About Bias failure detection, clipping indication and active protection.
SHEETS 6, 7.

SHEET 6, bias failure detection.

Sheet 6 can be read in conjunction with Fig 5 at my page on bias stabilization.
Sheet 6 includes the circuit method for detection of excessive Idc in any one
or more 6550 in the 300Watt amp.

The dc operation of the tubes in any tube amp is very important for good
music and active protection is needed to prevent mental anguish with large
repair bills.
The dc operation should be considered a separate operational to the ac
or signal operation.

DC operation and twelve Ek bias test points.
The top left part of Sheet 6 shows 12 test points and 12 R&C networks
each with with 47k, 15k, and 220uF, and all this is mounted on a board
400mm x 50mm mounted inside a long side of the chassis. Test points
4 to 15 are accessible via 12 x 6mm holes in chassis side.

Checking bias is OK?
This is done with NO signal present and a minute or more after amp has
been turned on.

An amp owner can use a $10 digital multimeter to measure Vdc at each
test point. The meter should be set to 200Vdc range and red probe
held onto test point without also touching chassis, and black probe to
a nearby chassis screw.
Each Ek should be between +22dc and +26Vdc.

An owner can see if all tubes are conducting dc current, and if any are
too little or too much. The +22Vdc to +26Vdc is what I measure with 12
EH6550 made in about 2001, and which have had less than 500hours of use.
I assume replacement 6550 made in 2014 would measure similarly.  

When Ek = +23Vdc, The Vdc at CFB winding is about +0.73Vdc. Vdc across
each Rk 500r = 22Vdc approx and average Ikdc = 44mAdc.
My schematic SHEET 4 at 300W amp power supply
shows the screen
Idc flow in each 6550 = 4mAdc average, +/- 10%.
So, if we measure Ikdc = 44mAdc, then Iadc at idle = 40mAdc.

The idle Ea of the tube = B+ at CT less measured Ek test point, = 512V - 23V =
Idle Pda of 6550 = Ea x Ia = 489V x 0.040A = 19.56Watts.
Idle Pdg2 = Eg2 x Ig2 = 364V x 0.004A = 1.45Watts.
Total Pda+Pdg2 = 21.0Watts which is very much lower than the max Pda for 6550.

If one of the 12 6550 has Ek of +30Vdc, while most others were +23Vdc,
then that tube could be faulty, and it might worry us. It could be replaced.
We could calculate total Pda at idle = 26Watts. But this isn't much worse than 21Watts
so we can conclude that this tube is warmer than the rest but it may be still serviceable.
If you replaced it you'd hear ZERO sound change.

But nobody can be relied upon to constantly measure voltages and make
numerous confusing calculations and come to muddled conclusions.

Therefore is is highly desirable to have a simple circuit which turns off the amp if one

or more 6550 has Pda that has risen to say 35Watts which is getting close to the
rated maximum Pda for Russian 6550 or KT88 at 42Watts. If ever this happens,
it means Ikdc would be about 80mA and Ek test point = +41Vdc. The Ig2 will
probably have risen to 7mAdc, and Ia = 74mAdc and Pda = (512 - 41) x 0.074
= 34.8Watts.
Tubes which begin to overheat at idle may have suffered gas leaking past glass-pin
joins or have more than say 5,000 hours of use, or have some other random failure
internally. Before they eventually stop working the Eg1 does not control the Ia and it
begins to increase and the tube Pda exceeds 42Watts and suffers thermal run-away.
Maximum current when "saturated" can be 500mA, but with cathode Rk of 500r,
the Ek could possibly rise to +250Vdc, with the resistance developing 125Watts
of heat. But the 15Watt 500r Rk for each 6550 is then likely to fuse open.
But well before the Ek rises so high with a fault the protection circuit will turn off the

Between 1994 and 2014, I must have supplied a hundred or more Russian made
6550,KT88, KT90 and others. One owners used his 5050 amp so much that a
replacement set of 6550 was needed. One owner with 8585 had 8 x KT90 and
one failed after he'd used it from 2005 to 2011, each day, 4 hours minimum.
I estimate 8,000 hrs+ The other 7 KT90 tested as if they were new, with no
discoloration of the silver gettering and no positive grid current at idle, and all
like they were brand new.
The protection circuit worked perfectly, and the LED bias indication worked to
tell the owner something was wrong just before the tube failed.

Generating a suitable error dc voltage.
Each 6550 cathode feeds an RC network with 47k + 15k in series to 0V with
220uF bypassing the 15k. All cathodes have up to 90Vac because they are
connected to the CFB windings of OPT.  This high Vac needs to be filtered
away leaving a suitable +5.7Vdc at top of each 22uF. There are twelve
+5.7Vdc signals which are all connected by 12 diodes to a single rail at Z. 

The forward voltage drop across diodes is about 0.55Vdc so expect point Z
Vdc = +5.15Vdc. The real Z Vdc is determined by the one 6550 which has
the highest Ek. The point Z drives a high impedance base input of an emitter
on SHEET 7 which shows the active protection circuit.

SHEET 7.  Bias fault SCR and Clipping indicator.

Point Z Vdc is fed to a high impedance input of emitter follower buffer
Q4&5 in Sheet 7 above. The low impedance emitter Q5 output is applied
to a 5.6V zener
diode in series with R8 1k8 and to gate of SCR1, a C106D.

Normal idle emitter Vdc = +3.9Vdc. The zener voltage = 5.6V, so no current
can flow and SCR gate is at 0V. The scr gate must rise to +0.68Vdc to switch on
so point Z must rise to +7.6Vdc. There will be 0.111mA in R7 68k, and one of
15k will have 8.2Vdc, so 0.546mAdc so 47k must have 0.657mAdc, so Ek
must be +39Vdc, with Ikdc = 76mAdc, or nearly twice the idle current.

If more than one 6550 conduct too much Idc then many Ek might rise so the
threshold Vdc where SCR1 is turned on slightly reduces which is favourable.

Protection by turning amp off internally.
When the SCR1 turns on, it turns on the Relay 1 in Sheet 3 PSU which opens
the Neutral line to OPT1 primary, so the amp remains turned off, yet with
mains switch still turned on. Owners must fix the problem lest it keep
happening. But the amp is reset by turning off mains switch and the +12Vdc
Protection rail quickly reduces and SCR1 turns off. When mains switch is turned
back on in say 3 seconds or longer, the amp Vdc rails all are re-established.

LED indication.
The SCR1 also controls the red and blue LEDs on power amp chassis.
The blue LED draws little current through Relay 1 and to 0V, and when blue
is lit it indicates all is well. When SCR1 conducts, the blue LED is turned off.
The red LED is turned on, indicating a fault, and stays lit until SCR1 stops
conducting during amp re-setting.

Clipping indication.
On Sheet 7 there are Q1&2 Darlington pair bjts which are turned on by a
positive going peak signal of more than +1.2Vpk. This very low current signal
is derived from V1 anode via R1, C3, R4. When the amp clips, the sine wave Va
at V1 anode develop voltage peaks which exceed the maximum Va when no
clipping occurs. The large sudden rise in Va is divided down R1 and R3, Vdc
excluded, and Q1&2 will flash on and off with signal peaks at clipping.
When Q1&2 turn on, Vc is pulled negative to turn on Q3 which turns the red LED
This action does not affect the SCR1 operation. So the red LED can flash
at clipping, and also display when the SR1 is turned off, when there cannot
be any clipping because whole amp is turned off.

AC Operation and Dynamic Bias Stabilization.
Now look at Sheet 6, bottom half, which shows the 12 cathode biasing
networks, one 500r and one 1,000uF for each 6550. 
The negative ends of six 1,000uF cathode bypass caps on each side of PP
circuit are commoned at rails V and W.
These rails pass signal currents with opposite phase through R1, R2, 2.35r.
The peak positive going signal voltage between V&T and W&U is used to turn
on Q1 and Q2 when Vbe exceeds +0.6Vpk. This only happens when the amp
makes class AB power when the increase in peak Ia for 6 x 6550 can be up to
2.4A, while the decrease is only from 250mA at idle to to zero mA when 6550
cut off in class AB.

While ever the bjts turn on, they bypass collector current through R7 & R8,
16r, so that any rise in Ek due to rectifier effects of high peak cathode current
limits Ek to no more than +3Vdc more than idle Ek, and this rise is not enough
to turn on the SCR1. 

There is square wave signal less than +/- 0.8Vpeak across each of R1, R2
when the amp makes high class AB Po. The increase of overall THD due to signal
at R1 & R2 is difficult to measure. During class A the Q1 and Q2 are dormant,
and have no effect whatever.
The arrangement allows the steady Idc bias condition to be monitored without
having Ikdc flow in R1 and R2 which might stop Ek rising which would prevent
protection circuit from working.

More about LEDs indications.
The +12Vdc rail on SHEET 7 is brought to amplifier chassis from PSU
via umbilical cable. with white plug. If the plug is not plugged in, SCR2
in Sheet 3 PSU turns on which turns off green LED at PSU and turns on
red LED. The blue and red LED on amp chassis will not turn on because
the white plug is not connected to PSU. Making sure white and black plugs
are plugged in correctly will fix the problem after turning mains off, then on
again after 3 seconds.

If an owner leaves his amp to make a phone call, and a fault turns the
amp off, when he returns he will find blue LED on amp is off, red is on
and at PSU green is off and red is on, to tell him there is a problem.
The amp will appear to be cool, nothing is damaged. This condition could
continue for years if house mains remain on without interruption. Amp will
try to reset itself but if the problem remains the amp will be automatically
turned off again.
Obviously, sensible ppl always turn off their amps if they go out, or do not
want to listen to music.

If the protection PSU rail does not come up to +12Vdc after turn on,
blue and green will not light up. There is no protection. Amp will still work
and dynamic bias stabilizer will work but only fuse protection exists.
A fault in protection circuits should be found lest the amp work without
any active protection. But simple low voltage solid state circuits as I use
are far more reliable than any tube circuits.

Circuit board used for protection.
I only use boards in tube amps for protection circuits which have small
R&C and a few solid state parts all running cool. The board with 12 RC
filters and 12 test points was easily constructed on a 4mm board with
hooked solid wires for tracks with R and C and other parts surface
mounted and soldered between wire tracks. Hand made boards are
more rugged than most mass produced printed circuit boards which can
develop cracks in delicate tracks.

Each of the 12 test points is a recessed countersunk 4mm phillips head
wood screw in the circuit board. Each test point recessed and accessible
from outside the chassis through a two rows of 6mm holes. The test point
positions correspond to the output tube positions. The Ek of all 6550 can
be checked in a minute. Such dc bias measurement is always done
without any ac signal present.

If the amp keeps turning off soon after warm up, it may be possible to
find out which 6550 is conducting too much Idc cathode current by
quickly measuring Ek to see which one rises to a high enough Vdc
to trip the protection circuit. An audio tech might disconnect the wire
from point Z to SCR driver to give more time to observe Ek.

A faulty tube will easily be spotted.
You are at

300W amp active protection.

300W amp input/driver and output stages. 
300W amp power supply.  
300W amp dynamic bias stabilization.
300W amp power vs load graphs.
300W amp images, tubes with blue glow, and more views of amps.

Back to Power amplifiers.

Back to Index Page.