So, for the 6N11, 6922, 6DJ8, and ECC88 it drops the gain (it already has plenty anyway?) and decreases distortion. Is that right? What is the gain after removing the cap?
AndrewT, you're right about phase margin. Although in theory 1 degree should be enough to ensure stability, I'd rather have at least 60.
daniel, you do not need input buffer, because the tube's input impedance is quite big. If you want to see how the calculation is made, take a look here:
Designing Common-Cathode Amplifiers
daniel, you do not need input buffer, because the tube's input impedance is quite big. If you want to see how the calculation is made, take a look here:
Designing Common-Cathode Amplifiers
It will work, the gain will be equal with -R14/R3. However, the input impedance will be set by the value of R3, being 10k. I would remove the R11-C8 and R13-C10 and add a resitor between noninverting input and ground. The value has to be identical with R3 in order to minimize the input voltage offset. Also, 220pF between inputs will help removing pops and clicks.
AndrewT, you clearly are a better engineer than me and I have a question for you. I've read a ton of documentation and I couldn't find an explanation for this. Why the inverting configuration is stable also at lower than unity gain?
AndrewT, you clearly are a better engineer than me and I have a question for you. I've read a ton of documentation and I couldn't find an explanation for this. Why the inverting configuration is stable also at lower than unity gain?
I don't know if that is true.
I don't use inverting.
The only place I have used inverting is CFP filtering and second stage of an active RIAA preamp.
btw,
the experts here say that 60degree phase margin is not enough. I originally thought >=45degree prevented oscillation as long as the slope was <=9dB/octave at cross over.
But, there seems to be consensus that 80 to 85degrees phase margin gives best reproducibility of the various waveforms in the passband.
the experts here say that 60degree phase margin is not enough. I originally thought >=45degree prevented oscillation as long as the slope was <=9dB/octave at cross over.
But, there seems to be consensus that 80 to 85degrees phase margin gives best reproducibility of the various waveforms in the passband.
Ok,
It sounds like we are getting somewhere here.
The phase margin stuff is all "greek" to me so someone will have to decide on the proper margin?
I assume what we are talking about somehow involves prevention of oscillation due to the feedback signal versus the input signal "overlapping" at some point?
as far as the junction between the tube and chip,
Because the plate of the tube is at +100VDC Potential a bi-polar cap is needed to allow only the AC signal to pass. Otherwise By By chip!! The size of this cap is determined by the input IMPEDANCE of the chip. simple CR High Pass filter.
The GAIN of the tube is determined by the factors in the link RATZA posted. It is an EXCELLENT source of material. I started a few years ago in Tubes and used Aiken's stuff along with RCA RC-30 as the "BIBLE".
I now use calculators for the math. I can give you links later but it is imperative that you understand the calculations before using the "short way".
The capacitor that most affects the GAIN and SOUND of a common cathode amp is the Cathode Bypass cap Ck (C was already used for Capacitor so Cathode is Kathode)
Removing it allows negative feedback at the level of Rk, inserting it decreases the feedback 6db/octave above the F3 formed by the cap/resistor value.
So you see it is a compromise between GAIN and Frequency response. In addition the LARGE values needed to get a decent response (unless it is a mid-treble amp) require an Electrolytic cap, ie; not Audiophile quality.
Most TUBE designers looking for QUALITY and LINEAR response will AVOID Electrolytics at all cost and if more Gain is needed will insert another stage. For a Gheetar amp they are fine.
Daniel,
To clear up some other things.
1: We want to drive a higher impedance with a lower impedance not equal or the other way arround.
2: DO NOT USE a Headphone output. Line Level will ALWAYS be cleaner as the Headphone output is looking for a 32R or so load not 10 - 100K.
3: I have checked about 7 different sound cars ALL had output impedance of 10K or greater. Lowest Vout was about 200mV RMS @ 1Khz.
4: Gain is MULTIPLIED not ADDED. So a gain of 10 in the chip amp and 16 in the tube gives total gain of 160 X input RMS or in the case of the lowest card I tested 32V.
ie: .200 x 16 = 3.2VRMS 3.2x10 = 32VRMS
The above is EXACTLY WHY I need to know what the optimum design for the chip amp is.
5: I appreciate the defference being given to the tube amps "Dynamics" but in all honesty tubes are not much different than Chips overall. While a tube amp is a WHOLE LOT GENTLER when it comes to clipping and distortion, a WATT is a WATT.
For purposes of this project lets assume I am NOT building a woofer amp. I want a chip amp that can drive full range. We can "adapt" the design to match whatever range we want by playing with the input or place a filter between the input and a cathode follower buffer stage if need be. Lets make as "optimum" a chip amp we can. Maybe it is two different circuits for the chip amp. A low end optimized one and a fullrange/mid/treble optimized version. You guys are the chip gurus, let me know.
It sounds like we are getting somewhere here.
The phase margin stuff is all "greek" to me so someone will have to decide on the proper margin?
I assume what we are talking about somehow involves prevention of oscillation due to the feedback signal versus the input signal "overlapping" at some point?
as far as the junction between the tube and chip,
Because the plate of the tube is at +100VDC Potential a bi-polar cap is needed to allow only the AC signal to pass. Otherwise By By chip!! The size of this cap is determined by the input IMPEDANCE of the chip. simple CR High Pass filter.
The GAIN of the tube is determined by the factors in the link RATZA posted. It is an EXCELLENT source of material. I started a few years ago in Tubes and used Aiken's stuff along with RCA RC-30 as the "BIBLE".
I now use calculators for the math. I can give you links later but it is imperative that you understand the calculations before using the "short way".
The capacitor that most affects the GAIN and SOUND of a common cathode amp is the Cathode Bypass cap Ck (C was already used for Capacitor so Cathode is Kathode)
Removing it allows negative feedback at the level of Rk, inserting it decreases the feedback 6db/octave above the F3 formed by the cap/resistor value.
So you see it is a compromise between GAIN and Frequency response. In addition the LARGE values needed to get a decent response (unless it is a mid-treble amp) require an Electrolytic cap, ie; not Audiophile quality.
Most TUBE designers looking for QUALITY and LINEAR response will AVOID Electrolytics at all cost and if more Gain is needed will insert another stage. For a Gheetar amp they are fine.
Daniel,
To clear up some other things.
1: We want to drive a higher impedance with a lower impedance not equal or the other way arround.
2: DO NOT USE a Headphone output. Line Level will ALWAYS be cleaner as the Headphone output is looking for a 32R or so load not 10 - 100K.
3: I have checked about 7 different sound cars ALL had output impedance of 10K or greater. Lowest Vout was about 200mV RMS @ 1Khz.
4: Gain is MULTIPLIED not ADDED. So a gain of 10 in the chip amp and 16 in the tube gives total gain of 160 X input RMS or in the case of the lowest card I tested 32V.
ie: .200 x 16 = 3.2VRMS 3.2x10 = 32VRMS
The above is EXACTLY WHY I need to know what the optimum design for the chip amp is.
5: I appreciate the defference being given to the tube amps "Dynamics" but in all honesty tubes are not much different than Chips overall. While a tube amp is a WHOLE LOT GENTLER when it comes to clipping and distortion, a WATT is a WATT.
For purposes of this project lets assume I am NOT building a woofer amp. I want a chip amp that can drive full range. We can "adapt" the design to match whatever range we want by playing with the input or place a filter between the input and a cathode follower buffer stage if need be. Lets make as "optimum" a chip amp we can. Maybe it is two different circuits for the chip amp. A low end optimized one and a fullrange/mid/treble optimized version. You guys are the chip gurus, let me know.
ted,
That is a very short sighted solution.
Does building someone elses design teach me anything besides soldering skills?
In fact when it comes to Chip Amps I am hardly an amatEUR, I am a beginner!
I am not stumbling from design to design, I am investigating different possibilities.
Perhaps a CONSTRUCTIVE post would have been more appropriate.
remember EXPERTS built the TITANIC!
That is a very short sighted solution.
Does building someone elses design teach me anything besides soldering skills?
In fact when it comes to Chip Amps I am hardly an amatEUR, I am a beginner!
I am not stumbling from design to design, I am investigating different possibilities.
Perhaps a CONSTRUCTIVE post would have been more appropriate.
remember EXPERTS built the TITANIC!
Hi,
Dan asked for my help here but I'm extremely limited in my knowledge of tubes. I can say that IMO the tube frontend in not needed and that it complicates the design for no real benefit. Clipping will occur in the chipamp, as usual so there is no gain there.
Using the chip at it's lowest gain invites instability. Running it in inverted mode (AFAIK) will not enable a lower gain.
Maybe marrying a tube frontend to a discrete output is the best approach? This has been done before (my K10A, designed by GK is a good example, but in class A output).
Dan asked for my help here but I'm extremely limited in my knowledge of tubes. I can say that IMO the tube frontend in not needed and that it complicates the design for no real benefit. Clipping will occur in the chipamp, as usual so there is no gain there.
Using the chip at it's lowest gain invites instability. Running it in inverted mode (AFAIK) will not enable a lower gain.
Maybe marrying a tube frontend to a discrete output is the best approach? This has been done before (my K10A, designed by GK is a good example, but in class A output).
Yes, The tube front end is NOT Needed, It IS a source of distortion, and it's gain is just an "Add-on", BUT there are some benefits to the tube.
It allows us to have a higher input impedance than the chip.
It provides a little bit of "warmth" to the sound IMHO
It GLOWS
For my purposes it "Tubifies" the amp so it is more palatable to me in my ALL TUBE system.
This is more of a "just for kicks" type project, We can live with running a gain of about 20 in the chip amp. Designed properly we can run it so that the Tube "CLIPS" before the chip, thus softening the clipping and protecting the circuitry and speakers.
Again, this is just my humble opinion and remember "Opinions are like, Sphincters we all have them and they all stink"
OK, so this means that it doesn't matter what configuration we use, the chart remains the same. This means that the chip will oscillate also in inverting configuration. I realize now that the only difference between gains of inverting and noninverting configuration is that the inverting one can have gain lower than unity. Actually, it makes sense, the opamp works the same, the only difference being the 180 phase shift.
That LM3886 has a terrible time with its input for some reason. Most LM3886 users report significantly improved performance with any sort of preamp or buffer. Perhaps they're using an MP3 player and gain of 50 on LM3886 doesn't go so well? So, this project is the inverse--drop the gain on LM3886. Its almost like "use something else" except that we didn't.
Sort of. But, the point is lowest gain while still maintaining good results. I think that inverted mode does usually facilitate slightly lower gain in non-unity-stable opamps, but results can vary considerably between different examples.
Absolutely. There's no doubt at all. But, the point here is non-linear amplification, simply because the majority of recordings are compressored, and can sound fake. Since we really are just driving a woofer, some inspired dynamics along with a "clear" bass is the project. It may be able to fit an overlarge bass wave into room that doesn't support that playback.
Considering source and room, the project has a chance to decrease distortion. I think that the odds are poor, the expense is small, the gamble is worth the potential results, and the financial risk will decrease if given some inspired engineering. Inspired engineering? Well, now that's the reason why I emailed you.
And, thank you!
Not cool! Please don't run off all of the prodigiously talented engineers.
P.S. Because of the aggressive spike system aboard the larger National Semiconductor power opamps, like LM3886, they're not easy to deal with. Any help is good and probably necessary. System symmetry with a tube amp system is on topic; however sphincters are severely off topic.
Last edited:
How to apply it?
I can heat up the soldering irons pretty quickly and give it a try (if I knew what it was that I was trying).
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.