about this.
Mainly because of the coupling capacitor which is there in many circuits .
The consensus seems to be that with a higher Zout in the driver stage and using a R load on it , the coupling cap has a lesser influence in the performance. Right ? If so then a CCS on the anode of the driver should be a shade better still.
I can't see a clear conclusion from the discussion . It appears that the standard R loaded driver would give a better performance (?).
I haven't followed all the arguments in this thread but I don't understand what coupling caps have to do with a source follower driver.
The circuit I'm used to seeing is a source follwer direct-coupled to the grid of the OP tube. The MOSFET is capacitively coupled to the previous stage and its gate is negatively biased. That sets the voltage at the source and, thus, the OP tube grid.
Putting a coupling cap between the source and the OP tube grid seems futile, because it would take away most of the benefits of a low impedance driver.
The circuit I'm used to seeing is a source follwer direct-coupled to the grid of the OP tube. The MOSFET is capacitively coupled to the previous stage and its gate is negatively biased. That sets the voltage at the source and, thus, the OP tube grid.
Putting a coupling cap between the source and the OP tube grid seems futile, because it would take away most of the benefits of a low impedance driver.
ray_moth said:
Precisely so, that would be the way to do it. However, Ashok posted a schematic on the top of this page which started the discussion of what happens when you try to drive A2 through a coupling cap, as in that schematic the coupling cap is between the follower and output tube grid.
Ha .........now we are talking ! The cap should have been moved to the gate of the MOSFET. That should settle the need to remove the cap from the source of the MOSFET and dc couple it to the 300B. OK , so I think I'll go ahead and try that .
It WAS an interesting discussion in any case. Thank's to everybody .
It WAS an interesting discussion in any case. Thank's to everybody .
ashok said:
You can take a MOSFET driver from my Alligator project; though it was not breadboarded yet it should work well.
http://www.diyaudio.com/forums/showthread.php?postid=1451297#post1451297
Ricky,
As you've seen, there is no "Yes or No" answer to your post. When you ask a question that involves a subject that's been "kicked around " by tube enthusiasts for at least fifty years, and well might be a subject of discussion for the next fifty, you'll get many replies. If you read carefully, in those replies you'll find many good points discussed and a lot of valuable information provided by knowledgeable people. That's why this is such a great forum!
As you've seen, there is no "Yes or No" answer to your post. When you ask a question that involves a subject that's been "kicked around " by tube enthusiasts for at least fifty years, and well might be a subject of discussion for the next fifty, you'll get many replies. If you read carefully, in those replies you'll find many good points discussed and a lot of valuable information provided by knowledgeable people. That's why this is such a great forum!
When multi-grid power tubes were develped, they seemed the answer to a maiden's prayer - easy to drive and bags of power compared with triodes.
The marketing people made the most of it, of course. While distortion from pentodes/beam tetrodes was considerably worse, negative feedback could tame them and result in a very respectable performance - certainly acceptable when compared with other components in the system.
The performance of radios, gramophones and other signal sources of the day, and the speakers that were generally available and affordable, left a lot to be desired. A well-designed pentode amp was easily the strongest link in the chain.
When better signal sources and more refined speakers came on the scene, the undesirable characteristics of pentodes became more noticeable but their reputation was saved by the introduction of distributed load/ultralinear topology, yeilding more triode-like quallity but stiil with pentode-like power capabilities. The period of acceptability of the pentode was thus extended for hi-fi.
During this fallow period for the power triode, there was virtually no development, at least for audio use. You could always triode-strap the pentode but you'd lose a lot of power and it wouldn't sound any better than ultralinear, by the standards of the day. The old 300B and 2A3 tubes were big and oldfashioned and hard to get, which made them unsuitable for commercial purposes. Most speakers at that time were also insufficiently sensitive for the low-powered 2A3.
Eventually it began to dawn on some informed people that the triode, under the right circumstances, offered a valid alternative to the pentode. Because negative feedback was rarely if ever used in early audio amps, triodes had to be very linear. The 300B, used with the older, more linear voltage amplifying triodes, in combination with the much better signal sources that had become available, enabled excellent sound quality to be achieved with no (or very little) negative feedback. Such an amp exhibited a livleiness and immediacy that made feedback-bound pentode amps seem dead by comparison.
The future of the pentode seems to be secure because of its usefulness in the music industry. However, for hi-fi use, it is questionable whether it will be able to offer a genuinely useful alternative to the SS amps of the future. Triodes definitely offer an alternative, arguably a highly desirable alternative, but pentodes . . ?
The marketing people made the most of it, of course. While distortion from pentodes/beam tetrodes was considerably worse, negative feedback could tame them and result in a very respectable performance - certainly acceptable when compared with other components in the system.
The performance of radios, gramophones and other signal sources of the day, and the speakers that were generally available and affordable, left a lot to be desired. A well-designed pentode amp was easily the strongest link in the chain.
When better signal sources and more refined speakers came on the scene, the undesirable characteristics of pentodes became more noticeable but their reputation was saved by the introduction of distributed load/ultralinear topology, yeilding more triode-like quallity but stiil with pentode-like power capabilities. The period of acceptability of the pentode was thus extended for hi-fi.
During this fallow period for the power triode, there was virtually no development, at least for audio use. You could always triode-strap the pentode but you'd lose a lot of power and it wouldn't sound any better than ultralinear, by the standards of the day. The old 300B and 2A3 tubes were big and oldfashioned and hard to get, which made them unsuitable for commercial purposes. Most speakers at that time were also insufficiently sensitive for the low-powered 2A3.
Eventually it began to dawn on some informed people that the triode, under the right circumstances, offered a valid alternative to the pentode. Because negative feedback was rarely if ever used in early audio amps, triodes had to be very linear. The 300B, used with the older, more linear voltage amplifying triodes, in combination with the much better signal sources that had become available, enabled excellent sound quality to be achieved with no (or very little) negative feedback. Such an amp exhibited a livleiness and immediacy that made feedback-bound pentode amps seem dead by comparison.
The future of the pentode seems to be secure because of its usefulness in the music industry. However, for hi-fi use, it is questionable whether it will be able to offer a genuinely useful alternative to the SS amps of the future. Triodes definitely offer an alternative, arguably a highly desirable alternative, but pentodes . . ?
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.