High Quality Push Pull 6B4 Bias-Shift Amplifier

[WD5JKO]

  Hi All,

   I am attaching a document written back in 1947 from "audio Engineering" written by JR Edinger at Brook Electronics.

This AMP manages 30 watts from a pair of 6B4's which as about twice normal power, and they achieve this was a bias-shift scheme. In addition they go over audio amp "listening fatigue" from higher order harmonics, and high intermodulation distortion. They also touch upon the pit falls of high amounts of global negative feedback, and triode versus beam power tube sound.

I find the document very interesting, and applicable to what we deal with today with our plate modulators.

I hope the group finds this interesting.

Since I never attached a file before on this forum, in case it doesn't go through, here is the link to the file on the net:

http://www.tubecad.com/2007/12/15/High-Quality%20Audio%20Amplifier%20With%20Automatic%20Bias%20Control.pdf

Regards,
Jim
WD5JKO

[W3RSW]

Very interesting.  I just love reading about stuff that wans't too antiquated when I was a jn.

That amplifier would do justice to any modern plate modulated rig, a DC bias amp, no less.

Too bad most of the distortion in those days was from the record itself, the steel needle pickup, and the speaker(s)/cabinet.

Many on this board remember the little needle cup built into the record changer board of a consol.   You never trusted someone else's needle. (hmm, the more things change....   )  You always put your own in after careful inspection of the lot with your youthful 20 power eyes.  Oh yes, if in doubt you compared one to one of mom's sewing needles to make sure it was very sharp.  Parabolic rounding of the tip had not yet entered your vocabulary... as if you would believe such trash anyway.  The sharper , the better.

If you were really lucky the ol'man left out a pack of new ones.  ...or so he let you think? .   We'll never know.

I remember our first GE VRII, with (gasp) a sapphire needle.  A later model had a swivel mounted diamond and sapphire.   Much literature discussed 0.5 vs. 1 micron tips. 

[w4bfs]

Jim ... thanks for sending this one along ... Brook amps are not too well known, nowadays ... Paul Klipsch used a 15W model of theirs when he hit the road demonstrating his Klipschorn in even some rather large auditoriums ... if you have not heard a quality horn loaded system driven by a properly running single ended directly heated triode amplifier, and you can still hear, then you've missed a treat ... oh yes, the majority of often quoted even order harmonic distortion comes from an overdriven amp ... 73   ...John

[W2XR]

Yes, the Brook was a very high quality amplifier. Triode audio output stages in high-fidelity amplifiers were basically extinct in the late '40s when the Brook was introduced, having been largely superceded by the popular British Williamson circuit that used beam power tubes, although the initial version of the Williamson did use triode-connected 807s or KT-66s as I recall. Soon, most Williamsons were sporting 6L6s and their other beam power tube variants. Shortly thereafter, the very popular Ultra-Linear circuit was rolled out; I think this was developed by Acrosound and popularized by Dynaco, but I could be mistaken here.

The Brooks was probably the only consumer-grade power amplfier that could be easily adapted to use Western Electric 300Bs. Other professional-grade power amplfiers used 300Bs, such as those made by Western Electric, and Fairchild made several versions for driving the cutting head in a recording lathe.

Paul Klipsch made the famous comment when he first introduced his Klipschorn, that "what the world really needs is a clean 5 watt amplfier!" When driving an extremely sensitive/efficient loudspeaker of approximately 105 dB SPL at 1 meter in a living room environment, that statement is largely true.

I know of a local old-time ham who was given a new-in-the-box/never used Brook power amplifier about 10 years ago. A contractor he was familiar with was cleaning out an old home and found the unit there in the attic. He asked this ham if he was interested in it. He jumped at it, knowing full well what the Brook was. He was running a single (mono) homebrew Klipschorn that he had built back in the early '50s, driven by a homebrew Williamson amplifier. Talk about being lucky!

Beside amateur radio, homebrew vacuum tube audio is my other electronics passion.

73,

Bruce

[k4kyv]

Good transient response is accomp-
lished by extending the frequency range
and having negligible phase shift
between input and output of an
amplifier.
Transformer resonance can cause
slight oscillations well above the audible
range. If the phase shift at these
oscillation frequencies is sufficient to
cause the feedback to become positive,
regeneration will take place, resulting in
sustained oscillations and overloading
of the amplifier at super-audible fre-
quencies. This can be avoided by
reducing the gain of the amplifier
above  the useful frequency range, or by
preventing the occurrence of these
oscillations by proper design of the
output transformer.
The transformers are completely free
from saturation or leakage reactance
effects from 25 to 20,000 cycles at any
power up to maximum output. The
low frequency response is flat within 2
dB to 3 cycles. The extremely low
frequency response in the amplifier is at-
tenuated in the first stage to eliminate
the effects of transients in tuning a radio
receiver, or phonograph turntable ec-
centricity, or rumble. Some amplifiers
for industrial application have made use
of this good low frequency response.
The low-frequency  attenuation is ac-
complished by using a condenser in
series  with the input to the grid of
the first  tube. The complete schematic
is shown in Fig. 4.

This is consistent with United Transformer Company's recommendation that the frequency response capability be at least one octave above and below the target frequency response of the amplifier.  For example, if you are aiming for good voice quality at 100-5000~, the transformers in the amplifier should be at least flat from 50 to 10,000 Hz.

This explains why the old 1930's RCA broadcast transmitters boasted of a flat frequency response all the way to 30~, when there were few, if any, broadcast receivers in that era capable of reproducing undistorted frequency response down that low.

It might be worth experimenting to see if this circuit works with a pair of 2A3's or 6B4's as class B drivers.  If so, it might be used to clean up the audio in the KW-1.  Under  normal conditions, the single pair is marginally capable, at best, of driving the 810's to full output without distortion.  This is probably the reason for the splatter filter.  But to make this work, I suspect one would need to increase the plate voltage on the 6B4's and/or increase the step-down of the driver transformer, which as the article says, must have extremely good coupling between primary and secondary with near zero leakage reactance.

I have found that running the 2A3's in my transmitter above their recommended 300 plate volts, even with the bias increased to correspondingly reduce plate current and maintain the same plate dissipation, reduces tube life.  The schematic of the amplifier doesn't show the DC plate voltage on the tubes, and rapidly skimming the  text, I didn't see any mention of it.

[ke7trp]

My Grandpa built an amp very close to this.  As a young kid I heard the term "Brook amp".  At his house in NewJersey, He had 3 sets of Klipschorns.  He had a set in the main room, A single with the matching "stereo" small speaker in his office, and two in the basement.

All of these models where the corner style. They built into a Corner of the room.  He had a 100Watt per channel stereo in the main room. I remember being in the basement playing as a kid and having Both Horns come on FULL blast. It was deafening. All I could do was run up stairs. What I found was two cousins and a Brother laughing it up!  I got 200 watts of power into two huge Horns in a concrete basement..

Later, I wired a Sony Walkman to the Corner horn in the office one summer afternoon. The output of the walk man was enough to have my Grandma ask me to turn it down.

He also built two Citations.  They where housed in Closets behind his "Concrete folded Horns".   These where built into the house and contained univrisity Drivers and horns.  They also used some kind of High voltage Tweeter that required its own power supply.

When I turned 25, I realized that my Grandpa had two mint Citations and the Brook amplifiers.  I called him up to ask him for them.  He said "what in the hell do you want those for?  Last month I set them out on the curb, They where gone in 20 minutes"

[KM1H]

Interesting coincidence.

I was just reading a 4/54 CQ article by Bill Orr on using a 304TL in a bias shift circuit. He further states that the bias shift procedure goes back to a 6/35 article in Radio Magazine using a pair of 50's to get 25W out in Class A.  Another CQ article for 10/50 described a parallel pair of 304TH's with a syllabic bias shift circuit.

I'll scan and post the CQ articles if there is interest.

With the later introduction of big tetrodes the concept should fit those as well an use up all those half soft at RF 4-1000A's laying around

Carl
KM1H

[k4kyv]

The 2A3 will perform better with fixed bias than with cathode bias.  Reason is that for full output you need to run them in class AB1 service.  With class AB1 the plate  current kicks up on voice peaks.  That causes the bias to become more negative just as the tubes begin to draw more plate current, just the opposite of what is needed to maintain output at minimum distortion.

With cathode bias the undistorted output is rated by RCA at 10w per pair.  With fixed bias they are rated at 15 watts per pair.  But the circuit just described shifts the bias to be less negative when the tubes draw more plate current.

[Bill, KD0HG]

The audio gear upstairs is a homebrew Williamson stereo amp, pair of KT-88s per channel driving a set of Klipsch Forte II speakers. 50 watts will blow you out of the house.

I have to question why speakers evolved into such super inefficient boxes. The Bose 901s cost $1,300 a pop, eat up to 500 watts per channel, and ultimately don't sound as good as the Klipsch line, IMO.

Doesn't matter..My kids are totally happy listening to MP3s.

[k4kyv]

I have to question why speakers evolved into such super inefficient boxes. The Bose 901s cost $1,300 a pop, eat up to 500 watts per channel, and ultimately don't sound as good as the Klipsch line, IMO.

So they can make them "small".  Large bass-reflex boxes with tuned ports evolved into "bookshelf" models.  They just fill a small box with fibreglass insulation and seal it tight with no port or anything.  The old infinite baffle concept, compressed garbage compactor style, into a tiny box.  That insulation has a lot more mechanical friction than air, so most of the audio energy goes into compressing, decompressing and ultimately heating the fibreglass.

I checked out a Bose display room once, and was amazed at the sound that was coming out of a pair of tiny boxes, each about 4" cube, suspended from the ceiling on cords.  I was also impressed by the price, almost $2000, and I don't think they even displayed the amplifier, probably hundreds of watts, used to drive them.

Just like radios.  The public has been led to believe that smaller is better.

Doesn't matter..My kids are totally happy listening to MP3s.

Just as kids in the 70's were totally happy listening to hissette tapes.  Again, smaller is better, regardless of what comes out from the product.

[WQ9E]

The audio gear upstairs is a homebrew Williamson stereo amp, pair of KT-88s per channel driving a set of Klipsch Forte II speakers. 50 watts will blow you out of the house.

I have to question why speakers evolved into such super inefficient boxes. The Bose 901s cost $1,300 a pop, eat up to 500 watts per channel, and ultimately don't sound as good as the Klipsch line, IMO.

Doesn't matter..My kids are totally happy listening to MP3s.

Bill,

I have been putting together a file of the construction articles for the folded corner horn speakers that appeared in various issues of Radio and TV News and a future retirement project is to build a pair along with a pair of amps using a pair of NOS UTC LS series output transformers I picked up at a hamfest.  In the meantime, the vintage audio equipment in the radio room is from the 70's era and consists of a Marantz preamp, amp, and tuner along with an Akai GX-400D SS reel to reel and a later vintage  Marantz CD player.  The speakers are a pair of Bose 901 Series 3 which provide more than comfortable volume from the 70 watt per channel amp.  The later Bose 901 (Series 3 which came out in the mid 70's and later) are efficient ported enclosures but the originals do require a lot of power.

But before the K-horn type enclosures I am building replica bass reflex floor cabinets to go with my RCA AR-77 and Halli SX-28.  When winter rolls around I tend to spend more time in the woodshop than on the radio bench.

Rodger WQ9E

[WD5JKO]

I was just wandering around the web site where I originally found the Brooks amplifier article written in 1947. To my surprise a fellow named John Broskie did a thorough analysis of the Brooks amplifier design. This was done in 2007, just a few years ago. Here is the link:

http://www.tubecad.com/2007/12/blog0128.htm

Enjoy!
Jim
WD5JKO

[W3SLK]

Bill said:

The audio gear upstairs is a homebrew Williamson stereo amp, pair of KT-88s per channel driving a set of Klipsch Forte II speakers. 50 watts will blow you out of the house.

Didn't you do an article in ER about that amp? I seem to remember you analyzing the KT-88's, 6L6's, EL-34/6CA7's. I thought that you said the EL-34's performed the best. Nice amp none the less.

[Bill, KD0HG]

Bill said:

The audio gear upstairs is a homebrew Williamson stereo amp, pair of KT-88s per channel driving a set of Klipsch Forte II speakers. 50 watts will blow you out of the house.

Didn't you do an article in ER about that amp? I seem to remember you analyzing the KT-88's, 6L6's, EL-34/6CA7's. I thought that you said the EL-34's performed the best. Nice amp none the less.

Yes I did, thanks.

The amp is still sitting upstairs in the 'Great Room'.  You know, the older you get, the faster time goes by and it's been almost 20 years up there.

I'm listening to Miles Davis on it. Someday, I hope my kids appreciate it.

[N3DRB The Derb]

It may just be me, but I cant hear the difference between a cd and a 256K VBR AAC rip of the same song. I can tell a big difference on a 128K rip - lots of stuff goes down the drain.

There's practically no tessitura at all !  

But how much quality does one need to listen to Afroman, anyway?   

[WBear2GCR]

It seems to me that what the Brook amp is comes down to a "voltage regulator circuit" where the Vref is the Vdrop across the cathode resistor, integrated by the time constant capacitor.

This presents a problem in actual operation.

Since the text indicates that the time constant is on the order of 6 seconds this means that it will work fine for gradually increasingly louder program material, or where one turns up the volume gradually. It fails miserably when one is dealing with a loud transient, since it simply isn't fast enough to track.

Seems to me that it would be a better idea to simply make a class AB2 amplifier in the first place. Or if you must have class A (some of us must... for hi-fi applications) to use bigger tubes or more tubes (paralleled).

If one reduces the time constant, say down to 1 second, or 1/2 second, it's still too slow to track transients (drum hits, 1812 overture, etc...). If you make it fast into the milliseconds, I suspect it works very badly as it starts to look like some sort of LF modulation to the tube and maybe the output iron as well... clearly if the tubes don't track well if you modulate the bias quickly then there is a DC offset in the output iron to contend with... and if that DC offset is happening at say <10Hz it looks like an AC signal at some frequency. Which, is still too slow to track any transient!

This in short is the inherent problem with all "bias tracking" schemes - they lag, and are useful for longer passages that are louder than the average program material, and not much else.

Other schemes have been devised to handle this sort of scenario, one of them is the use of a "composite" tube, consisting of a triode in class A, with another tube (often a pentode, not sure why) that is biased in class B or more like class C. The second tube acts like the second rail in a solid state class H amplifier, conducting only when the level reaches beyond some threshold. Until then then is a usual class A amp. Of course there is the hand-off area that looks a little like "crossover distortion" as for a brief period both tubes are still conducting so there is a gain non-linearity there. But in this scheme they share the same output iron so that is mitigated to some extent by the transformer being somewhat of an integrator... some feedback is usually employed. This too is not a new scheme...

Imo, build the right size amp for the job, skip the monkey buisness as it rarely has many benefits - otherwise it would still be popular with the DIY high-end crowd today.

                        _-_-bear


Derb, CD is inherently a flawed storage medium... it really isn't that good to begin with...so ur comparing two not-so-good things... 

[WD5JKO]

Bear,

   You make many great points! When I first saw the bias-shift technique used in the Brook amplifier, my thoughts pondered the R-C attack and delay times, and whether there would be periods of overdrive with a quick rise in signal level, and other AGC like effects that might be audible.

   This technique may have serious issues with true high fidelity as we know it today, but in a AM modulator where the dynamic range needs are far lower, the Brooks approach just might be the ticket.

   One ham radio application I am thinking of is my Gonset G50. The G50 has dual 6L6 class A single ended Heising modulation. On my G50, I have made many modifications, see reply 7 at:

http://amfone.net/Amforum/index.php?topic=18961.0

   I use two values of cathode resistance, a higher value for receive and a lower value for transmit. I think a Brooks circuit like the transistor circuit at the bottom of the following link just might be the ticket for my G50, scroll down to the bottom for circuit:
http://www.tubecad.com/2007/12/blog0128.htm

   So do you think that for an application such as my G50, is the Brooks circuit worth constructing? With my G50 as it is now, the cathode bias voltage rises as I approach 100% modulation leading to "limiting" when the input sustains long enough. I can only hit 100% + modulation on the occasional voice peak; anything longer than that moves the cathode bias upward.

Jim
WD5JKO

[KM1H]

When I got back on 6M AM about 20 years ago it was with a G-50. No amount of circuit diddling gave full audio. I picked up an Eico modulator at the next hamfest and it made all the difference even with its flaws which I didnt know about then. A few years later a Clegg Zeus and Interceptor came home and are still in use.

Carl
KM1H

[k4kyv]

To get greater percentage of modulation you need more than just bias shift.  Either increase the plate voltage or reduce the modulation transformer step-down.  Shifting the bias will pull the instantaneous plate voltage only slightly closer to zero at maximum plate current, and this is what defines the peak a.f. output from the tube.  Modulation percentage capability is pretty much defined by modulation transformer ratio and the modulator plate voltage/final amplifier plate voltage ratio.

In the case of class B driver service the peak AF voltage output is again limited by DC plate voltage, and bias shift may only slightly increase the peak output. To take advantage of this increase, the driver transformer step-down ratio needs to be increased accordingly, so that the peak af voltage capability to the grids is the same (assuming that the amplifier initially was capable of fully driving the class B modulator).  Increasing the voltage output capability while increasing step down will improve the internal resistance of the driver stage and lower distortion.  It is a waste of driver capability if the modulator stage reaches saturation on peaks before the driver stage saturates.  Ideally, they both should hit saturation at precisely the same time.

Better driver regulation can be accomplished by increasing plate voltage while increasing transformer step-down.  But I tried that years ago, increasing the plate voltage of 2A3's from 300 to 400 volts while reducing plate current from 40 ma/tube to 30 ma/tube, maintaining the rated plate dissipation.  Within a few months, a quad of good 2A3's each tested weak.  I replaced the 2A3's, returned to 300 volts after substituting a new a class B driver transformer that was rated to carry the full 80 ma of plate current each side of primary midtap to the 2A3's at that voltage.  After several years of use, I tested  those 2A3's and they were all still good.  The only thing I can think of that killed that first quad was the increased plate voltage.  RCA must have had a good reason to give them a maximum voltage rating of 300.

Perhaps this technique would be more useful for using a pair or quad of triode  connected 6L6's, than for 6A3's or 6B4's.  The latter are too hard to come by to risk shortening their life.

A similar bias-shift concept has been used for AM linear operation to allow higher carrier output from a  given tube.  At resting carrier levels, the PA tube functions more as a class C amplifier with higher efficiency and lower heat dissipated by the plates.  But under modulation it shifts more to class B linear operation for undistorted output.  Sort of like controlled carrier, except that the carrier output level remains constant, shifting the efficiency of the tube rather than carrier output to keep plate dissipation within a safe limit.  IIRC, this is what they call a "class B-C" linear.

[N3DRB The Derb]

400 jolts on them things is right at the edge. have you checked the plate current on em? I think the highest Pv I've ever seen on a published design using 2/6A3's was 320 volts. I think you might be are piss beating them into submission. 

below 400 watts a pair will do OK, above, you gotta have 4, unless you have everything perfect, and your ratios are perfect.

[WBear2GCR]

Jim,

The only way the bias shift is valuable is if the circuit is class A for a PP circuit. So why not just fixed bias the tube for AB? Preferably find a way to run it in AB2 for some extra drive and output...?

I am on a dialup tonite, so not going surfing for the links - but you say your rig is SE - so I'm not sure how bias shifting will buy you anything... maybe I can look tomorrow... took about 5 mins to load this page!! No kidding!

                      _-_-bear


Edit: ok, you could do electronic bias shifting in the cathode like Broskie shows, but since I don't know the details of the modulator/audio amplifier stage that apparently does both jobs in the Gonset, I'm not sure if it makes sense or not, or if it will work. I looked at the links 

[WD5JKO]

Brian,

  I was looking for the info on the 6B4, and here is what I found:

http://www.triodeel.com/6b4g.gif

 Says max ratings: 450V, 100ma, 15 watts AND "The tube cannot be exploited at two or more limited conditions".

  I'm not sure, but maybe this is for the current production Sovtek 6B4G..

http://store.triodestore.com/sovtek6b4g.html

Bear, thanks for looking! Isn't dial up fun? 

Jim
WD5JKO

[WBear2GCR]

The Ruskie tubes are generally speaking new designs, based upon the old USA tube types.

In some cases where a nearly equivalent Ruskie tube already existed they rebranded the type - but many of them are in essence new tube designs and have different ratings than the NOS.

             _-_-bear

Jim, I read the bit where Gonset wanted to use the class A modulator scheme to assure a smooth clipping characteristic, as in "clipper/limiter" for the transmit side - more "talk power" and all that. If ur not interested in that feature, it might make perfect sense to go AB2 and just get more power, put in the 3 diode limiter and wail! 

[k4kyv]

AND "The tube cannot be exploited at two or more limited conditions".

Any explanation as to what that is supposed to mean?  Looks to me like someone did a poor job translating from Russian, or tried to write it in English with poor command of the language.

[WBear2GCR]

AND "The tube cannot be exploited at two or more limited conditions".

Any explanation as to what that is supposed to mean?  Looks to me like someone did a poor job translating from Russian, or tried to write it in English with poor command of the language.

Don't exceed:

plate voltage AND current
plate voltage and screen voltage
plate voltage and screen current

etc...

ie. you might get away with pushing the limits, if ur careful...

Just my impression...

            _-_-bear