INTRODUCTION: After listening to the Manley 300B and my experience with the Svetlana directly heated tubes in a circuit that has no audio output transformers I strongly suggest you pay close attention to this article.

Dr. Gizmo


by Eric Barbour

        Ever compared a really good tubed line stage to a passive preamp? Even the best possible design, using tubes with the lowest distortion plus some negative feedback, tends to cause a distinctive "soft" coloration in the sound. Unless you really crank in the feedback, a tube preamp using common indirectly-heated tubes will  have at least a little of that "softness". Many tube equipment users feel that this effect is a major component of tube sound. You can get very close to real transparency, but at some cost. Probably, most VALVE readers agree that piling on negative feedback is not the best policy!

        In my previous tests of small-signal tubes, some experiments with UX-201A and 01A radio triodes showed me something interesting. The original UX-201A had a thoriated tungsten filament, while the later 01A had an oxide-coated filament. Otherwise, these tubes are supposed to be exact plug-in replacements for each other. And although my distortion tests showed comparable levels of THD for both types, and both types had similar "tubey" distortion, the older version seemed to have much clearer treble and greater detail than the 01A. Subsequent listening comparisons of the SV811 and SV572 triodes versus oxide-filament triodes (such as the 300B) seemed to support the impression that a thoriated filament is more detailed in apparent sound than an oxided filament. Little or no serious comparative listening had been previously given to this difference, especially when the tubes are used as small-signal amplifiers. So there is not much in the way of previous "design art" to follow. (As one dedicated DIYer bluntly put it, "only the truly desperate would use directly-heated tubes in a preamp!" And he has a point. DH tubes are VERY microphonic.)

        Small thoriated-filament triodes suitable for low-power use are no longer in production and are becoming difficult to obtain. The recent arrival of Svetlana's SV572 series, however, gives us some interesting experimental possibilities. These triodes, although power types rated for 125 watts dissipation, are extremely linear. They also come in four different values of mu, allowing some leeway for the requirements of preamp design as well as power-amp design. Although not really intended for small-signal use, my experiments showed that the SV572-10 would make an excellent line-level amplifier at low plate voltages, provided that power-supply noise and microphony issues were addressed.

        An interstage transformer would give the best possible performance. Unfortunately, such transformers are usually very expensive or inferior in quality. Electra-Print Audio's recent introduction of a high-quality interstage transformer for SE circuits was perfectly timed. Their 3533-2 tranny has a 1:1 ratio and is rated for a 10k ohm primary load. And unlike many interstages, it was designed for SE operation using a single driver tube, with up to 20 mA DC idle current in the primary. Electra-Print's Jack Elliano was kind enough to loan me a pair of 3533-2s for use in an experimental line stage. Since they would be a perfect load for an SV572-10, I decided to try building the simplest circuit possible, while staying within reasonable parameters for electrical performance.

        An important thing about this preamp has to be made very clear. The plate, filament and bias supplies MUST be as low in AC hum and noise as possible. The plate power is only 200v, which turns out to be more than enough for a line stage of this type. The "open-frame" power supply shown is a common industrial part, readily available in surplus or from Mouser Electronics. It is more than enough for this stereo preamp. One issue: the manufacturer rates the AC ripple voltage of the supply at 1% of the DC output, even though it is electronically regulated. This is equivalent to 2v RMS, an unacceptably high value for audio circuits. If the builder prefers, a more conventional source of plate supply can be fabricated from a 215v plate transformer, rectifier and filter. Regardless of its source, heavy-duty filtering of the 200v DC is called for. One RC filter (R1 and C10) gets the ripple from about 1.5v RMS down to less than 20 mV. Then, in the preamp chassis, we go through a pair of large Hammond chokes into C2 and C4, further reducing the ripple to much less than 100 uV. Remember, any AC ripple on the plate power will appear on the output at the same level.

        Even greater filtering is needed for the filaments of the SV572-10s. ANY noise riding on the filament supply will be amplified by the mu of the tube, because it will act as a grounded-grid amplifier. We have to provide VERY clean DC, and several amps of it. This is very difficult using only a simple transformer-rectifier-filter circuit, so I would prefer a switching power supply. But the 572's filament needs 6.3 volts and such supplies are scarce and expensive. What to do?

        Relax, there is a very low-cost shortcut. In Sash Ohtsuka's MJ article about the SV572 triodes, we find that Sash tested the emission of the filament with less than 6.3vapplied to the filament for heating. Sash commonly does this with tubes he tests for MJ, and has found that it is a good measure of the quality of a new tube. As it turned out, operating at 5.04 volts DC caused the plate current to drop only 10%. The decline would be much more if this were a conventional oxide-filament tube or an indirectly-heated tube, often 20-30%. It appears that thoriated filaments are much more tolerant of low filament voltage than oxide emitters. A 10% decrease might be an issue if we were attempting to get maximum  power from the triode. However, since we are only going to operate this tube at 200v and 15 mA, such a small decrease will be hardly noticeable. The plate resistance will rise slightly; again, hardly enough to notice. And a major bonus is an enormous lifetime from the SV572-10. In normal operation, a life of 5000 to 10,000 hours can be reasonably expected from the filament of an SV572. In this circuit, with the filament running at well below its normal temperature and emission, life should be comparable to any super-premium preamp tube, or even solid-state devices.

        We're in luck! Now we can use a cheap 5-volt personal-computer power supply. I went to a local surplus dealer and bought a used Epson 286 PC for all of $10. The hard drive had already been removed, yet the unit appeared to be in like-new condition. (Maybe it was never used at all--simply junked as obsolete before its time, a common occurrence here in Silicon Valley.) By removing the motherboard and floppy drive, plenty of space was made for the 200v supply and other components. Unlike more recent PCs, this one had a case that was almost entirely heavy-gauge steel. And alreadymounted inside was a like-new power supply, capable of +5v at 12 amps, +12v at 2 amps and -5v at 100 mA. With suitable filtering, this unit makes an excellent remote power supply for the line stage.

        On the main preamp chassis are the tubes, interstage transformers, chokes L1-L4, and some filter capacitors, plus a stereo volume control, input and output jacks and other small components. Chokes L3 and L4 are special Electra-Print models, designed to clean up the 4-amp DC used on filaments like those in the SV572 series. Combined with C8 and C9, the resulting hum and noise appearing across the filament of each tube is so low that my test equipment has trouble reading it consistently.

        Microphonics in the SV572 are a serious matter, so we took serious steps to reduce its influence. The tube sockets were mounted on rubber shock mounts, with extra-long wiring connections to allow the sockets to "float" somewhat. The whole preamp chassis was put on shock mounts. We found that remoting the power supply was the best policy. Not only did its heavy steel box help to cut induced hum to the preamp stages, there was also some mechanical vibration from the power supply and its (very quiet) cooling fan, which could reach the tubes if the fan was in the same enclosure. Finally, putting large Pearl tube coolers on the SV572-10s made the tubes very resistant to airborne vibrations.

The controls on noise and microphony were so successful that another problem cropped up. Even withall these extreme measures, 2 mV of 60 Hz AC hum still appeared in theoutput. Some experimentationrevealed the source: environmental hum was being picked up by the cores ofthe 3533-2 transformers! Jack Elliano was very helpful, and offered to pot the 3533-2s in massive welded steel cans. Once these heavily-shielded versions were installed, residual hum dropped greatly.  The prototype preamp, when completed, showed about 200 microvolts of noise on the audio output jack; comparable to many commercial preamps I've seen in the past. Most of this noise is above 20 kHz, being switching noise from the filament supply. Even a sensitive power amp, such as a vintage Heathkit or Harmon-Kardon, should be compatible with this circuit.

preamp, when completed, showed about 200 microvolts of noise on the audio output jack; comparable to many commercial preamps I've seen in the past. Most of this noise is above 20 kHz, being switching noise from the filament supply. Even a sensitive power amp, such as a vintage Heathkit or Harmon-Kardon, should be compatible with this circuit.

There you have it: a preamp that practically screams "audiophile extremist". It is an excellent test bed for tweaks of the acoustical variety. More important, it demonstrates the amazingly clear, clean sound of the thoriated filament triode at its best, and with not a trace of feedback. This is the kind of DIY project that you will want to show off to friends and family, it's both sonically remarkable and physically attractive.

ELECTRICAL PERFORMANCE OF THE SV572-10 LINE STAGE ---------------------------------------------------------------------------- Voltage gain at maximum volume setting into 1-megohm load: 9.2 Distortion at 1 kHz, 10v RMS output (28.2 volts peak-to-peak), with 1 megohm load:         THD--0.075% left channel, 0.10% right channel         IMD--0.42% left channel, 0.38% right channel         with 10k ohm load: THD  0.39% Frequency response, 1 megohm load, volume control at maximum:         -1 dB            46 Hz            34 kHz         -3 dB            16.5 Hz          85 kHz Frequency response, 10 k ohm load, volume control at maximum:         -1 dB            30 Hz            53 kHz         -3 dB            11.4 Hz          190 kHz Available dynamic range above 1 v RMS  at 1 kHz, 1 megohm load: 42 dB

REFERENCES -----------------

Hisashi Ohtsuka, "SV572 Test Report: Modernized 211--New DH Audio Power Triode", MJ AUDIO TECHNOLOGY magazine, September 1997, page 108. (English translation available from Svetlana Electron Devices--ask for Technical Bulletin No. 28.)

Eric Barbour, "Audio Preamp Tubes: A Survey", GLASS AUDIO magazine, 6/97 issue, p. 1.

ABOUT THE AUTHOR: --------------------- Eric Barbour is an applications engineer with Svetlana Electron Devices. He has been a staff editor with VACUUM TUBE VALLEY magazine since its founding in 1995, and has been a regular contributor to GLASS AUDIO magazine since 1991.

SIDEBAR: --------------------

Listening Impressions of the SV572 Preamp

Equipment used:  Dynaco CDV-Pro CD player (stock), Dave Wolze custom SV572 SE amps, B&W DM 110 speakers. Comparison based on CD player going straight into amps vs CD player to pre-amp to amps.

by John Atwood, One Electron Co.:

Effect on tonal balance: very neutral. Subtle expansion of soundstage, more "depth" to sound.  A more dynamic sound. Little noticeable hum, although hum and buzz from the Wolze amps may have masked pre-amp noise. Fan noise from power supply is objectionable. Microphonics from tubes is significant when chassis is tapped, or while pre-amp is warming-up. They did not cause problems when listening at normal listening levels. Conclusion:  Very good tonal characteristics, very clean.  Microphonics and power-supply fan noise keeps this from being a must-have preamp.

by Charles Kittleson, Editor Vacuum Tube Valley Magazine:

Over the last several months, Vacuum Tube Valley has had the opportunity to audition Eric Barbour's SV572 single-ended line stage. Most listeners, including myself were impressed with the detail and clarity of sound.  This preamp outclasses the "baby tube" 12AX7 and 6DJ8 line stages we have heard by the dozen.  The soundstage is huge with great depth of field. Eric's SV572 line stage, using the Electra-Print interstage transformers, brings a whole new meaning to detailed, clean sound.  I nicknamed this preamp "Big Clear" because of these qualities.  The only downside of this design is the SV572's sensitivity to vibration that can generate microphonics and instability if if the tube is not carefully isolated from any type of movement or vibration.  Through the correct use of tube dampers, chassis layout and rubber mounting of the sockets, this characteristic can be minimized.




Back to The Triode Guild               Back to Meta-Gizmo   

e-mail Dr. Harvey "Gizmo" Rosenberg:

Copyright© 2005™ and Dr. Harvey "Gizmo" Rosenberg      All rights reserved.
All the material contained within the above articles may not be reproduced without his express permission