Better than SET? The radically different 535SE Headphone Amplifier/Preamplifier

Discussion in 'Product Announcements' started by liamstrain, Jun 2, 2016.

  1. dsavitsk

    dsavitsk Friend

    Pyrate
    Joined:
    Oct 2, 2015
    Likes Received:
    1,616
    Trophy Points:
    93
    Home Page:
    I was staying out as I don't want to butt in on someone's thread, but I'll comment on this. The basic DSHA circuit (used in the DSHA1, DSHA2, Ravenswood, Apsara, Walnut, etc.) is differential. All differential circuits are balanced, not all balanced circuits are differential. I prefer differential circuits as I feel they generally offer advantages over single ended (and various non-differential push pull circuits), and very few drawbacks (and we can discuss that elsewhere). Steve feels differently I think.

    Yes, Steve and I go back. He was very helpful to me getting the cases made on my first batch of amps. We probably agree on more than we disagree (particularly about the use of transformers for gain), though we decidedly disagree on some fundamental stuff :) And I think he is mad at me, or thinks I am mad at him, or both. Not sure.
     
  2. liamstrain

    liamstrain MOT: The Audio Guild

    Joined:
    May 31, 2016
    Likes Received:
    149
    Trophy Points:
    33
    Location:
    Atlanta
    Home Page:
    Thanks for chiming in @dsavitsk! Glad to see you here.
     
  3. liamstrain

    liamstrain MOT: The Audio Guild

    Joined:
    May 31, 2016
    Likes Received:
    149
    Trophy Points:
    33
    Location:
    Atlanta
    Home Page:
    Just for a little historical perspective, the 535's industrial design was inspired by the old "blackface" radios that were common in the 1920s, like this Adams-Morgan Paragon.

    [​IMG]

    We wanted to remain as true to the inspiration as possible, complete with the engraved front panel. The typeface used was very common and was produced by manual pantograph engraving machines. These machines used brass lettering templates. A stylus was guided through these templates which was then reproduced in the material by the cutter head the diameter of which determined the line width of the text. The brass letter templates were made in just one size, and the pantograph mechanism would be adjusted to determine the size of the lettering on the work piece.

    This same typeface was also commonly found in old schematic and other drawings as similar pantograph machines were used as lettering guides in the drafting trade.

    However as common as this typeface was throughout the '20s and '30s and into the '50s and '60s, we were not able to find its equivalent in a digital typeface. Some were close, but always off in some way, such as the style of the numerals.

    So we reached out to our friends at Able Engraving & Design in Surrey, UK. We had worked with Able in the past on a previous project and knew that they still had some old manual pantograph type machines which used that very same typeface. Able engraved the entire character set on a sheet of black/white Traffolyte which we then scanned and painstakingly reproduced digitally in our graphics program.

    Here is a photo showing one of Able's manual machines in action. In the upper left, you can see the worker tracing the stylus on the brass lettering templates. The work piece is the white block in the vice on the work table. Note that the motor is at the bottom of the machine and that the cutter head is driven by a belt and pulley system.

    [​IMG]

    Old radios such as the Adams-Morgan typically used a material called ebonite for their front panels. Ebonite is a hard material produced using natural rubber. It was invented by Charles Goodyear in the late 19th century. We have worked with ebonite in the past and while it can produce a beautiful result, sanding and polishing it is rather labor intensive as is the paint filling of the lettering once it has been engraved.

    For the 535's faceplate, we chose to use a material called Traffolyte. Traffolyte was developed in the late 1920s by Metropolitan Vickers of Manchester, England. Vickers was located in Trafford Park, hence the trade name Traffolyte.

    Traffolyte is a phenolic laminate (essentially Bakelite) that was made specifically for engraving. It uses multicolored layers such that engraving through one layer exposes the color of the layer below. This negates the need for subsequent paint filling and is far more durable and long lasting.

    Here is an old Traffolyte advertisement from the 1930s or '40s.

    [​IMG]

    We hope that gives you guys a little more background into some of the materials and design choices we are making. More than just an homage, we really do intend to make this an heirloom.

    Chassis drawings are with fabricators, and transformers going in the queue.
     
    Last edited: Jun 7, 2016
  4. bazelio

    bazelio Friend

    Pyrate
    Joined:
    Nov 16, 2015
    Likes Received:
    3,417
    Trophy Points:
    113
    Very, very cool!
     
  5. eastboundofnowhere

    eastboundofnowhere Facebook Friend

    Contributor
    Joined:
    Mar 28, 2016
    Likes Received:
    202
    Trophy Points:
    43
    Are you guys doing preorders, anything like that? This looks interesting and I keep getting interested in the low volume solid state stuff just after it becomes unobtainium. Really, really like that it is also a line stage.
     
  6. liamstrain

    liamstrain MOT: The Audio Guild

    Joined:
    May 31, 2016
    Likes Received:
    149
    Trophy Points:
    33
    Location:
    Atlanta
    Home Page:
    Glad you are interested. We are still figuring out the sales step. We want to get the pre-production units out into the hands of testing and reviewers. But after that, we may try to do a pre-order, or possibly a kickstarter to get production quantities going. Still a few ducks to get in a row first. :)

    I expect if we get enough pre-order interest on the forums, we can try to put together an early batch.
     
  7. Hands

    Hands Overzealous Auto Flusher - Measurbator

    Staff Member Pyrate MZR
    Joined:
    Sep 27, 2015
    Likes Received:
    12,285
    Trophy Points:
    113
    Location:
    Colorado
    Home Page:
    Since you mentioned testers and reviewers, and chance you could loan a unit out to a small handful of folks here to share our thoughts privately or publicly? If that's a possibility, maybe reach out to one of the staff members in private.
     
  8. liamstrain

    liamstrain MOT: The Audio Guild

    Joined:
    May 31, 2016
    Likes Received:
    149
    Trophy Points:
    33
    Location:
    Atlanta
    Home Page:
    @Hands - that's definitely a possibility. Once we are a bit closer, I'll reach out. Thanks :)
     
  9. liamstrain

    liamstrain MOT: The Audio Guild

    Joined:
    May 31, 2016
    Likes Received:
    149
    Trophy Points:
    33
    Location:
    Atlanta
    Home Page:
    For those interested, Steve has provided the full schematic for the amplifier section of the 535SE

    (Note: Please see Post #42 for an explanation of these updated documents)

    [​IMG]
    (PDF Download)
     
    Last edited: Jun 14, 2016
  10. Priidik

    Priidik MOT: Estelon

    Pyrate BWC
    Joined:
    Sep 27, 2015
    Likes Received:
    2,153
    Trophy Points:
    113
    Location:
    Estonia
    Love the simplicity! Thumbs up.
     
  11. SoupRKnowva

    SoupRKnowva Official SBAF South Korean Ambassador

    Pyrate Contributor
    Joined:
    Sep 26, 2015
    Likes Received:
    4,249
    Trophy Points:
    93
    Location:
    Austin, TX
    I'm showing my lack of EE knowledge here(or more accurately my just enough to look even stupider amount of knowledge), but since it's a solid state output, why's there a cap in the signal path after the current amplification?
     
  12. TwoEars

    TwoEars Friend

    Pyrate
    Joined:
    Feb 2, 2016
    Likes Received:
    2,006
    Trophy Points:
    113
    Audio signals are basically AC so the cap is going to be virtually invisible to the audio signal, what the cap however does is block DC voltage. Without the cap you'd get a constant DC voltage through your headphones and a ground short by the looks of it.

    The trickier question is how they managed to ballance the componets in the amplification stage to get it to work properly, just trial and error? It's actually a pretty simple design but the choice of resistors etc will be crucial. Intriguing...
     
  13. SoupRKnowva

    SoupRKnowva Official SBAF South Korean Ambassador

    Pyrate Contributor
    Joined:
    Sep 26, 2015
    Likes Received:
    4,249
    Trophy Points:
    93
    Location:
    Austin, TX
    Yes, I'm aware that caps block dc, but basically everyone agrees they degrade the signal, so I'm wondering why you would design a solid state output stage to need one...unless there is something else that necessitates it, hence it was a matter of compromise.
     
  14. HAL9000

    HAL9000 Almost "Made"

    Contributor
    Joined:
    Apr 4, 2016
    Likes Received:
    233
    Trophy Points:
    63
    Location:
    Fairfield, IA
    Seems like a killer design! Put me on the interested in pre-ordering list...
     
  15. liamstrain

    liamstrain MOT: The Audio Guild

    Joined:
    May 31, 2016
    Likes Received:
    149
    Trophy Points:
    33
    Location:
    Atlanta
    Home Page:
    From Steve:

    It's not an issue of tube or solid state, but one of topology. The output stage of the typical solid state amplifier is what is called complementary push-pull. Solid state devices come in two polarities, negative, or N-type and positive, or P-type. You can use both of these types in a complementary fashion to create an output stage in which the signal is split between the two, the N-type handles the positive half of the signal and the P-type handles the negative half of the signal.

    Such a topology uses what's called a split power supply, which has a positive voltage, a negative voltage and ground. The positive voltage powers the N-type devices and the negative voltage powers the P-type devices. This allows you to have an output stage that has zero DC offset voltage and therefore can drive the load directly without the use of a coupling capacitor.

    Vacuum tubes on the other hand have just one polarity, the equivalent of solid state's N-type device. There is no complementary device. At least not until someone invents a vacuum tube that uses protons instead of electrons as the charge carrier. ;-)

    And since the 535's inspiration was the classic SET vacuum tube amplifier, I chose to use only N-type devices. When you do this, you are left with just a few options.

    You could power it using a split power supply which would still leave you with a small but significant DC offset voltage. You could use a capacitor, but since the amplifier will be driving relatively low impedance loads when driving headphones, the capacitance needs to be quite large in order to preserve adequate low frequency response. So large that it would be impractical to use a film capacitor. Instead you need to use an electrolytic capacitor. And because the DC offset will be smaller than the average signal, you would need to use what is called a non-polar electrolytic capacitor. This is done by connecting two polarized electrolytics back-to-back as electrolytics are polarized by nature. And there's little that sounds worse than a non-polar electrolytic capacitor with a DC bias that is smaller than the average signal level. You can liken it to crossover distortion in a class B amplifier.

    Another option would be to use a DC servo to bring the offset voltage to zero. But this is rather messy and inelegant in my opinion.

    I chose the third option. Which was to use a single supply (V+ and ground) just like the classic SET amplifier and to set the output DC offset to half the supply voltage, or in this case, 16 volts. This way I could simply use a regular polarized electrolytic that would be biased at a level higher than the maximum signal level.

    Some people turn their nose up at the thought of an electrolytic capacitor being used in the signal path. But there have been a lot of bad implementations of them as I have described above. However when using a good quality electrolytic designed for audio use coupled with proper biasing, they can sound absolutely wonderful. Certainly far far better than the steel core, current-carrying, air-gapped output transformer of the classic SET vacuum tube amplifier. I've always been amused by those who shun any use of capacitors in the signal path because they supposedly distort the signal, but love the sound of of SET amps whose output transformers are distorting the signal literally orders of magnitude more than even a mediocre coupling capacitor.

    Hell, Nelson Pass used electrolytics to capacitively couple the input and the output of his original Zen amplifier and it still sounds wonderful. That's because those capacitors in the Zen were properly biased.

    I hope this helps answer your question.
     
  16. liamstrain

    liamstrain MOT: The Audio Guild

    Joined:
    May 31, 2016
    Likes Received:
    149
    Trophy Points:
    33
    Location:
    Atlanta
    Home Page:
    Hey there @TwoEars - What do you mean by "amplification stage"? The transformer? Because that's where all of the signal (voltage) amplification is.
     
  17. jelt2359

    jelt2359 Friend

    Pyrate
    Joined:
    Nov 8, 2015
    Likes Received:
    157
    Trophy Points:
    43
    Thanks for all the info. Sidetracking a bit from all the discussion about the design... How does it sound, especially when compared to a classic SET amp?
     
  18. TwoEars

    TwoEars Friend

    Pyrate
    Joined:
    Feb 2, 2016
    Likes Received:
    2,006
    Trophy Points:
    113
    I was thinking about the part with the JFET and ndn transistor later on. Very unique design. I guess this is the part which is equivalent to a grid/cathode/plate in a tube amp. I guess my question was how you managed to find the correct values for all the resistors in the design, just trial and error? Such as the R103/R104/R203/R204 resistors. And did you have a lot problems finding the "right" JFET and so on.
     
  19. bazelio

    bazelio Friend

    Pyrate
    Joined:
    Nov 16, 2015
    Likes Received:
    3,417
    Trophy Points:
    113
    So you bias FET to the base current of the BJT. This emitter follower is now (intentionally it seems) less linear than a traditional BJT Darlington and you've added coloration through low order distortion as seen in the distortion plot. I wanna hear it. :)
     
    Last edited: Jun 13, 2016
  20. liamstrain

    liamstrain MOT: The Audio Guild

    Joined:
    May 31, 2016
    Likes Received:
    149
    Trophy Points:
    33
    Location:
    Atlanta
    Home Page:
    So far the only in depth listening has been done by Steve - and of course, we're all a bit biased. :) Judging it as a whole experience (rather than dissecting individual characteristics), we feel like it is the kind of thing that reminds you of SET, while still being very much it's own identity and character. But we are all subjective creatures, and our impression may not be yours.

    Your question is exactly why we are working to get the pre-production units out into the hands of those who can give more objective reviews and measurements to share with the community before pre-orders begin. We are all very curious to see what people think of it, but I personally suspect that it's going to make a lot of people very happy.

    From Steve:

    Yes, a transistor's base/emitter/collector or gate/source/drain in the case of FETs is basically the equivalent of a tube's grid/cathode/plate. Fundamentally transistors are triodes too (I should probably have some t-shirts made up with that on it). Triode simply means three electrodes. And both tubes and transistors use one electrode to control the current flowing through the other two electrodes in order to achieve amplification and other functions. Just that in tubes that current flows through a vacuum and in transistors it flows through silicon.

    But it ain't easy getting electrons to flow through a vacuum. They cling tenaciously to the cathode. So you have to shake the little porkers violently to get them to let loose. That's what the heater in a vacuum tube is for. To heat the cathode to the point that it liberates some electrons. Then you can you get them to scoot through the vacuum. But this joyous little cruise is very short-lived. Before you know it, their little heads are smashed to bits when they hit the plate. Horrible.

    For me, this revelation was akin to watching cattle being slaughtered and prompted me to start an organization called PETE, People for the Ethical Treatment of Electrons. I think the way vacuum tubes treat electrons is barbaric and we need to put a stop to it. Silicon allows us to put electrons to use in a much more natural, less violent and humane fashion.

    Joking aside, the resistors you mention are actually the least critical components in the circuit. Their values were largely arbitrary when this circuit was first breadboarded and tested.

    R103/203 are gate stoppers, the equivalent of grid stoppers in vacuum tube circuits and are used to reduce the risk of local oscillation. You could probably reduce the value by tenfold and still be safe.

    R104/204 could arguably be left out. In switching applications where the transistors need to turn on and off quickly, that resistor is used to help the second device switch off faster. But in linear applications such as this where the devices never turn off, it can help linearize the changing signal current in the first device a bit, but it's not really going to affect performance in any meaningful way. It just remained in the schematic largely out of habit.

    As for the JFET device, I've long been a fan of the Toshiba 2SK170 and it worked wonderfully in this circuit so I stuck with it. Toshiba has discontinued that device but Linear Systems now makes an improved version in the form of the LSK170 and that is the device that will be used for production.

    The key to understanding the output stage is that it is not functioning as a signal amplifier where component selection is far far more critical and a tweak here and a tweak there can have a significant effect on performance. It is an emitter follower (the equivalent of a cathode follower). It has a gain of one meaning that it only handles the signal after it has been fully amplified. It's sins are small and are not amplified by the inherent gain of the device as it would be if it were being used as such. Instead it's just the opposite. It's sins are reduced by the inherent gain of the device. This allies you much greater leeway in terms of parts selection.

    The most critical part in this circuit is the input transformer where all of the signal amplification takes place.
     

Share This Page