2024, Chapter 4
Not Just a Pretty Face
I’ll admit it: Vali 3 was almost just a cosmetic update.
I mean, it
is pretty. Just look at it:
Oh yeah, sure, you could say it wasn’t just cosmetic. I had to completely change the board layout, to move to the new standard established by Magni+. The headphone jack went to the other side. The power supply was completely reconfigured. The gain switch moved to the front.
But, bottom line, the planned “Vali+” was the Vali 2++ in a different box. The BOM was exactly the same, except for chassis pieces.
In one way, that felt totally right. Vali’s evolution has been a positive one, and in Vali 2+ or 2++ form, it was the quietest, most flexible and capable inexpensive tube hybrid amp out there. It was a super-solid intro to the world of tubes, for a hundred and a half dollars. It was very polished and established, a super-safe choice.
But:
really just the same thing in a pretty box?
Nah.
I
couldn’t do that.
And with that decision, I started an odyssey that consumed months of time, and led down some very crazy rabbit-holes. Which is why the saying goes, “If it ain’t broke, don’t fix it.”
Continuous Improvement
Anyone new around here is probably saying, “wait a sec, what’s this deal with Vali versions, why is there a stupid thing called a 2++, and what the heck is going on? Primer please?
Okay, here we go. Since inception, Vali has had 4 inceptions:
- Vali. The original Vali used two 6088 “pencil” tubes inside the chassis. It had issues with microphonics, and had issues with tube matching, but with sorbothane and hand matching, they worked well, and sounded very good. Used a 45V tube rail, which was about all the 6088s could take. It was still a very simple amp, with a single gain level (about 4) and no preamp outputs. We still repair them from time to time, but in general they seem very reliable. Replacing original Vali tubes is still $20 from us.
- Vali 2. Vali 2 broke from the original Vali in a big way. It got an external 6BZ7 tube to facilitate tube rolling, a new gain stage (the not-yet-named Coherence™), a gain switch, 60V tube rail, preamp outputs, and a whole new look. Really a nice little amp, though it could be a bit noisy (hummy) with some tubes.
- Vali 2+. Evolved the Vali 2 formula with lessons learned from Magni 3+: added a driver stage, redid the tube heaters, upped the tube rail to 70V, and tweaked a bunch of small things. The result was a super-quiet, super-refined tube hybrid amp we could have just kept making forever. Except…we used the universe’s supply of inexpensive 6BZ7 tubes. Sooo…
- Vali 2++. It’s a Vali 2+ that uses a 5670-style tube, which we can still get inexpensive variants of. It also allows the clinically insane, like myself, to run stupid things like WE396As without socket adapters. It has a stupid name because the chassis are labeled “Vali 2,” so going to “Vali+” would have been weird.
So, when we came up with the idea of the “advanced form” or “recurve” chassis on Magni+ and Modi+, I figured it would be a good idea to see if it would work on Vali as well. This is basically engineering boredom—there was no need for this, other than to make the line look the same.
It did, however, give me a chance to really up the chassis game.
I mean, we’re really getting into this stamping thing, which allows us to make much more interesting shapes. And “interesting,” in terms of tubes, can go far beyond “interesting.” It can help us solve a problem that has plagued tube designs since the beginning—the problem of heat dissipation.
Tubes run hot. That frequently surprises people new to tubes. But to engineers, it isn’t surprising at all. Every tube has a component called a “heater.” And it does exactly what it says: it heats the tube. Specifically, it heats the cathode, to create the cloud of free electrons that enables a tube to work.
Aside: the British term “valve” is a better description of a tube, because that’s basically what it is. By varying the voltage on a control terminal (the grid) you can control the flow of electrons from the cathode to the anode. So it’s a valve. A thermionic valve. Makes total sense.
Aside to the aside: the sharp-eyed will have noticed I mentioned “the flow of electrons from the cathode to the anode,” which, in most electronic notation these days, is backwards. In actuality, it’s 100% correct. You get positive potential from a shortage of negative charge—as in, a shortage of electrons. So the electrons would flow from the negative side (the cathode) to the positive side (the anode), which is opposite the convention we use today. And that’s OK. No need to panic. Everything keeps working; no need to turn your batteries around.
But, back to heaters, and tubes getting hot. Most small signal tubes will dissipate something between 2 and 4W (watts) as pure heat, just from the heater. They’ll also dissipate more watts from the current flowing through the tube, so it’s not unusual for a small signal tube to dissipate 5-8 watts at idle, just sitting there. Compared to a modern IC at 50-400mW, yeah, tubes are hot.
And, again, this is fine. Hot electronics aren’t necessarily bad, unless they aren’t designed to withstand the heat. Some modern discrete devices are rated to run 175 degrees C continuous. That’s crazy hot. But if you keep them away from heat-sensitive parts, like electrolytic capacitors, everything is fine.
Aside: cue the “cooking the capacitors, they’re only rated for 2000 hours!” comments. Wrong. Look into capacitor derating, and you’ll find that most stuff run at 45 degrees C or so, even with bog-standard 85 degree rated caps, is gonna go 50,000-100,000 hours without an issue. That’s a loooooooooooooooong time. Like turn it on and leave it on for 5-10 years time. I have tons of electronics from the early 1990s that have never been “recapped” and operate just fine.
But hey, when you can get rid of heat more effectively, that’s great. It’s just that cooling a component usually comes at a cost—a cost in terms of heatsinking, or size, or fans.
But when you can do it for free—you do it, absolutely.
And that’s what the sophisticated stamping and recurve chassis allowed us to do. Because we weren’t stuck in a 2D folded metal world anymore. We could create complex shapes. Shapes that allowed us to more fully expose the tube and keep heat out of the chassis. Shapes, to be frank, inspired by Lyr+, and its elaborate “funnel” to help keep it cool.
What’s more, we could do it at no cost, by stamping the shape into the chassis itself. That’s a huge win.
So I was really excited about that, and then inspiration struck: what if we moved away from the standard perforation pattern we’d been using since, like forever? What if we did something more expressive, with slots, like we did on Folkvangr?
Yeah. And holy moly, the next Vali really turned into a looker. I personally think it’s one of the best looking amps we’ve done at any price, especially when used with a shorty NOS 5670 or WE396A tube.
And for a while, I was happy with Vali just being a pretty face. I put a “Vali+” on the board, and prepped to introduce this new, cooler-running, cooler-looking Vali and call it a day.
But then I got bored.
What-Ifs and Dead Ends
One of the most horrible and wonderful things about engineering is that there’s always something to do. Always something new to design, or always something you can apply new learning to.
So, while we started planning for a cosmetically-pretty-but-otherwise-the-same Vali+, I started wondering:
Is this enough?
Surely I could do more?
Nah, it looks really cool.
But what about higher volts?
What about DC coupling?
What about…well, just a few tweaks?
And so that’s how I found myself, several months ago, putting aside a bigger, gnarlier project to do a quick redesign of the Vali+, a redesign that gave it several features I always wanted, including a higher voltage tube rail (100V) and DC coupling for the input and output.
This was a bit of a hail mary. It was more complicated (about 20 more parts, not insignificant in a simple design like Vali), it was space-challenged, and it had some guesswork on my part that hadn’t been fully tested.
But the boards came in, and I excitedly stuffed them, hoping to reach even higher levels of sonic nirvana with this new supercharged Vali.
Except…
…except it went pretty much as you expected: it didn’t work.
A couple of my assumptions were bad. One of them was critically bad, and did not allow this new version of Vali to run with a full 2V RMS input. It simply didn’t have enough headroom in the front end to make it work, at least not at unity gain.
Now, there are ways around this, of course, because we worked around it with Lyr 3 and Lyr+. Those products are wayyyyyyy more complicated, though. So I stared at it a bit, guessed at a couple of things that might approximate what Lyr 3 was doing, and threw one on the board for the next prototype.
Aside: for those wondering “why don’t I build these guesses in the air, rather than going for a board?” it’s simple: boards of Vali 3’s level are very inexpensive and very fast to get these days. A lot of times it’s better just to go straight to board.
Aside to the aside: for those wondering, “why don’t I simulate this before doing anything physical at all?” excuse me for being old skool, and for preferring to find out what’s gonna go whomper-jawed on the board right away, rather than being surprised later.
The bad news is that didn’t work either.
So I went on to another prototype, which was much closer to Lyr+, and also more parts. The bad news started in layout. It was now complex enough that it needed to be a 4-layer board, rather than the 2-layer of the original Valis. Not a dealbreaker these days, but also more work.
The bad news continued in prototype: yeah, this worked, but it had two achilles heels: high noise floor, and popping when switching gains. The way I’d stacked the supplies caused too much residual noise, and it was audible. Worse, switching gain resulted in a massive pop—as in, like about half a volt. Not acceptable.
Now, there were other ways around these problems, but I’d begun to remember why Lyr 3 and Lyr+ were big and complex: because it had big and complex work-arounds for big and complex issues. One choice was to essentially make Vali 3 exactly the same as Lyr 3…except that it might not have fit in the chassis at all, and pushed the price up significantly.
Sigh.
Another choice was to scale back my expectations. Higher rail voltage is great for tubes—they run more linearly at higher voltages. DC coupling is also nice, but very hard to achieve all the way through with tubes. Lyr 3 and Lyr+ are the only tube products we’ve made that achieve this feat, and Lyr+ is the only one to go completely capacitor-free, even in the feedback network, and that’s done only with the help of a bunch of sensors and microprocessor oversight.
So maybe lose the DC coupling? That would get rid of the huge complexity of an additional bipolar supply. I could still stack the power supplies to get a higher tube rail.
And that’s what I did: I gave up on DC all the way through with Vali 3. And, in doing so, it allowed a stunning simplification. As in, Vali 3 is only a couple of different capacitors and an extra diode away from a Vali 2++…but those few extra components mean it gets 100V at the tube anode, rather than 70V!
That’s a huge change for very small investment!
What’s more, it means Vali 3 can sell for exactly the same price as Vali 2++. So, despite years of inflation from the original 2014 Vali 2, and despite all the changes that make it a quieter, more capable amplifier, Vali 3 actually sells for less than Vali 2!
Aside: Yes. The original Vali 2 was $169. We eventually lowered the price to $149 when our production got high enough. And the all-new Vali 3 stays at $149. Yes, even in these inflationary times.
Now, of course, upping the voltage on the tube also changes the gain, and the operational point, and lots of other things, so I ended up tweaking a bunch of values. We ended up with an amp that had slightly lower overall gain, but pretty much the same distortion. So a little more linear—but lower feedback means about the same in the end.
So What’s the Big Deal?
The big deal with Vali 3 is that it’s an even better way to try out tubes, and see if they’re something you want to get into, at a very low price. It’s quite simply the most serious, most capable entry-level tube hybrid amp out there. Nothing else has a serious 100V rail, nothing else has Vali 3’s high power output, nothing else has its extremely low noise floor, nothing else is backed by a company that’s been around for 13 years, making things in the USA, and keeping prices as low as possible.
Now, we can get into tons of arguments about “well, tubes don’t measure good, and therefore they are not good,” which yeah, you can accept as gospel and never question. Or, you can get a Vali 3, and for a hundred and a half bucks, listen for yourself, and decide if tubes are for you.
The result might surprise you. Yes, even with 1000x the distortion of a Magni. Yes, even with a glaring bit of Luddite-itry sticking out of it. Yes, even if it doesn’t make any sense.
Oh yeah, and in case it matters…
it’s also a looker!
