You bet.
Tons of blanket statements posted around here which then lead straight to unneeded and useless bloody wars.
Tons of blanket statements posted around here which then lead straight to unneeded and useless bloody wars.
I do know where you're coming from. Was there myself until about 10 years ago (a short time at my age). In a discussion about re-entrant distortion (Baxandall/Putzeys models) I was banging on about how it couldn't apply to locally degenerated single device stages, (emitter deneneration or what's now called a "Schade" feedback for some reason). I'd never measured it, and I do have very good measuring equipment, so I was pretty confident.
Nelson Pass was kind enough to show a comparison between a single MOSFET common source amplifier stage with and without a degenerating source resistor. Re-entrant distortion was obvious in the spectrum of the degenerated version. I was dead wrong.
My overarching point is that "Four legs good, two legs bad" is much too broad, and fundamentally wrong. Do we want to go back to the long loop feedback designs of the Golden Era? Not so much. But should we be afraid of the F-word? Also, not so much.
All good fortune,
Chris
Nelson Pass was kind enough to show a comparison between a single MOSFET common source amplifier stage with and without a degenerating source resistor. Re-entrant distortion was obvious in the spectrum of the degenerated version. I was dead wrong.
My overarching point is that "Four legs good, two legs bad" is much too broad, and fundamentally wrong. Do we want to go back to the long loop feedback designs of the Golden Era? Not so much. But should we be afraid of the F-word? Also, not so much.
All good fortune,
Chris
You have to differentiate between global negative feedback (GNFB) around multiple stages, and local NFB. In general, there is no reason to use GNFB in a tube phono or line preamplifier. The already low subjectively pleasing odd-order distortion products will be even lower, but the more annoying even-order distortion will go up a bit. The sound character will change anyway.
On the other hand, a local feedback (often realized by an unbypassed cathode resistor in a common cathode stage) will straighten the transfer curve of the tube and reduce its gain at the same time. It will keep the "tubey" sound character more than the GNFB does, IMO.
On the other hand, a local feedback (often realized by an unbypassed cathode resistor in a common cathode stage) will straighten the transfer curve of the tube and reduce its gain at the same time. It will keep the "tubey" sound character more than the GNFB does, IMO.
I must disagree. As an example, the sun is a feedback system. Is luminousness, internal temp, internal pressure, are all connected through a tight feedback loop. Change in Helium content changes the temp and the pressure to a new equilibrium, until it runs out of control range and then boom!.
We humans (any living being) cannot live without feedback. Without feedback from the rising CO levels in our blood to the brain we wouldn't even bother to breath to bring it down again. Check out Homeostasis. The whole discussion around climate change is nothing more than a system finding a new equilibrium after some system input (CO2) is changing, through feedback.
Feedback governs anything that is a system the universe.
Jan
If course, there are many that argue that every triode has internal feedback anyway. The only true way to have no feedback is to use pentodes.
Cheers
Ian
Cheers
Ian
This is the line amp I use, 12AH7 with CF, no NFB, feeding MoFo which is also no NFB.
My other system uses ACA (NFB) and B1(no NFB), plus DIY Pearl I (no NFB).
To me the gear with no NFB produces the better sound, the more natural sound.
I've listened for so many years to FB amps and wish someone told me there is a different way.
All these arguments re. NFB vs NNFB are so tiresome, they usually bring the usual suspects that love to argue for arguments sake and derail the main topic.
So I would say to them if you want to argue go and start your own thread.
To the OP I would say: go ahead, build a no NFB line amp with the tubes of your choice that will match the amp after that and listen.
A few tube suggestions if I may:
6SN7(or the noval equivalent 6CG7), or even better 12SN7
12B4A
if in need of very high gain E88CC
Have fun.
Stan
Attachments
Pete Millett’s low mu preamp is a really nice sounding one if you don’t need a lot of gain. Simple triode gain stage, the CCS adds a few extra parts but still a pretty easy build.
I've been here a while now, and I've seen a lot of tube preamp designs come along. I've also learned how to model these things in SPICE, which can be revealing.
While you can't rely on simulations to tell you the absolute real world THD level to expect from any given circuit, you can use simulations to tell you the *relative* levels of THD of this vs. that operating point or this vs. that change of load on a given circuit.
With those caveats in mind, I'm surprised that certain rather high distortion circuits are so popular. I believe there are some people who like the sound of an amplifier that adds a bit of tube 'sweetening' through dynamics compression and/or a bit of added THD of the 'right' kind. I know, because I like that on some kinds of music. Yes, small group jazz music, especially 1950s and '60s recordings, and especially on vinyl.
However, I also like large scale symphonic works in 24/96 digital. My experience is that little 5 watt triode amps with no global NFB loop driving small-ish loudspeakers (usually 88dB to 91dB sensitivity 1W/1m) fall apart on big orchestral blasts, even at moderate listening levels.
I know people who listen to nothing but small group jazz and classical/baroque chamber music recordings on vinyl (and 78s). SE triode amps with no gNFB lend themselves well to those styles. But then I bring some digital Mahler and there goes that.
Everything is a compromise.
While you can't rely on simulations to tell you the absolute real world THD level to expect from any given circuit, you can use simulations to tell you the *relative* levels of THD of this vs. that operating point or this vs. that change of load on a given circuit.
With those caveats in mind, I'm surprised that certain rather high distortion circuits are so popular. I believe there are some people who like the sound of an amplifier that adds a bit of tube 'sweetening' through dynamics compression and/or a bit of added THD of the 'right' kind. I know, because I like that on some kinds of music. Yes, small group jazz music, especially 1950s and '60s recordings, and especially on vinyl.
However, I also like large scale symphonic works in 24/96 digital. My experience is that little 5 watt triode amps with no global NFB loop driving small-ish loudspeakers (usually 88dB to 91dB sensitivity 1W/1m) fall apart on big orchestral blasts, even at moderate listening levels.
I know people who listen to nothing but small group jazz and classical/baroque chamber music recordings on vinyl (and 78s). SE triode amps with no gNFB lend themselves well to those styles. But then I bring some digital Mahler and there goes that.
Everything is a compromise.
Last edited:
The DC coupled cathode follower is a good solution, besides standard SRPP etc., but unlike what is commonly reported on the book theory Ra1 and Rk2 doesn't have to be necessarily matched due also to the unavoidable small mismatching on the tube's sections, so one should "modulate" the values...this can be observed on LTspice sims too. Oftentimes it is suggested the add of a few hundred ohms before OUT and Rk2.
Last edited:
Just a little addition:
The big difference between open loop and FB amps, is that as I see, the feedback amplifiers have a certain restraint to them, less lively, that is, tube amps.
- just closing the loop will (might ...) alter the spaciousness.
- a triode in its structure uses local feedback, that is, the anode voltage as it is amplified has an effect on the grid; I am not pertaining to the Miller Capacitance at higher frequencies, but the load line is formed with some kind of feedback in it.
- And kathodes may or not be bypassed.
The big difference between open loop and FB amps, is that as I see, the feedback amplifiers have a certain restraint to them, less lively, that is, tube amps.
- just closing the loop will (might ...) alter the spaciousness.
It's possible, but hardly worthwhile. You have to use triodes, and you end up with a fairly high output impedance. Don't forget, you can't use cathode-followers - or cathode resistors - as these imply/incorporate feedback. Expect high low-order distortion, and fairly indeterminate gain.
It's not clear to me how you can get precise gain without NFB. Precise channel matching for example.
It is possible, but at the cost in some ways of simplicity. Allen Wright did it with his Realtime Preamplifier - the gain of the phono stage, in particular, is set by the Vbe characteristics of the input transistors and their emitter resistors, and the source impedance for the RIAA filter network is to a good approximation set by the anode resistors, provided the load provided by the volume pots is a constant and precise 50K. So if you have set the phono EQ precisely by trimming the passive components (not that I did that with mine) it will stay that way, even as the valve characteristics change with time.
This circuit also, it seems, has very low distortion, despite having no loop NFB.
Alex
It seems like by and large the responses are defining NFB a lot more rigedly than I think the original post intended. If I am correct then a line amp of the type the OP requested you simply build a simple common cathode stage with a nice low mu triode and an oprional cathode follower. Yes the CF is a feedback amplifier but it is local and in the form of a buffer. If you want to eliminate the follower just use a very low Rp tube like 12B4 as opposed to something like a 6SN7 and take the output from the CC stage load resistor.
What I would suggest is to build some different topologies and see what you like best. 12B4 CC, 6CG7 or similar with and without cathode resistor bypass into a CF, 12AY7 direct coupled to CF with output fed back to input (buffered anode follower). These being listed in order of least FB to greatest. This way you can decide based on experience rather than an assumed dogma. I plan to experiment with some of these myself. I suspect my old ears will not detect a great deal of difference but it will be interesting to find out.
What I would suggest is to build some different topologies and see what you like best. 12B4 CC, 6CG7 or similar with and without cathode resistor bypass into a CF, 12AY7 direct coupled to CF with output fed back to input (buffered anode follower). These being listed in order of least FB to greatest. This way you can decide based on experience rather than an assumed dogma. I plan to experiment with some of these myself. I suspect my old ears will not detect a great deal of difference but it will be interesting to find out.
I was thinking just that... I'll take the second order over the third any day, especially if the minimum amount of the second order does some good masking of the third order. It's all about the harmonics profile, not the THD figure itself. That said, sometimes I also want as accurate as possible, so minimising all of that.
I suppose some people like the added 'sparkle' or perceived 'dynamics' some odd order brings. While this can be true for some styles of music, I find that kind of response fatiguing very rapidly. It robs instruments from their original timbres.
To the OP, you're going to have a lot of fun doing that. Here are some recommendations:
- start with low voltage. I am doing that using a salvaged socket that I soldered breadboard leads to and a breadboard, small 6V AC transformer for heaters, 12V-13.8V linear Reg PSU for the plates. Currently, I am experimenting with only a single triode in the tube (12AX7, 12AT7 and Triode 2, which means Pins 1, 2, 3, 4, 9). There's a lot of configurations you can test just with this. I have added a little MOSFET follower to test with headphones.
This allows you to explore without any expensive mistakes. When you're confident enough, and if need be, you can switch to high voltage builds but be really, really careful there. Things like cap discharge, pre-installing probes, one hand behind your back, dim bulb tester, etc... are mandatory here.
Sims are useful and fun too, although not absolutely necessary. Sometimes I do these and explore there, sometimes I find interesting things on the breadboard.