Schiit Happened: The Story of the World's Most Improbable Start-Up
- Thread starter Jason Stoddard
- Start date
-
- Tags
- schiit audio schiit-audio
wink
His amps are made out of recycled beer cans
and his source from tomatos.
- Joined
- Apr 13, 2009
- Posts
- 20,312
- Likes
- 3,656
bigro
1000+ Head-Fier
- Joined
- Dec 18, 2014
- Posts
- 1,056
- Likes
- 538
The Slowest Part of ordering from schiit is fedex. They have devised a way teleport your pile of schiit right to your door.
artur9
Headphoneus Supremus
- Joined
- Apr 25, 2015
- Posts
- 3,817
- Likes
- 10,547
Tuco1965
Headphoneus Supremus
- Joined
- Dec 28, 2012
- Posts
- 4,545
- Likes
- 192
jsiegel14072
100+ Head-Fier
- Joined
- Aug 29, 2014
- Posts
- 256
- Likes
- 225
RazrLeaf
New Head-Fier
- Joined
- Jan 6, 2015
- Posts
- 44
- Likes
- 29
I suspect that my Christmas Gift to myself is about to be released...
bigro
1000+ Head-Fier
- Joined
- Dec 18, 2014
- Posts
- 1,056
- Likes
- 538
My Birthday Gift to me was a BiFrost Multi Bit. I hope I don't have to explain to my wife my self gifting again. I do give my self the best gifts however.
FrivolsListener
500+ Head-Fier
- Joined
- Jul 22, 2015
- Posts
- 909
- Likes
- 1,099
My Christmas gifts to myself were a Gumby and Sennheiser HD-800s.
tonykaz
500+ Head-Fier
- Joined
- Feb 2, 2014
- Posts
- 778
- Likes
- 388
Ugly stuff brewing up lately,
I'll stand with Schiit
Tony in Michigan
I'll stand with Schiit
Tony in Michigan
AudioBear
Headphoneus Supremus
- Joined
- Mar 3, 2007
- Posts
- 1,656
- Likes
- 1,021
Tony
What ugly stuff? Schiit is beautiful....
What ugly stuff? Schiit is beautiful....
Dalgas
Head-Fier
- Joined
- Jul 11, 2015
- Posts
- 82
- Likes
- 25
No surprice there! My experience is that most old cd recordings are far more fun than most new cd-recordings - and even surpasses most 24/96 og 24/192 recordings. My favorite reference cd is "Tormé" by Mel Tormé in a remastered cd-version of the original 1958 Verve album. It don´t get much better than that!
I have just bought a 30 years old Philips CD 304 mkII - with a CDM1 drive! It is totally recapped and with a new TDA1543 DAC-section. Oh boy - it is heaven!
Regards from another "fossil"
Jason Stoddard
Sponsor: Schiit Audio
2015, Chapter 19:
The Most Difficult Design Brief
Okay, so let’s imagine for a moment that amplifier design briefs are conducted kinda like hiring a sketchy spaceship on a certain desert planet—as in, at a table in a seedy bar, with both the marketing wonk and the engineer keeping their cards as close to their chests as possible. Such a design brief might go something like this:
“I need an amp,” says the marketing wonk, casually, as if the company can survive on nice thoughts and well-wishes, as if they don’t really need, well, like, a product—a damn good product—that people actually want to buy.
The engineer sighs. “What kinda amp?” he answers, after a time, knowing that there are a thousand ways to give the marketing wonk what he wants.
“A good amp, and cheap,” replies the marketing wonk.
“How many watts, how many channels, powered by what, what application, how big, what kinda numbers you want, THD, noise, you know…details?” the engineer rattles off, crossing his arms.
“It’s a headphone amp, maybe a watt, and small, desk-friendly,” the marketing wonk says.
“That’s doable,” the engineer says, thinking, this is a cakewalk, throw one of the big TI headphone drivers in a box with an op-amp and wall-wart and I’m done in time for dinner.
“But I was thinking, well, you know, about something special, like fully discrete,” adds the marketing wonk, not looking up from his drink.
The engineer clears his throat and says nothing, thinking: Now you’re talking real amplifier design. Thermal stability. Bode plots. Testing with reactive loads. Tons of parts. Only a company like Schiit would be crazy enough to do something like that on an inexpensive desktop amp.
“Price?” the engineer asks, finally.
“Hundred and fifty, hundred and seventy, tops,” the marketing wonk says, looking hopeful.
The engineer keeps his poker face, quickly calculating the cost of a complex discrete product with a ton of parts. “Anything else? You want a DAC in there too? Navigation system? Slow-cooker and microwave? Remote control?”
“No, no,” the marketing wonk waves his hand, as if to cut through the engineer’s rapid-fire onslaught of ideas. “Except…”
“Except what?” the engineer prompts.
“Except, we were also thinking, well, like, maybe a tube…” the marketing wonk trails off, smiling hopefully at the engineer.
“A tube?” the engineer asks, speaking very slowly, as if to a child. “As in, a vacuum tube? As in, you want this to be a tube amp?”
The marketing wonk nods.
The engineer, laughing uproariously, pulls out his blaster and blows the marketing wonk to atoms.
The Challenge of Inexpensive Tube Amps
Okay. I had my fun. Let’s now get down to it. And that “it” is this: inexpensive tube amps are a huge design challenge. That is, if you want to do them right.
And they’re an even bigger challenge if you want to do them right, make them flexible, and make them act like, well, a modern solid-state amp.
But I’m getting ahead of myself. First, let’s talk about what makes a tube amp so expensive. Tube amps are fundamentally different than solid state amps, and many of those fundamental differences equate to higher cost. Things like:
Output Transformers. Many tube amps use output transformers to drive low-impedance headphone loads, or speakers. Tubes are fundamentally high-output-impedance, high-voltage, low-current devices (in the context of 32 ohm headphones or 8 ohm speakers), and output transformers translate high voltage and low current to lower voltage and higher current. One problem: output transformers are not cheap—in fact, good output transformers usually cost much more than a power transformer. Which, in itself, is one of the most costly things in any product. So, add the cost of two good output transformers to an amp design, and watch the MSRP skyrocket. That’s why many inexpensive tube amps are OTL, or Output Transformer-Less designs. Like Valhalla 2. Through careful selection of tubes and design of the output stage, Valhalla 2 can drive many low-impedance, high-sensitivity headphones credibly…but it will never be able to drive them like a tube amp with output transformers can. Luckily, Valhalla 2 is absolutely stellar with high-impedance headphones, which need less transformer magic.
Tube Outputs. And, let’s face it, simply using output tubes (rather than solid-state outputs in a hybrid amp) is costly. The tubes can be expensive themselves, plus you’ll need a lot more heater current to run them (see below). That’s why Step 2 in cutting tube amp cost is usually losing the output tubes and making the design a hybrid. There’s no shame in this—a well-designed hybrid can be very, very good. But it is, in general, a cost-saving measure.
High Voltage Supplies. To run tubes right, you need high voltages—think hundreds of volts. Tubes need a lot of voltage to run linearly and give their best performance. This means high-voltage capacitors, custom transformers, high-voltage-rated parts, maybe even through-hole parts exclusively, depending on the voltages you’re shooting for. When you’re running 200V rails to get 100V on the plate (like we do in Valhalla 2, Lyr 2, and Mjolnir 2), the whole design is going to be more costly than an amp running 30V or so (a typical solid state design). Unfortunately, many inexpensive tube amps are “starved plate” designs, running at 12-24V…which is a recipe for high distortion and a very colored amp design.
Tube Heaters. And the power supply keeps getting more complex when you talk tubes. You also have to take into account the tube heaters, and the current you need to run them. Hey, those glowing tubes glow because they are actually hot. It’s not cosmetic. Tube heaters do exactly what you’d think: they heat up the cathode so there’s a convenient cloud of electrons to make them work. No heater, no worky. The problem is, tube heaters can draw 300-600mA of current at 6.3V (AC or DC)—each. That’s automatically 2-4W of power dissipated just in the heater of a single tube if you’re running AC heaters…and even more if you’re running DC heaters. This is a pretty beefy power supply in itself. As far as we know, no inexpensive tube amp has ever just thrown a tube in there and lit it up with an LED…which would make the tube literally just for show.
So now you know a little bit about why an engineer might moan and groan when asked to design a good, inexpensive tube amp. Sometimes, the “good” and “inexpensive” can seem to be mutually exclusive.
“And yet you did a pretty good job with Vali,” some of you are saying. “That’s a helluva good sounding little tube hybrid, for hella cheap.”
Yes. And it did sound very, very good.
That is, if your cans happened to fit a fairly narrow profile (relatively inefficient, full-sized headphones that did not reveal the high noise floor), and if you were tolerant of a big design foible with the subminiature tubes (high microphonics), and if you didn’t need a preamp out, or gain switching, or wanted to roll tubes…
Yes, it sounded great. But it was a very limited amp. It wasn’t for everyone. It was no substitute for the do-all Magni 2 and Uber.
Which is why I started to ask myself: What if we could make an inexpensive tube amp…without the limitations of Vali?
Yes, I know. Get out the blaster.
Adventures in Seconding the Vali
As I mentioned, Vali had plenty of obvious things we could do to improve it. Adding a gain switch and preamp outputs were the two most top-of-mind. But simply adding a gain switch and preamp outputs wouldn’t alter the basic character of Vali…and it wouldn’t address the real limitation of the design, which was the microphonic tubes.*
*Although we got very good at sorting out the microphonic Valis and keeping them out of customer hands, they were a huge liability in the shop. First, the tubes had to be tested, sorted, and matched, then installed in a product, then burned-in for a few days (an unusual amount of time for an inexpensive product), and then, when they failed—usually by continuous ringing—they had to be manually desoldered, replaced, and the whole process started all over again. When a run of 1000 Valis yielded 150 that had to be reworked—after pre-sorting the tubes—you can begin to understand why this was such a disruption.
So, it was apparent, very early on, that the best way to improve Vali would be to change the tube.
But what tube? And what would be the ramifications of changing tubes?
The original Vali used an unusual tube: the 6088 subminiature pentode, in triode-strapped mode. These submini pentodes were odd in many ways:
For the purpose of choosing a new tube, the first two oddities are the most important. Could we find a tube that worked at Vali’s low plate voltage (about 45V) and wouldn’t cause the power supply to self-immolate with the amount of heater current it needed?
Furthermore, could we maybe run the heaters via AC? We had a 14VAC wall-wart. A small drop would get that into the range where two series 6.3VAC heaters would work, or 12.6V heaters in parallel.
We started out by looking at some very similar tubes—different subminiature pentodes, specifically Russian rod pentodes. Rod pentodes were close enough in terms of heater and plate voltage to drop in and tweak. These devices have one huge advantage over Vali’s old tubes: they are pretty much non-microphonic. However, they were also noisier. And their operating point wasn’t super-conducive to a headphone amp (higher gain, hard to get rid of the gain and noise because of the direct heated cathode.) And, they were very high distortion.
So we moved on. We looked at some very cool subminiature Russian triodes…but those needed 400mA of heater current at 6.3V. This would be a helluva load on our power supply, and, with the tube inside the chassis, the Vali could substitute as a portable grille (seriously, you’re looking at 60 degrees C or so).
So we moved on again. This time, I started thinking, “You know, what we really need to do is hang the tubes outside the chassis, then the heat from the heater won’t be a real problem. So, we looked at some 7-pin pentodes that only used about 150mA of heater current each. That could be viable, but I was nervous about the supply of those NOS tubes.
So we moved on again, looking at different tubes (12AU7, etc) that worked well at low volts and had 12V heaters. I didn’t like the performance of those tubes, and their new-production nature meant those tubes would be costly.
Deep into the project, I was stuck. The real problem was the power supply. Using a single 14VAC wall-wart, yes, I could step it down to run 12.6V heaters, and use a linear voltage multiplier for a high-voltage rail, or I could use a switching supply to derive a regulated 6.3VDC for a 6.3V tube, but neither of those options appealed. What I really needed was a power supply like I used in the higher-end amplifiers, with multiple AC voltages for multiple purposes. Making one single AC voltage try to do everything was an exercise in compromise.
But when you’re using a wall-wart, you really could only get one AC voltage on its barrel jack. That’s the rub. Two connection terminals = one AC voltage. Done and done.
At that point, I almost decided to leave well enough alone. After all, Vali was selling well. It was a good amp for many headphones. It just wasn’t really, well, universal or flexible.
Except…
What if our wall-wart guys could do a wall-wart with two AC voltages? And have something like a DIN plug connector, rather than a barrel jack?
If they could do that, then I could run the tube heaters off AC at 6.3V, so there wouldn’t be any power dissipated by DC regulators…and also get an AC voltage high enough to rectify or multiply into a proper tube plate voltage.
Now, I’d never seen a wall-wart like that before, but it was worth a query. I fired off an email, expecting to receive a “sorry, we can’t do that,” response. Instead, I got a picture showing a wall-wart with a DIN plug connector. “Like this?” the email asked.
Aha, now we were talking! I arranged to get a sample of the wall-wart made up.
A Vali With 2 Wall-Warts….and a Bizarre Gain Stage
While I was waiting for the wall-wart, I realized that I could get an idea of the performance of the power supply simply by using one of our 6VAC wall-warts and one of our 16VAC wall-warts with a voltage multiplier…and so that’s how I started hacking up a Vali.
And, while I was at it, I figured, Why not just use a 6BZ7 tube, like Lyr 2?
We had an excellent supply of them, and while it might look silly sticking out of a small amp like Vali 2, it would be a good test case to see what we could do.
And it’s a good thing I started there, because it quickly showed me one of the limitations of even a two-AC-voltage design: there would have to be some compromise on how high a plate voltage we could run. 100V like the Lyr 2 or 120V like the Valhalla 2 just wasn’t going to work…because we also needed that same AC voltage to supply Vali 2’s solid-state output stage.
So, I started looking on how we could maximize voltage on the plate. What we ended up with is a truly bizarre combobulation of a tube and a PNP transistor (this is a hybrid, remember). But the transistor is NOT the plate load…the plate load is still a resistor, though a much smaller one than you might expect. I haven’t seen this topology elsewhere, but that doesn’t mean that someone hasn’t done this before. Most good analog ideas have.
So what does this weird gain stage get us? It gets us a full 60V on the plate, which is plenty to get good performance out of 6DJ8 tubes and tubes in the same family. Now, 60V may not sound like a lot, but in the context of inexpensive amps that sometimes use 6-12V on the plate, it’s a huge, huge difference. It also means we can fully swing the discrete output stage into clipping, so the output stage is highly efficient. Also, with a new biasing arrangement, the output emitter resistors are almost 5X smaller than some of our other amps, for lower output impedance.
Sounds like a ton of design and optimization work for a very simple stage, right? That’s because it was.
And, it wasn’t over even after I’d gotten the first dual-voltage wall-wart and PC board prototypes for Vali 2. Because, in the process of testing which tubes we could use in the design (besides the stock 6BZ7), I found that the new combobulated tube-PNP hybrid gain stage ran at different operating points for each tube we used. Some were just a little bit off…but some were different enough that the amp’s power output would be affected.
So, I added a bias-centering servo. Which makes the little Vali 2 a fairly complex example of a modern hybrid: a tube for voltage gain, discrete bipolar transistors for output stage, an integrated circuit op-amp for bias control…on top of a dual-AC-voltage, high-voltage+heater power supply with regulated rails for both the tube and bipolar output stage.
Sounds complex? Yes. To achive simplicity and flexibility in use, internal complexity is usually required. Not that this is a bad thing, nor is Vali 2 overly complex (in fact, it uses only about 10 more parts than the original Vali.
A Thoroughly Flexible, High-Performance, Affordable Tube Hybrid
So what’s the end result of all of this design work? An amp that is, by all measures, a significant step up from Vali. It has:
So how does it sound? Well, Mike, I, and the rest of the (formerly skeptical) shop think it leaves Vali in the dust. But the final judgment, of course, is yours.
And when you get right down to it, sound or not, flexibility or not, the main thing I hope to communicate with this chapter is simple: amp design isn’t simple or easy—even if it’s “just” a revision of an existing product. When you go off the reservation and start talking about discrete design, or tubes, it’s a much bigger undertaking than you might think.
Perhaps even more interestingly is that in Vali 2’s case, there was no design brief. No discussions in seedy off world bars. Hell, we didn’t even have the amp in the overall “2015 plan.” It was always a, “Well, let’s see what we can do with this.” As I mentioned, I came close to simply letting Vali soldier on for a few more years. But in the end, the relatively slow pace of development put me in a “no pressure” mode that allowed me to explore some really different designs…from the weirdo wall-wart to the combobulated gain stage.
And in the end, I think Vali 2 is really a big step up for inexpensive tube amps. But hey, I’m biased.
The Most Difficult Design Brief
Okay, so let’s imagine for a moment that amplifier design briefs are conducted kinda like hiring a sketchy spaceship on a certain desert planet—as in, at a table in a seedy bar, with both the marketing wonk and the engineer keeping their cards as close to their chests as possible. Such a design brief might go something like this:
“I need an amp,” says the marketing wonk, casually, as if the company can survive on nice thoughts and well-wishes, as if they don’t really need, well, like, a product—a damn good product—that people actually want to buy.
The engineer sighs. “What kinda amp?” he answers, after a time, knowing that there are a thousand ways to give the marketing wonk what he wants.
“A good amp, and cheap,” replies the marketing wonk.
“How many watts, how many channels, powered by what, what application, how big, what kinda numbers you want, THD, noise, you know…details?” the engineer rattles off, crossing his arms.
“It’s a headphone amp, maybe a watt, and small, desk-friendly,” the marketing wonk says.
“That’s doable,” the engineer says, thinking, this is a cakewalk, throw one of the big TI headphone drivers in a box with an op-amp and wall-wart and I’m done in time for dinner.
“But I was thinking, well, you know, about something special, like fully discrete,” adds the marketing wonk, not looking up from his drink.
The engineer clears his throat and says nothing, thinking: Now you’re talking real amplifier design. Thermal stability. Bode plots. Testing with reactive loads. Tons of parts. Only a company like Schiit would be crazy enough to do something like that on an inexpensive desktop amp.
“Price?” the engineer asks, finally.
“Hundred and fifty, hundred and seventy, tops,” the marketing wonk says, looking hopeful.
The engineer keeps his poker face, quickly calculating the cost of a complex discrete product with a ton of parts. “Anything else? You want a DAC in there too? Navigation system? Slow-cooker and microwave? Remote control?”
“No, no,” the marketing wonk waves his hand, as if to cut through the engineer’s rapid-fire onslaught of ideas. “Except…”
“Except what?” the engineer prompts.
“Except, we were also thinking, well, like, maybe a tube…” the marketing wonk trails off, smiling hopefully at the engineer.
“A tube?” the engineer asks, speaking very slowly, as if to a child. “As in, a vacuum tube? As in, you want this to be a tube amp?”
The marketing wonk nods.
The engineer, laughing uproariously, pulls out his blaster and blows the marketing wonk to atoms.
The Challenge of Inexpensive Tube Amps
Okay. I had my fun. Let’s now get down to it. And that “it” is this: inexpensive tube amps are a huge design challenge. That is, if you want to do them right.
And they’re an even bigger challenge if you want to do them right, make them flexible, and make them act like, well, a modern solid-state amp.
But I’m getting ahead of myself. First, let’s talk about what makes a tube amp so expensive. Tube amps are fundamentally different than solid state amps, and many of those fundamental differences equate to higher cost. Things like:
Output Transformers. Many tube amps use output transformers to drive low-impedance headphone loads, or speakers. Tubes are fundamentally high-output-impedance, high-voltage, low-current devices (in the context of 32 ohm headphones or 8 ohm speakers), and output transformers translate high voltage and low current to lower voltage and higher current. One problem: output transformers are not cheap—in fact, good output transformers usually cost much more than a power transformer. Which, in itself, is one of the most costly things in any product. So, add the cost of two good output transformers to an amp design, and watch the MSRP skyrocket. That’s why many inexpensive tube amps are OTL, or Output Transformer-Less designs. Like Valhalla 2. Through careful selection of tubes and design of the output stage, Valhalla 2 can drive many low-impedance, high-sensitivity headphones credibly…but it will never be able to drive them like a tube amp with output transformers can. Luckily, Valhalla 2 is absolutely stellar with high-impedance headphones, which need less transformer magic.
Tube Outputs. And, let’s face it, simply using output tubes (rather than solid-state outputs in a hybrid amp) is costly. The tubes can be expensive themselves, plus you’ll need a lot more heater current to run them (see below). That’s why Step 2 in cutting tube amp cost is usually losing the output tubes and making the design a hybrid. There’s no shame in this—a well-designed hybrid can be very, very good. But it is, in general, a cost-saving measure.
High Voltage Supplies. To run tubes right, you need high voltages—think hundreds of volts. Tubes need a lot of voltage to run linearly and give their best performance. This means high-voltage capacitors, custom transformers, high-voltage-rated parts, maybe even through-hole parts exclusively, depending on the voltages you’re shooting for. When you’re running 200V rails to get 100V on the plate (like we do in Valhalla 2, Lyr 2, and Mjolnir 2), the whole design is going to be more costly than an amp running 30V or so (a typical solid state design). Unfortunately, many inexpensive tube amps are “starved plate” designs, running at 12-24V…which is a recipe for high distortion and a very colored amp design.
Tube Heaters. And the power supply keeps getting more complex when you talk tubes. You also have to take into account the tube heaters, and the current you need to run them. Hey, those glowing tubes glow because they are actually hot. It’s not cosmetic. Tube heaters do exactly what you’d think: they heat up the cathode so there’s a convenient cloud of electrons to make them work. No heater, no worky. The problem is, tube heaters can draw 300-600mA of current at 6.3V (AC or DC)—each. That’s automatically 2-4W of power dissipated just in the heater of a single tube if you’re running AC heaters…and even more if you’re running DC heaters. This is a pretty beefy power supply in itself. As far as we know, no inexpensive tube amp has ever just thrown a tube in there and lit it up with an LED…which would make the tube literally just for show.
So now you know a little bit about why an engineer might moan and groan when asked to design a good, inexpensive tube amp. Sometimes, the “good” and “inexpensive” can seem to be mutually exclusive.
“And yet you did a pretty good job with Vali,” some of you are saying. “That’s a helluva good sounding little tube hybrid, for hella cheap.”
Yes. And it did sound very, very good.
That is, if your cans happened to fit a fairly narrow profile (relatively inefficient, full-sized headphones that did not reveal the high noise floor), and if you were tolerant of a big design foible with the subminiature tubes (high microphonics), and if you didn’t need a preamp out, or gain switching, or wanted to roll tubes…
Yes, it sounded great. But it was a very limited amp. It wasn’t for everyone. It was no substitute for the do-all Magni 2 and Uber.
Which is why I started to ask myself: What if we could make an inexpensive tube amp…without the limitations of Vali?
Yes, I know. Get out the blaster.
Adventures in Seconding the Vali
As I mentioned, Vali had plenty of obvious things we could do to improve it. Adding a gain switch and preamp outputs were the two most top-of-mind. But simply adding a gain switch and preamp outputs wouldn’t alter the basic character of Vali…and it wouldn’t address the real limitation of the design, which was the microphonic tubes.*
*Although we got very good at sorting out the microphonic Valis and keeping them out of customer hands, they were a huge liability in the shop. First, the tubes had to be tested, sorted, and matched, then installed in a product, then burned-in for a few days (an unusual amount of time for an inexpensive product), and then, when they failed—usually by continuous ringing—they had to be manually desoldered, replaced, and the whole process started all over again. When a run of 1000 Valis yielded 150 that had to be reworked—after pre-sorting the tubes—you can begin to understand why this was such a disruption.
So, it was apparent, very early on, that the best way to improve Vali would be to change the tube.
But what tube? And what would be the ramifications of changing tubes?
The original Vali used an unusual tube: the 6088 subminiature pentode, in triode-strapped mode. These submini pentodes were odd in many ways:
- They didn’t require a very high plate voltage—30-40V would be fine for them. You could probably even get away with a 24V supply and 12V plate, as some inexpensive amps do.
- They didn’t require a lot of power for the heaters—only 20mA at 1.25V (yeah, only 25mW to the heaters…big difference from 2W for a 6DJ8-style tube.)
- They were direct heated, so the cathode and the heater were inextricably intertwined, which led to some limitations when it came to looking at a way to add a gain switch to the amp.
- They were inherently noisy (hiss), which limited their application as a preamp stage.
- They were very microphonic (diiiinnnnnnnnnnngggggggggg….) which, as above, limited their application as a preamp and for sensitive headphones.
For the purpose of choosing a new tube, the first two oddities are the most important. Could we find a tube that worked at Vali’s low plate voltage (about 45V) and wouldn’t cause the power supply to self-immolate with the amount of heater current it needed?
Furthermore, could we maybe run the heaters via AC? We had a 14VAC wall-wart. A small drop would get that into the range where two series 6.3VAC heaters would work, or 12.6V heaters in parallel.
We started out by looking at some very similar tubes—different subminiature pentodes, specifically Russian rod pentodes. Rod pentodes were close enough in terms of heater and plate voltage to drop in and tweak. These devices have one huge advantage over Vali’s old tubes: they are pretty much non-microphonic. However, they were also noisier. And their operating point wasn’t super-conducive to a headphone amp (higher gain, hard to get rid of the gain and noise because of the direct heated cathode.) And, they were very high distortion.
So we moved on. We looked at some very cool subminiature Russian triodes…but those needed 400mA of heater current at 6.3V. This would be a helluva load on our power supply, and, with the tube inside the chassis, the Vali could substitute as a portable grille (seriously, you’re looking at 60 degrees C or so).
So we moved on again. This time, I started thinking, “You know, what we really need to do is hang the tubes outside the chassis, then the heat from the heater won’t be a real problem. So, we looked at some 7-pin pentodes that only used about 150mA of heater current each. That could be viable, but I was nervous about the supply of those NOS tubes.
So we moved on again, looking at different tubes (12AU7, etc) that worked well at low volts and had 12V heaters. I didn’t like the performance of those tubes, and their new-production nature meant those tubes would be costly.
Deep into the project, I was stuck. The real problem was the power supply. Using a single 14VAC wall-wart, yes, I could step it down to run 12.6V heaters, and use a linear voltage multiplier for a high-voltage rail, or I could use a switching supply to derive a regulated 6.3VDC for a 6.3V tube, but neither of those options appealed. What I really needed was a power supply like I used in the higher-end amplifiers, with multiple AC voltages for multiple purposes. Making one single AC voltage try to do everything was an exercise in compromise.
But when you’re using a wall-wart, you really could only get one AC voltage on its barrel jack. That’s the rub. Two connection terminals = one AC voltage. Done and done.
At that point, I almost decided to leave well enough alone. After all, Vali was selling well. It was a good amp for many headphones. It just wasn’t really, well, universal or flexible.
Except…
What if our wall-wart guys could do a wall-wart with two AC voltages? And have something like a DIN plug connector, rather than a barrel jack?
If they could do that, then I could run the tube heaters off AC at 6.3V, so there wouldn’t be any power dissipated by DC regulators…and also get an AC voltage high enough to rectify or multiply into a proper tube plate voltage.
Now, I’d never seen a wall-wart like that before, but it was worth a query. I fired off an email, expecting to receive a “sorry, we can’t do that,” response. Instead, I got a picture showing a wall-wart with a DIN plug connector. “Like this?” the email asked.
Aha, now we were talking! I arranged to get a sample of the wall-wart made up.
A Vali With 2 Wall-Warts….and a Bizarre Gain Stage
While I was waiting for the wall-wart, I realized that I could get an idea of the performance of the power supply simply by using one of our 6VAC wall-warts and one of our 16VAC wall-warts with a voltage multiplier…and so that’s how I started hacking up a Vali.
And, while I was at it, I figured, Why not just use a 6BZ7 tube, like Lyr 2?
We had an excellent supply of them, and while it might look silly sticking out of a small amp like Vali 2, it would be a good test case to see what we could do.
And it’s a good thing I started there, because it quickly showed me one of the limitations of even a two-AC-voltage design: there would have to be some compromise on how high a plate voltage we could run. 100V like the Lyr 2 or 120V like the Valhalla 2 just wasn’t going to work…because we also needed that same AC voltage to supply Vali 2’s solid-state output stage.
So, I started looking on how we could maximize voltage on the plate. What we ended up with is a truly bizarre combobulation of a tube and a PNP transistor (this is a hybrid, remember). But the transistor is NOT the plate load…the plate load is still a resistor, though a much smaller one than you might expect. I haven’t seen this topology elsewhere, but that doesn’t mean that someone hasn’t done this before. Most good analog ideas have.
So what does this weird gain stage get us? It gets us a full 60V on the plate, which is plenty to get good performance out of 6DJ8 tubes and tubes in the same family. Now, 60V may not sound like a lot, but in the context of inexpensive amps that sometimes use 6-12V on the plate, it’s a huge, huge difference. It also means we can fully swing the discrete output stage into clipping, so the output stage is highly efficient. Also, with a new biasing arrangement, the output emitter resistors are almost 5X smaller than some of our other amps, for lower output impedance.
Sounds like a ton of design and optimization work for a very simple stage, right? That’s because it was.
And, it wasn’t over even after I’d gotten the first dual-voltage wall-wart and PC board prototypes for Vali 2. Because, in the process of testing which tubes we could use in the design (besides the stock 6BZ7), I found that the new combobulated tube-PNP hybrid gain stage ran at different operating points for each tube we used. Some were just a little bit off…but some were different enough that the amp’s power output would be affected.
So, I added a bias-centering servo. Which makes the little Vali 2 a fairly complex example of a modern hybrid: a tube for voltage gain, discrete bipolar transistors for output stage, an integrated circuit op-amp for bias control…on top of a dual-AC-voltage, high-voltage+heater power supply with regulated rails for both the tube and bipolar output stage.
Sounds complex? Yes. To achive simplicity and flexibility in use, internal complexity is usually required. Not that this is a bad thing, nor is Vali 2 overly complex (in fact, it uses only about 10 more parts than the original Vali.
A Thoroughly Flexible, High-Performance, Affordable Tube Hybrid
So what’s the end result of all of this design work? An amp that is, by all measures, a significant step up from Vali. It has:
- Higher output power
- Gain switching
- Preamp outputs
- Tube rolling
- Easily replaceable tube
- A single tube to keep rolling costs down
- By far the most sophisticated power supply in its class
- A unique tube gain stage
- A unique discrete output stage
- No starved-plate design or integrated buffers
- And, like the Ubers, a brushed aluminum top and knob
So how does it sound? Well, Mike, I, and the rest of the (formerly skeptical) shop think it leaves Vali in the dust. But the final judgment, of course, is yours.
And when you get right down to it, sound or not, flexibility or not, the main thing I hope to communicate with this chapter is simple: amp design isn’t simple or easy—even if it’s “just” a revision of an existing product. When you go off the reservation and start talking about discrete design, or tubes, it’s a much bigger undertaking than you might think.
Perhaps even more interestingly is that in Vali 2’s case, there was no design brief. No discussions in seedy off world bars. Hell, we didn’t even have the amp in the overall “2015 plan.” It was always a, “Well, let’s see what we can do with this.” As I mentioned, I came close to simply letting Vali soldier on for a few more years. But in the end, the relatively slow pace of development put me in a “no pressure” mode that allowed me to explore some really different designs…from the weirdo wall-wart to the combobulated gain stage.
And in the end, I think Vali 2 is really a big step up for inexpensive tube amps. But hey, I’m biased.
Users who are viewing this thread
