Looking for a simple tube preamp build

[rongon]

I use the 4S with cathode follower, still have a lot of gain. Maybe i must use 2 x 12au7. View attachment 1072297

The 12AX7 will have gain of about 30X to 40X.
The cathode follower will have gain of about 0.9X.
35*0.9 = 31.5X gain (30dB).
We want something like 3X gain (9.5dB).
You would need to introduce about 20dB of negative feedback to reduce the gain of that 12AX7 down to 3X.

If you substitute a 12AU7 for the 12AX7 first triode, then the gain will be about 10X (20dB).
Again, we want something like 3X gain (9.5dB).
You would need to introduce about 10dB of negative feedback to reduce the gain of that 12AU7 down to 3X.

In both cases, you will need to introduce negative feedback to get the gain down to 3X (9.5dB).

 

[rongon]

Great explanation. My technical understanding is limited so I tend to approach things differently. On my Boogie Factor 1626 preamp I have volume controls on both the input and output. This was done for purely practical reasons.

A pair of mono pots allows me to alter the L/R balance because my listening position is often not centered between the speakers. A stereo stepped attenuator allows me to adjust the overall volume without altering the L/R balance once it's been set.

I have no idea about the technical implications. Care to comment?

1. A pot on the input reduces the incoming signal but does nothing to help signal to noise ratio.
2. A pot on the output reduces the outgoing signal, including noise, but the resulting output signal has more distortion having been previously amplified and then its amplitude reduced.

So the end result of using two pots is that you've degraded both the noise performance and the fidelity of your amplifier.

As I said earlier, if you don't need much gain, why start with a tube that has much more gain than you need and then dump most of the signal to ground?

Yes, but excess gain can be put to good use through the introduction of negative feedback. Unfortunately...

There seems to be a widespread aversion to any form of negative feedback imposed on a circuit (while forgetting the internal feedback mechanism in all triodes, or the degenerative feedback produced by leaving a cathode load resistor unbypassed), along with the fact that there are practically no suitable triodes with mu = 2 or 3.

So now we're stuck.

There are no triodes that have mu = 2 or 3.
Most triodes have mu of 10 to 20. (6V6-triode has mu = 8 to 10, 6N6P has mu = 15, 6SN7 has mu = 20.)
We're not allowed to use negative feedback to reduce the gain of the triode.
That leaves us with voltage dividers (volume controls) to basically shunt the signal to ground.

Who can solve this problem?

 

[100decibels]

Like this maybe?
3X gain...

View attachment 1072294

Any article related to this schematic? Never saw diodes on a tube preamp or amp before.

 

[rongon]

Any article related to this schematic? Never saw diodes on a tube preamp or amp before.

Yes!

First -- Triode with local parallel negative feedback:
https://www.valvewizard.co.uk/localfeedback.html

Discussion of DC-coupled cathode follower showing use of diode (or neon lamp) for protection from excessive cathode to heater voltage:
https://www.valvewizard.co.uk/dccf.html

Here's an example of a tube circuit that uses the opposing polarity zener diodes on the output to protect solid-state amp inputs downstream from startup DC voltage overload:
https://www.valvewizard.co.uk/phonopcb2.html

 

[Koifarm]

There is no problem with negative feedback to get less gain. I sounds really good.

 

[jean-paul]

It is also a necessity or a lesser evil to do it.

 

[FlaCharlie]

So the end result of using two pots is that you've degraded both the noise performance and the fidelity of your amplifier.

There are no triodes that have mu = 2 or 3.

Well, some method of controlling the volume is necessary so I don't see how there is any way to eliminate the shortcomings entirely.

Would it be better to run both the input and output controls as wide open as possible and use the variable (digital) output of a DAC or iPad to control the overall volume?

The 1626 has a mu of 5, so it's relatively close. Perhaps there are others that are a bit lower?

 

[rongon]

The 1626 has a mu of 5, so it's relatively close. Perhaps there are others that are a bit lower?

It looks like the rp of 1626 is 2500 ohms, which means its Zout will be about 2k ohms in typical usage, with bypassed cathode resistor or fixed bias. That's marginal for use into a 10k ohm load (as is common with powered speakers these days), but if you don't mind the triode being loaded down a bit (and the higher level of harmonic distortion that would result) then it could work. If the cathode is left unbypassed, Zout will be too high to use with present day Class D amps and powered speakers.

EL86-triode has fairly low mu and very low rp (<1k ohms). That's a possibility.

12B4A is a popular choice. That has very low rp (<1k ohms) and low mu.

45 DHT has low mu, but then you have the added expense of making the filament supply quiet (or else hummmmmmm).

2A3 same problem as 45, but high plate current demand, Nice, low rp though.

Some people like 4P1L in triode, but now things get complicated (it's microphonic and again it's directly heated so there is the added complexity of a good DC heater supply that will still 'sound good').

Another possibility is the 12P17L pentode wired triode. However that one seems to have THD that varies with level. I don't know if it would 'sound good' used as a line level preamp. Probably worth a try, though.

And all this to avoid using negative feedback. That's cool. Hopefully the extra work results in a better sounding end result.

The title of this thread is "Looking for a simple tube preamp build." What exactly is "simple"? When is something too complicated to be "simple"?

 

[stenak]

My favorite are the 12B4A tube amp. Single ended, not to much gain and sounds great!

 

[rongon]

https://www.audioasylum.com/forums/tubediy/messages/15/159939.html

https://www.audioasylum.com/cgi/t.mpl?f=tubediy&m=158008

I would add a proper 1M ohm grid leak resistor on the 12B4A, instead of letting the pot be the grid leak. What happens if the wiper loses contact with the track? But that's a quibble.

I'd also reduce the value of the pot to 50k, or perhaps 20k, depending on the Zout of my sources. Another minor quibble.

 

[stenak]

https://www.audioasylum.com/forums/tubediy/messages/15/159939.html

https://www.audioasylum.com/cgi/t.mpl?f=tubediy&m=158008


View attachment 1072416

Thats the one!

 

[100decibels]

https://www.audioasylum.com/forums/tubediy/messages/15/159939.html

https://www.audioasylum.com/cgi/t.mpl?f=tubediy&m=158008


View attachment 1072416

I would add a proper 1M ohm grid leak resistor on the 12B4A, instead of letting the pot be the grid leak. What happens if the wiper loses contact with the track? But that's a quibble.

I'd also reduce the value of the pot to 50k, or perhaps 20k, depending on the Zout of my sources. Another minor quibble.

To pin 3/L and 3/R is not clear to me, to the + of the 100uf 100v cap?

 

[FlaCharlie]

It looks like the rp of 1626 is 2500 ohms, which means its Zout will be about 2k ohms in typical usage, with bypassed cathode resistor or fixed bias. That's marginal for use into a 10k ohm load (as is common with powered speakers these days), but if you don't mind the triode being loaded down a bit (and the higher level of harmonic distortion that would result) then it could work. If the cathode is left unbypassed, Zout will be too high to use with present day Class D amps and powered speakers.

EL86-triode has fairly low mu and very low rp (<1k ohms). That's a possibility.

12B4A is a popular choice. That has very low rp (<1k ohms) and low mu.

45 DHT has low mu, but then you have the added expense of making the filament supply quiet (or else hummmmmmm).

2A3 same problem as 45, but high plate current demand, Nice, low rp though.

Some people like 4P1L in triode, but now things get complicated (it's microphonic and again it's directly heated so there is the added complexity of a good DC heater supply that will still 'sound good').

Another possibility is the 12P17L pentode wired triode. However that one seems to have THD that varies with level. I don't know if it would 'sound good' used as a line level preamp. Probably worth a try, though.

And all this to avoid using negative feedback. That's cool. Hopefully the extra work results in a better sounding end result.

The title of this thread is "Looking for a simple tube preamp build." What exactly is "simple"? When is something too complicated to be "simple"?

Not sure where I got the figures, it may have been a post by the owner of Bottlehead, but I read somewhere that the plate resistance of the 1626 is ~1.9k and, based on that, I calculated the output impedance as ~1.4k. But, again, I'm not real technical so maybe I'm off a bit.

The output impedance, whatever it actually is, wasn't really a major factor for me because I knew I'd mostly be using it with tube amps. I just wanted it to be usable with SS or Class D since I do have a couple of them, but I don't really care if it's optimal.

My SS amp has an input impedance of 20k. I have no idea what it is for the Aiyima A-07, but I assume it's lower than 20k. Bottom line, it seems to work well enough and it's certainly lower than many more commonly used preamp tubes.

Perhaps an earlier poster expressed an aversion to using NFB, but it wasn't me. I've used NFB in amp builds but I've never thought about NFB in a preamp. Other than using an un-bypassed cathode resistor or diodes, I'm not even sure how it would be implemented. My preamp uses battery grid bias, though.

Certainly, everyone has a different definition of simple. But I figure a single triode per channel qualifies. That's about as simple as it gets. If I was using SS or Class D all the time then I'd consider adding a cathode follower section but a simpler, single triode anode follower, design can work with the right tube choice.

For me, the cost of the tube is also part of the equation in pretty much everything I build. While I do have a 45 SET and a PP 6B4G, I'd probably not be interested in using them in a preamp. The ones I have were purchased years ago when their prices were more reasonable. I did try the 6A5G (indirectly heated 6B4G) when I was breadboarding. It sounded good but I liked other tubes better.

Years ago I built a 12B4A pre and liked it but many of them are very microphonic. I had to try 15 or 20 tubes in order to find a pair that were reasonably quiet. It's the only tube I've ever used that started howling when music was playing.

I haven't tried the 4P1L or the indirectly heated 12P17L, although I have some of each. I did try the 2P29L, which is a DHP and it sounded very nice in triode. For me (YMMV) the 1626 just had something special. The two that were close behind were the 6V7G and the 2P29L. Here's the thread about the 2P29L:

https://www.diyaudio.com/community/threads/2p29l-preamp.336620/

So, any comment on my earlier question?

Would it be better to run both the input and output controls as wide open as possible and use the variable (digital) output of a DAC or iPad to control the overall volume?

 

[rongon]

The plate resistance of any triode is somewhat variable (even though it's considered to be one of the 'tube constants'). If you take a triode and run it with a low B+ voltage and high plate current, its rp will be on the low side. If you take that same triode and run it with a higher B+ voltage and with low plate current, its rp will then be higher (sometimes significantly higher).

Here's a comparison of how rp differs at two different operating points for triode type 5687:

The rp for type 5687 can be 3k ohms, or it can be 1.75k ohms, depending on how you measure it. So, which is it? The answer depends on the operating points you choose in your design.

Perhaps an earlier poster expressed an aversion to using NFB, but it wasn't me. I've used NFB in amp builds but I've never thought about NFB in a preamp. Other than using an un-bypassed cathode resistor or diodes, I'm not even sure how it would be implemented. My preamp uses battery grid bias, though.

There have been a couple of preamp designs posted here using shunt local NFB from plate (output) to grid (input). Another way of doing that is to use two common cathode amplifier stages into a cathode follower, and take feedback signal from the cathode follower cathode (output) back to the cathode of the first gain stage. I've seen that called a 'Ring of Three' topology.

The advantage of this circuit is that it's non-inverting. Gain is about 6X (16dB) as shown. In simulation, the circuit has super-low THD and quite low Zout, but it rings a little if I try to apply much more negative feedback than this. I don't plan on trying to build it, but it was an interesting exercise.

Would it be better to run both the input and output controls as wide open as possible and use the variable (digital) output of a DAC or iPad to control the overall volume?

I'm not sure, honestly. What happens to signal-noise ratio when you use digital means to reduce the gain from your DAC by >30dB? I'm sure it's better than putting resistor voltage dividers at both the input and output of your preamp. But how will you like the sound produced? I'm unable to answer that.

I'd rather design a preamp with appropriate gain for the application, or go without any preamp at all.

In my current setup I'm using an Intact Audio "Autoformer Module" inductive attenuator (https://intactaudio.com/atten.html), without any line level preamp. My DAC and (hybrid FET-tube) phono preamp go straight to a 4-input selector switch and Intact Audio attenuator, and finally to a pair of active speakers.

I've been toying with the idea of adding a line preamp with 6 to 9dB of gain and balanced outputs. But that's not actually necessary, and I'm relaxing after going through designing and building a couple of phono preamps over the last 6 months or so.

A setup I used to have depended on the gain from a line level preamp. I had a 2-stage push-pull 2A3 amp that required almost 4V peak to reach full power. I used a single common-cathode 5687 per channel for the line level preamp, driving said PP 2A3 amp. It actually sounded quite nice, but the line preamp is unusably noisy (too much gain) with typically sensitive power amps (or my powered speakers). When the PP 2A3 mysteriously broke, the 5687 line preamp went on the shelf, and it's still there. (I need to try to fix that 2A3 amp. I really liked it.)

 

[deafbykhorns]

That looks very much like the John Hogan SRPP 6SN7 preamp I use to have. I loved the sound of it. I agree Steve: As these speakers are, more or less, in my face I don't need that much gain, however, I have noticed all these years, that gain = dynamics and I fear I would not get that with just a buffer.

Look at the 01a tube if you don’t need much gain
That preamp will tame any amp
It’s the best I’ve heard to date

 

[rongon]

To pin 3/L and 3/R is not clear to me, to the + of the 100uf 100v cap?

You know... I missed that. I see what you mean.

Pin 3 is the center tap of the tube's heater (filament).
Pin 1 is the tube's cathode.

Type 12B4A has a center tapped heater so that it can be run with the two halves of the heater wired in series for operation with 12.6V 0.3A, or in parallel for operation with 6.3V 0.6A.

In the schematic, there is a 100uF capacitor connecting from pin 3 (heater CT) to ground. I don't know what that's doing. Is that a drawing mistake, or is that a cathode bypass scheme I've never seen before? Or is that supposed to reduce noise from the AC heater? I really don't know.

There is a 500 ohm resistor from pin 1 (cathode) to ground. It looks like that 500R resistor is not bypassed by the 100uF capacitor, which means the gain of the triode will be reduced and its rp will be increased by way of current feedback (degeneration) as AC signal current travels through the resistor.

 

[rongon]

Look at the 01a tube if you don’t need much gain
That preamp will tame any amp
It’s the best I’ve heard to date

01-A plate resistance is about 10k ohms. The OP mentioned powered speakers, which I assume will have input impedance of about 10k ohms, or possibly lower.

It looks like 01-A mu is listed as 8. That's about the same as an EL86-triode, but the EL86-triode has rp below 1k ohms.

Is the sound of a 01-A spoiled by adding a follower (buffer) on its plate (output)?

 

[deafbykhorns]

01-A plate resistance is about 10k ohms. The OP mentioned powered speakers, which I assume will have input impedance of about 10k ohms, or possibly lower.

It looks like 01-A mu is listed as 8. That's about the same as an EL86-triode, but the EL86-triode has rp below 1k ohms.

Is the sound of a 01-A spoiled by adding a follower (buffer) on its plate (output)?

Even better, put a TVC at the output
I did a custom from Dave Slagle
At Most of my listening levels, TVC brings impedance below 1k

 

[rongon]

I have the older, 4-control version of Dave's TVA. I do like it a lot.
How does an autoformer present a high impedance load to the preceding tube stage if it's in parallel with the input impedance of the amplifier after the TVA? I don't understand how that can be, since the a resistance in parallel with a large inductance is still going to reduce the load impedance presented to the driving stage... I think.

 

[deafbykhorns]

I’ll try to dig up the thread, Dave helped me with this issue. I have a spice simulation on my other computer on various loads. I discovered the issue when running freq response of the preamp before adding TVC. It was crazy how different cables changed the freq response before the mod. I also used Ales gyrator circuit. Impedance was 600ohms at unity gain. Ale also has a nice buffer that works well with this tube