Megatron Electrostatic Amplifier
Jul 12, 2015 at 9:02 PM Post #16 of 37
Yeah DC coupling still has the bypass caps, but I guess I rationalize it by telling myself that putting boutiques there actually hurts the performance so it's "better" because of the lack of dependency. I think I would also use the russian parts if I were to do this again.
 
I have had the most mileage from the input tubes so I think we're in agreement. I swapped out half the RFT octet for that very reason. I have to go fetch my toshibas just to double check. On that note an interesting variant of the Megatron might be to use tube CCS's everywhere instead of solid-state parts. I was thinking the ac-coupling also allows me to use other interesting input tubes like c3m/c3g (what I happen to have on hand). BTW it's been a while since I looked at the schematic but I think the feedback is local, on the output stage only. It's consistent if the input stage makes the most difference.
 
I think you would do good justice to the Megatron if you did a point to point version Frank. Depending on what parts I have left over I might attempt to make a simpler version with a BHSE supply and russian caps. I'm not sure how sensitive the output stage is to temperature drift though.
 
Kevin also said his direct-drive DHT amplifier is going to be structured like the megatron, pretty curious as to how that's going to look...I can't really picture the schematic at the moment.
 
Jul 13, 2015 at 12:18 AM Post #17 of 37
 I probably won't build a Megatron, but I would definitely build Kevin's direct-drive DHT amp if the materials costs can be kept to a couple of $k. Probably not his approach, but the only way I can think to do this for a reasonable amount would be to use Russian tubes like the 4P1L in the front and middle, and 6P21S as the HV outputs. The gain of these types is limited, so there wouldn't be much room for feedback, but you might not need it with DHT's.  Something like the EML types that have previously been discussed would take the project out of my price range.
 
If I was to build a "Megatron" type amp it would look a little different than Kevin's. Sounds strange for a "die hard" tube person like me, but unless I could hear the difference, I'd use solid-state CCS plate loads on everything. The filament power demands for your eight EL34's exceeds those of my 845 amp.
 
I too have thought of using something like D3A or a C3G as the front end. I'm sure he has his reasons, but I don't understand using the 12AU7/12AX7 combo in an amp of this caliber.
 
Finally I'd think about getting rid of any sort of phase splitter in the front end. Do we really need that LTP any more? Almost all modern dac's have balanced outputs now. Makes everything a lot simpler just to forget about any single-ended inputs.
 
Speaking of capacitors, my current project is a  balanced DHT OTL amp designed for the HD-800. Input tubes are 112A with CCS plate loads. The 112A grids are run straight off the balanced outputs of my Metrum Hex. No LTP or any other type of phase splitter.   Output tubes are EL156 in cathode follower with CCS  loads. Coupling between the two stages is direct. Output caps are Russian military PIO. By running the 112A's in filament bias, all 4 DHT's and 4 EL156 run off a single 5A/15V filament supply. B+ is tube regulated similar to what Audio Note uses .Everything very simple. Sounds far better than it has a right to. Having fun playing with different output caps.
 
Jul 13, 2015 at 1:22 PM Post #18 of 37
  Finally I'd think about getting rid of any sort of phase splitter in the front end. Do we really need that LTP any more? Almost all modern dac's have balanced outputs now. Makes everything a lot simpler just to forget about any single-ended inputs.

I love the ED-1 Signature and Pawel HP-1 filters for acoustic ensemble, both which are single ended. Ideally it would all be balanced but it just takes one component (and some people still listen to Vinyl)
 
Finished the Megatron amp board (one of the simpler amp boards Kevin designs which is nice) and am finishing the PS. Just hunting around for a case, I like a cold war era instrumentation look. 
 
Jul 14, 2015 at 7:11 AM Post #19 of 37
You guys already saw the initial, ugly comparison setup, but I'll include a picture of the KGSSHV in the state which it was tested:
 
 

 
Megatron vs KGSSHV, 009
 
I recently finished a KGSSHV, so I figured I'd throw it in here before I sell the unit. I don't think the KGSSHV needs any introduction either, it's probably the most ubiquitous Stax amplifier that isn't manufactured by Stax at this point. Features:
 
Onboard
2SA1968
2SC4686
BHSE power supply
-/+400V
 
There are a fair amount of differences from the Megatron so I'll just be evaluating things from a sytem perspective. If it was a matter of practicality the KGSSHV would be better without question. After installing an Omron relay for PSU delay the 5 minutes I use to warm up the tubes is excruciatingly long. With an amplifier like the Megatron I also don't see the point in keeping the amplifier on 24/7. After my testing though I think it could actually be done, with proper ambient temperature control. 
 
Album:
Kalmah - The Black Waltz
Florence and the Machine - How Big, How Blue, How Beautiful
Garnet Crow - Garnet Crow Best of Ballads
Sam Smith - In the Lonely Hour
Evangelion 2.0 OST
 
Bass: Megatron has more impact than the KGSSHV, mostly noticeable on bass kicks. I think some of this can be attributed to a slight bump/coloration in the Megatron. On the KGSSHV, the treble emphasis gives these same bass kicks a "sheen" which gets pretty fatiguing after a while. In general the KGSSHV seems to have pretty compact bass. Extension on both amplifiers seems to be similar. 
 
Mids: Female vocals are more nasal on the KGSSHV. Psychologically there seems to be less dynamic contrast in the vocals when there is some sort of buildup. KGSSHV vocals are by comparison are more two-dimensional, there tends to be the perception that things are emanating from a flat wall in space. This is also meant to imply that there is more of a stereo separation effect to voices in general. Vocal reverb on the Megatron is excellent, or at least worth pointing out while I'm listening to pop ballads. Multiple voices, whether by arrangement or artificial, are clearer on the Megatron.
 
Treble: KGSSHV is more strident. Similar to the BHSE, initial attack on cymbals, shakers, etc. the initial impact seems to be accentuated. The phenomenon sounds a lot more unnatural on the KGSSHV though. There's a definite emphasis on the treble in the KGSSHV that seems to affect the characterization of the other bands as well.
 
Obviously there's the question of whether stridency or sibilance is accurate, but I'll let others make the connection there. The general impression is that the Megatron does things more effortlessly, whether it be because of its coloration or additional output power. The main difference seems to be how the KGSSHV and Megatron present vocals.
 
It should be noted I had a lot more difficulty matching listening levels because the treble impact threw me off quite a bit. I spent much more time fiddling with the knobs between tracks. That being said, the KGSSHV is probably 75% of the way there- you're really dumping power/tubes/expense into that last mile of finesse. I think this matters more for extended or "critical" listening sessions, whatever that means. It's really hard to ignore how dynamic the higher-end amplifiers sound though, pretty easy to draw a parallel to the tree of knowledge. 
 
I think these two data points against other amplifiers addresses one of the points I raised earlier- just because something has tubes/transistors in it doesn't necessarily make it sound more "tube-like/solid-state"...unless it was designed poorly. I think it's important to evaluate each characteristic after everything's built and done.
 
Jul 14, 2015 at 8:41 AM Post #20 of 37
Megatron vs. KGST, 009
 
I've had the KGST in the shop for some repairs for a while, might as well compare the two before I ship it back out. 
 

 
For an amplifier that was in development for a while, the KGST has only really surfaced in the past few years. I won't speak to the design details (because I can't), but it features the same input stage as the KGSSHV and a newer,simplified output stage using 6s4a tubes. These tubes are actively loaded with the 10m90s regulator, something people will be seeing a lot of in the next wave of electrostatic amplifiers. Special versions of the KGST exist, with features that include a cascoded CCS, or the option to use 9RA6 tubes. I plan to implement both of these in my next build, whenever I get around to it. To paraphrase spritzer, it was an amplifier that was created with the 009's in mind. I assume that has something to do with the way the gain is implemented on the output. Doesn't require a filament soft-start which is nice. 
 
I don't remember the exact features of this particular build, either GE or Sylvania 6s4a's inside. 
 
Maybe I can earn more audiophile points by including cover art? I heard at 100 I can exchange them for a free personalized brandy glass, and I get double if I listen to albums whose languages I don't understand. Albums:
 
Francesco Guccini - Radici

 
Omara Portuondo - Buena Vista Social Club presents Omara Portuondo

 
KachitagariTV - twitch.tv/topangatv

 
 
Bass: Megatron has more reverb, KGST has more extension. The Megatron's bass seems to be a little more coherent, and by this I mean that there's more of a perception that it is "centered". This might have something to do the soundstage and general imaging character.
 
Mids: Megatron has more body in male vocals. Performance seems similar. Soundstage-wise vocals on both amplifiers seem to predominantly occupy center stage. The Megatron seems to have better separation between instruments and vocals. 
 
Highs: The Megatron seems to have accentuated decay when I'm listening to the transients on stringed instruments. It seems much more natural on the Megatron. While the KGST competes with its imaging the lack of decay ultimately comes off as a bit "artificial".
 
From a macrodynamic perspective the Megatron has a slightly wider soundstage. the KGST and Megatron are similarly competent at projecting an image forward, I didn't notice a huge difference.
 
While I have chosen to review these amplifiers with the 009, I am among the users who feel the 009 has a sound signature that needs to be messed around with a little bit on the equipment side of things. Even with the supposedly "fixed" versions I think there's still some excess treble energy that Spritzer and Kevin addressed really well. The KGST is a  great, lower-cost alternative to the Megatron. This is probably one of the spots where I'd feel compelled to troll a few tubes and see how I could change the sound signature. For extended sessions I'd still say the Megatron is a clear winner, not because of the treble energy but the reduced stereo separation. This is something I have found to be fatiguing historically and this time is no different.
 
I think in this case people should be excited for the "new generation" of electrostatic amplifiers coming out. As I wrote before the Megatron occupies a unique spot in the DIY lineup because of its topology and design choices, and from what I've been writing it seems to have some equally unique strengths over the other amplifiers. Definitely worth looking into, whether it's via comission or otherwise. 
 
I'm still in the process of building other amplifiers (listed beforehand), so if I still have the Megatron around I'll make some similar comparisons. I think at this point there's a clear trend (or maybe I'm just making one), so there should be enough info to make an educated decision if you rely on written material. If you guys want to know anything else please let  me know.
 
Jul 14, 2015 at 1:48 PM Post #22 of 37
I really love the KGST. Small design, low part count, easy to build, few boutique/hard to find parts, and uses a not hard to find old TV 9 pin triode. Sound wise I'll be able to compare it to a Megatron (boards done, no chassis), KGSSHV (boards done, chassis almost completed), DIY T2 (chassis done, boards almost done) and Carbon (nothing yet), but at the moment just have a SRM-006T and SRM-232S. I'll also be able to test with a couple 007's Rossliew.
 
The first thing you notice is the more powerful, defined and substantial bass. This is surely due to the regulated power supply. The second thing I notice is greater details and 'depth' of the sound stage, everything is more well localized and placed.
 
I'm finding it a little tweaky too. I've got a load of JAN tubes and one went belly up after only a few months. Second, it was a little hard to set and maintain a tight balance/bias it seems. Nothing horrible, now I've got it to where it starts at +- 10VDC (compared to Stax amps which will be twice or three times that) and then settles to +-4VDC, but fiddling with that has taken some time and I haven't been able to get zero. Finally, I also get a low level buzz/noise that intermittently comes in until it fully warms up - after 15 minutes. 
 
On the circuit I'm trying to understand these SS/tube Stax circuits. Here's the schematic, I think it's basically a three stage design, the input gain/phase splitter is a common source amplifier with a CCS load and feedback directly from the output stage, which drives a common gate amplifier. That then feeds the triode output stage which is a classic tube gain and buffer which is DC coupled to the headphones. The high capacitance, high impedance, high voltage and low current drive of a tube matches great with an electrostatic, so it can be all DC coupled with a servo keeping everything around 0VDC.  Anyone please help and correct my take on this, there's probably (maybe large) mistakes with this. 
 
To my mind this kind of design - only possible with electrostatic headphones, is an ideal in amplifier design. We get SS and triodes, both doing what they do best. The SS provides lots of low noise gain which we need to bring it up to hundreds of volts. As I mention tubes direct couple to electrostatic headphones beautifully and the whole thing can be DC coupled. Minimal design, wire with gain and all that. CCS loads are ideal for tubes and transistors, and the advantage of electrostatic headphones (compared to speakers) is the output voltage is more sane, you don't have to get to the kV range. Really just perfect in terms of engineering and sound. 
 
Anyhow it's a treat of an amp.
 
Jul 14, 2015 at 2:16 PM Post #23 of 37
  Well written and a pleasure to read, A.!
 
If only you have an 007 to compare between the Megatron & KGST, that would be most perfect for myself personally :)

 
Yeah I have a mk1 I just kept the 009 plugged in for consistency. I guess it wasn't shown in the pictures. I prefer the 007 to the 009 by a wide margin, I just haven't really introduced other headphones into this because then I'm just reviewing headphones, which has certainly been done to death here.
 
The impressions are largely the same, just that the discrepancies in the bass are more obvious because there's no dead spot in the 50-60hz region, and generally the bass is more coherent. The 007 also responds better to the high current output, whatever that means. That's kind of what I was getting at when I was talking about the loudness problems, usually the treble is the first thing to become overbearing but this has never been the case with the 007s + Megatron. 
 
It really comes down to what you want to prioritize in the sound, and what kind of reference you've been using up to now. If that reference is a KGST then relatively speaking you will most likely perceive some lower frequency coloration and a more complex soundstage if you switch to the Megatron. I think the coloration can probably be tuned somewhat with tubes but I have yet to try. I'm actually going to pick up some of my alternatives today, so I'll try swapping a  few and see what happens. No impressions though, rolling tubes to compare an amplifier to itself isn't worth the effort at the moment. If it doesn't sound like trash then I'll leave them in.
 
I would say if you know you want to roll tubes, etc. etc. and you need a heater then the Megatron is a no-brainer. Beyond that you probably need to listen to one. After these comparisons I would hesitate to call it neutral, but there's clearly something about it that is very enjoyable. 
 
My Carbon parts should be arriving towards the beginning of next week, maybe I'll be able to compare its prototyped form to the Megatron. I know it's priced alongside Birgir's Carbons so there's probably some merit in doing that. I need to finish that before I start reading Frank's posts otherwise I'm going to end up building something totally orthogonal
 
Jul 15, 2015 at 11:49 AM Post #24 of 37
   I probably won't build a Megatron, but I would definitely build Kevin's direct-drive DHT amp if the materials costs can be kept to a couple of $k. Probably not his approach, but the only way I can think to do this for a reasonable amount would be to use Russian tubes like the 4P1L in the front and middle, and 6P21S as the HV outputs. The gain of these types is limited, so there wouldn't be much room for feedback, but you might not need it with DHT's.  Something like the EML types that have previously been discussed would take the project out of my price range.
 
If I was to build a "Megatron" type amp it would look a little different than Kevin's. Sounds strange for a "die hard" tube person like me, but unless I could hear the difference, I'd use solid-state CCS plate loads on everything. The filament power demands for your eight EL34's exceeds those of my 845 amp.
 
I too have thought of using something like D3A or a C3G as the front end. I'm sure he has his reasons, but I don't understand using the 12AU7/12AX7 combo in an amp of this caliber.
 
Finally I'd think about getting rid of any sort of phase splitter in the front end. Do we really need that LTP any more? Almost all modern dac's have balanced outputs now. Makes everything a lot simpler just to forget about any single-ended inputs.
 
Speaking of capacitors, my current project is a  balanced DHT OTL amp designed for the HD-800. Input tubes are 112A with CCS plate loads. The 112A grids are run straight off the balanced outputs of my Metrum Hex. No LTP or any other type of phase splitter.   Output tubes are EL156 in cathode follower with CCS  loads. Coupling between the two stages is direct. Output caps are Russian military PIO. By running the 112A's in filament bias, all 4 DHT's and 4 EL156 run off a single 5A/15V filament supply. B+ is tube regulated similar to what Audio Note uses .Everything very simple. Sounds far better than it has a right to. Having fun playing with different output caps.

 
I think he opts for the LTP because of the PSRR, not because of what can or can't be plugged in upstream in 2015. I'm guessing you feel it's overkill with the active loads? 
 
Jul 15, 2015 at 9:26 PM Post #25 of 37
 
Speaking of capacitors, my current project is a  balanced DHT OTL amp designed for the HD-800. Input tubes are 112A with CCS plate loads. The 112A grids are run straight off the balanced outputs of my Metrum Hex. No LTP or any other type of phase splitter.   Output tubes are EL156 in cathode follower with CCS  loads. Coupling between the two stages is direct. Output caps are Russian military PIO. By running the 112A's in filament bias, all 4 DHT's and 4 EL156 run off a single 5A/15V filament supply. B+ is tube regulated similar to what Audio Note uses .Everything very simple. Sounds far better than it has a right to. Having fun playing with different output caps.

Frank, this OTL project of yours sounds interesting (sorry for derailing the thread)...any pics to share?
 
Jul 16, 2015 at 1:55 AM Post #26 of 37
   
I think he opts for the LTP because of the PSRR, not because of what can or can't be plugged in upstream in 2015. I'm guessing you feel it's overkill with the active loads? 

 
In a commercial or public fully balanced design you pretty much have to go with either a LTP input or an input transformer. Building a personal amp, I would make it with either single-ended or balanced inputs, but not both. I like circuit simplification.
 
Not sure you can use CCS plate loads with a LTP if you have a CCS in the tail. I'd think they'd fight each other. Usually you see resistive plate loads in a LTP because you need the maximum impedance of the CCS  in the "tail" for balancing purposes.  I can't see there'd be any "real world" PSSR issues in any of Kevin's designs.
 
Nice write-up about the sonics of the Megatron and how they contrast with solid-state designs. The Megatron should be of significant interest the 15% of audiophiles (like me) who prefer a more classic/analog presentation than that provided by many modern "neutrality" based amps. A welcome addition to the stable of electrostatic amps.
 
Jul 16, 2015 at 11:42 AM Post #27 of 37
   
In a commercial or public fully balanced design you pretty much have to go with either a LTP input or an input transformer. Building a personal amp, I would make it with either single-ended or balanced inputs, but not both. I like circuit simplification.
 
Not sure you can use CCS plate loads with a LTP if you have a CCS in the tail. I'd think they'd fight each other. Usually you see resistive plate loads in a LTP because you need the maximum impedance of the CCS  in the "tail" for balancing purposes.  I can't see there'd be any "real world" PSSR issues in any of Kevin's designs.
 
Nice write-up about the sonics of the Megatron and how they contrast with solid-state designs. The Megatron should be of significant interest the 15% of audiophiles (like me) who prefer a more classic/analog presentation than that provided by many modern "neutrality" based amps. A welcome addition to the stable of electrostatic amps.
 

 
You can use CCS's in the loads and tail if you implement something with a higher impedance in the cathode, so it "dictates" where the output DC lies when you tune it. If I remember the gist of it correctly in Morgan Jones' book he cascodes the tail in order to achieve this. Maybe something worth exploring on  the turret strips, and certainly with my C3ms...
 
Jul 19, 2015 at 1:40 PM Post #29 of 37
Very good information in this thread..
 
beerchug.gif

 
Nov 16, 2015 at 10:37 PM Post #30 of 37
Having built both the KGSSHV and the Megatron (the red one in your pictures), I do like the Megatron a little more with  the 009's.
The soundstage may be a little better in the KGSSHV, but you are right that it sounds a little more shrill in the treble than the Megatron.
Thanks for starting this thread.
 

Users who are viewing this thread

Back
Top