ef83/6BK8 circuit examples

[ioaudio]

used in old tube tape recorder for automatic level control.
interesting tube?

 

[PRR]

> interesting tube?

Yes. Only tube I ever saw that boasted vari-Gm for A.F. use. (Most are made for radio signals.)

Also called 6BK8. TheTubeStore.com wants $5 each.

> used in old tube tape recorder for automatic level control.

Have you seen such a schematic?

 

[ioaudio]

it was used in the tape recorder i took apart - grundig tk19 automatic ;-
i´ll try redrawing the mic pre/leveling section.

edit: only schematic i found regarding the ef83 so far:
http://freespace.virgin.net/phil.crooks/images/sch13.gif

 

[ioaudio]

 

[PRR]

The Phil Crooks guitar-amp schematic is using the EF83 just as an amplifier, not a variable gain stage.

Your tape recorder plan-- hard to tell. Are they applying control voltage at the cathode, or is this just audio feedback?

EF83 datasheet

In the data-sheet: what is "VR" on page 2?

It seems to me it must be Control Voltage. But a suggested schematic would help clarify what you do.

It seems to be a single-ended AGC tube. The wonderful feature is that the plate DC voltage shifts only ~15V from max gain to min gain. The shortcoming is that there is only about 16dB of gain reduction shown.

The DC shift (thump) with GR is still larger than the signal voltage. However if we spent $10 for two tubes and used them push-pull, the thump could be very low, even with a simple differential load.

 

[ioaudio]

not shure but dont you adjust vr (which seems -) with the 150 ohm pot?

edit: sorry: i´ve no clue what i´m writing about-still got to learn so much...

so do i have to look for any rectified signal which comes back to the ef83 to control vr?

 

[PRR]

Sorry to be cryptic; crazzzy day at work.

> rectified signal which comes back to the ef83 to control vr?

You got it.

The tube is specified to give varying gain without large distortion.

Of course we could run a steady DC voltage to it, to control the gain. Or trim a DC voltage with a potentiometer to control the gain. But it is (used to be) so much simpler/cheaper and cleaner to just run the audio through a pot to change gain. (Later came audio chips for TV sets that changed gain with a DC voltage, but that's a lot easier/cheaper to do.)

So the usual reason we would bother with a voltage-controlled variable-gain tube is so a lightning-fast electric signal (rectified from the output) can change gain and reduce peaks faster than the human hand.

Another odd thing about the schematic you traced: It seems that the control voltage is fed to the cathode through a low-resistance voltage divider. It is more common to feed control voltage to the grid, a very high impedance point. Driving the low-R cathode means LOTS of rectifier power which means a power amplifier. If I read that right, to get 16dB gain reduction, you need about 20V at the cathode, almost 20mA of current in the resistors, maybe 40V at the top of the divider, 40V*20mA= almost 1 Watt of power. If that is fed rectified audio directly, distortion is gross; you need a fast-attack slow-decay filter. That means the rectifier has to take enough power in the short attack to hold-up all through the decay. So the peak power needed from the amplifier is 10 or 100 Watts!

So there has to be something more.

You do not show any grid resistor on pin 9 of the EF83. There has to be one. The sheet says it could be as large as 3 Meg. Maybe that is where the control voltage is applied? Then your "feedback from ECC82" is probably just a DC bias voltage: they wanted to jack-up the EF83 cathode so it did not gain-reduce on small signals, and that ECC82 might just be a convenient source of a low clean DC voltage.

I am ordering a TK19 schematic from a small German shop. It may take weeks to pay, scan, ship, etc, but I like German service manuals from the 1950s and 1960s.

 

[Butterylicious]

[quote author="PRR"] It may take weeks to pay, scan, ship, etc, but I like German service manuals from the 1950s and 1960s.[/quote]

Who doesn't?
Picture of my Grundig TK-830 R2R

 

[ioaudio]

[quote author="PRR"] Then your "feedback from ECC82" is probably just a DC bias voltage: they wanted to jack-up the EF83 cathode so it did not gain-reduce on small signals, and that ECC82 might just be a convenient source of a low clean DC voltage.[/quote]

edit: insert a 1Mohm resistor between A-A

 

[bcarso]

Ah that's more like it. The far-right-hand resistor is probably bigger than 220 ohms---maybe 220k? As PRR conjectured the gain control bias is indeed applied to the left hand grid.

Nun ist Alles (nearly) klar.

 

[ioaudio]

you are nearly right: i forgot 1Mohm in series with the 220 Ohm to the right!
(the signal is split after the 220 Ohm going to...i dont know where to...)

unfortunately i unsoldered most parts&wires before redrawing :-(
i did it with the nearly empty pcb and this foto:

http://groupdiy.twin-x.com/albums/userpics/10055/PICT0016.JPG

prr´s wunderbare explanations pointed me to this.

 

[bcarso]

Another note: that diode-capacitor junction needs some resistor to establish a d.c. path, unless the diode is a zener. The latter may well be the case, and the zener voltage is involved in setting the threshold of compression/limiting.

 

[ioaudio]

im not shure if its a zener. thats what it looks like:

 

[ioaudio]

finally here is a schematic for the grundig tk 145(similar to tk19 automatic):

http://groupdiy.twin-x.com/albums/userpics/10055/Grundig_TK145.pdf

 

[bcarso]

I don't recognize that part in the picture, but the missing resistor whose absence would defeat the simple diode operation is indeed shown on your new schematic.

 

[PRR]

> indeed shown on your new schematic.

So is a lot more.

Note that the control voltage off the diode toward the EF83 grid is also dribbled into the preceding EF86 grid (through 15M+15M/10M). Two stages are under AVC. That explains why they could get away with just 16dB of gain reduction in the EF83: while the EF86 is not rated as a vari-gain tube, its gain will drop as the grid is pulled down. If it can drop 10dB-15dB of gain before it distorts (it only needs about 1V output to push the EF83) we have ~30dB of GR, all we can ever use.

The gain seems high. EF86 and EF83 each have gain over 100. The triode after that is a 12AX7 type, gain probably over 40. 100*100*40= 400,000. The load is a power tube with peak input about 8V. 8V/400,000= about 20 microvolts! Surely R24 (1K trim pot) is set-on-test to give a lower overall gain.

Ah wait: the 12AX7 stage has feedback, and may only give gain of 2 or 3, while loading the EF83 stage to max again like 60. Then 100*60*3= 20,000. And we will see that the limiting voltage is 20V pk at the 12AX7 plate. So input sensitivity is 1 milliVolt, perfectly reasonable.

The first stage of the side-chain amp is biased "off". The cathode is held up around 20V by the 68K++3K3+5K+1K+120 resistor string. That means the 12AX7's output has to be above 20V peak before the sidechain passes a signal.

 

[ioaudio]

PRR, thank you for your explaining the circuit!

im really learning a lot these days, i´m really excited about all the circuits.
i wonder if older circuits generally tend to use higher-ohmic resistors/pots more likely than modern ones?

while mixing a track today i came up with the idea of the matrix summing compressor. something like a multi channel compressor, with a matrix side chain controled by every channel. wired thought, so i´ll think about it.

here is another tube i got today:

http://www.radiostation.ru/tubes/ef11_tf.pdf

 

[ioaudio]

sorry to bumb this thread one more time...more questions...

in the side-chain amp around the ecc81:

http://groupdiy.twin-x.com/albums/userpics/10055/Grundig%20TK145.pdf

does the 1µ before the diode define attack time and the 10µ after the diode define release?

 

[PRR]

> does the 1µ before the diode define attack time and the 10µ after the diode define release?

Ummmm...

To a first approximation, the 1uFd is "just a DC-block coupling cap" and the 10uFd does both time-functions. Attack is set by the resistance of the diode, the tube, and other things. Release is set by the large value resistors around the GR tubes' grids.

Ignoring the 1uFd cap: the Attack resistance is roughly the ECC81 cathode impedance (~1K) for small changes, and roughly the 8K resistor for very large changes.

But the 1uFd coupling cap makes it more complicated. A 1uFd coupling cap can't transfer its whole voltage to a 10uFd cap in one half-cycle; it can transfer less than 10%. So it can make a few-volt change in one half-cycle, but it takes many half-cycles to arrive at final value.

That means many half-cycles at the start of a transient are under-controlled. However it also means gain won't plunge on a single-cycle transient, and it means control is slower for loud bass than loud treble, reducing some types of limiter artifacts. It is actually pretty sloppy: this is NOT what you need for disc-cutting or for perfectly clean digital recording. Tape is pretty tolerant: this scheme seems to be slow so it will keep the average level at a nominal level where peaks won't splatt much, not a peak limiter like the Fairchild or AudiMax.

It is hard to read some values. If that is 6,8M, then the release time is up around 50 seconds! The attack time is around 10mS to 80mS, plus a dozen cycles which adds ~10mS midrange, 100mS bass.

Corrections welcome.

 

[gyraf]

[quote author="ioaudio"] interesting tube?[/quote]

Used in the K&H UE-100 as output driver - in triode mode:

http://www.gyraf.dk/schematics/schematics.html

Jakob E.