diyAudio

diyAudio (http://www.diyaudio.com/forums/)
-   Tubes / Valves (http://www.diyaudio.com/forums/tubes-valves/)
-   -   300B (or other DHT) biasing options (http://www.diyaudio.com/forums/tubes-valves/207968-300b-other-dht-biasing-options.html)

tomchr 2nd March 2012 07:29 PM

300B (or other DHT) biasing options
 
2 Attachment(s)
Folks,

I'm trying to understand something here.... I've been messing with a 300B amplifier design for 2+ years now and am now back to square one. My current design sounds too harsh for my taste.

Output stage: There are several options for biasing a 300B (or any other DHT for that matter). Some people swear that one topology sounds better than another. I'm trying to understand what would cause the difference.

For the sake of simplicity, I'm limiting the options to three; fixed bias, cathode bias, and Loftin-White (cathode bias but with the cap to B+). Attached shows the schematic for the three. I have considered using a CCS, but that opens up a different can of worms.

I intend to set up a test bench for all three configurations and try them one by one. For those of you who have already made this experiment, what were your findings?

~Tom

Eli Duttman 3rd March 2012 01:43 AM

IMO, you need to add a 4th option to your list, combination bias.

The advantages of combination bias include: resistance to runaway, a reasonable amount of heat generation in the cathode resistor, and (if the smallish cathode resistor is unbypassed) some linearization of the O/P tube. The 2A3 is more linear than the 300B. Remember that Harvey Rosenberg, a lover of the type, unambiguously stated that 300Bs put on a show.

DHTs are vulnerable to runaway. When simple "fixed" bias is employed, a small value is essential in the grid resistor value.

tomchr 3rd March 2012 01:58 AM

Combination bias? So some amount of the Vgk is by dropping voltage across a cathode resistor and the rest applied to the grid? That sounds intriguing. How much of the total bias would you suggest to apply by cathode and fixed bias respectively, then? Like 80/20 split between fixed/cathode? What's a good starting point?

~Tom

Eli Duttman 3rd March 2012 02:27 AM

100 Ω as the cathode resistor makes setting the "idle" current a cinch, but might not provide enough protection against runaway. 150 Ω is not especially difficult as a test point from a math perspective, while offering a fair improvement in the runaway protection dept.

It is possible to cap. bypass a portion of the total cathode resistance. That allows for superior stabilization of the operating point, without an excessive increase in the effective RP from local current NFB (degeneration). 100 Ω bypassed by 470 μF. stacked on top of an unbypassed 100 Ω might be a very effective setup. There's plenty of protection against runaway, the math is easy, and the amount of degeneration reasonable.

Magz 3rd March 2012 03:01 AM

What about putting a CCS on the plate to give protection against runaway?

andyjevans 3rd March 2012 09:32 AM

I've just been doing some 300b biasing. I had insufficient HT, so I had to look at combination bias. Right now I have 388v on the 300b anode, 24v on the cathode with a 407 ohm cathode resistor bypassed with a 100uf polypropylene cap. 60mA of current, so 364v a-k and 21.5W dissipation. The 300b is fed by a 46 driver through a 126C Hammond interstage, and the bias voltage is applied to the bottom of the 126C secondary. Very good interstage, incidentally.

At the same time I was biasing the input stage I looked at biasing the 46. Battery grid bias was clearest, but not as smooth as cathode bias. However, bypassing the cathode on the 46 was not successful. There was a small gain in clean treble, but that was very much offset by a monotonous blurring of the whole sound - the harmonics and timbre of acoustic instruments were less rich, and the whole sound was frankly boring. An electrolytic was worst (quite awful) and a polypropylene cap was better but still nowhere near as involving for the listener as no cap. Bypassing the cap with smaller teflons made no difference.

So I'd say the worst choice would be cathode bias with an electrolytic. For the output stage bypassing with a good film cap is acceptable, but I would look for other options for input stages. For the bias supply itself, I'm not satisfied. Just a cap resulted in a slightly shrill treble. CRCRC with electrolytics and 1K resistors was a little better but treble still dirty. Changing the last cap to a 47uF polypropylene was better, but treble still not silky smooth. I currently have UF4007 diodes, but no doubt schottkys would be better. So as I say, I'm not satisfied with the bias supply yet. I actually posted this week on this subject but got no useful replies. I've seen an EZ81 choke input supply used for this, and some have used batteries (!!!). So as things stand, I'd go with cathode bypass with a polypropylene cap (NOT electrolytic). With combination bias the sound is clean with good bass, but this may be due to other factors.

For my preamp I use filament bias which is the best I know. Look at the "26 pre amp" thread - posts around 1 and 2 March where Rod Coleman suggests a way of using filament bias with bigger tubes like 46. Maybe 300b - haven't considered this. But I will say that even with Rod Coleman's brilliant filament supply boards which are now widely used, I like to add choke input to the supply.

Andy

tomchr 3rd March 2012 07:40 PM

Andy,

I agree that film caps are preferable to electrolytic caps in many cases, though, I've had good experience with the Nichicon KZ-series of audio electrolytic caps. They don't come in very high voltages, though. Currently, I'm using some Solen 47 uF, 630 V polypropylene caps.

I'm familiar with Rod's filament regulator. Personally, I've gone the switchmode route (see my website). I've tried both constant voltage and constant current. I've never been able to tell a difference between the two in listening tests or measurements and constant voltage is less expensive and much more efficient (especially when using switchers). So I went that route. But filament regulators are a different can of worms. One can at a time... :)

I think I'll get my butt into my lab and wire up some circuits. I'll keep you posted. Thanks for your input.

~Tom

SpreadSpectrum 3rd March 2012 08:01 PM

I've never built an SE am so feel free to ignore me, but maybe some of my experiences will help. I did an experiment a few years ago where I built a push-pull amplifier with triode-connected KT88s. I drove them with direct coupled mosfet followers but grid current was limited with a 1k stopper at the KT88 grid. Fixed bias was applied to the gate of the mosfets. I cranked up some music, turned the volume way up, and watched amp output on a scope. The music was really up loud and I would didn't sound perfect but I thought it was pretty good. Shockingly, the amplifier was grossly distorting (I was well into clipping I could see on the scope) but still sounded decent, not great but not nearly as bad as I thought it would have by looking at the scope. I became a big fan of instantaneous overload recovery that day.

I later built a unity-coupled push-pull amp (McIntosh output circuit) and as a result of my driver design and the huge drive requirements I had a lot of 2nd harmonic in the amp output(for a push-pull amp). The amp measured well otherwise and I think I actually asked you at the time what you were getting for THD from your 300B amp and it was similar to what this amp was putting out at 1W. The problem is, when I played an electronic-type song with female vocals (very complicated song), the 's' sounds in the vocals just sounded off. I was guessing that it was just the bent transfer characteristic of the amp making IMD with complex music. I am in the process of rebuilding that amp right now and am trying to really get that transfer characteristic straightened up so that I can listen to things other than 'lady singing with piano'.

I would start with the question of why the current amplifier sounds harsh by taking some measurements (maybe you already did this, but really characterize the behavior of this thing) and try to address whatever the problem is. You could have an overload recovery issue or maybe you listen to interesting music and hate IMD, who knows?

45 3rd March 2012 09:16 PM

Quote:

Originally Posted by tomchr (Post 2930704)
Folks,

I'm trying to understand something here.... I've been messing with a 300B amplifier design for 2+ years now and am now back to square one. My current design sounds too harsh for my taste.

Output stage: There are several options for biasing a 300B (or any other DHT for that matter). Some people swear that one topology sounds better than another. I'm trying to understand what would cause the difference.

For the sake of simplicity, I'm limiting the options to three; fixed bias, cathode bias, and Loftin-White (cathode bias but with the cap to B+). Attached shows the schematic for the three. I have considered using a CCS, but that opens up a different can of worms.

I intend to set up a test bench for all three configurations and try them one by one. For those of you who have already made this experiment, what were your findings?

~Tom

I haven't tried the 3rd option but, at least on chart, using a good driver tube could be better than the other two but I can't tell for sure.
I wouldn't choose any of the other two. I would rather use a direct-coupled cathode follower. It is not so complicated if you don't demand too much in terms of positive grid drive. Many Audio Note amps use this solution but don't really use positive grid operation so much, just a little bit.
The reason for this is strictly related to the 300B and similar power tubes, not all them though. The 300B looks really good on chart but is quite demanding in terms of driving. All troubles and "hard work" are on the driver. If you think that one pair of 2A3's in parallel has total 30W plate dissipation without any problem, about 10000 umhos transconductance, 400 ohms internal resistance and no more than 110V peak-to-peak for light A2 drive is required you can see the difference! I guess the harsh sound you talk about is related to the driver. No matter what driver and bias you choose a 470K grid resistor and a coupling cap are not good enough for me. It's really a no-no...

Cheers,
45

tomchr 4th March 2012 05:33 AM

Quote:

Originally Posted by 45 (Post 2932097)
I would rather use a direct-coupled cathode follower. It is not so complicated if you don't demand too much in terms of positive grid drive.

Personally, I prefer a source follower for this. No issues with grid drive.... :)

Quote:

Originally Posted by 45 (Post 2932097)
The 300B looks really good on chart but is quite demanding in terms of driving. All troubles and "hard work" are on the driver.

No kidding!

Quote:

Originally Posted by 45 (Post 2932097)
I guess the harsh sound you talk about is related to the driver.

Or the speaker, actually...

Quote:

Originally Posted by SpreadSpectrum (Post 2932000)
I've never built an SE am so feel free to ignore me, but maybe some of my experiences will help. [.....] I became a big fan of instantaneous overload recovery that day.

Your input is always welcome. Yeah... I too have experienced the nasty effects of blocking distortion. That's what happens when the DC coupling cap to the 300B discharges during a transient, thus, moving the DC bias point of the 300B on recovery. It causes the speakers to emit some really rude sounds. Hence, my use of source followers as well.
However, with the present setup where I'm driving my 300B through an LME49811-based amplifier with 120 Vpp, 5+ A current capability blocking distortion is really not an issue... :) The amp can't actually deliver enough voltage swing to drive the 300B into A2.

Quote:

Originally Posted by SpreadSpectrum (Post 2932000)
I was guessing that it was just the bent transfer characteristic of the amp making IMD with complex music.

Do you happen to have measured data for the IMD? I don't... But it would be interesting to see.

Quote:

Originally Posted by SpreadSpectrum (Post 2932000)
I would start with the question of why the current amplifier sounds harsh by taking some measurements (maybe you already did this, but really characterize the behavior of this thing) and try to address whatever the problem is. You could have an overload recovery issue or maybe you listen to interesting music and hate IMD, who knows?

Well... I hooked up my sand amp. It's based on an LME49811 and is about as low THD as modern sand allows. 0.005 % THD up to clipping, basically. I still get the harshness to the sound - though not as much as with the tube circuit I had yesterday. So I suspect much of the harshness is due to the speaker. But I think I'd have to measure the woofer and tweeter separate to figure that out. It's possible that either is emitting rude sounds during the crossover transition.

~Tom


All times are GMT. The time now is 02:55 AM.


vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2014 DragonByte Technologies Ltd.
Copyright 1999-2014 diyAudio


Content Relevant URLs by vBSEO 3.3.2