[patrick-turner]

Phil replied to my comments with a good post...

** You have seen one faulty WW pot in a BWD141 - probably the only faulty
one that ever existed.

^^^^Well, who is to ever know?

" I was cleaning up my workshop when I discovered I had been given an old
BWD Electronics
141A audio oscillator which had many problems. "

**There is a big clue there ....

Wire wound pots can be notoriously jerky with Vo levels.
But OK when not being moved, when they are then usually noisless.


** I have always found WW pots to be highly reliable and long lived.

^^^^In General, I agree 100%, and for example I use WW pots for biasing amps where failure can lead to smoke. But even then I put a R between wiper and negative end of pot so if the pot fails the grid bias goes more negative.

1. The voltage control on my most used bench PSU has a 10 kohm WW pot made
by IRH in Sydney, Australia - been in daily use for nearly 30 years, still
works perfect.

^^^^Wonderful.

2. The two fine adjust controls on my Wien bridge notch filter are 200ohm 10
turn WW types - allows up to -100dB notch depth at 66, 996 and 5855 Hz.
They are 30 years old too, got them the same time I bought my Sony CD101
player.

^^^^I have a bridged T LC notch filter with two pots, one for course nul adjustment and and the other for fine, and then I have a tuning cap at the 1kHz oscillator to slightly vary the F and it all works well to get good rejection of 1kHz,. After the notch filterI have a bandpass filter with op-amps pass only 2kHz to 11kHz. The carbon pots work fine, and are not surrounded by loops of FB.

3. I have two Compton dual gang WW pots, same as the one PT is so down on.
One is 40 years old and the other 24 years old. They both work perfectly.

^^^^Wonderful. I know why I am down on the Comptons, and why you ain't. And maybe why the guy gave me the BWD 141. Much work had been done on it, to no avail.....

**To overcome the issue with fine frequency adjustment, I fitted 500ohm, 10
turn WW pots in series with one of the sections of each Comptom pot - which
is 2 x 15kohms. Makes it easy to set any frequency up to 50kHz, +/- 1Hz.

^^^^Good practice.

Stability is good enough to work with the -100dB notch filter, which at 996
Hz has a notch width of about 0.02 Hz.

^^^^I don't know how wide my notch filter is at the -3dB point, probably 2kHz, but that's OK, and I don't know what the notch width is at deepest null, maybe it is similar, but for where I have a signal of say 3Vrms and THD 0.001%, then what one cvan see is some dithering of F which seems to be caused by slight noise somewhere and there is some AM and FM going on. But one rarely ever sees less than 0.001% in tube gear becaus4e the noise usually swamps the THD. BUT, in my last trial of a tube amp to use on my latest WB oskar laytar, I DID MEASURE less than 0.001% with signal of 5Vrms. My 1 kHz oscillator makes less than 0.001%. I didn't believe it at first, but then I did the maths on the gain reduction and sure enough the calculated THD was as low as I measured.

^^^^Electronics World had an article on measuring THD with two cascaded notch filters with very slightly different F for the notch null, so that the only control needed for measuring THD at 1kHz was slight F adjstement which wasn't much used. But my own THD checker works fine for me. The beauty of a bridged T LC notch filter is that its passive, and presents a 5ko load to a power amp, and you can put in 200Vrms without smoke.

^^^^I have a high Z input buffer but that's all solid state and protected against any Vin above about 8Vrms.

^^^^Getting the WB oscillator with tubes to oscillate properly between 500kHz and 1Mhz has been difficult because of unpredictable bothers with biasing and gain at such high F, and with stray C, and the minimum C needed in the WB network to overcome stray C elsewhere is about 130pF. This means that a tuning gang to cover 100kHz to 1MHz must vary from 1,300pF to 130pF.

^^^^But I have 4 x triple gang caps with each gang 400pF to 13pF. The C value can be 2,400pF to 102pF. But to the mimum C value, enough trimmer C must be added to so that minimum total C including stray C and tube C is around 250pF, and maximum C is then 2,500pF, and then the R value is trimmed, but for 100kHz to 1MHz, the R = 636r, and then Zin for the network at Fo = 1,347r, and if the Vo = 7Vrms, then I in max = 5.2mA. The NFB network with lamps seems to work OK so far because caps give smooth F change, but load is 1k2, so Iin = 5.8mA, and then I have a 1k0 pot to power so another 7mA, so total I needed = 18mA, plus there will be a feed to a Schmitt trigger, so say 20mArms.
So load to be powered by the oscillator = 7 / 0.02 = 350r.

^^^^But the White follower I made with two EL86 in triode have Ea = 130V, Iadc = 53mA, and do give very fine perfomance with loads as low as 200R.
I found changing load from 400r to 200r caused only 1mV of Vo drop because of the NFB.

^^^^Perhaps the Technics output transformerless output stage with a 3rd EL86 triode may make a better output follower buffer because the drive to lower SET is not derived from current flow in top tube, but both tubes always get the same Vg-k regardless of load.

^^^^I have some work to do to finalise circuits and prepare the webpage to explain myself better than time allows here with mere words which are not enough because a schematic is needed.

Patrick Turner.


..... Phil