• WARNING: Tube/Valve amplifiers use potentially LETHAL HIGH VOLTAGES.
    Building, troubleshooting and testing of these amplifiers should only be
    performed by someone who is thoroughly familiar with
    the safety precautions around high voltages.

What am I doing wrong? (300B HF Boost)

I’ve been working on my 300B amp and after some subjective listening tests, my main issue is the roll off on the top end… sure enough, it drops off after 10k… So after some experimentation I decided to “voice” it away by playing with the bypass scheme of the second stage… Seems to have flattened things out…. I lose some input sensitivity but it was very sensitive to begin with…

Now… this is where I pause and think
- There must be a reason i’ve never seen this implemented anywhere else that i’m aware of… what am i not considering?

- I feel like this isn’t addressing the main issue of “something something miller effect” it’s remedial… But the only “real” solutions seem to require unobtainium loctal tubes, interstage transformers that I don’t have space for, or some sort of mosfet follower that I don’t think I can easily implement into this topology…

- there might be something else in my circuit causing the rolloff

Here is my quickly drawn schematic… Missing is the details of the HT supply - it’s fairly conventional. What I did draw is the filament supply since once again, I did something I don’t see in other schematics and thus might be a factor in my HF rolloff behavior:

A0942B72-5CF4-445C-9BCA-68A80E6478AA.jpeg


Before (purple) and after (green) bypass tweak:
C642B059-2135-442F-B4D2-5BD4DD76BD6E.png


How does it sound? I don’t know… at this point I have “tunnel hearing” and don’t trust my ears anymore.

Any advice or comments would be appreciated! thanks!


Updated schematic:
415AF2CF-61FC-4EFD-8B1F-1F778C43900C.jpeg
 
Last edited:
  • I tried a different OPT, from an Elekit TU-8800, it was very similar but marginally worse.
    • Maybe I need to try a smaller OPT? So far my tube experiences have been with smaller OPTs powered by pentodes.
    • I can also swap this amp's Lundahl LL1664 in a different amp I have with the (smaller) OPTs to see if it rolls off the top end, to rule out the possibility of it being the transformer.
  • I tried a different set of 300Bs
    • there was a difference in output level
    • no difference in frequency response.
  • Initially started with the "typical" 6SN7 topology... Was not satisfied for the same reasons others are not satisfied with it, such as not enough drive.
    • My constraint is that the driver needs to be ONE tube per channel. Limited by the EH Scott 510 chassis I'm working with.
    • I figured that maybe if I moved to a different type of front end (such as this 6AW8A) it would take care of the HF rolloff issue, while much improved over the 6SN7, the top end looks the same.
    • Perhaps I need to explore more front ends?
  • Might be a matter of taste?
    • Accustomed to pentode amps with smaller transformers,
      • Prefer pentode mode over UL or even Triode in these amps. I like the "glassy" top end pure pentode mode appears to have.
      • I just figured the "triode" mode in these amps just "suck" because they aren't a "real" triode. Which I found to be a bit too "polite" sounding.
    • The music I listen to...
      • Tends to be hard, dynamic, with a lot of top end action (busy high hats) electronic music, psych rock, etc..
      • Haven't tried this amp with my jazz collection yet, or what little folk or classical I own.
    • Maybe I just need to stuff this amp away with my RSPs (sort of like a Canuck 401k lol) when my tastes change/mature...
  • Expectations
    • This has been my project for the past two months - I think I had a natural tendency to expect a lot from it. Especially since I've applied everything I've learned from building three prior amps.
    • 300B hype - Because of how this tube is regarded as the "holy grail" of tubes, I think I was expecting to be blown away even with a lackluster generic 6SN7-driven topology.
Q1 doesn't look right.
Quite possible - But that part appears to be doing its job. it's just supposed to be an NPN running as a zener booster as a rough form of regulation to keep the filament voltage between 4.98 and 5.1v.. If you're talking about the pin numbers yeah - I just picked the first NPN schematic symbol I found in KiCad which may have a completely wrong leg assignment. oops. I went this way instead of of a VR because I don't have quite enough voltage headroom coming out of the PT.. and from what I understand, this doesn't need to be completely ripple-free, just not as obtuse as 5VAC.
 
Last edited:
I've the same issue with a pair of Radiotron 300B amps I built. Tapers off more than yours do.

I ha
RadiotronSE2A3-PS.jpg2.jpg
Radiotron w mods.jpg

Advice I've received from a trusted internet friend is to triode strap the 6C6 which should boost the top end at the expense of sensitivity (which I can afford). I am planning on working on that this winter as I am happy with the 3C24 amps I built from the same guy's schematic.
I don't know if that's applicable to your situation.
 
Honestly? It's a single-ended, no-feedback 300B amp, and you're expecting it to play "hard, dynamic" electronic music with a ton of high-frequency information. That's not typically a winning combination. ;-) Regardless of the claims made by transformer makers, a low-transconductance triode with an SE output tranny carrying standing current is going to be frequency-limited.

I suggest you try some feedback. Local feedback from the output tube to the driver might help, but IMO the output tranny is your limiting factor. Try a small amount of global feedback, like 6dB. Elekit uses it, Grant Fidelity used it in their 300B amp. Remove the tweak on the driver cathode and use the extra gain to apply a small amount of global feedback and see what you get.
 
  • Like
Reactions: 1 user
triode strap the 6C6
A single pentode is one of the next things I intend to try, but I don't have any 6C6 on hand and finding them as a pair seems a bit of a challenging endeavor.. I'll have to explore other pentodes.

As far as Q1 I'm confused how it works. It can't be shorted across the rails like a shunt.
That's pretty much how it works, Q1 start conducting and shunts whatever is above Vz, (which makes me realize, there should be a resistor going from the base of Q1 to the negative rail). By the time it goes through the 1H choke and to the filament, it's between 4.98 and 5.1v depending on the tube. Full disclosure this is just something I found on google and tried out. The reason why I chose to do this is because my voltage is in the awkward region where no series device will get me to where I need to be. First time I did this without the transistor, but the lone zener was dissipating too much, with the transistor, this whole bit runs nice and cool. I get the feeling that the Bridge Rectifier I hastily chose (GBP406, not GBU8J) is wasting/dissipating a lot.. thermal cam shows 60-70+ degrees C
Do you high frequency measurement at the different stages.
I have not... but I would like to, I haven't investigated a safe way to probe those points and into my soundcard... any ideas? schematics? (short of buying an expensive USB scope/Analog Discovery)

I suggest you try some feedback. Local feedback from the output tube to the driver might help, but IMO the output tranny is your limiting factor. Try a small amount of global feedback, like 6dB. Elekit uses it, Grant Fidelity used it in their 300B amp. Remove the tweak on the driver cathode and use the extra gain to apply a small amount of global feedback and see what you get.
Will try this recommendation! I'll see if I can get some NFB working with the Lundalhl transformer, if not, I still have the Elekit TU-8800 opt which has a NFB winding, but I was saving that for a different project.
Oh, also, what speakers are you driving?
A DIY thing which comprises of an Eminence Delta Pro12A woofer, JBL 2445 w/2380A horn for midrange, and Beyma CP25 as the tweeter... I've spent at least 2 years refining this particular speaker set up and sounds good with most amps I throw them on. But I very conservatively approximate at least 95db... (limiting factor being the woofer which is 99db/w which runs through a 0.35ohm dcr choke... I even made the compromise of "downgrading" to iron core just to get down from 0.66 ohm of the air core I was using before. Still A/B listening between the two) The biggest challenge and surprise was the tweeter, started with the Beyma CP21/f but it got on my nerves, tried a whole bunch of other tweeters which were too narrow in dispersion, splurged on a set of JBL 2405 tweeters wasn't satisfied... ultimately ended up with the Beyma CP25 which sounded the most natural to me, one of the cheapest tweeters I tried! Sometimes you don't get what you pay for lol... anyways, i'm getting off topic here
 
Last edited:
You're not off-topic at all. The speakers are critical to the combination and the ability of the amp to drive them. They may be very efficient, but depending on the crossover complexity and the impedance characteristics, the amp may not be able to develop enough power at the frequency extremes to give you the sound you want.

I am not technically proficient enough to suggest exactly how to implement the feedback, but a rough guess would be a 100K pot from the positive output tap back to the input tube's cathode, and dial it down until you get a small amount of feedback. And be sure you ground the negative output terminal.
 
I am not technically proficient enough to suggest exactly how to implement the feedback, but a rough guess would be a 100K pot from the positive output tap back to the input tube's cathode, and dial it down until you get a small amount of feedback. And be sure you ground the negative output terminal.
That's ok - I'll look at a bunch of other schematics and steal, mix, and match as I have been with the rest of this amp :)
You're not off-topic at all. The speakers are critical to the combination and the ability of the amp to drive them. They may be very efficient, but depending on the crossover complexity and the impedance characteristics, the amp may not be able to develop enough power at the frequency extremes to give you the sound you want.
Good point! Maybe I'll do an impedance sweep of the system as a whole, run some in-room measurements and see if the issues correspond with with impedance peaks and dips...
For the music mentioned try some 6L6/6L6GC PP amp
That's an upcoming project.. never done a PP amp before... but I got a pair of Hammond 1650N for a good price that i've been sitting on waiting to be turned into an amp... But I don't want to give the impression that I'm a bass-head... while I like music that tends to be bass-heavy, I actually prefer to listen to it with leaned-out bass, but full, snappy midrange and detailed highs. Melodies and shimmer! Can't have too much bass anyways because I'm in a condo.
 
  • Like
Reactions: 1 user
Administrator
Joined 2004
Paid Member
Honestly? It's a single-ended, no-feedback 300B amp, and you're expecting it to play "hard, dynamic" electronic music with a ton of high-frequency information. That's not typically a winning combination. ;-) Regardless of the claims made by transformer makers, a low-transconductance triode with an SE output tranny carrying standing current is going to be frequency-limited.
Funny, I do just that with mine and am down just 1dB @ 40kHz. (D3A hybrid mosfet mu follower into fixed bias EML300B into MM amorphous core OPTs).

Low source impedance drive is what is required. Measure bandwidth of each stage to determine where the problem lies.

Check the ESL/ESR of your output stage cathode bypass capacitor over frequency.

@WntrMute2 I would remove C4 and C5 and remeasure. The high driver source impedance is not helping either.
 
  • Like
Reactions: 1 users
Funny, I do just that with mine and am down just 1dB @ 40kHz. (D3A hybrid mosfet mu follower into fixed bias EML300B into MM amorphous core OPTs).

Low source impedance drive is what is required. Measure bandwidth of each stage to determine where the problem lies.

Check the ESL/ESR of your output stage cathode bypass capacitor over frequency.

@WntrMute2 I would remove C4 and C5 and remeasure. The high driver source impedance is not helping either.

Like I said, it's typically not a winning combination. ;-) With a more complex and refined (not to mention more costly) build I'm sure you can get better bandwidth. Good suggestion to check the 300B cathode bypass cap.

The Elekit 8600 achieves good bandwidth with a not-too-complicated circuit but employs global negative feedback with the Lundahl outputs.
 
  • Like
Reactions: 1 user