Tube DIY Asylum

Do It Yourself (DIY) paradise for tube and SET project builders.

Return to Tube DIY Asylum


Message Sort: Post Order or Asylum Reverse Threaded

Neutralization of Miller capacitance

128.125.179.206

Posted on March 11, 2017 at 15:29:24
sser2
Audiophile

Posts: 2229
Location: Pittsburgh USA
Joined: July 30, 2003
Miller capacitance can be regarded as frequency-dependent negative feedback from plate to grid that reduces gain at high frequencies. In one of the old publications I have read about a technique allowing neutralization of this effect in differential amplifiers by connecting capacitors equal in value to Cpg between grids and opposite plates. These extra capacitors add positive feedback numerically equal to the negative feedback of Miller capacitance, thus neutralizing HF rolloff caused by the latter.

If it works as intended, isn't it a solution to driver problem for PP triode output stage? Like 2A3 PP driven by 6SL7 without loss of bandwidth. Or it is too good to be true?

 

Hide full thread outline!
    ...
RE: Neutralization of Miller capacitance, posted on March 11, 2017 at 16:00:06
Tre'
Industry Professional

Posts: 12230
Location: So. Cal.
Joined: February 9, 2002
"Or it is too good to be true? "

I have never tried it myself but those that have say it turns into a nightmare. I don't remember the details but you might find the threads with a search.



Tre'
Have Fun and Enjoy the Music
"Still Working the Problem"

 

RE: Neutralization of Miller capacitance, posted on March 11, 2017 at 16:06:26
megasat16
Audiophile

Posts: 207
Location: SoCal
Joined: April 15, 2015
In other words, you can add more gain to reverse where the roll off begins due to the Cmiller. I found the IT (some) have just the magic to do that. Some have increasing Vout as the F goes up (above 10K) and this is what you need to counter the Cmiller. It's obviously working out quite well in the 845 amp I am trying out now.

The header almost made me think you are talking about immigration for Mr. Miller.
.
.
.Thou shall not stand where I type for I carry a bottle of Certified Audiophile Air and a Pure Silver Whip.

 

I should have done better, posted on March 11, 2017 at 22:03:06
sser2
Audiophile

Posts: 2229
Location: Pittsburgh USA
Joined: July 30, 2003
searching the forum. I guess what you meant was the post by Lynn Olson (linked below).

Lynn's objection was that neutralization would only work under condition of perfect balance. Any imbalance (e.g. transition into Class B) will cause positive feedback exceeding negative feedback and causing bursts of oscillation.

This is a real problem indeed, but there are at least two ways to mitigate it. First, a differential stage can be forced into complete symmetry (pure Class A) by a CCS or a choke. Second, neutralizing capacitors can be made smaller than Cpg, which will cause partial rather than complete neutralization, but safeguard against excessive positive feedback.

Yet another source of problems could be imperfections in neutralizing capacitors introducing distortion at the critical signal juncture. However, since capacitances in question are relatively low, very high quality capacitors can be used. Vacuum caps come to mind.

 

RE: Neutralization of Miller capacitance, posted on March 12, 2017 at 10:29:28
Lew
Audiophile

Posts: 8931
Location: Bethesda, Maryland
Joined: December 11, 2000
I don't know about the method you mention, but if Miller Capacitance is a major concern, I would go to a pentode or a pair of triodes in cascode mode, the latter if high output impedance is tolerable.

 

RE: Neutralization of Miller capacitance, posted on March 12, 2017 at 13:40:42
sser2
Audiophile

Posts: 2229
Location: Pittsburgh USA
Joined: July 30, 2003
Yes, most definitely. But this is about driving a triode power stage using a triode driver. The common knowledge is that you need low Rp driver running at high idle current to overcome high frequency roll-off due to Miller capacitance. Neutralization should, at least theoretically, make it possible to use high mu high Rp drivers. Power triodes driven as easily as power pentodes - this should allow a lot of flexibility in circuit design. For example, one could think of a very simple two-stage 2A3 amplifier with overall gain in excess of 100.

 

RE: Neutralization of Miller capacitance, posted on March 12, 2017 at 16:53:42
Tre'
Industry Professional

Posts: 12230
Location: So. Cal.
Joined: February 9, 2002

One could think of a [differential] two-stage 2a3 amplifier with overall gain in excess of 100.

Neutralization caps only work cross coupled across a differential stage.

BTW the 7b4 has a mu of 100 but in circuit the stage has a gain of about 55.

In a differential amp, CCS plate loads on the driver tubes would get you a gain of 100 (7b4, 12ax7, etc.) if direct coupled or close to 100 if cap coupled with the largest grid resistors allowed on the output tubes.


Tre'
Have Fun and Enjoy the Music
"Still Working the Problem"

 

Grid resistors, posted on March 12, 2017 at 18:04:54
sser2
Audiophile

Posts: 2229
Location: Pittsburgh USA
Joined: July 30, 2003
Indeed, grid resistors of the output stage may limit driver's Rp. Grid chokes might help though.

As to RF oscillation due to positive feedback, a good 2A3 design, with or without neutralization, should incorporate grid stoppers, or RF plate chokes, or a combination of both. These measures should prevent oscillation.

PFB is not necessarily evil in an audio amplifier, as has been discussed on this forum.

Here is a link to a practical high quality 2A3 amplifier utilizing neutralization in both driver and output stages:

 

RE: Grid resistors, posted on March 12, 2017 at 19:06:59
Tre'
Industry Professional

Posts: 12230
Location: So. Cal.
Joined: February 9, 2002
I like the choke feeding the IT to help force differential behavior in the driver stage.

The output tube cathode resistor with a fuse bypass in case of neg bias supply failure is a neat trick.

Thanks for the link.


Tre'

Have Fun and Enjoy the Music
"Still Working the Problem"

 

One more interesting thing, posted on March 12, 2017 at 21:40:03
sser2
Audiophile

Posts: 2229
Location: Pittsburgh USA
Joined: July 30, 2003
is the choice of driver tube. The author claims that 6C4 was the best available driver tube at the time (1950). 6C4 was primarily designed as oscillator. It is similar to 6J5 or 6SN7, but with one important distinction: its plate-to-grid capacitance is 60% less. According to the article, this low Cpg, together with Miller neutralization, is important to keep ringing in phase inverting transformer above the audio range. 6C4 is used at 280 V, close to its rated Ua, and biased to its maximum plate dissipation of 3.5 W.

 

Why not just use a 6SN7?, posted on March 13, 2017 at 12:11:36
Ralph
Manufacturer

Posts: 3307
Location: Minnesota
Joined: April 24, 2002
Contributor
  Since:
April 1, 2002
No worries getting bandwidth with one of those. You can get bandwidth higher than the output transformer will go...

Don't forget to put a proper CCS in the cathode circuit of whatever tube you ultimately use. Otherwise differential performance will suffer. A proper CCS will have two stages; a single stage will leave performance on the table.

 

6SN7, posted on March 13, 2017 at 14:15:47
sser2
Audiophile

Posts: 2229
Location: Pittsburgh USA
Joined: July 30, 2003
is indeed a good driver for 2A3, but under many scenarios an additional voltage amplification stage may be necessary in front of 6SN7. If 6SL7 can be used instead, this additional stage can be dispensed with. Less is more, especially if you consider pain-in-the-butt stage coupling problems.

 

RE: I should have done better, posted on March 15, 2017 at 14:08:41
Lee of Omaha
Dealer

Posts: 1325
Location: Omaha NE
Joined: September 8, 2006
I like the way you're thinking.

I suspect (zero evidence) that vacuum capacitors (1) would sound better than anything else (no dissipation or "memory") and (2) would be microphonic (maybe treat them with clay or loctite on the outside?).

You also couldn't use them in much because of their size/unit energy (capacitance*voltage).

 

Vacuum capacitors, posted on March 15, 2017 at 17:13:53
sser2
Audiophile

Posts: 2229
Location: Pittsburgh USA
Joined: July 30, 2003
I have no doubt that vacuum capacitors are closest to ideal. One of the reason why tubes sound better than transistors is that their interelectrode capacitances are of highest quality, whereas those in semiconductors are of the worst.

What I meant weren't purposeful vacuum capacitors, like those that HAMs use, which are indeed too bulky to be useful in this role. The simplest neutralizing capacitor would be a dud tube of the same type, however, with tubes like 2A3 this is also clumsy. But there are smaller tubes with relatively high inter-electrode capacitances that won't take too much space, like 7-pin 6 volt full wave rectifiers or sub-miniature pentodes.

Microphony problem in tubes used as vacuum capacitors is the same as it is in tubes used as amplifiers. A 6X4 or a sub-miniature pentode would be virtually free of microphony if used for neutralization in a 2A3 power stage.

 

Re. positive feedback and RF oscillation, posted on March 15, 2017 at 17:29:55
sser2
Audiophile

Posts: 2229
Location: Pittsburgh USA
Joined: July 30, 2003
Whether or not Miller capacitance neutralization would cause instability is an open question. With high transconductance tubes like 2A3, RF oscillation is inherent to the circuit. Accordingly, neutralizing capacitors may actually block oscillation rather than promote it. That is the opinion of Mr. Marshall, the author of the Audio Engineering article.

 

6SL7, posted on March 20, 2017 at 00:20:05
drlowmu
Manufacturer

Posts: 9706
Location: East of Kansas City
Joined: January 10, 2005
Even better to drive the 2A3 in a direct couple with a mu of 100 Tube. A 6SL7, (a) besides not having the higher mu, has (b) an inherent degrade in the solder connections of all the leads, going into the Octal socket's eight pins.

Nice to avoid (a) and (b) by using other NON-octal based tubes.

Listening-wise, the mu of 70 6SL7 won't be quite as much fun to hear, as it will fail to have the subjective positive qualities of a mu of 100 driver, particularly at low to medium listening volumes.

None of the available driver tube choices are perfect, but the 6SL7 is less perfect, more compromised from the start, than other tubes one can choose.

Jeff Medwin

 

RE: 6SL7, posted on March 20, 2017 at 01:36:42
GeorgePope
Audiophile

Posts: 18
Location: Havant , UK
Joined: March 15, 2017
Perhaps I am stating the obvious but :

What about the inherent degrade in the solder connections of all the leads, going into the 2A3 socket's four pins ?

George

 

RE: 6SL7, posted on March 20, 2017 at 08:53:34
drlowmu
Manufacturer

Posts: 9706
Location: East of Kansas City
Joined: January 10, 2005
Hi George,

The simple answer is : there are NO pin alternatives to the 2A3 pin arrangement.

The two-stage DC circuit's primary gain is from the driver, where driver choice and implementation is most urgent when seeking perfection.

My entire response addresses, and is entitled " 6SL7" and all pin comments were referring to an octal socket's pins. That would be the 6SL7, where alternative better choices exist.

As examples of three "non-octals", with linear graphs and mu of 100 : 1/2 12AX7 as a 9 pin mini, 6AV6 with symmetrical plate as a 7 pin mini and diodes tied to plate, and a 7B4 seated in a loctal socket.

There are compromises in any and ALL choices of high mu driver tubes for a direct coupled JJ 2A3-40 amp.

For example, the latter two tubes I mentioned are more microphonic than the first tube mentioned, but when employed just one time on the input, its OK in real-world use. The latter two tubes have symmetrical plate-filament-grid structures. All three tubes given as examples, are mu of 100, thus are preferable, normal listening-wise, to a 6SL7 octal-based tube IMHO.

Jeff Medwin

 

RE: 6SL7, posted on March 20, 2017 at 09:30:46
GeorgePope
Audiophile

Posts: 18
Location: Havant , UK
Joined: March 15, 2017
Ok , whatever you say , although a 2A3-40 is a 300B with folded filament and the driver tubes you suggest do not make sense for use with this tube . It will work but I suspect significant HF loss and major issues when driving the 2A3-40 grid close to 0V . A 12AX7 (and I'm sure this applies to 7B4) is not really a good driver unless it is driving a pentode . In fact it has higher Rp and is much less linear than 6SL7

I think you're barking up the wrong tree regarding this mu of 100 thing and also your idea about 6AV6 . If you tie the diode plates to the triode plate , the cathode will start pulling quite a substantial current . The triode section in terms of signal will be 'shorted out' .

Usually redundant diode sections are tied to the cathode .

George

 

Connecting Diodes on a 6AV6.........?????, posted on March 20, 2017 at 16:53:52
drlowmu
Manufacturer

Posts: 9706
Location: East of Kansas City
Joined: January 10, 2005
A normal E.E. will tie the diodes to ground. LOL, one good EE I trust has already told me to do so. Another tells me, last night, the two diode sections, tied to ground, will lose high end, more-so than when tied to the plate. You suggest to tie diodes to the top of the cathode. I don't know who to listen to for sure.

Separate point: you fail to look at the driver requirement in any realistic way George IMHO. I hope you realize, I am talking about a two stage amp that is directly coupled, Driver's plate to Finals control grid. This makes for an ease of driving the Finals, as a direct couple. If it were a cap couple, ( the worst way to couple two stages ) you would be correct, your analysis, view. If it were to be a transformer couple, ( far better than a cap couple ), you may still be a correct. But I am talking of NO devices at all to " drive through " between the two stages , except a few inches of high quality wire, silver, silver coated, etc, etc.

Direct coupling greatly eases the burden on the initial driver stage. A huge percentage of DIY amps in the Orient are Loftin White variants, over several decades now, mainly....because it sounds good. Isamu Asano's amps, Nobu Shishido's, all used mu of 100, low current triodes.

An EE type will typically look at tube choice from their textbook basis only - in term of maximum swing as it approaches 0 VDC of bias. That is hardly how the circuit operates IMHO!!!

I totally agree, you are correct about when "approaching zero" , but you are considering the wrong things !! Its like a MD operating on Cancer, when all the patient really needed is to eat fresh fruits and lots more organic vegetables.

In the real world, 99.999 percent of the time, the mu of 100 tube is centered at about 1.5 VDC of self bais at idle, running maybe -0.5 to -2.5 VDC of bias, playing music at normal ( even loud ) levels in the home. I design and optimize for that parameter, normal use, on a 98 dB ALTEC VOTT or ALTEC 604.

I doubt you have listened to such a circuit, particularly an optimized one, as I describe, much, if any at all.

From your comments, and understanding,you certainly wouldn't ever design it. It would require thinking outside of the box. Most trained E.E.s typically can not mentally do that, due to their schooling, training and thought processing, .... poor fellas and gals !!

A mu of 100 driver is clearly more fun, listening-wise, executed as good as possible, then is a mu of 70 driver!! Herb Reichert "got it", when he used the term " presence " of the performers " being in the room ", at normal listening levels - during RMAF 2016 last October. I happened to be there, listening with him, when he visited. He " got it" upon listening. A mu of 70, or 50 driver, can't do that.

You likely have little or no experience with this circuit, well executed. I can understand that. I have substantial listening experience of this George.

When looking at a 12AX7, 6AV6, or a 7B4, as a DCed 2A3 driver - also include your 6SL7, I would suggest all four are very similarly and nicely linear, in the ranges they will operate in for this circuit, 99 percent of the time. I am excluding " zero bias " conditions, because it is simply backwards to what occurs in the real world use George.

So...get with it, think out of the box, not textbook and formula rigid . The truth will set you free. Results is what counts, not formulas and theory that does not totally explain all that is happening !!!

As for the question of what to do with the 6AV6 diodes, tie them to plate or ground, or top of the cathode as you suggest, I am entirely "open" to see what others, who think they have an answer, say about this.

I have not yet used the tube. I do plan to, driving a Type 46, next month for two friends' DC amps, and I would be open to LEARNING from experienced technicians, on what to do with the two diodes, how and why.

So far, two people I respect, have given me opposite answers. You suggest a third approach, the cathode. What do I know ??? I was a business major.

Let us see what smarter than I, more experienced people, can suggest, and WHY they suggest it. Fun !

Thanks for posting,

Jeff Medwin

 

Four legs good, eight legs bad???, posted on March 20, 2017 at 17:54:16
cpotl
Audiophile

Posts: 734
Location: Texas
Joined: December 6, 2009
"Perhaps I am stating the obvious but :

What about the inherent degrade in the solder connections of all the leads, going into the 2A3 socket's four pins ?"

Indeed, a very good question. I didn't understand his response, which was "The simple answer is : there are NO pin alternatives to the 2A3 pin arrangement." I don't think it adequately addressed your question, though.

Your point, I think, was that in each case there are wires coming through the tube envelope and then presumably soldered onto the pins, and so why is one case supposedly worse than the other?

Chris

 

RE: Four legs good, eight legs bad???, posted on March 20, 2017 at 18:00:54
used-hifi
Audiophile

Posts: 529
Location: Surprise AZ
Joined: March 18, 2003
For the finals I solder direct to the tube pins NOT to the crummy socket pins, this is an area that no body talks about that can lead to dynamic signal losses IMO




Lawrence

 

RE: Connecting Diodes on a 6AV6.........?????, posted on March 21, 2017 at 04:25:44
GeorgePope
Audiophile

Posts: 18
Location: Havant , UK
Joined: March 15, 2017
You do not want the diodes to be conducting or have any electrodes floating . Connecting the diodes to plate will not work . Connecting to cathode or as you suggest 0V in fine

Apart from that , I fail to see any relevance within your post .

George

 

Wow!, posted on March 21, 2017 at 12:01:27
Lee of Omaha
Dealer

Posts: 1325
Location: Omaha NE
Joined: September 8, 2006
That is an interesting idea! A vacuum tube, unpowered, as a vacuum capacitor. You could pack it in mortite or something, as it wouldn't get hot.

 

RE: Four legs good, eight legs bad???, posted on March 22, 2017 at 04:33:05
PakProtector
Audiophile

Posts: 11347
Joined: May 14, 2002
why stop there? turn the socket pins into squares and use the wire wrap connection? if solder is bad, this will be a quantum leap forward in performance.
cheers,
Douglas

Friend, I would not hurt thee for the world...but thou art standing where I am about to shoot.

 

Page processed in 0.041 seconds.