Balancing P-P O/P stages (long post!)
Cathode self-bias has an inherent balancing problem (but as, we know,
two resistors are much better than one!) With two resistors to either
side of the balance pot, with the slider grounded, the individual
cathode resistances to ground vary with balance pot setting. Thus,
measuring differential voltage across the cathodes does not help and,
of course, making them equal is quite counterproductive.
One goal is to get all DC bias out of the OPT primary (another is
equal drive to clipping - not addressed here but see below.)
It's easier to do this with a circuit that allows a centre-zero on a
meter at balance rather than switched current metering in the cathode
legs or a switched voltmeter across equal sensing resistors.
If the OPT had equal DC resistance primary sections one could put a
voltmeter between the two plates and, via the bias control (either
self bias or fixed bias), set it to zero for equal plate current.
Unfortunately, the OPT plate sections are generally not of equal DCR.
They are also very "B+ hot".
Anyway, I tried this recently on my "Scratch 10" amplifier (5AR4, 2 x
6v6 (2 x 470 ohm cathode resistors plus a 100 ohm balance pot), 6SN7
(with CC cathode cct), EF86) :
1. Measured the DC resistance of the two OPT primary sections (using
an AVO Model 8 analog multimeter. Note: digital DMM's can give errors
due to the high winding inductance.)
2. Added a series resistor to the lowest resistance side (plate to
OPT) to bring the total up to that of the other side (it is a small
resistance - see below)
3. Bypassed this resistor with a large capacitor (positive towards B+,
of course)
4. Placed AVO meter on a high DC voltage range across the plates
5. Powered up amplifier
6. Reduced meter DC voltage range until the unbalance deflection was
seen
7. Adjusted bias balance pot until differential voltage read zero.
8. Current in each O/P tube was now virtually equal.
Results
1. OPT coil resistance was was about 150 ohms and 175 ohms each side
of primary CT.
2. Added resistor to 150 ohm section - 22 ohms was to hand (close
enough)
3. Added bypass cap - 1000 uF, 16 VDC (-3dB corner frequency with 22
ohms is about 7 Hz, seems low enough.)
4. Put initial meter range on power up to 100 VDC full scale
5. Reduced meter range for balancing to 2.5 VDC full scale
6. Easily balanced the O/P tube currents with 100 ohm pot - much
easier than with interative voltage measurements across the OPT
primary sections as I had been doing.
7. Checked power output on 'scope and VTVM (across 8 ohms): OK, had
the same peak clipping (equal each side) and class AB X/O distortion
as before when overdriven at both 40 and 400 Hz.
I should add that I run an eclectic mix of NOS and well used O/P tubes
though this "bench" amplifier, from 6K6's to 6L6's (mostly using an
external variable B+ supply from 250 to 325 VDC), some tubes are quite
mismatched and a few sorry specimins below 50% emission - so it's
good to have a quick balancing act l
IM****TANT! - there are some serious practical and SAFETY problems with
this method...
1. SAFETY: the meter is some 300 VDC above ground - not safe for the
average user
- might use a permanent, switched, diode-clamped, isolated, center
zero meter installed in the chassis but I'm not sure if meter
movements can take 300 VDC from coil to case - my guess is no!
2. There is a very large differential signal voltage across the two O/
P tube plates, i.e. to the meter:
-- must ensure that no residual hum, noise or signal is present while
taking measurements; perhaps pull out the phase splitter tube before
test power-up in most circuits.
3. You get a large meter kick on power-down if it is left connected on
low range:
-- if probes are used you will have removed the meter; if croc-clips
are used, remove or switch range back to over 100 VDC before power-
down
-- any meter must be out of cct. when amplifier is being used, also at
power up/down. A large AC range volt meter could be left in (I tried
this) but to what purpose? VU meter?
Question: Have I missed a simple, permanent "centre zero" method that
is entirely in the "safe" cathode area? I do not recall ever seeing
one and nothing comes to mind. Actually, I was going to apply the auto-
balance Blumlein "garter" bias but I'm a bit reluctant to concede the
20 VDC or so of useful B+ that it takes up. For this see
http://www.tubecad.com/2005/May/blog0046.htm
Disclaimer: there are lethal voltages around tube circuits. Do not do
any of the above unless you are qualified by experience with tube
equipment design, build and repair. Don't use the above balance
method except on the bench under controlled conditions.
All comments welcomed.
Cheers,
Roger
|
