|
Audio Asylum Thread Printer Get a view of an entire thread on one page |
For Sale Ads |
63.246.183.22
I've been thinking about the many different ways to build a control amp - or line stage - for quite some time. Most of the designs I've seen or simulated in SPICE exhibit way too much gain, and nearly all invert the signal. The few that don't invert usually accomplish the feat through an overly judicious application of NFB to two cascaded stages. I'm not NFB-phobic, but I do avoid it when possible. Using it strictly for the purpose of reducing gain from, say, 20dB to 6dB doesn't sit well.
The bottom line is that typical control amps all seem to have a downside, so I'm considering a new approach for my next build. I've also decided voltage gain isn't necessary. All my amps can produce full output with well under 1V RMS input, and all my sources are capable of considerably more than that. This opens the door for a high quality cathode follower. A choke loaded follower would be my preference, and I just happen to have a small stash of one type of choke that might work well in this application. It's roughly 70H - more than sufficient for use in a low Z circuit like this - and it's been proven in testing to have a very smooth impedance curve. The only potential issue is the DCR, roughly 2.2K.
To see why this is a problem, consider the curves below for a triode-connected 6W6. I want the stage to idle at 25mA. Using a 300V supply, 25mA will flow with approximately 36V grid bias. This leaves +264V anode-to-cathode. The operating point is shown with a red dot on the curves. This combination of values, 36V @ 25mA, requires a 1.44K cathode resistor. Obviously, using a 2.2K choke here will cause considerably less current to flow than is desired.
Two techniques for resolving this come to mind. One is the addition of a shunt resistor across the choke, reducing the DC resistance of the cathode circuit. This would load the cathode's AC signal significantly, rendering the choke nearly worthless. I believe a better approach would be to bias the return lead (bottom) of the choke with a negative supply. The supply only needs to provide the idle current of the stage, and cleanliness isn't overly critical. In fact, AC at the output of the supply will be isolated from the cathode by the choke, and with the cathode operating at such a low impedance, attenuation will be significant. In the example, approximately -13V will do the job. A representative circuit is shown below.
This leads to my question, namely whether anyone here has seen a commercial example of this technique. Related to that, of course, is whether anyone here has ever tried it.
--------------------------
Buy Chinese. Bury freedom.
Follow Ups:
In addition to the octal types listed below, I tested a 12B4a this evening. As expected, it's much "hotter" than the others. Where the 6K6 plate dissipation was about 3.5W with a +300V supply, the 12B4a dissipates 4.5W. That's approaching the limits of the tube, and the additional dissipation likely does nothing for sonic quality. Adding a 1K resistor in series with the choke reduces dissipation to 3.2W, which is still a little more than I'd like to see. With or without the additional resistance, this tube really needs less B+, probably something closer to +250V. I'll need to consider this when I'm deciding whether the low Rp of this tube might be advantageous.
--------------------------
Buy Chinese. Bury freedom.
Try a 12A4...:)
Douglas
Friend, I would not hurt thee for the world...but thou art standing where I am about to shoot.
I've come up with a list of tubes that will probably work in this application without a negative bias supply under the chokes. I won't bore the group with all of that, but I did try the first tube tonight, a 6K6GT in triode mode. Here are the preliminary results:With exactly +300 applied between the anode and ground, the choke drops 30.1V. It's actual resistance is close to 2.3K, so that equates to 13mA through the choke and tube. At this current, the 6K6 is dissipating 3.5W. This is somewhat less than half the tube's maximum rated dissipation.
With 20V pk-pk applied to the grid (the maximum output of my function generator), the cathode produces 17V pk-pk into a 12K load. That's a modest 1.4dB loss. Frequency response is very good. 10Hz is down just a little (sorry, just viewing the scope for now, no numbers yet). On the high end, it's flat to 500kHz, then there's a gradual roll off that appears to be roughly 3dB at 1MHz. This is confirmed by the virtually perfect response to a 100kHz square wave. I have roughly a dozen more tube types to try - 6L6, 12B4, 6AS7, 6V6, etc. I think this particular combination will be hard to beat though.
The schematic of the test circuit is below:
--------------------------
Buy Chinese. Bury freedom.
Edits: 05/19/17
I'm almost surprised the 6K6 doesn't run cold but I haven't looked at operating points for a 6K6.
Anyway,is that a 33uf on the cathode or a 3.3uf?
"For every complex problem there is an answer that is clear, simple, and wrong" H. L. Mencken
Michael, you do realize that I said "under the choke," right? Such a bias supply would increase the tube's conduction.
The output capacitor is 33uF. It was selected for the purpose of testing levels and frequency response.
--------------------------
Buy Chinese. Bury freedom.
I did realize that but excuse me for being a thick in head here.What do mean by under the choke? The resistance of the of cathode is lower then the choke?
"For every complex problem there is an answer that is clear, simple, and wrong" H. L. Mencken
Edits: 05/23/17
I meant that the cold end of the choke should be disconnected from ground, and a negative bias applied there. This is a solution for a tube type that doesn't otherwise conduct sufficiently due to the relatively high DCR of the choke.
--------------------------
Buy Chinese. Bury freedom.
I saw that principle applied in some circuits.Anyway,all I could think of was the above ground DC source we apply to filaments to kill hum in high gain circuits.
"For every complex problem there is an answer that is clear, simple, and wrong" H. L. Mencken
Hook it up and tell us what it sounds like!
That will take more time. I'm confident this will prove to be exceptionally transparent, with very little sound of its own.
--------------------------
Buy Chinese. Bury freedom.
Well, if you are going to share the design when you are done, I will be looking forward to seeing it.
Yes, I'll put it into the public domain, so to speak. I'm debating exactly how much circuitry to include in this project. Much as I like the simplicity of separates, the functionality of a full-featured preamp can be very convenient.Incidentally, I didn't have any 12B4a 9-pins on hand, so all the tests of the cathode follower circuit were with octals that I thought would bias correctly with the DCR of this choke. I was surprised to discover that none of the tubes required socket wiring changes. They were all plug-and-play except for appropriate changes to filament voltage. Here are the tubes I tested. Note that all the pentodes were wired in triode mode:
6K6 RCA
6V6 RCA
6N6C (Russian 6V6)
6P3S (Russian 6L6GB)
6P3S-E (Russian 5881)
6L6GC Sylvania
1626 Ken-RadThat's right - the little ST-bottle 1626 often used in "Darling" amps plugs right in. Performance differences between these tubes were minimal, mostly having to do with idle current and output impedance. The 6K6 idled at the highest current, but the 6P3S-E and 6L6GC exhibited the least drop in output voltage when heavily loaded (substituting 1.5K for the 12K load). I also tested every tube with a load consisting of the 12K resistor and a 1500pF capacitor in parallel. In every case, 10kHz square wave response was only slightly degraded by the capacitor, and there were no indications of instability or visible waveform distortion.
I have to say, the 1626 would look pretty cool in this application, and a pair of large, shiny 6P3S-Es wouldn't be bad, either. Strictly from the standpoint of functionality though, the 6K6 is my favorite. It idles at roughly half its rated dissipation, and it draws very little filament current for a power pentode in this class.
--------------------------
Buy Chinese. Bury freedom.
Edits: 05/22/17
mine is about as neutral and 'not there' as I have seen...and by your standards, it is *LOADED* with silicon.
With a triode cathode follower, running it cascode, with a MOSFET on top serving as the screen grid is quite useful. It leaves the triode with a nearly constant voltage across it( if you make the circuit properly), and the very high effective plate resistance leaves it quite immune to noise( like PS ripple ). Changing the voltage the gate is at, is comparable to changing the g2 voltage on a pentode, and is quite useful for setting idle current in a system where the cathode resistance is not a parameter you can change.
cheers,
Douglas
Friend, I would not hurt thee for the world...but thou art standing where I am about to shoot.
Douglas, can I see a schematic?
Thanks.
Tre'
Have Fun and Enjoy the Music
"Still Working the Problem"
send me an email and I can get one for you...:)
cheers,
Douglas
Friend, I would not hurt thee for the world...but thou art standing where I am about to shoot.
"the very high effective plate resistance leaves it quite immune to noise( like PS ripple )"
Don't think I've seen that technique. Won't a high plate resistance degrade the output impedance of the follower?
--------------------------
Buy Chinese. Bury freedom.
It is the gm of the lower element that is of concern. The usual triode-on-triode cascode eats up plate voltage like a fiend...and the MOSFET nearly totally eliminates this.
cheers,
Douglas
Friend, I would not hurt thee for the world...but thou art standing where I am about to shoot.
I'm not sure we're on the same page. A cathode follower won't work if a CCS is supplying its anode voltage.
--------------------------
Buy Chinese. Bury freedom.
"A cathode follower won't work if a CCS is supplying its anode voltage."
Why not?
Tre'
Have Fun and Enjoy the Music
"Still Working the Problem"
A CCS at the anode will block the AC current path through the tube. Very little will reach the cathode.
--------------------------
Buy Chinese. Bury freedom.
So then how does a common cathode stage feed by a CCS work?BTW. A CCS (or a choke) at the cathode will block the AC current path through the tube in the same way as a CCS (or choke) at the anode will.
But isn't the AC current path through the load what we're interested in?
Tre'
Have Fun and Enjoy the Music
"Still Working the Problem"
Edits: 05/26/17 05/26/17
"So then how does a common cathode stage feed by a CCS work?"
The AC current path (loop) is through the load, into the anode, through the tube, out the cathode, and into ground (the return for the load). The CCS delivers DC current to the anode, but is not in the AC loop.
"BTW. A CCS (or a choke) at the cathode will block the AC current path through the tube in the same way as a CCS (or choke) at the anode will."
In the case of the cathode follower, the AC current path is through the B+ supply into the anode, through the tube, into the load, and through the load into ground (the return for the supply). The CCS below the cathode is not in the AC loop, it serves only to return DC current to ground.
--------------------------
Buy Chinese. Bury freedom.
Just to illustrate what I mean, the two diagrams below show the AC current paths in a cathode follower and a common cathode amplifier. Installing a CCS (or high-impedance choke) anywhere in either loop will prevent AC current from flowing through the respective load.
--------------------------
Buy Chinese. Bury freedom.
What is the current path for the upper output of this circuit?
Tre'
Have Fun and Enjoy the Music
"Still Working the Problem"
Well, the mu follower isn't quite so clear-cut. In a circuit using typical values, about 80% of the AC load current derives from the upper triode. Roughly 18% flows into the load from the lower triode through its anode resistor, R1. The remainder, one or two percent, reaches the load through Cg2 and Rg2.
This all changes if a high impedance choke or CCS is substituted for R1. That breaks the AC current path from V1 anode to the load. Now, nearly all the AC current will be delivered by V2, with a small contribution through Cg2 and Rg2.
--------------------------
Buy Chinese. Bury freedom.
How about these stages?
Tre'
Have Fun and Enjoy the Music
"Still Working the Problem"
I had to model the first stage of that circuit to be sure what was happening. It appears to be a follower, not a CCS as I first supposed. Nearly all the output signal current is supplied by the MOSFETs. Seems pointless to have a tube in the circuit at all. Why not just build a MOSFET amplifier and save the cost and trouble of a HV supply? Definitely not something I would ever do... :)
--------------------------
Buy Chinese. Bury freedom.
"It appears to be a follower, not a CCS as I first supposed. Nearly all the output signal current is supplied by the MOSFETs. Seems pointless to have a tube in the circuit at all."
I believe the tube provides gain.
Tre'
Have Fun and Enjoy the Music
"Still Working the Problem"
"I believe the tube provides gain."
It does, but so would a FET in that position. You know, it's only a small step from believing that a MOSFET used as a follower or CCS won't affect sonics to replacing all the tubes in the circuit with solid state. When a group consists of individuals who will readily argue which resistor, capacitor, diode, fuse or wire type sounds best, I think this is extremely hypocritical. Not referring to you, just the forum in general.
--------------------------
Buy Chinese. Bury freedom.
I see the sonics being affected more by a FET as a follower than as a CCS.
One more question. Is the output impedance value of the FET follower affected by the plate resistance of the tube below it?
Tre'
Have Fun and Enjoy the Music
"Still Working the Problem"
The more important question ought to be, 'what about SS sounds so different'? Their input capacitance varies depending on the voltage across them for one. But that does not seem to be of any interest; 'SS is bad', or some such and then complete dismissal...LOL
Tubes are not without significant limitations, which when crossed leave them sounding like crap. I suspect combinations that take advantage of the strengths are much more capable than either alone.
cheers,
Douglas
Friend, I would not hurt thee for the world...but thou art standing where I am about to shoot.
The Rp of the tube has almost no effect when it's driving a MOSFET follower. Nearly all the AC signal current to the load is supplied by the MOSFET. That's what sets the output impedance.
--------------------------
Buy Chinese. Bury freedom.
.
Have Fun and Enjoy the Music
"Still Working the Problem"
This exercise gave me a chance to review the current paths for these various configurations. I had forgotten that signal current in the cathode follower flows through the power supply. That means it's important to have a good quality cap in the last position, and to be sure the cap's full potential isn't degraded by semiconductors or poor quality components upstream. Maybe this is another reason some CFs have an unpleasant sound. A common cathode stage with a CCS or anode choke wouldn't be vulnerable to this.
--------------------------
Buy Chinese. Bury freedom.
Do the MosFets represent a CCS if the signal were taken off the plate of the 6c45?
Tre'
Have Fun and Enjoy the Music
"Still Working the Problem"
"Do the MosFets represent a CCS if the signal were taken off the plate of the 6c45?"
Yes, that configuration impresses an AC voltage differential across the gate and source, creating a CCS. The tube will provide nearly all the current to the load.
--------------------------
Buy Chinese. Bury freedom.
You have misunderstood. How is a MOSFET with its gate a fixed a certain voltage above the triode's cathode considered a CCS?
cheers,
Douglas
Friend, I would not hurt thee for the world...but thou art standing where I am about to shoot.
"How is a MOSFET with its gate a fixed a certain voltage above the triode's cathode considered a CCS?"
I never referred to a CCS in those terms. And before we go too far down this road, I want to reiterate something you already know. I never attach semiconductors to the signal path. The purpose of posting in this thread was to describe and demonstrate the benefits of a choke-loaded follower as a line stage. I'm not interested in discussing the addition of a transistor to this circuit.
--------------------------
Buy Chinese. Bury freedom.
What I described is a cascode, and labeled it as such.
It is utterly irrelevant as your sacred text, or golden tablets, or what ever, prohibit the use of SS devices.
cheers,
Douglas
Friend, I would not hurt thee for the world...but thou art standing where I am about to shoot.
I'm planning to use a HV bipolar regulator and large filter cap. That should provide a very low PS impedance for the tube. Would there be an advantage to the MOSFET over that? I don't need to drop voltage for any of the tube types I tested. It's a problem getting enough current the way it is, due to the relatively high DCR of the choke.
--------------------------
Buy Chinese. Bury freedom.
this is only a discussion if you are willing to answer a question. You can answer or not...
cheers,
Douglas
Friend, I would not hurt thee for the world...but thou art standing where I am about to shoot.
What type of amplifier is this?
I assembled a Vacuum Tube Audio kit line-stage with this circuit. It sounds exceptionally nice but it has way too much gain. Fortunately, my power amplifier has input potentiometers to reduce the incoming signal. This also reduces noise to inaudible levels and my system sounds better than ever before. Previously, I was using a Pass Labs X1 line-stage on its minimum gain setting of +4-dB and I still had to reduce the incoming signal to my power amplifier slightly. However, I prefer the sound of my new vacuum tube line-stage compared to the more expensive Pass Labs line-stage. It just sounds more natural and transparent to me.
Best regards,
John Elison
I am using the cathode follower stage of this preamp by itself as a buffer with selector switch and volume control. A no gain preamp. I am using the 6n1p tube. Best preamp that I have had so far. If you don't need gain from your preamp, this is highly recommended.
Hi John,
I am not as knowledgeable as the other guys here but it seems to be an Aikido circuit from John Broskie.
You may read about this circuit and many other tube related things on his informative website.Glad you are enjoying your tube preamp builds and keeping the DIY spirit alive.
Regards,
Hakan
Edits: 05/18/17
"Small audible differences don't matter to me."
"It just sounds more natural and transparent to me."
I'm not trying to be an asshole, I'm just trying to get you to think.
BTW The second stage is a cathode follower with an active load* (* the triode in the cathode circuit).
Tre'
Have Fun and Enjoy the Music
"Still Working the Problem"
What does that add to performance, is what I mean? Seems to me you don't absolutely need it. Thanks.
Good question. I don't know.
Tre'
Have Fun and Enjoy the Music
"Still Working the Problem"
Wouldn't the voltage drop thru r6 provide added negative bias for the upper triode in addition to bias provided by direct coupling from previous stage? The designer's chosen means of fine tuning the bias for that stage.
I was thinking one could do all that with R7? But you're probably right.
Tre'
Have Fun and Enjoy the Music
"Still Working the Problem"
As I interpret the designer's intentions, R7 sets the current thru the system while R6 fine tunes the operating point or voltage drop across the upper triode. Probably not a single value of R7 that allowed all design goals to be met.
Would be interesting to know whose circuit this is.
.
Have Fun and Enjoy the Music
"Still Working the Problem"
Speaking of John Broskie, I wonder if he dislikes choke-loaded followers for some reason. I searched through his blogs for an hour or so today and discovered he has never done a write-up on a choke loaded CF. There are several places he delves into theoretical cathode chokes (zero ohm DCR) in explaining follower theory, but no place he discusses a practical, brick and mortar amplifier built using this technique. The closest he comes to even acknowledging this method is in a brief, off-the-cuff discussion 10 years ago of basic follower topologies. In that article, he states the following:
"...the cathode resistor can be replaced by a choke, with relatively high DCR, which will serve as a cathode resistor in DC terms, but function as a constant-current source in AC terms."
Broskie goes on to describe the one unique advantage to this approach, namely that "a choke will allow output voltage swings in excess of the [static] grid-to-cathode voltage. " That's a critically important concept, but one that's easily overlooked because he neglects to flesh out its underlying value. Specifically, this characteristic of the choke's operation allows the cathode signal to swing negative , greatly increasing the signal voltage and current available to drive the load, and it does so without the need for a negative power supply.
I'm actually rather surprised by Broskie's failure to further explore this technique. He otherwise seems to be an avid proponent of cathode followers, and he has devoted considerable space to their pursuit.
--------------------------
Buy Chinese. Bury freedom.
I asked because I faced precisely this question back when I wanted to implement a CCS for the CF driver stage of my OTL amplifiers. I needed the resistor in the R6 position in order to drop some voltage that would have otherwise been too much plate to cathode voltage for the CF tube above it. (The CF is strung between the upper and lower poles of a positive supply voltage and a negative supply voltage, respectively, for a total of nearly 800V between pos and neg.) I did install a resistor analogous to R6, but I have always wondered whether having a resistance there compromises function of the CCS to linearize performance of the CF.
Small audible differences are not important to me. Improvements, on the other hand, are always welcome.
Best regards,
John Elison
I was hoping to use genuine triodes in this line amp. However, if the follower is set up correctly, the tube itself won't be heard. That opens the door for pentodes to be used in triode mode. Searching through the curves for a match, the 6V6, 6K6 and maybe the 6AQ5 look as though they will bias to 10-15mA with the 2.2K DCR choke connected directly to ground. That would eliminate the need for a negative bias supply under the choke.
Curves for the 6K6 in triode are below. At 15mA, the choke will drop about 33V. This is accommodated with a cathode-anode voltage of 275V. Adding back the 33V lost across the choke, the PS must provide +308V.
Other advantages - the 6V6, 6AQ5 and 6K6 exhibit much lower Rp than a 6SN7, and they don't require the high filament current of tubes like the 6W6, 6BL7, etc.
This looks to me like it might be win-win. Anyone see a downside to the use of these tube types in this mode?
--------------------------
Buy Chinese. Bury freedom.
The 6V6 family makes excellent triodes. You could do far worse than sourcing NOS RCA 6005s. IMO, the Loctal 7C5 and 14C5 are also good choices.
Eli D.
According to the curves, the only "triode" in that family capable of giving good current flow without applying bias to my chokes is the 6K6. I bought a few of those many years ago, so maybe I'll test the circuit for DC characteristics this weekend. Wish there was a direct 12V equivalent to that tube...
--------------------------
Buy Chinese. Bury freedom.
Looks like the 6Y6 (but higher filament current) and 12B4 are also good candidates for this.
--------------------------
Buy Chinese. Bury freedom.
I got some supa clean nos el32 marconi from Canada. I used them triode connected in preamp. nice sound from these top cap tubes maybe they might be ẃorth trying the curves. Very cheap and plentiful.
Of course now I scrapped them for a pair of KT77s to drive the 572Bs but I'm not sure how a single triode connected 6W6 is going to have too much gain as a linestage?
"For every complex problem there is an answer that is clear, simple, and wrong" H. L. Mencken
'SN7s would allow me to reduce the idle current to 10-15mA, and they won't require as much negative bias. The 6W6 has a much lower Rp, but the 'SN7 will still present an output Z of a few hundred ohms in CF service. That's low enough for anything I'll need to drive.A control amp of this type should exhibit approximately -0.5dB voltage gain. It will provide a large current gain by virtue of the impedance transformation. It should also (hopefully) be entirely transparent.
--------------------------
Buy Chinese. Bury freedom.
Edits: 05/17/17
Jack
Do you want this for a line stage in a preamp or s gain stage in an amp?
"For every complex problem there is an answer that is clear, simple, and wrong" H. L. Mencken
I'll probably include switchable tone controls, switched AC outlets, etc.
--------------------------
Buy Chinese. Bury freedom.
It has everything you are looking for along with tone switching and controlled AC outlets.It is an Anode follower however,it has a phenomenal four tube phono stage and you could convert the output to an cathode follower.
The factory spec alone is incredible and when you upgrade it with Jim's kits,the S/N and bandwidth improves even more.The circuit looks scary at first glance but you later realize it isn't all being used at one time.It is one of the most correct sounding preamps you will ever hear. There is also the Cit 4 which is simpler,but you would have to put in tone switching.
"For every complex problem there is an answer that is clear, simple, and wrong" H. L. Mencken
Edits: 05/19/17 05/19/17
Those old beasts are a bit on the rare side. It is also relatively easy to outperform them...even with something simple.
cheers,
Douglas
Friend, I would not hurt thee for the world...but thou art standing where I am about to shoot.
They need upgrading and some circuit revisions.I have seven on them actually.One is a carcass tho.I said it wasn't an anode follower.It is an anode follower.I can't believe I mistyped that. Anyway,if you are looking for accuracy,these are one of the best tube preamps for that.One would be cool to open up and mod in no holds barred way.
"For every complex problem there is an answer that is clear, simple, and wrong" H. L. Mencken
Edits: 05/19/17
~!
The Mind has No Firewall~ U.S. Army War College.
Yes it is.
"For every complex problem there is an answer that is clear, simple, and wrong" H. L. Mencken
One problem with all old preamps of that era is that moving parts like switches and pots are usually heavily worn. My time is better spent designing from scratch (using newer, better parts and technology) than chasing all over the planet looking for replacements. Building new also provides an opportunity to remedy all the issues that engineers of the day either didn't know about or were forced to accept due to the bean counters. I've learned some of the latter the hard way with ham gear. You'd be surprised what was allowed to come out of a transmitter or RF amplifier made in the '50s or '60s, simply because the design was vetted by a factory without access to a spectrum analyzer or 100 MHz scope.
--------------------------
Buy Chinese. Bury freedom.
Bypass those switches and if you look at the circuit,it isn't all working at once..Now the Citation 4 has tone controls and you can put tone switches in it's place.They do have a bypass switch.If you clean and polish the wafer switches,you are usually good to go.What some people do is put the modern gold contact switches in place of the wafer switches and then use mini toggles in place of the slide switches.
"For every complex problem there is an answer that is clear, simple, and wrong" H. L. Mencken
and they claimed the power transformer was four times the size of the original which is nonsense. For five grand,it got great reviews but they copied the original and put in better switches and a MC step-up trafo.
"For every complex problem there is an answer that is clear, simple, and wrong" H. L. Mencken
nada aqui
--------------------------
Buy Chinese. Bury freedom.
We need some dynamite.
"For every complex problem there is an answer that is clear, simple, and wrong" H. L. Mencken
nada aqui
--------------------------
Buy Chinese. Bury freedom.
A 5687 or a 6H6Pi is such an improvement over the 6SN7 for what you have in mind for it.
cheers,
Douglas
Friend, I would not hurt thee for the world...but thou art standing where I am about to shoot.
As it turns out, none of these tubes will be convenient to use due to biasing requirements. A 6K6 in triode mode will likely require no negative bias supply at the bottom of the choke, and output Z will be only a few hundred ohms. That's the direction I'm headed for the moment. :)
--------------------------
Buy Chinese. Bury freedom.
With a standing current of 25mA through the choke that has a DCR of 2.2K there's a 55V drop. The potential at the tube's cathode will then be 42V (55V-13V). The grid bias for the chosen OP needs to be at 36V; it would seem that neg supply should be at -19V. where am I wrong?Also - and I haven't given this much thought yet so I could be wrong, but shouldn't the input shunt resistor be referenced to the negative supply instead of ground in order for the tube to operate as a CF? If that's the case, then of course an input cap would be needed. What say you?
Edits: 05/16/17 05/16/17
"it would seem that neg supply should be at -19V. "
Yes, it should be -19V in the example. My mistake, thanks for pointing this out.
"shouldn't the input shunt resistor be referenced to the negative supply instead of ground in order for the tube to operate as a CF?"
The negative supply is at AC ground, so the resistor can be connected to either point. The difference, as you noted, is that an input coupling cap will be required if the resistor is connected to the negative supply.
--------------------------
Buy Chinese. Bury freedom.
"The negative supply is at AC ground, so the resistor can be connected to either point."
Obviously I haven't seen the PSU schematic and I wouldn't want to make any assumptions, but I'm a bit confused by this comment. Could you please elaborate a bit on the AC ground perhaps with a schem? I don't want to hijack this thread by taking it off on a tangent, so if you want me to start a new thread I'll be glad to do it.
PS. About three years ago I had devised a similar circuit for driving a 300B. It used the 5687 as a CF, DC coupled to the grid of the power tube with an appropriate (by taking the 300B's biasing into consideration) negative supply at around -100V, very similar to what you show in your schematic above minus the coupling cap. All this to say, that my interest in this is more than academic.
The drive circuit in my 211 SETs is similar to what you're describing. however, there are many differences relative to the line stage I'm proposing. For one thing, the grid resistor of the 211 driver tube isn't grounded. It's connected to a variable source of negative voltage, and that's how I set bias current through the 211. Also, unlike the line stage, the grid must be driven by the pre-driver through a coupling cap.
Simplified circuit:
Regarding the negative bias supply for the line stage, its output is a virtual short to ground for AC. A representative circuit is shown below, but many topologies can serve the same purpose. The output of the supply provides a fixed DC voltage to bias the cathode of the follower. However, any AC signal connected to the output will be essentially shorted to ground. Even without the output filter cap, the supply exhibits a very low output impedance at all audio frequencies. This serves to "anchor" the bottom of the choke for AC, but allows DC current to flow through the choke.
--------------------------
Buy Chinese. Bury freedom.
I assume that there is a cap to ground from the -19 volt point of the negative supply.
Tre'
Have Fun and Enjoy the Music
"Still Working the Problem"
If you don't need the gain, why not just build a board with multiple pairs of input jacks and a selector switch and go directly into the amp from one set of output jacks? Might also want to include a volume pot.
I have run my cd player directly into my amp and it sounds pretty good.
Edits: 05/16/17 05/16/17
Most of my future amplifier builds won't have volume controls. So, I need a control amp that can adjust the volume and also perform an impedance transformation. Otherwise, I would have a 100K or 500K volume control driving a length of cable. That's no good.
--------------------------
Buy Chinese. Bury freedom.
I'm curious. What would happen if you put a volume control with each set of input jacks? Also why no volume control on the amps? I like the extra fine tuning you get with dual controls.
Edits: 05/16/17
Vinnie, there are several reasons not to cascade volume controls in the way you suggest. For one thing, every stereo volume control exhibits errors in channel-to-channel tracking. Then there's the noise of wipers and carbon tracks, as well as the effect a variable input resistance has on frequency response due to Miller effect. These issues can be significant, even with only one volume control, and they're cumulative when the controls are connected in series in the audio chain.
--------------------------
Buy Chinese. Bury freedom.
The output impedance of a 100k pot adjusted to the -6db point will be 25k.
25k driving the Miller capacitance of the first stage will likely cause a loss of high frequencies in the audio band.
Tre'
Have Fun and Enjoy the Music
"Still Working the Problem"
If it above 12khz I probably can't hear it anyway. One of the joys of being 70!
"Otherwise, I would have a 100K or 500K volume control driving a length of cable. That's no good."
100% correct! A buffered control center is an excellent, practical, solution. Passive control centers require VERY low capacitance cabling between them and the power amplification. Equate VERY low to short and unshielded, which creates its own set of problems. TANSTAAFL will, forever, be with us.
Eli D.
Or use an autoformer based volume control. I've been using one for years. The new modules from Intact Audio are really nice.
High sensitivity, wide dynamic range, low distortion, and smooth frequency response. Pwk
http://www.itishifi.com
.
Have Fun and Enjoy the Music
"Still Working the Problem"
Most of them seem to cost more for a pair than I spend on the rest of the parts for an amp!
vishay cermet Oring sealed trimmers. they are the little black knobs with one white spot on and are very good for level setting used with a fake law resister. they are affordable and can be relied on to stay stable over large periods with out turning them.(years) I think they may be out of production as some popular values are increasing in price and newer conductive plastic versions of dubious quality are being added to the farnel catalog. The original cermet version are so clean and are comparable to a fixed resister in sound quality imho.
IME, no matter what is behind it, if I were going to go and get nutty I'd start with one of Dave's autoformers and a good 2-gang switch. Then I'd put something that can amplify and buffer behind it. A core of Supermalloy, with a small gap is likely ideal...:)
cheers,
Douglas
Friend, I would not hurt thee for the world...but thou art standing where I am about to shoot.
.
Have Fun and Enjoy the Music
"Still Working the Problem"
nt
High sensitivity, wide dynamic range, low distortion, and smooth frequency response. Pwk
http://www.itishifi.com
You are missing my point. If I was spending $4k on the rest of the amp parts it would make sense, but if I am building an amp for $500 and using say alps pots, they are unlikely to be the item that would most affect the sound. The money would be better spent on better opts or beefing up the psu.
Edits: 05/17/17
I built one in this general flavor. Instead of a choke I have a bi-filar output in the cathode. Also, instead of the triode, with its poor current PSRR, I stuck a MOSFET on top in cascode to improve this and deliver minimum input capacitance. I am running a 5687 at 20 mA.
Where do those chokes you have go self resonant?
cheers,
Douglas
Friend, I would not hurt thee for the world...but thou art standing where I am about to shoot.
I searched for resonance by driving the chokes with signal through a 100K resistor. Monitoring with a scope, there's a very slight dip in amplitude just above 20kHz using that method. I assume that's the self-resonant frequency. Driving the chokes with 10K ohms, the dip is almost impossible to detect, and it disappears with a 1K source. FWIW, I also measured inductance and resonance with 30 mADC flowing, and there was no noticeable difference. Based on all this, I expect these chokes to provide a frequency response of at least 10Hz to 50kHz (+/- 0.5 dB) in cathode follower service.
Yes, PSRR is a concern in this configuration. The PS will need to be absolutely clean. I'm also thinking a 12V tube will be a better choice than the 6W6. A well-regulated 12.6V DC supply for the filaments could also be used to bias up the bottom of the choke. That would kill two birds.
Was the primary DCR of your transformer sufficiently low that you didn't need to bias the output stage like I'm describing? How did it sound?
--------------------------
Buy Chinese. Bury freedom.
IIRC, you want to look for the peak in voltage, where the L and shunt capacitance have their maximum impedance. Like the tuning of a choke with a parallel cap to tune this at 120 cps for filtering work.
cheers,
Douglas
Friend, I would not hurt thee for the world...but thou art standing where I am about to shoot.
Yes, SRF should create a peak. Maybe I would have seen that if I'd swept higher than 100kHz. The dip is clearly a different form of resonance, but like I say, it was too small to be of concern.
--------------------------
Buy Chinese. Bury freedom.
It is the first Parallel resonance you are looking for. Look lower...:) It should be above 120 cps, and is *VERY* unlikely to be above 2,000 cps.
cheers,
Douglas
Friend, I would not hurt thee for the world...but thou art standing where I am about to shoot.
The dip would seem to be series resonance. In series LC circuits, X L and X C cancel, at the resonant frequency. That leaves only DCR to impede current flow.
Eli D.
You're correct, the dip would be a series resonance. Sorry, typing too fast, had a brain lapse.
--------------------------
Buy Chinese. Bury freedom.
It is the first Parallel resonance you are looking for. Look lower...:)
cheers,
Douglas
Friend, I would not hurt thee for the world...but thou art standing where I am about to shoot.
There are no discernible peaks in the voltage that appears across the choke when I sweep it. That only happens when an external cap is attached.
--------------------------
Buy Chinese. Bury freedom.
What Douglas is talking about indeed happens midband typically around 1Khz and will not show up in a typical frequency sweep because it is at the self resonant point where the impedance is the highest. Other interesting things to note about this is that at the self resonant frequency the impedance is purely resistive and most LCR meters sample at 1Khz and spit out incorrect results for audio chokes since they are sampling near the SRF. I think this behavior was the origin of the myth that some amount of DC is required to "turn a choke on" and the lack of DC current (in a gapped device) is the reason a handheld LCR meter gives a reading quite different for what is printed on the can. One other reason handheld meters typically spit out seemingly incorrect numbers is the AC excitation level. The norm of a handheld meter is usually 100mv @ 1Khz and even the meters that can sample at 120hz will give low readings for power transformer primaries. This isn't because the meter is incorrect, it is just measuring the wrong thing. In order to know the correct inductance of a power transformer you need to measure it at 120 / 240V... anyways sorry for the slight tangent and back to the regularly scheduled discussion
In order to see the SRF you simply put a large valued resistor in series with the choke and a scope across the choke and sweep frequency for the highest value of AC across the choke. Again this does not show up in a traditional sweep because for our uses the value of the choke is 10X+ the value of the source giving flat response throughout the area where the self resonance occurs.
dave
"put a large valued resistor in series with the choke and a scope across the choke and sweep frequency for the highest value of AC across the choke."
I appreciate your input, but I described this exact process in my first reply above to PakProtector:
"I searched for resonance by driving the chokes with signal through a 100K resistor. Monitoring with a scope, there's a very slight dip in amplitude just above 20kHz using that method. I assume that's the self-resonant frequency. Driving the chokes with 10K ohms, the dip is almost impossible to detect, and it disappears with a 1K source. FWIW, I also measured inductance and resonance with 30 mADC flowing, and there was no noticeable difference. Based on all this, I expect these chokes to provide a frequency response of at least 10Hz to 50kHz (+/- 0.5 dB) in cathode follower service."
Funny, I never thought this would devolve into a discussion about the characteristics of the chokes. If anything, I expected at least a few inmates to tell me how awful a CF will sound. :)
--------------------------
Buy Chinese. Bury freedom.
I know that using a choke in this application is "cool", but I wonder why not use a CCS, instead? In my experience, a CF with a CCS sounds way better than a CCS with a cathode resistor (never tried a choke), and a CF with a CCS and a constant voltage strap (the SLCF, a la Allen Wright) sounds better yet. In fact, the latter reaches a state of transparency, in my opinion. AW knew his stuff.
It has been some time since I tried this...but I recall picking a resistor that left 80% of the signal voltage across the choke at its self resonant point. This of course being the parallel LC point that delivers a peak...
I did not find one I was not disappointed in...LOL
cheers,
Douglas
Friend, I would not hurt thee for the world...but thou art standing where I am about to shoot.
"I did not find one I was not disappointed in...LOL"Do you mean choke performance or the amplifier using a choke?
--------------------------
Buy Chinese. Bury freedom.
Edits: 05/17/17
The choke's low resonant frequency. That applied to so-called 'vintage' units I measured. It left me directing my winder to use a few of the capacitance reduction techniques which yielded not-so-subtle improvements in the SRF...:)
cheers,
Douglas
Friend, I would not hurt thee for the world...but thou art standing where I am about to shoot.
We must be missing something here in the discussion. With your test setup you should see both attenuation at low frequency (from the finite inductacne) and at high frequency (from the shunt capacitance) The dip you are seeing above 20Khz is not the behavior of the SRF and at high frequencies you should be able to track down the point where the capacitance dominates. (typical -3dB point with a 6dB slope just like happens at LF.)
Now in between those two points you will see a peak in amplitude across the choke and that happens at the SRF point. If the peak is not showing up then it may be hidden in the "passband" of the lowpass and highpass filters at which point you need to increase the source impedance in order to "flush it out"
I just swept a choke with 100K and 500K source Z and the 100K SRF was harder to find and with the 500K the SRF peak was clearly there.
I do understand your meaning.
"With your test setup you should see both attenuation at low frequency (from the finite inductacne) and at high frequency (from the shunt capacitance)"
Yes, that's how it behaves.
"you should be able to track down the point where the capacitance dominates. (typical -3dB point with a 6dB slope just like happens at LF.)"
I monitored the rolloff while sweeping by hand. It becomes evident above 20kHz (just above that very slight dip I mentioned), so I didn't quantify it.
"Now in between those two points you will see a peak in amplitude across the choke and that happens at the SRF point."
I've already repacked these for temporary storage, but maybe I'll break one out and test again. 500K or 1 Meg might reveal the self-resonant frequency. As you said earlier, none of this will create performance issues when the choke is in a circuit with an operating impedance of less than 500 ohms.
--------------------------
Buy Chinese. Bury freedom.
Agreed on the performance thing... Its a wonder chokes find any use in audio given the far superior measured results from a CCS ;-)
This discussion seems to come up frequently. For me, it's an easy decision. Hate semiconductors, love stored energy. Voila!
--------------------------
Buy Chinese. Bury freedom.
but you do need a negative supply voltage, for sure.
The stored energy in a choke allows the cathode to swing positive and negative equally, even without a negative supply. This creates a huge dynamic range, and it eliminates the deficiencies designers often unknowingly introduce into followers. Bottom line - it's neither necessary nor desirable to contaminate your music with silicon.
--------------------------
Buy Chinese. Bury freedom.
Thx again.
you've taught me something valuable, and for that I thank you, but I am talking about a tube CCS, with two haves of a high Gm triode in cascode. (I use ECC99 most often.) No silicon need apply. But again, I realize you do need a negative supply for said tube CCS.
I am wondering why is it that I have not read more about using a choke in a CF before this, either here or on the various relevant other websites? And doesn't a robust CCS achieve the same goal of allowing the cathode to swing with equal amplitude in both directions? Albeit, the parts count is much higher.
can't resist diverting things a bit. I find that a triode on the bottom, and a full pentode on top does quite well. Given the *GENERAL* characteristics of a pentode approximate a cascade, it is not going too far IMO to call this one a triple...LOL They do eat up some voltage, that is for sure, but with the right bottom triode selected, can do a pretty fine job driving a 1000R load( If you want to stack the deck in a linestage shoot out listening session, bring an amp with a nutty low input grid resistor ). Tell them at some point what happened...:)
cheers,
Douglas
Friend, I would not hurt thee for the world...but thou art standing where I am about to shoot.
Thx
Not just a pentode, but a pentode stacked on a triode to form a composite; a cascode. It is all about creating some floating voltage references. You'll want a large current set resistor for the lower triode, and you need one for the upper pentode's grid so the lower element has enough voltage to work under. A CCS driving a resistance works well, and floats adequately. And then there is the pentode's screen supply; dropping R from B+ and a cap bypass works well. Be careful of cap-to-case capacitance; no need to shunt stuff to ground...:)
cheers,
Douglas
Friend, I would not hurt thee for the world...but thou art standing where I am about to shoot.
Probably why I did not get your point the first time.
Honestly, I've never built and auditioned a high-voltage vacuum tube CCS. Whether it would be the equal of a choke, sonically speaking, I don't know. I do think that simpler is usually better, and the choke certainly wins that contest.Regarding the lack of literature on this subject, I can think of a few possible reasons. First, cathode followers aren't too popular these days. That's due in large part to a legacy of poor designs. It's much more difficult to create a good CF than a standard, common cathode stage. If they're not done right, they sound awful. Then there's the fact that a CF has no voltage gain. That often relegates it to secondary functionality such as tape out jacks and signal loops in guitar amps, applications in which sonic purity generally takes a back seat to practicality. Finally, most good DC-capable audio chokes are produced for use at the anode. Circuit impedances are higher there, so the requirements for the choke are more stringent and more costly. A decent pair of plate chokes can easily run $200-300, and that puts them out of range for many builders.
I paid less than $25 each for the chokes I intend to use in this project. Except for the high-ish DCR, they're probably as good as anything out there for this purpose. They could also be used as plate chokes, assuming the tube has an appropriate Rp. Keep an eye on eBay if you want to give something like this a try on the cheap. :)
--------------------------
Buy Chinese. Bury freedom.
Edits: 05/19/17
"Bottom line - it's neither necessary nor desirable to contaminate your music with silicon."
My apologies, and I do hope I don't start start a big interruption to the flow of this thread, but I am always so puzzled when I read statements like this one. Unless your chosen sound source is old gramophone records, you are surely listening to music that has already been "contaminated" by a huge amount of silicon? Does one more silicon item, in itself, have that much effect one way or the other?
Another possibility is that the vacuum tube stage is adding in something to the sound, in which case indeed, one might find it doesn't do that as well if silicon is included. I find that a perfectly good reason for preferring to use a tube amplifier, though I'm not sure all tube lovers would accept this premise.
Chris
"you are surely listening to music that has already been "contaminated" by a huge amount of silicon? Does one more silicon item, in itself, have that much effect one way or the other?"
If that's the criteria you're using to qualify the value of audio equipment, I'd say you might be in the wrong forum. There's a lot more research and online activity in the solid-state genre, and it's much simpler to build. I build with tubes; you can use whatever parts make you happy. :)
--------------------------
Buy Chinese. Bury freedom.
"If that's the criteria you're using to qualify the value of audio equipment, I'd say you might be in the wrong forum. There's a lot more research and online activity in the solid-state genre, and it's much simpler to build. I build with tubes; you can use whatever parts make you happy. :)"
I'm not disagreeing with the idea of choosing to build with tubes; that's what I do also, in general. The only thing I'm puzzling about is the notion that silicon devices in some way "contaminate" the music. My point, in this regard, is that the sound source one is commonly using (a CD) has been produced using silicon all the way from the microphones to the output of the CD player, and so why should one additional silicon device make any appreciable difference to the level of "contamination" of the music?
Perhaps I am taking words like yours, or other similar ones I have seen elsewhere, a bit too literally. As I say, I have no problem at all with understanding the concept that one wants to build using just tubes, and no silicon. But I was puzzled by the "contamination" justification, which seemed to overlook the vast number of silicon devices that the music signal had already passed through.
Anyway, carry on with the thread, which I find quite interesting, and don't let me divert it down these side alleys!
Chris
LOL...I just giggle when I hear that, 'contaminate with silicon' idea. I have built a tube CCS a few times, and while some of the performance parameters are not as good as the SS implementations, the critical ones certainly are. Things like a lack of voltage variable capacitance for one.
Also, to say that a choke is a better load because it stores energy and can operate with less voltage across it only matters if delivering an adequate amount of voltage is expensive; either to the design implementation as a whole, or outright Cash outlay...:)
cheers,
Douglas
Friend, I would not hurt thee for the world...but thou art standing where I am about to shoot.
I'm with you on the stored energy thing but from a purely objective viewpoint, when I look at the impedance curve of a choke I quickly start twitching in fear. It wasn't until I became comfortable that a large value but varying load isn't remotely close to being a problem in the circuits we are dealing with. Then learning to ignore the flat earthers that pound their chest about constant impedance loading finally allowed me to sleep at night after a long day of enjoying music.
dave
as a certified flat earther, the worry about changing loading only extends to the discontinuity the grid exhibits when going positive. The change is not small...LOL Cathode drive, here I come.
cheers,
Douglas
Friend, I would not hurt thee for the world...but thou art standing where I am about to shoot.
When a pentode or FET voltage follower is the object of the exercise, I'll take a choke over a constant current sink, "in a NY minute". I'm not saying a CCS can't work, but "head butting" between entities, each of which is "constant" current, is something (IMO) to avoid.
Eli D.
Must beg to differ...a cathode follower( or source follower, be it single element or cascode) is only a constant current if its lower gate source voltage is fixed by a resistive element. When it is riding a solid sink, it is a variable current delivery device, and will settle into a constant gm operation due to it not caring about what the voltage across it is( the horizontal plate lines ).
Nothing to worry about IME...:)
cheers,
Douglas
and please consider 'cascade' a typo for 'cascode' when ever you see me writing it.
Friend, I would not hurt thee for the world...but thou art standing where I am about to shoot.
In fact, a cathode follower more closely approximates a constant voltage source. Using a constant current device (in my case, a choke) at the cathode is exactly the right thing to do.
--------------------------
Buy Chinese. Bury freedom.
"If anything, I expected at least a few inmates to tell me how awful a CF will sound. :)"
Implementation details matter . Well executed voltage followers are among the most transparent active circuits extant. If somebody thinks otherwise, let them butt heads with SY over on DIY Audio.
FWIW, I'm partial to FET (source) followers, as high transconductance/ low O/P impedance is readily available.
Eli D.
FAQ |
Post a Message! |
Forgot Password? |
|
||||||||||||||
|
This post is made possible by the generous support of people like you and our sponsors: