Kit Review: Gakken Vacuum Tube Amplifier

[TankaaKumawani]

Tankaa Reviews:  Gakken Otona no Kagaku (Science for Adults) Vacuum Tube Amplifier

I ordered this little kit for about $175 from Amazon.  While it was available at a lower price on eBay, I wasn’t quite ready to trust the seller.  If I only had the presence of mind, I would have picked it up at the low, low price of 9300¥ ($100 or so at the time) at either Tokyu Hands or Yodobashi-Akiba back during the summer of 2010—there’s nothing I can do about it now.  Let’s move on to the goods.

The kit itself comes in a corrugated paperboard box that’s 22 by16 by12cm.  Upon unpacking it, we are greeted by the original Japanese manual, followed by the bottom of the chassis and the various parts, with the tubes in their own little box.  At the very bottom of the box, we find the top case and the main board.

On the top side, we see heavy output transformers, four ceramic sockets, the power LED, and a mysterious metal can.  Said can contains a DC-DC boost converter, evidenced by the nearby filter caps and toroid inductor.

Flipping the board over reveals that Gakken really doesn’t want us poking around in Switchmode Land.  Since I don’t have my soldering rig set up, this enigma will have to wait.  Is it a flyback?  Is it a boost switchmode?  That’s an answer for the next episode.  Or Dave, if he decides to buy one of these.  (The kit itself is a screw-together affair.  I’ll probably solder what connections I can later.)  It doesn't look like a DIP from the bottom, maybe it's an SMT on a board?


Let’s Assemble!

First, I’ll put together those horns.

The plastic stampings were just a little bit askew and didn’t quite fit together, but a bit of elbow grease gets stuff into place.  I’ve had the same issues with snap-together model kits, it’s probably a mold wear issue.

With that out of the way, we move on to the chassis assembly.  Push things in, screw them down, so on and so forth.

After that, we go to work on the battery contacts, which really deserve solder.  On the other hand, that would anchor the case to the boards in a relatively permanent fashion…
(No picture due to limited number of hands.)
Time to close up this case.  I wish they had used plastic card instead of cardboard…and proper rubber feet.  That’s easy enough to fix, though.



Aww, look at the cute little Chinese pentodes.  I wonder who stamped these out and the possible year of manufacture (1960 for the 1B2/1S5s and 1964 for the 2P3/3A5s  Actually, is that Cyrillic?  It seems that the Wan Hung Lao outfits have been in business for longer than I thought.)…the input pentodes seem to be a bit more worn than the output tubes.  The getters look shiny, so let’s set the tubes in place.

Okay, almost done.  Hook up the speakers…

Time for some Coltrane.  The little drivers and horns struggle with lower frequencies, and the trumpet is a bit overemphasized.  It’s decidedly not a hi-fi device, but it’s a neat little project all the same…and quite a bit of sound for 100mW/channel.


VERDICT:  Fun and simple introduction to glassfets, but with that import markup and the less-than-stellar speakers, it's better to roll your own device.

EDIT:  Did some paper calculations, the best possible output from a 3A5 in a Class A configuration at 45V is around 120mW.  Interesting.

Well, the next step would be a bit of homebrew.  I’m pretty sure the LT1073 can deliver 45V at low currents for B+ voltages, and the little battery pentodes sell for $3-$5 (USD) as tested new old stock.  I wonder if I could build something a bit heftier…say, Class AB with tone controls, balance control, feedback, and 500-700mW/channel.  I could probably get an off-the-shelf set of Dayton Audio speaker enclosures and some 100mm full-range drivers.  That smells like a project, doesn’t it?

ADDENDUM:  It can drive Dad's old A25 speakers at a reasonable volume at about 60-70% or so, it tends to clip at higher volume levels when delivering power to those.

I decided to take a look at it with the oscilloscope...that waveform seems a bit fuzzy.

And with the input at ground:

Well, looks like Gakken's filtering wasn't up to the task.  25kHz, 36mV ripple.  I guess I'm going to be looking for some 100V electrolytics and mapping out relevant parts of the V2 schematic that aren't in the manual.

[Whales]

I'm surprised at how little power this thing outputs.  Albiet the directional horns focus the output more, leading it to appear louder.

How does it sound with different speakers?  Is there anything obviously wrong?  I'm leaning towards believing the horns are the biggest culprit.

Thanks for the review

[dannyf]

hat waveform seems a bit fuzzy

Oscillation or interference (from a smps?)?

It can still sound nice, if the speaker rolls off at high or your hearing doesn't extend that far.

[djacobow]

The oscillations on those waveforms are not pretty.

I built one of the original "K-12" amps from this guy: http://www.s5electronics.com/thome.html

Must have been almost ten years ago. Anyway, nice little amp, easy to build, and easy to mod to improve. The thing about these amps is that they use these old TV tubes. Easy to find NOS since there is not that much demand for them.

Prices seem to have gone up, though. I remember mine cost about $125 and came with 4 11MS8's.

[TankaaKumawani]

@Whales:  It can drive Dad's old A25s to a reasonable volume (86db/W aperiodic damped port), but has limited frequency response and starts to clip above 80% volume.  Not bad for something of its size, though.

@DannyF:  I can still hear CRT flybacks.  The rolloff covers most of the stray harmonics.  (It seems to be coupling between PC board traces, so bodge wires aren't a solution.)

@djcobow: Those are nice little kits, especially the 8LS.

I was thinking about trying to do something with switchmode boost converters and subminiature pentodes (5902 or the rare-but-efficient 6495 missile tubes) in Class AB1 or AB2 operation as a follow-on, with tone and balance controls.  I think it would be a useful experiment in terms of power supply design, I've been screaming for an applications example since I took my first Power Electronics course.

[Richard Crowley]

That amplifier appears to have a HF oscillator DC-to-DC converter that creates the "high" voltage for the tube/ valve plates.
That oscillation imposed on the audio waveform is from the insufficiently-filtered "B+" power rail.

The DC-to-DC converter appears to be inside the shield in the middle of the PC board.
You can observe on the bottom of the board what may be the footprint of a high-frequency transformer.

They must have a pile of those old 1B2 and 2P3 tubes to get rid of.

[TankaaKumawani]

Yeah, I was expecting that to be the case.  (I'm looking at an LT1082 for my design once I get off of my lazy butt and read up on that darn design theory, unless there's a higher-frequency external switch design available in through-hole packaging.  90-100V should be enough B+ for the 6.3v subminis.)

Yeah.  The 1B2s don't make much sense (1S5 equivalent, lower performance than a 1T4), until you realize that their radio kits use them as detectors because of the diode.  The 2P3 (3A4 or 3S4 equivalent, it's hard to tell) power pentode is probably the best option as far as low-voltage output tubes would go.  (There's one or two low voltage subminis capable of matching them, but they're probably pretty rare nowadays.)

[Richard Crowley]

I would think that simply adding some filter/bypass capacitance to the appropriate nodes would knock out that noise.  You should be able to completely eliminate the hash with $5 worth of capacitors.

[TankaaKumawani]

Yeah, I've got a few high voltage mylars that might do the trick.  I'll look into bodging something in later this week.

[Richard Crowley]

We're talking about power-bus filtering here. 10s or even 100s of uF is typical.  I don't think that low-value mylars would do the trick. 
Although, granted, at that high ripple frequency, you wouldn't need near as much capacitance as down at mains power frequencies (50 or 60 Hz).

[N2IXK]

How much plate voltage does this thing run on anyway?  1S5 and 3A4 tubes were originally used for battery powered radio applications, where plate voltage came from a "B" battery, at perhaps 90V or so.

[Richard Crowley]

How much plate voltage does this thing run on anyway?

Good question. The "documentation" is extraordinarily vague.  It will take actual measurement with a DMM to determine what voltage rating the capacitors need to be. 

Of course, if you are messing about with old firebottles, maybe it would be more "authentic" to use a VOM.
But then, after all, you would have to refurbish and calibrate the VOM before you could use it.

I was getting ready to scoff at the characterization "heavy output transformers" until I read the spec that is was only good for 100mW. 

[TankaaKumawani]

From what I've heard (and their other kits, which used a stack of 5 9V cells for plate voltage), roughly 45V on the plates tops.

By "high voltage capacitors," I mean something with a greater dielectric strength than the 25V electrolytics I have in my parts bin.  So probably better than 50V.

[Alex Eisenhut]

Of course, if you are messing about with old firebottles, maybe it would be more "authentic" to use a VOM.

Also more correct, usually voltages are given as measured by a VOM, at least on old schematics.

[N2IXK]

Of course, if you are messing about with old firebottles, maybe it would be more "authentic" to use a VOM.

Also more correct, usually voltages are given as measured by a VOM, at least on old schematics.

Or a VTVM in some circuits...

[TankaaKumawani]

Haha.

I'll just use my Keithley 199 bench multimeter for now.  It's out of cal, but it's close enough for this sort of work.  (I have my old Radio Shack Special VOM (mostly nostalgia) and a Wun Hung Lo handheld DMM.  I'll probably get a mid-end Fluke handheld sooner or later, and that rather impressive Mastech LCR meter next time they're on sale.)