On Monday, 17 February 2014 08:53:17 UTC+11, John L Stewart wrote:
This note describes a simple, low cost power supply improvement. It
allows reduction of the power frequency ripple component (60 Hz in North
America & 50 Hz in Europe) by up too 20 db. The problem this
modification resolves occurs only in center tapped, full wave rectifier
"John L Stewart"
( snip )
Depending on the loading conditions of a triode power stage, from 2/3 to
4/5 of this ripple voltage will be applied to the load. Unfortunately,
most loudspeaker systems have one or two
resonance's below 100 HZ. One of these could be at or near the power
frequency. Presto!! We have objectionable hum in the listening room.
Figure One illustrates an ordinary center tapped full wave power supply.
Two sides of the HV winding measured 38 & 40 ohms.
** FFS - just add 2 ohms to the 38 ohm winding.
Problem gone.
Dodgy 5U4s etc not withstanding.
SET amp loonies sure are a bunch of neurotic old women.
..... Phil
I'm not so sure an added 2r to a winding with 38r would fix the "problem".
What if there is slight difference in turns of the CT winding? What if there
is Idc imbalance in the two output tubes?
----------------------------------------------------------------
The 2R will help. But if a directly heated rectifer is used then it could be further improved by the pot I shewed across the 5V winding. That way most all the power frequency ripple can be elimnated. But it don't matter much in a guitar amp. Helps a lot for HIFI tho!
------------------------------------------------------------------------
Now, consider Quad-II amps which have about the worst PSU of any amplifier ever made and which looks exactly like an "accountant's delight"
There is a GZ32 charging 16uF and the B+ at approx 140mA Iadc is applied to the anode winding CT on OPT. The Vr at 16uF is approx 18Vrms! However, CMR reduces this to negligible levels and any differential Vr across OPT anode winding appears at cathode windings where the local CFB reduces it. The KT66 screen supply has a 20H choke followed by a second 16uF which then reduces the 18Vrms
by a factor 0.008 to approximately 0.15Vrms. Having a very well filtered screen supply is more important than a well filtered anode supply.
The GNFB then reduces any residual hum further to be what Peter Walker considered low enough, and was less than the THD measured. This assumes that the rectifier has equal Ra anodes, and tubes are new and have equal Idc.
Where the tubes are old and Idc imbalanced, then THD and the hum can rise to be the same as if there was no NFB in the amp.
My solution to the "Quad problem" has been to completely revise the Quad PSU using Si diodes feeding say 150uF then 150ohms to 470uF mounted where tube rectifier once was, then use individual R&C biasing on each KT66, and then the amp works fine with less Idc. The screen choke can be retained, but bypass cap is 100uF. Thus makes the amp dead quiet, even if output tubes are badly matched and a fault is present.
Where there is a CLC filtered supply, the Vr can usually be much reduced by adding a C+R series network across the choke, arranged so that C is resonant with the *working* choke inductance at twice the mains F. R should be about 22r. The damped parallel resonance usually reduces the main harmonic of 2H
in ripple ( 100Hz or 120Hz ) by 12dB. Other H are not much reduced but then the
following C2 of the CLC filter has an increasingly lower XC thus shunting the higher ripple H.
-----------------------------------------------------------------------
The massive CLC filter with resonant L is nice. But it is still 12 db less effective at the power freq than at the main ripple freq. A simple FET regulator or source follower refed to a string of Zeners would do a better job. And you can have that system ramp up slowly so that full PS HV does not appear on the power tube G1s at switch on. And a current limiting NTC Thermistor in the PT primary cct also helps.
http://www.ge-mcs.com/en/temperature...t-limiter.html
-----------------------------------------------------------------------------
Many Guitar amps have a hum nulling pot so hum from heaters can be nulled but often is has no effect. Some allow balancing of the Idc in each output tube.
Few guitar amps have CLC anode filtering.
--------------------------------------------------------------------------
I checked that in Aspen Pitman's ' The Tube Amp Book, 4th Edition'.
About half the amps listed showed some kind of choke in the HV section. But seldom tell us how many 'Heinrichs'. So difficult to tell how effective they are.
If the plates in a PP amp are run straight off the raw DC then ripple side bands on the progam material are a sure thing. But in a guitar amp don't matter much anyway.
---------------------------------------------------------------------------------------
Hum seems only objectionable if the volume is turned up. Switching noise of diodes can often be heard as pulses at 100Hz or 120Hz. But for hi-fi, owners expect dead quiet when volume is up and methinks one has to go to more trouble to get quietness than either Phil or John suggest. I have also repaired and re-engineered many amps where the earth path has had to entirely re-wired, including position on chassis for green-yellow Earth wire from mains input wiring. a connection lug should be well away from say a bolt holding the PT to chassis.
Noise is not always easily reduced.
Most amps with CT HT windings don't make much more Vr than amps with single HT windings with 4-diode bridges or voltage doublers with single HT winding and two diodes - all things being equal with current to capacitance ratios.
Keeping ancient electrolytic caps and tube rectifiers in ancient amps is not good practice and it is always an up-grade to change to Si diodes and healthy value filter caps. Especially with SE amps, where noise is not reduced be CMRR.
Many SE amps have triode output tubes which means the primary load is in series with lower tube Ra and the B+ rail so that 2/3 to 4/5 of Vr is across the OPT as JS mentions above. So if Vr across OPT primary = 0.5V, and OPT TR = 25:1, then Vr hum at sec = 20mV and 12dB GNFB might reduce it to 5mV but its far too much hum for hi-fi, so the Vr at OPT MUST be much less, so there is no cheap solution and the dopey cost avoiding hobbyist MUST let moths from his wallet.
RDH4 has other solutions of using twin T R&C networks for B+ hum reduction
but none apply to output stages with high Idc.
The other benefits of having a LARGE value C between OPT B+ connection and 0V is to give a very low impedance anchor to the connection and this minimizes IMD generation in tubes caused by having high amounts of LF appearing in series with the higher F. It applies especially to SE amps and to those working in Class AB - including Quad-II, which works in AB with RLa-a only 4k0 when you have a load of 16r connected to the OPT when strapped for 16r.
There is less hum when PP OP tubes work in class A. Nearly all PP amps made now
work in in class AB and owners insist on great bass at high levels and the use of 470uF caps between OPT CT and 0V is very common. 35uF = 91r at 50Hz, 470uF = 6.7r at 50Hz.
Many amps now proudly display the very large C value fancy brand electrolytics above the chassis. But so often that alone does little to change their other dreadful shortcomings with noise and other issues.
Patrick Turner.