300B NFB SET Integrated Amp
Copyright 2008 © Troels Gravesen


The project shown here is based on the construction found at: http://www.triodedick.com/. Click "Projects" and then "Stanley en Oliver". Only available in Dutch.

In 1986 I bought the book "Rörförstärkare" (valve amps) by Jean Hiraga and for the first time read about the queen of triodes, the 300B. This book is a Swedish translation and was sold by Danish "High Fidelity" magazine. In the years to come I built quite a range of valve preamps, but only one power amp, a push-pull EL84 that really wasn't very good: Loyes, "Grand Amateur 1960". I finally bought a secondhand PP EL34 from Luxman, MQ70, which I had until the early Nineties. For some 8 years I wasn't doing much hifi and had some unfortunate solid state amps and a CD player while all my vinyl was in the attic. Around 1998 I bought the CLIO measuring system, which really was what matured the idea of launching a website on DIY loudspeakers. There's a time for everything and from the last two years of work on high-efficiency speakers, I needed a wider range of low-wattage valve amplifiers, thus also a reference 300B SET amp - 22 years after Hiraga's book. Better late than never.

After some serious coupling cap tests, this amp delivers some of the most enjoyable sounds I've ever had. Despite 8 watts maximum output it runs the DTQWTs very well and I'm always looking forward to setting it up again after having used my other amps for testing speakers.

The 300B triode. Some references on the history of the 300B tube: http://www.antiqueradio.com/300B_01-98.html, http://www.stereophile.com/features/229/

Thorsten Loesch has these comments on the JJ 300B:
Made in the Slovak Republic in the former Tesla Valve factory this Valve has been around a while. The Valve has a substantial ceramic base and in general appears well made. Internal structures are a little different from normal. However the connection between Glass Bulb and base was worryingly loose and lacking in rigidity. I felt distinctly unhappy handling these Valves. While no valve with a separate base should be pulled out by its glass bulb this lack of solidity is worrying and possibly has an influence on the sound too. Both valves where identical so I do not think this was an isolated fault. The valves where pretty new but run in for around 100 Hours. Sonically this valve failed to convince, sounding bright, harsh and edgy. Soundstaging was lacking too. One positive side, the bass was quite strong and well defined but the overall tonal balance was subjectively objectionable and faintly unpleasant. This valve may have needed "burn in", however my experience has so far been that few valves change their sound dramatically after burn in. Moreover, the friend who bought them still reports similar sonics many 100 hours later. Microphonics are about average.

Bright, harsh and edgy... I couldn't disagree more. My JJs have a rock-solid base and are detailed/transparent with a wide soundstage and excellent tonal balance. For reference I had the AVVT and some cheap Chinese valves. The AVVTs have a more forward/up-front sound, not a good match with this amp - to my ears. Point is: Read any review on valves with a big grain of salt - including mine. There may be an awful lot of circumstances determining the outcome of a given construction.

Steven R. Rochlin has this on the JJ valves:
Although a bit microphonic, it seems to have the best balance of smooth mids with upper and lower frequency extension in my beloved Wavelength Audio Cardinal amplifier.  If fact i love it so much it now pretty much permanently resides on it as part of my main reference system, only giving way to the WE300B when i desire a more, um, seductive sound.  Ambience and imaging are excellent as it inner resolution too!  This is the latest 300B to arrive here in my humble abode and also the best so far for direct replacement on a normally run 300B amplifier.   Some of you might worry about a bit of that blue glow on the top, but then again so did the WE300B have that at the beginning too.  I feel there is just too much emphasis on that top blue glow.  Yes, it does mean a few things, yet it's not necessarily gonna make your system sound like junk, burn down a small village, and release a nuclear warhead on your government.  Sheeesh, relax and enjoy the music! The JJ Electronic 300B seems to have all the things i love about 300B tubes!  What more can I say then that?  May I humbly suggest you try the JJ Electronic 300B?   You too might find yourself in love while enjoying the music.

Right? I mean, take your pick on valves.

A complete set of transformers (five) were bought at
AE Europe/Holland including these JJ/300B.
Other tubes are 12SL7GT and NOS Siemens D3a. Someone told me the 7788 (E810F) were better than
the D3a, so I bought a pair. To my ears the D3a are the best, but difference is small, if any.

Construction pics
(back to top)

Left: The massive walnut wooden frame for the 300B amp was a generous gift from
John Eekels/Holland.
Right: Only a test set-up to view final appearance. Top panel is 3 mm alu.

Left: Binding posts and attenuator from Taiwan, phono sockets from Hong Kong, all trafos from
AE Europe/Holland,
resistors, caps and tube sockets from
TAD/Germany + local shops, tubes from everywhere; quiet a global project.
Right: 300B heaters from
Tentlab/Holland, quiet expensive, but no hum. Very important with 95 dB speakers!

Left: First all heater supplies and signal cables. All cables used are silver plated copper in teflon.
Teflon is great compared to e.g. PVC as it doesn't curl up during soldering and you need a lot of heat here and there.
Right: High voltage lines shaping up.

Left: This chunky mains trafo has a lot of secondary windings! Tube amps need a lot of different power supplies. Siemens D3a tubes in front.
5U4G rectifier tubes at rear. The lay-out is for a 12SL7 and obviously 6SL7 can be used, but order main trafo with 7V/2A tab rather than 14V/1A
and use 7806 regulator. I don't know why Triode Dick used 12SL7 as the 6SL7 - from my google search - is available everywhere. I had my 12SL7
from the US.
Right: Preliminary amplifying stage with whatever coupling caps I had a hand. Silver/gold caps have been ordered.

Left: Ready for turning on the power switch. Right: No smoke or bad smell!!
Time to measure voltages and see if I could reproduce the values given by Triode Dick and John Eekels.

Bottom panel.

(back to top)

Bold red is John Eekels measurements. In circles my measurements, left and right channel. Measurements done with shorted input.
Fluke true RMS meter used. My mains supply was 235 volts this evening, rather high.

CLIO measurements

5 volts RMS into 8 ohms resistor @ 1000 Hz.

The CLIO can be useful for amps too. Here THD = 2% @ 5 volts RMS into 8 ohms, 1000 Hz.

I calculated:
THD at 1000 Hz:
1% THD @ 0.84 watt into 8 ohms
2% THD @ 5.1 watt into 8 ohms
3% THD @ 6.6 watt into 8 ohms
5.4% THD @ 8 watt into 8 ohms

At 8 watts into 8 ohms the amp starts clipping. All as expected.

The coupling caps
(back to top)

It is claimed that only the 0.47 uF coupling cap has a significant impact on performance, where the 2.2uF and 0.1 uF caps are of minor
importance. Buying high-quality 2.2 uF is seriously expensive, so until later I have used a 2 uF Sonicap here.
For the 0.1 uF a Jensen copper should do well.

Left: Auricap 0.47 uF/600V. Right: Mundorf silver/gold/PP, 0.47 uF/1200V.

Left: Russian FT3 teflon caps, 0.22 uF/600V. Right: Hovland 0.1 uF/600V.

Left: VCaps, tin/teflon. Right: 2.0 uF Sonicaps replaced by 2.2 uF silver/gold caps (prototypes).

Left: Russian Silver Mica, here 40 nF for C1. 178 nF used for C3, see tweak section below.
Middle: PIO K40Y-9, 220 nF/1000V.
Right: Russian PIO (green), K42Y-2, 220 nF/500V

Had to wait some time for the silver-gold coupling caps. This amp is enormously sensitive to the quality of 0.47 uF capacitor. Put in a low cost Solen cap - and you have an amp you can listen to for hours and hours; smooth as velvet. Add a 100 $ silver/gold and all of a sudden you have treble extension and detail. Jantzen Superior cab did well too but I like these Z-caps better for speaker crossovers.

Before entering some comments on sound, let's see what others have found (always a good idea!):
My Capacitor comparisons Mundorfs, VCap, Sonicap Platinum, Auricap, etc
A good-sounding capacitor is essential!

A review with some surprising results:
Tempo Electric The Great Capacitor Shoot-Out

Interesting review from Chinese magazine (in English):
AV Frontline Magazine, November, 2006

Apart from the very expensive V-Cap Teflons, consensus is not the first thing that comes to mind after reading these reviews. As always: Read a lot and do your own tests.

It can be quite amusing reading what people come up with in terms of describing sound. I've collected a range of expressions here.
"- infinite shades of resolution riding on the oh-so-smooth sonic flow": How about that?

Caps tested:
1. Auricap 0.47 uF/600V
2. Mundorf silver/gold/PP, 0.47 uF/1200V
3. Russian FT3 teflon caps, 0.22 uF/600V
4. Hovland 0.1 uF/600V
5. Solen MKP-FC, 1 uF/650V
6. Jantzen Audio Superior Z-cap, 1 uF/800V
7. VCap, tin foil/teflon, 0.1 uF/600V.
8. Russian Silver Mica, here 40 nF for C1.
9. Russian Silver Mica, 178 nF for C3.
10. PIO K40Y-9, 220 nF/1000V.
11. Russian PIO (green), K42Y-2, 220 nF/500V

The tests here were done with the 300B amp driving the DTQWT speakers and front-end was vinyl/DV20X-high/Transcendent RIAA w. Mundorf silver/gold/PP coupling caps. Digital was Shanling T80 with silver/gold coupling caps and WE396A tubes in buffer stage.

To cut a long story short:

The Solens have already been commented. Smooth and easy on the ear, but (seriously) lacking in treble balance and resolution.

Russian FT3: Smooth, smooth as described by Jon, I couldn't agree more on his characterisation of this cab. Transparency is quite good and if you have trouble with sibilance, try this cap. Not my cup of tea.

Auricap: An overall smooth sounding capacitor with fairly good resolution - if it wasn't for the Hovlands just being better.

Hovland: This is a cap I could live with. Transparency is fine and overall tonal balance excellent if it wasn't for the Mundorfs being better.

Superior Z-cap: I need to try this again, because I couldn't decide how I felt about this cap. I love them for speakers, but as coupling cap.... not sure.

Mundorf Silver/Gold: There's no free ride when we're entering the very high-end range of capacitors. These silver/gold/pp have all the transparency I could ask for; they have the right tonal balance and listening to female vocals I found out I had to further fine-tune the VTA on my LINN tonearm! If vertical tracking angle was less than perfect the Mundorfs will clearly tell you from emphasized sibilance generated by poor cartridge alignment. Solen/FT3/Auricap would to some extent smooth things out here being more forgiving in upper mid/lower treble but we don't want caps to "repair" shortcomings in other places.

V-Caps: The king of caps - or is it? Fortunately we can get away with 2 x 0.1 uF for this 300B amp, some 100 US $. These caps are heavy despite their average size, but tin foil is significantly heavier than alu foil. When I play the VCaps, I miss the up-front aggressiveness of the Mundorf silver/golds and when I play the Mundorfs I miss the calm, authoritative presentation of the VCaps. Having to compromise, these VCaps probably are the best of them all. I don't find much difference in terms of transparency between the two; they are both extraordinary good. What I will try is having a blend of both from phono cartridge to power amp. In all I have 3 coupling caps here, two in the phono stage and one in the 300B amp. So, the set-up could be Mundorf + VCap (RIAA) and Mundorf - or - VCap in the 300B amp.

Russian Silver Mica, here 40 nF for C1. 178 nF used for C3.
PIO K40Y-9, 220 nF/1000V.
Russian PIO (green), K42Y-2, 220 nF/500V
The three caps here were all tested against the Mundorf Silver/Gold cabs and these PIO caps really aren't bad at all. Surprisingly good, actually. The Mundorfs may excell by a fraction in overall resolution, but I might as well pick one of these two PIOs for smooth and balanced presentation.
The Silver Micas are some chunky beasts (178 nF) and to cut a long story short, I might as well leave these caps in the 300B amp for good. These cabs can be stripped from the metal case (magnetic) and put into some wooden containers filled with wax. I got two of the 178 nF for 17 US $. Two 220 nF Mundorfs are ~82 US $.


This cap (C2) is claimed to only have minor significance to the overall presentation and I have been running the 300B for a long time with some 2.0 uF Sonicap leftovers from the Audio Mirror amp. Now this C2 is directly connected to the most important cap in the whole construction, C3. C3 needs to be the very best we can afford, but being under constant AC, like the C2, I have my doubts that C2 will not have an impact on the performance of C3. I'm not an expert in valve circuitry, so I have to try out stuff - and listen. The Sonicaps were replaced by silver/gold caps - and I think it pays off.

Swapping caps is a never ending exercise. We easily end up wanting one cap for a particular recording and another cap for another recording. No cap will be the best for all circuits and whatever is in front of your amp and behind. Tuning a complete system takes a lot of compromises in making the best match of components from our phono cartridge/CD-player to the final speakers.

I'm really happy I made this 300B amp. It's the best I've ever had although the Audio Mirrors are getting close with the latest mods. The 300B differs significantly from all other amps I have had. It has this expansive, holographic midrange I never heard from any other amp and from good recordings it can bring you close to the music like few other amps can do - that is - if connected to high-efficiency speakers. Don't count on 90 dB or less speakers to experience what 8 watts from 300B SETs can do. They need more, a lot more to reveal the dynamics it is capable of! The Audio Mirror amps have better low-end punch, but I can live with the more soft presentation of the 300B.
My 300B amp can run the
TQWT and DTQWT to ear-splitting levels without serious distortion - something the MAE1 amp couldn't (only TQWT tried). I guess the oversized trafos of the 300B amp plays a role here. 1½ watts difference can't be the whole story, but I'd like to hear the MAE1 again with some good OTs driving the DTQWT speakers.

300B mods
(back to top)

Mail from an experienced 300B builder (and my comments in red)

#1) The most important (sonically) capacitor is the 300B Cathode capacitor and the Powersupply capacitor for the 300B Stage, because they are such large values and electrolytics, even Black gates are quite bad.
Connect the best quality 22uF/450V (value quite critical, 23.8uF is theoretical ideal)  capacitor you can locate between the 300B Cathode and the top of the Output transformer and connect a resistor equivalent to 1/4 of the DCR of the Output Transformer but at least 47R between the +B capacitor and top of the output transformer. This 22uF capacitor now closes our output current loop at most frequencies (above around 300Hz), instead of the series connection of the two electrolytic capacitors, which help out at the very bottom end. See the Lynn Olson slides on currentloops at
www.nutshellhifi.com to understand the goings on.
Tweak #1 was tried and left on for about two weeks.
Trying modifications like these, the best thing to do is to stay with the modification for quite some time and the remove it and listen again. That is, if the modification is not immediately apparent in terms of improved sound. And this tweak wasn't. No improvement could be heard, neither immediately nor after 6-8 days of burn-in. After two weeks I removed the capacitor and restored the original set-up and again no change to the sound could be detected.
After a few hours of listening I came in doubt whether there was a slight midrange warmth associated with the 22 uF in place, thus I inserted the cap again and left it in for experiments #2-4.

#2) The second most important capacitor is the coupling cap, as you know. I use nonmagnetic silver/mica ones.
Silver Mica caps, 178 nF, were bought and replaced 0.47 uF Mundorf Silver/Gold caps. The micas were stripped from their metal case.

#3) The THIRD most important one is the bootstrap capacitor in the Mu-Follower, C1. This is another coupling capacitor. If I use mu-Followers I use 10nF Silver/Mica with a 4.7M Gridleak resistor instead of 470K/100n.
Silver Mica caps, 40nF, were bought and replaced 0.1 uF V-Caps. The micas were stripped from their metal case. Gridleak resistor was changed accordingly.

#4) The 2.2uF is also important, but most important here is a sufficient value, 2.2uF is definitely marginal, at best. I would personally prefer to clamp the screen grid with at least ten times as much...
22 uF is a lot! But having the 22 uF at hand from #1 tweak. I inserted this one. All other things "as-is" (Mundorf, V-Cap, etc.). No change to the sound could be heard from increasing the value 10x.

With tweak #1 in place, tweak #2-3-4 were all added at once and some 10 hours of burn-in was allowed before sonic evaluation began.
First of all, had I known these silver micas, I'd probably not bought the Mundorfs. That's how good I think they are. The Mundorfs may offer a fraction better transparency, where the micas seem to produce a more "naked" presentation, i.e. colouration seems very low. The micas do not in any way smooth things out, make things sound "nice", rather adds a fraction of presence compared to the Mundorfs. These two caps appear so close, it's really hard to tell the difference, and I don't think I'd be able to pick one from the other in a blind test.
After having had tweak #1-4 on for several days I restored the 300B back to the starting point and plugged it in. For some strange reason my brain wants to state
some difference, but I cannot pinpoint any significant change to the sound from going back. Again I don't think I would be able to pick one from the other in a blind test. And blind tests can be very tricky and the few times I have tried, I often find myself confused after a few minutes because you tend to swap back and forth too quickly, too many times, and after a while we can't hear a damned thing, in particular when trying to evaluate minute differences between coupling caps. I have to listen to a particular cap for a considerable length of time before making a change, and while a range of recordings are still in my aural memory. Choosing the best cap - from a range of really good caps - is a matter of taste and may depend on how the overall system in put together.
While doing the tests described here, I had the opportunity to borrow a TRAM w.o.t. line stage. After building the Grounded Grid line stage, I though there'd be peace on the line stage front for a while. This turned out not to be the case and despite a modest priced coupling cap (Obbligato) in the TRAM, this amp has a midrange transparency I didn't think possible. The TRAM amplifying stage is extremely simple, while the power supply is not, being valve rectified and valve regulated. Point in bringing this up here is that another circuitry may require a completely new cap test - with completely new results. The better the basic circuitry, the less critical our choice of coupling caps may be. We'll see once I've cloned the TRAM line stage.

From left: 40 nF silver mica, 178 nF silver mica, Mundorf Supreme silber/gold, V-Cap. Micas still in metal housing.

#5) Another thing to try is to add a gridchoke on the 300B to the Mix...

Tweak #5 and #6 will have to wait for another 300B project.

#6) Finally, I am not the greatest fan of the 12(6)SL7 Mu Follower. It is okay, but tends to sound a little mechanical and artificial. Given you already have a D3a (about the best driver tube for a 300B Amp I can think of) in your amp, why not use it? Wire it as triode with 100R from both screen grid and suppressor grid to the Anode. Use a 22K/12W anode load resistor (Mills), 240R cathode resistor and a suitable bypass capacitor (at least 140uF). I use 80uF/35V Film Capacitors, two in parallel, but these are virtually impossible to get, as are the Rubycon SWR series which includes 100uF/35V. Given the capacitor issue, if you like, find a 2V large size button cell (Lithium Type) plus socket and connect it to the grid of the D3a, negative to grid, parallel it with a 0.01uF capacitor of the highest grade (I use Silver Mica) and a 4.7MOhm resistor. Then connect the cathode to ground. Change Battery every two years or so, together with the tube. This simple triode stage to my ears is much better sounding than the Mu Follower with 6SL7 or 12SL7. If you don't like it, you can easily convert back.
Anyway, have fun. NN

Tweak areas

Tweak #1

Tweak #1 again, this time with 22 uF/630 V Obbligato caps and 47R Kiwame resistors.
Maybe - and only maybe - it adds a slighly lush tone to the midrange, where the un-modded situation sounds more
clean and - well, right. I'm sure I wouldn't be able to pick one from the other in a blind test. This tweak just simply doesn't pay off.

Left: Bogart at the counter and lady on the floor playing with flying saucers!? Right: I guess we don't...

Return to intro page.

(back to top)