SE conjunctive filter

[EFK]

I've never used a conjunctive filter however I have a particular SE project currently which I may wish to experiment with. One thing I'm curious about - does it make any difference to which 'end' of the OT (B+ source or tube plate) the capacitor is attached? In other words, most descriptions put the capacitor on the tube plate and the resistor to the B+ supply. Is this mandatory for some reason, or might these be reversed - cap off the B+ and resistor off the plate?

I'm also curious as to why the typical resistor value utilized is 1.33 times the OT primary? It seems most folks stick to this formula but I have read here or there of others using slightly different values, i.e. a 1/1 ratio with the OT or some variation. Nothing dramatic, just a variation. Where does this 1.33 value come from - I mean, what is the purpose of this ratio?

[Chuck H]

I was curious about the purpose of the ratio myself. I can see not getting too low because it DOES skew the impedance @ frequency. But 1:1 or anything higher should be fine. The circuit doesn't care which component is attached to which end. IME (which includes some failed circuits) you'll want to triple the B+ voltage for the cap rating and definitely use a resistor good to higher voltages. Shunt filters love to fail (ask DR.Z). My original circuit used a cap at double the B+ and a standard wire wound resistor with whatever the typical voltage rating is. The cap shorted and the resistor turned brown but was still working when I removed it. I replaced the circuit with a 2KV rated cap and a "high energy" higher voltage rated resistor and have four amps with this circuit out there with no failures. My personal proto of this model gets pounded on regularly at full tilt and is holding up well. And I often run it into a resistive attenuator, which means that the HF, with the shunt filter in place, may be about two thirds the rated load and maybe one tenth or less what the amp would see without the filter if playing into a normal speaker. But no problems with reliability or tube life.

[EFK]

Thanks Chuck. I can see it being prone to failure - I'm waiting on some components that should be here today, caps rated @ 6KV and 15W cement resistors. Should be fine for a SE EL34 hahahaha. I'm a big fan of overkill, especially for amps built to be cranked all the time.

I'm primarily interested in it as a protection against flyback spikes and frankly don't really want it to affect the overall tone too much. As such, I don't plan on using caps much larger than .001 or .0015 at most. However, this got me to thinking, I don't really know if there is a point of diminishing returns so to speak in which the cap may be too small in value to offer full protection against such spikes. Going to have to research this one a bit more.

[Chuck H]

I hate to say it, but I think the max voltage spec on most cement type resistors is lower than the spike voltage in that circuit. Which may or may not be a problem.

If you look for a max voltage spec for the resistors you ordered, and it's not indicated, that's a pretty good indication that it's not a high voltage resistor.

[Steve Conner]

That is a valid point.

I think as far as those cement blocks are concerned, the low resistance ones are wirewound, the high resistance ones are metal oxide. Don't know the resistance value where the change happens, though, and what implication it has for the voltage rating.

Specialised high voltage pulse rated resistors are available, but a big old carbon comp might also be a reasonable choice.

[exclamationmark]

Carbon comp resistors are essentially just lumps of material, so no arc-over would occur internally (unless it's by some weird dielectric breakdown mechanic), and you'd need like 20kV for creepage to occur externally. Virtually all modern resistors are just some sort of material, spiral wound around a ceramic core, which I assume would have close tolerances internally and cause bad things to happen at over voltage conditions.

[trobbins]

The cap voltage can also be constrained by using a MOV in parallel - the MOV cap is often specified, but typically in the 100-300pF range if you keep to a small disk (ie. 7-10mm) and a high voltage rating. And you could put two MOV's in series to get higher voltage rating and half the capacitance (use min VDC 1mA rating and keep sufficiently higher than B+, which may require something like two 300V types).

A zobel filter preferably maintains a constant total effective load impedance as frequency rises. A choke or primary impedance will rise with frequency until the first self resonant frequency, and so the filter frequency range may start influencing from say 1-10kHz, and may have a significant effect in dampening the SRF in the region of say 30-100kHz. I'm not exactly sure why some have chosen exactly 1.33 ratio, although I find the accuracy of 1.33 and the typical range of OT impedances and resistor values to be a bit ?? There is certainly no need to use a lower R.

[Gingertube]

Zobels (Conjunctive Filter) design depend upon where it is used and what for.
That is, are you trying to adjust predominantly the phase or the amplitude?

For a feedback stabilisation scheme where it is used across an anode load in a earlier stage of the amp you want to adjust amplitude response while leaving the phase response largely intact and so you use a zobel resistor of 1/10th the load resistor the zobel is across and adjust the cap to suit.

If you were mostly trying to adjust phase response while leaving amplitude response the same, you would choose the capacitor value to have have an impedance of 1/10 of the load resistor you are shunting (at the highest frequency of interest) and adjust the resistor to suit.

For Zobels across the primary of a SE output Tranny (or anode to 0V on a push pull, or anode to screen on an Ultralinear etc.) you are trying to adjust both amplitude and phase response. The resistor value can vary but in most cases will end up being between x1 and x2 of the Ra value. I generally start with a value of 1.4 x Ra (Root 2). I can justify that using "cork sniffer" calculations but won't bother, just take my word that it is a good place to start. Use that value then choose the best cap to go with that resistor and for fine adjust of the end result, then trim that resistor up or down a "smidge". You may well end up at 1.33.

Cheers,
Ian

[EFK]

Hmmm. I obviously do not understand exactly what is going on with the spikes. If the resistor is being exposed to the full voltage of the spikes, then what is the purpose of the cap? Obviously it is blocking DC - are the spikes AC? Or I guess, the spike's high voltage can originate on either end of the OT winding? Or is the high voltage rating requisite in the event that the cap fails? Going to have to look into this. I do have some 5W resistors w/ 1000VAC dialectic rating (for 1 min, according to the data sheet) and I could string a couple together to meet the requisite resistance and also increase the voltage rating.....

Practically, at the base level, what kind of watt rating should I be looking for here with a SE EL34 amp? I am reading all kinds of stuff, some people using 2W or 3W resistors and some going to 15W or higher. ??? Can this actually see the full output power of the amp?

Just as a side note, I seem to have *another* problem that I am only mentioning here because it is dealing w/ the same amp. This amp has now eaten 2 meters, no s**t. What do I mean by that? I mean, when checking plate voltage (which is at 434 last I measured it), this amp has burned out 2 meters. Granted, they were cheapies, but that has NEVER happened to me previously. Black clipped on ground, I go to check the plate voltage and I notice a funky smell and the HT fuse blows. When I open up the meter, there is all kinds of burnt black crap around the traces for the dial. What the heck? TWO TIMES! The internal meter fuse never blew but there is burnt stuff on the circuitboards (meter) also. Two different brands. The second time this happened, I actually saw an arc to the probe before I ever even touched the plate pin with it.

[Chuck H]

I'll speculate that the impedance of the budjet meters is low enough to allow too much current to pass through the meter. Maybe they've been cheaped down a bit lately to increase profits. But on different brands??? Sure. These sort of manufactured items are often either made by the same company and put into different cases or competing companies make analogous changes in design to remain in competition.

The voltage spikes on the OT primary can be surprisingly high. How high? Not sure. A couple of guys here have taken measurements but I can't recall the exact results. Point is, it's the spikes that kill the resistor. You need a resistor that is not only capable of surviving in the higher voltage environment but also capable of surviving much higher spikes. Hopefully someone here with actual numbers will chime in. I just bought the burliest spike handling resistor I could locate in the Mouser catalog.

[Steve Conner]

The purpose of the capacitor is to stop the resistor from eating half of the amp's output power. It has a high impedance at audio frequencies, so those can't pass through the snubber. But spikes are composed of high frequencies, and those go right through the cap and are absorbed in the resistor.

The exploding meters suggest a problem with RF parasitic oscillations. Unless you're trying to measure the plate voltage while the amp is cranked, in which case the AC voltage is likely just too much for a cheap meter. If you have a neon lamp or one of those neon screwdrivers, does it light up when held near the power tube?

Or maybe even 434V is too much for them. Long ago I cratered a couple of DMMs while working on a tube radio transmitter, but that had over 750V B+. Maybe meters have got cheaper.

[pdf64]

When I got my first Fluke 30 years ago, I tried to measure the plate signal voltage of a fender reverb driver stage. I got a reasonable reading at low levels from a sig gen, but when I tried it with a guitar signal, it resulted in smoke release and a dead Fluke. Thankfully it was repaired under warranty.
Lesson learnt was never to put a meter probe on an overdriven transformer coupled plate.
Pete.

[EFK]

I will admit the first meter was getting old - for a throwaway type, @ a year, and I was using it under signal. But the second time, yes one of those harbor freight cheapies, did it with the amp just idling there. I kind of wonder if there is some kind of inaudible hf oscillation going on that is spiking the current up at that frequency? I don;t even have it ramped up to 100% at idle because I don;t think the plate voltage is going to allow it - so it's only around 48-49 mA at the moment.

Anyway - I wonder if the caps hanging off the plates directly to ground on silverface Fenders offer the same spike protection in the pass frequencies? I would suspect they must. I'm going to assume, however, that because nobody ever talks up the virtues of using those plate snubbers that they must dull the tone down more than a conjunctive filter...

Chuck do you have a brand name or series of the resistors you are using? I found Ohmite Super Mox that can handle some super serious voltage (!!!!) but Mouser isn;t stocking any close to the value I need and I suspect that they may be super seriously huge!!

[pdf64]

Putting a meter probe on to the plate could be sending it into oscillation (which may not be there normally) thereby overdriving the meter.
Also a cap (with no series resistor) across the OT primary may create a resonant peak and cause more trouble. The series resistor of the conjunctive filter may help to keep the Q factor / resonant peak low.
Pete.

[Chuck H]

Chuck do you have a brand name or series of the resistors you are using? I found Ohmite Super Mox that can handle some super serious voltage (!!!!) but Mouser isn;t stocking any close to the value I need and I suspect that they may be super seriously huge!!

Huh... I checked twice now and I don't see the resistor I bought. Now I'm wondering if it was dropped because it's not a good product?!? I don't recall enough specifics to go searching elsewhere other than Mouser. FWIW I used one of those aluminum housed chassis mount resistors (5k, 20W standard wire wound) for the shunt filter in a 4xel84 amp and it's working fine after almost three years. The owner only knows one way to set the controls. I call him the five and dime. All EQ's and presence half way and the volume at max.