Dedicated DIY Audio Thread

[jderimig]

I didn't want to pollute the excellent Dedicated Audio/Video thread with too much DIY tech, so with @hobie1dog blessing I will start this parallel thread to highlight any of mad audio scientist projects and tech.

I myself really enjoy Vacuum tube tech and there is a lot of good R&D being done with this old tech that I find fascinating. The latest is the invention of the UNSET amplifier topology. In the next couple posts I will detail what it is and what I plan to do with it in my next amplifier project.

But first some tube eye-candy. The big tubes on the top is what I have been collected (hoarding?) from ebay over the last few years. WWII era aircraft radio transmitting tubes.

 

[Hobie1dog]

I didn't want to pollute the excellent Dedicated Audio/Video thread with too much DIY tech, so with @hobie1dog blessing I will start this parallel thread to highlight any of mad audio scientist projects and tech.

I myself really enjoy Vacuum tube tech and there is a lot of good R&D being done with this old tech that I find fascinating. The latest is the invention of the UNSET amplifier topology. In the next couple posts I will detail what it is and what I plan to do with it in my next amplifier project.

But first some tube eye-candy. The big tubes on the top is what I have been collected (hoarding?) from ebay over the last few years. WWII era aircraft radio transmitting tubes.

Keep your DIY stuff right here, most of us really love it, we just can't build it.

 

[jderimig]

So what is this UNSET? To understand this we start with SET which is Single Ended Triode. SET amps are a high end audiophile thing which are characterized by very expensive amplifiers with very low power output. $3000 and up for a 3-5 Watts out are not uncommon.

The most popular SET amp uses the Western Electric 300B tube which was designed in 1933 to amplify telephone signals. Later they were used with AL Voice of the Theater speakers in movie houses. The tubes are still manufactured today, Western Electric restarted manufacture of these tubes to satisfy the audiophile demand. You can buy them here.

https://www.westernelectric.com/300b
$700 each for a 8W amp. Which are a bargain, NOS orginal 300B are going for as much as $1200 ea. Bet your sorry now your parents threw away grandpa's tube stash those many years ago.

However there are source of nice tubes that can be gotten very cheaply. Television sweep tubes were made by the million and were available at hamfests for 2 for a dollar. UNSET was developed by a pair of tube amplifier hobby guys who were looking for a way to get these TV sweep tubes to have the same transfer function (more or less) than the expensive 300B and 2A3 classic triodes.

 

[jderimig]

The problem with triodes...output power.

The triode (God's amplifier component) works by using grid voltage to modulate the flow of electrons from the cathode to the plate. If the operating point is chosen optimally you get a nice linear relationship between input and output. No negative feedback is required to get a low distortion amplifier. That is the main "selling point" of SETs, no nfb.

The problem is that the current is both a function of the grid voltage and the plate voltage. As plate voltage decreases in the AC signal swing, the current is also reduced which limits how low you can pull the plate voltage to the cathode voltage. This AC swing limitation limits the efficiency or power output from a triode. Even though otherwise its the perfect amplifying device. As a result, true SET amps need very efficient speakers, like horns or huge box volumes, which come with their own set of tradeoffs.

Which is why tetrodes and pentodes were developed. In the days before solid state all hi-fi's and guitar amps used pentodes as the power amplifier tube(s). The pentodes plate current is much less dependent on plate voltage so it can produce a larger voltage swing than a triode given the same power supply voltage. But Pentodes have tradeoffs as well.

A common modification to guitar amps to provide a switch from Pentode to Triode mode. This switch ties the screen grid of a Pentode to the plate which makes the tube operate as a triode. Even the tube itself is not a triode. It is often called the half-power switch.

https://www.premierguitar.com/diy/ask-amp-man/ask-amp-man-reduce-power-with-a-pentode-triode-switch
Didn't make to much sense to me. "This amp goes to 5...."

 

[Sooner Boomer]

A little something to read...

https://worldradiohistory.com/Archive-All-Audio/Audio_Collection.htm

 

[jderimig]

Pentodes: What's the problem?

The pentode adds a screen grid which effectively removes the plate voltage influence on the tube current. It was a great leap forward in tech at the time. Invented in 1926 with first pentodes in production in 1932. In 1936 the first popular beam tetrode, the 6L6 was introduced and became the basis for audio, rf transmitter, and TV sweep tubes used in modern times.

By holding the screen voltage at fixed point, current through the tube is almost totally a function of just the grid voltage. In fact, its characteristic curves look very much like current mosfet curves.



So what's the problem? It was found that a Pentode final output amplifier for some reason doesn't sound as good as a triode unless some negative feedback is applied. The effect is called the pentode sound. In my initial DIY 1625-tube amp I forgot to wire the feedback and the best description of the Pentode sound is a slight grating of female vocals. Some describe the sound as "sparkle". But the other qualities of pinpoint imaging, 3D depth soundstage, and super fast dynamics were wonderful. As nfb was applied this 'sparkle' was corrected but there was also an loss of the other positive attributes. So there was a tradeoff.

With a single ended triode or the pentode wired in triode more you did not need the nfb to get it all.

 

[davel]

It was found that a Pentode final output amplifier for some reason doesn't sound as good as a triode unless some negative feedback is applied.

So what did measurements indicate?

 

[jderimig]

So what did measurements indicate?

Generally pentodes with no feedback have many higher order 5th+ odd harmonic distortions. So nfb is required to tame them.

Generally the triodes have much much less and sound good with no negative feedback circuitry. However the action of the triode of reduction in current with plate voltage swing is really a feedback mechanism albeit an internal one.

 

[davel]

Generally pentodes with no feedback have many higher order 5th+ odd harmonic distortions. So nfb is required to tame them.

Generally the triodes have much much less and sound good with no negative feedback circuitry. However the action of the triode of reduction in current with plate voltage swing is really a feedback mechanism albeit an internal one.

Thank you. For me anyway, the measurements carry much more weight than some subjective 'sounds better' report.

 

[jderimig]

Thank you. For me anyway, the measurements carry much more weight than some subjective 'sounds better' report.

I understand, I used to be exactly there with you. Until I realized I don't listen to sine waves all day. My approach is to first get the amp to measure well then tweak till it makes me happy playing music while still measuring well.

My test if my feet don't involuntarily tap when listening to Neil Young's Southern Man then the amp is not right. Or other parts of my body moving with different musical material.

My first and current 1625 SE amp measures great. THD is below the resolution of my FFT. Its that with two forms of NFB which I can select by throwing a switch. One mode sounds better than the other. Both measure the same within my instruments capability. Then I go with my ears/brain. My speaker/room distortion is probably a couple orders of magnitude greater than my electronics.

 

[jderimig]

Tube curves and LTSpice (the Rocksim of electronics).

Tube audio is very popular in Japan and there is a very active community of tube amp developers there. In 2003 Ayumi Nakabayahsi published his method of modeling vacuum tube circuits with Spice. Prior to that the method was mixture of art and using basic published tube curves to locate basic operating points and tweaking from there. Since the Ayumi method tube circuits can be simulated pretty accurately.

Here is quick example of a Spice simulation of a vacuum tube. Below is a simple 6L6 model and we will model the tube's current characteristics in Pentode and Triode Mode.

FIrst Pentode Mode. Screen connected to a 275v dc supply.
On left pane is the circuit. On the right is current versus plate voltage at a family of equally spaced grid voltages. As you can see, the plate voltage only has a modest effect and grid voltage is the dominant control.



Next simulation is we are going to wire that tube as a triode by connect the screen to the plate. This is a classic triode curve where both the plate and grid voltages have strong effect on the current.



Next post I will talk about the practical effects these two sets of curves affect the performance of an audio amplifier.

 

[Hobie1dog]

You might as well be talking in Swahili as you lost me on the first post.

 

[FredA]

This is great stuff John - didn't know the spice models existed Too bad I don't read Japanese.
Keep up the posts

 

[Funkworks]

Next post I will talk about the practical effects these two sets of curves affect the performance of an audio amplifier.

Practical effects will be interesting. I wouldn't qualify as an audiophile today, but I still might get something out of this.

 

[jderimig]

This is great stuff John - didn't know the spice models existed Too bad I don't read Japanese.
Keep up the posts

In the Edge browser (default on my Windows laptop) a translate to English box happens...

Then, magic

 

[FredA]

Can you translate the pdf's?

 

[jderimig]

Can you translate the pdf's?

Unfortunately no, but the pictures are in english...

 

[jderimig]

Now we get to fun part and start landing the plane.
How do we get this wonderful audio signal power to our speakers?

Our speakers require AC power to create the pressure waves to our ears. The goal of any good audio power amplifier is to deliver power. AC power is simply the AC voltage swing X AC current swing.

The input signal to our tube comes from our pre-amp or DAC and is sent to the grid of the tube to swing the tube current up and down. Our goal is to maximize both the current and voltage swing at the plate of the tube. Below is our Pentode curve shown earlier with 3 options. Assuming we have a 540 volt power supply for the tube we would like operate on the green line because it provides the most voltage and current swing response to our input signal. The Red lines are also possible paths but the lower one doesn't swing current as much as the green and the upper doesn't swing as much voltage.

That line is called the load line and you will hear that alot if you are in a group talking about tube amplifier design. The load line is also an important topic with solid state also but not as much as with tubes as you will see.


What determines the load line is the load (speaker) you put on the tube. So what speaker load will give us this green line? So we see the current swing is about 210ma with a voltage swing of 400 volts. Its basically the slope of this line. So the load resistance is using ohms law: 400v/0.21A = ~2000 ohms. So just go out and buy a 2000 ohm speaker system and it will be perfect for this tube.

So ugh, we have a practical problem don't we?

 

[jderimig]

So the problem is there are not many nominal 2000 ohm speaker systems. So to use your favorite 8 ohm speaker we need to transform it to a 2000 ohm speaker to match it to the output tube. This is done by a heavy expensive transformer. Which is why tube amps have heavy expensive transformers. Now the quality of that transformer is very important to the sound. Good transformers are key. But in this thread we will not talk about transformers, this just explains why we need them with tube amps.

Interesting historical trivia. In the 1950's Philips introduced 600-800 ohm speakers to enable transformerless or much simpler transformer amplifiers. Also today you can get 600 ohm headphones. This allows the design of tube headphone amplifier without the need for a transformer.

Philips history here: https://www.vintage-radio.com/repair-restore-information/valve_philips-speakers.html

 

[Hobie1dog]

I have some Beyer dynamic DT990 headphones that I have owned for 30 years that are 600 ohm

 

[jderimig]

Why a triode or a pentode wired as a triode gets less power from a tube.

Take the 6L6 pentode curves shown earlier. Assume a 500v power supply and the red line is how much voltage swing we can get out of the tube. Almost 480volts


Now take that same 6L6 wired as a triode. The Red line is where we can operate that tube and we immediately see we can't pull down the plate voltage very low. We can only an AC swing of only 300v at a lower current at that voltage. Less voltage and current swing = lower power. But Triodes sound magical. Can we have our cake and eat it too? Next post.

 

[Hobie1dog]

I was going through a box today and found a pair of Tang band 4 in coax drivers which have a model number of W-4-937 s and I've never taken them out of the box. there's a YouTube video on these things, and at least played through headphones, they produce a marvelous sound on the video.

 

[jderimig]

I was going through a box today and found a pair of Tang band 4 in coax drivers which have a model number of W-4-937 s and I've never taken them out of the box. there's a YouTube video on these things, and at least played through headphones, they produce a marvelous sound on the video.

I was very close to looking at full range speaker project. Of course there are tradeoffs but some people swear by the tradeoff, no passive crossover, coherent sound through the midrange, great for vocals and jazz I have heard. I think coupled with active eq they could sound pretty nice.

 

[jderimig]

For those who may want to do their own measurements here is some well regarded software popular in the DIY space. Uses your PC soundcard to take all sorts of nice audio measurements, freq response, THD, speaker impedance etc. Shareware but only 39e for unlocked version...

http://audiotester.de/

 

[Hobie1dog]

I was very close to looking at full range speaker project. Of course there are tradeoffs but some people swear by the tradeoff, no passive crossover, coherent sound through the midrange, great for vocals and jazz I have heard. I think coupled with active eq they could sound pretty nice.

The sad part is that 90% of all "full-range" drivers are no where near full range. Most having very little output below 50hz. I've had several pairs through the years. ok for folk music, solo guitar, etc. They also normally only have 5mm of travel so they bottom out easily.

 

[jderimig]

The sad part is that 90% of all "full-range" drivers are no where near full range. Most having very little output below 50hz. I've had several pairs through the years. ok for folk music, solo guitar, etc. They also normally only have 5mm of travel so they bottom out easily.

Make an array of about a dozen per tower....

 

[Hobie1dog]

Make an array of about a dozen per tower....

the last set of speakers I built

 

[John Kemker]

This is my DIY turntable shelf for better isolation from an unstable floor. Two shelf brackets from Mayfield Ace in Covington, GA, four speaker isolation spikes from Amazon and a Mapleshade Audio platform. Heavy duty anchor bolts in the wall, the spikes in the horizontal portion of the brackets with the metal discs attached under the maple platform. I can do jumping jacks, now, without upsetting the turntable.

Since I already owned the Mapleshade platform, the rest of it was cheap compared to buying a premade platform from Amazon or some other source.

EDIT: Forgot to mention, the spare tonearms are attached to the wall with a tool holder from Amazon. Protects the tonearms and makes it handy to swap them out! (Schiit Sol doesn't have interchangeable headshells, like some other tonearms, but my first spare tonearm wasn't expensive and the second on was free. It's real easy to swap the entire tonearm to use a different cartridge.)

 

[FredA]

A spare tonearm rack is a first for me....awesome!

 

[jderimig]

The UNSET tube final stage. So here is the "new" tech to address the triode power problem. The inventor calls this the CED (Compound Electronic Device) which consists of a Pentode and a Mosfet. The grid of the tube is fed signal from the plate using a voltage divider to provide local feedback. Drive to the tube is done at the cathode using the Mosfet as a voltage follower. The result is:
* triode characteristic curves
* able to pull the plate voltage close to zero which yeilds pentode power levels with triode curves.
* driving the cathode with mosfet dispenses with the cathode resistor which dissipates power which is wasted.