Conjunctive filter info please?

[jojokeo]

Hello and happy Labor Day holiday.

I've been looking to get a better understanding of a conjunctive filter put accross the power OUTPUT trans at times? What is this for, what does it do, when does one want it or not in their build, etc...? Any help is much appreciated.

[Mojo2001]

Across a power trans, a conjunctive filter can be used to lessen ringing caused by diode switching impulses--more to the point with solid state diodes.

A cap is sized to let the resistor to ground appear to the impulses at a given high frequency but not at low frequencies.

If the cap wasn't there, there would be a lot of power lost in the resistor at 60hz frequency.


In an output trans, the conjunctive filter acts the same way...it lets a resistor be seen across the primary at a given high frequency. This counters the effect of transformers having rising impedance at high frequencies and  flattens out the impedance curve.


In both cases, it is a sort of frequency selective resistor used to counter resonant or frequency related qualities of inductors.

[HotBluePlates]

The RCA tube manual has information for the conjunctive filter listed as "corrective filter". It reads as follows:

     "A corrective filter can be used to correct the frequency characteristic of an output stage using a beam power tube or pentode when inverse feedback is not applicable. The filter consists of a resistor and a capacitor placed in series across the primary of the output tranformer. Connected in this way, the filter is in parallel with the plate load impedance reflected from the voice-coil by the output transformer. The magnitude of this reflected impedance increases with increasing frequency in the middle and upper audio range. The impedance of the filter, however, decreases with increasing frequency. It follows that by use of the proper values for the resistance and the capacitance in the filter, the effective load impedance on the output tubes can be made practically constant for all the frequencies in the middle and upper audio range.
     The resistance to be used in the filter for a push-pull stage is 1.3 times the recommended plate-to-plate load resistance; or, for a single-tube stage, 1.3 times the recommended plate load resistance. The capacitance in the filter should have a value such that the voltage gain of the output stage at a freqeuncy of 1000 cycles or higher is the same as the voltage gain at 400 cycles.
     A method of determining the proper value of capacitnace for the filter is to make two measurements of the output voltage across the primary of the output tranformer; first, when a 400-cycle signal is applied to the input, and second, when a 1000-cycle signal of the same voltage as the 400-cycle signal is applied to the input. The correct value of capacitor is one which gives equal output voltage for the two signal inputs. In practice, this value is usually found to be on the order of 0.05uF."

[Mojo2001]

OP was asking about RC network across POWER trans.

Forgot to mention...this is called a "snubber." Google that term and you will find boatloads of info.

[jojokeo]

Great info you guys. Can you give a "real working" description of what it does regarding the sound or tone vs not having it installed? My first impression is that it helps reduce parasitic problems?

[loogie]

In guitar amps I think alot of guys have done it primarily to remove the ice picky sound.  They show up in DR Z amps and others as well.  I've done some experimentation and basically what I hear is just a rollling off of the highs.  I think for me  best effect came when I rolled off as much low as I could stand in the pre and then made sure that the PI didn't clip before the output tubes.  The best thing to do is just try it.  Use a power resistor and maybe a hefty switch.  I switched mine with a cap in front of the switch.

[snoof]

The RCA tube manual has information for the conjunctive filter listed as "corrective filter". It reads as follows...

Thanks for that HBP, good info.

[HotBluePlates]

OP was asking about RC network across POWER trans.

Doh! The first time anyone ever heard the term "conjunctive filter" seems to be when Dr Z was interviewed for ToneQuest magazine. Everything else I have ever seen calls it a "corrective filter". And since everyone always asks about a conjunctive filter in relation to the output transformer, I just saw what I wanted to see. My bad...

But you do see them around chokes in power supplies occasionally.

[tubenit]

I have been reading up on the conjunctive filter idea and since I like to experiment, I thought I might try this out?

I am NOT sure what values to begin experimenting with?  I need a starting place since I'll need to order the cement resistor.

I have tried reading up on the internet about this in SE tube amps and the information has not been real helpful & in some places seems inconsistent?

Any suggestions?   What will increasing each of those values do?

With respect, Tubenit

[jojokeo]

From what I've read on AX84 and TAG you start w/ 10k and .01uf - .001uf.

In the Guit Ply Mag a couple months ago, there was info regarding this as done by a large Amp maker which suggested values determined by which output tube you were using.

[tubenit]

I guess I'll start with 5-10k & .01-.02 and see where it goes?

Reading on the net, I saw everything from 1-10k  and 250p to .047.  That wasn't much help.  

It did sound like it actually added smoothness instead of just losing high frequencies.  Not sure how it works though?

With respect, Tubenit

[PRR]

> The first time anyone ever heard the term "conjunctive filter" seems to be when...

It is a basic concept in Modern Filter Theory.

"Modern" is late 1930s, when the telephone company was putting multiple calls on single pairs with frequency multiplexing.

NOBODY but diehard geeks ever read those chapters.

And the way it is tossed around is "incorrect" by the original usage. You build a narrowband filter with a chain L-C-L-C-L-C-L. At the front and the back, you want to have a pure resistor, and instead you have a little leftover L which must be "conjuncted" with a little C. The reactances are assumed exact, and the conjunct is exact. That's very different from putting a snubber on a crazy PT winding and diode.

The R-C network on the loudspeaker transformer (see many AM radios) is more correct as a conjunctive filter. The speaker is R+L. An R+C makes the total impedance nearly resistive for all frequencies. (However many AM radios omit the R because in AM you probably want to short-out the static.)

[Ritchie200]

Hey Joe, you originally asked about a power transformer application and then talked about an output transformer application?  I'm not sure about the power side, but as someone noted, they are used frequently on SE builds to tame the fiz - especially with the EL84.  I'm not a big fan because (to me) it sounds like you are playing with a blanket over the speaker.  I think you can tweak the tone stack to accomplish the same thing - AND keep the tone more open sounding.  I played around with it on the Firefly with different values and finally gave up.  Different strokes...  There used to be lots of info about it on the AX site.

Jim

[jojokeo]

I corrected my original post to reference OT, thanks Jim it's been a while since I last saw this.

Okay, now that I'm home w/ the Mag & a bit time, it's in the GP August issue and they call it "Smoothing your sonic ride" w/ "The Snubber" written by Dave Zimmerman of Maven Peal Amplifiers. It says that as the power tubes begin to distort, everything in the power amp begins to distort including the pi and OT. They reference lower quality OT's w/ ringing "bell-like" qualities as a problem which can sound like unwanted fizz on the notes - especially bass notes. It cites high voltages during this and it describes the snubber as a sort of shock absorber. Anyway, w/out going into further detail it's supposed to remove unwanted high frequencies while leaving the others untouched. Their circuit shows it being used on PP amps but it will also work w/ SE I too. They solder the network right to the output sockets.
It describes the values based on tube types and quantities. You can get the "kit" sold by AES here - http://www.tubesandmore.com/ > then type in "snubber" in the search bar. They're from ~$3.5 - $5 each.

*tubenit - For two 6v6/6l6/el34/kt66 (6k6 for you) they're recommending 2-1.5k/5w resistors & two .01uF/600v orange drop caps - just try using one of each. (you won't need the 8.2M "balancing resistors" since your amp is only a SE type)
Personally, I gator-clipped a 6k/.01uF to try. I think using less like a .001uf may be a better way to go so as not to lose too much high end. I didn't experiment w/ it much at all so I didn't try dialing it in - I only wanted to try it on something quickly to check it out. I can see how it might correct some harshness especially w/ el84s. It's nice to have another "smoothing option" to add to your list.

[darkbluemurder]

I have soldered a Zobel network (basically the same thing as a conjunctive filter but on the speaker side of the OT) into a small stompbox size case across the input jack. I can simply add that to any amp with more than one speaker output jack. My findings:

1. It takes away highs as expected but it's pretty much like putting a blanket over the amp. A cut control works better IMHO, and it can be dialed out.
2. It has a more pronounced effect on amps without global negative feedback. On amps with global negative feedback it only makes a discernible difference if the presence control is turned up.

To be honest, to me it is pretty useless.

Cheers Stephan

[tubenit]

it sounds like you are playing with a blanket over the speaker

it's pretty much like putting a blanket over the amp

Gee guys, it doesn't sound very inspiring! I'd have to eliminate all my favorite amp describing adjectives like super, transparent, touch sensitive, etc......

And at the worst, (Jim) it would make my Tele sound dull and lifeless like a strat.
 

Oh well!   

With respect,Tubenit

[Willabe]

And at the worst, (Jim) it would make my Tele sound dull and lifeless like a strat.
 

Oh well!   

With respect,Tubenit

Ohhh,             there he goes again.                      


                       Brad       


                             

[jojokeo]

Found something else from TAG called the Paul Ruby Mod that's more transparent. Quoted from one of the guys, "I've built several 18 Watters (with both very good AND very questionable output transformers) and the Ruby mod made most of them sound better. Never paid attention to which OT was in the ones that had the biggest improvement.... I do recall on most of them that the change is very minor - just reduced the "fizz" on the decay of notes when cranked."
The trick is to measure your cathode bias voltage and go 1.5v to 2v higher w/ the zener although you can experiment w/ the zener's voltage for a bit more or less as wanted and see how it changes the swirl/buzz characteristics.

[jojokeo]

Found the Ruby mod article...

[tubenit]

Found the Ruby mod article...

Thanks for sharing that.

Tubenit

[phsyconoodler]

I'm wondering if the Paul Ruby mod might be good for smoothing out a 6V6 or 6L6 cathode biased amp as well?I rather doubt it because of the unique structure of EL84's.Some need the mod and others don't.

[Ritchie200]

it sounds like you are playing with a blanket over the speaker

it's pretty much like putting a blanket over the amp

And at the worst, (Jim) it would make my Tele sound dull and lifeless like ME.
 

Oh well!   

With respect,Tubenit

Nit, trust me, you do not need any kind of electronic or physical device to make yours or any other Tele sound dull and lifeless....those qualities are inherent to the personality of the guitar and the players who use them.  

Jim

[phsyconoodler]

Tele lifeless? Oh my! Ready to start a war are we?

[HotBluePlates]

Let's get the basic idea straight...

A speaker is not a constant 8 ohms (or whatever you have). It is likely that impedance at some midrange value (people often cite 400Hz), then rises as the frequency increases. This is because of the parasitic inductances become the dominant factor as frequency rises. There is also a spike in impedance to ~10 times the nominal value at the bass resonant frequency.

One way to smooth out the bumps is with negative feedback. A possibly earlier method, and one which you can use when feedback is not applicable (for whatever reason) is a corrective (or conjunctive) filter.

So you need to smooth out the rising frequency response due to the speaker's non-linear response. A cap has roughly the opposite response as an inductance, so you slap a cap across the transformer primary. You do that (as opposed to across the speaker/secondary) because the OT will have its own response anomalies. You want to fix the overall response in one shot.

The reason RCA never provides a formula is that they have no way to know what peaks/dips your speaker has. They suggest a starting point for the cap and resistor, but they advise you to adjust on test so that the output level of highs is approximately equal to that of lows.

EDIT: Forgot to mention, the filter will fix the rising impedance as frequency rises, but will still leave the big spike at the speaker's resonant frequency. True negative feedback addresses both, to the extent you have gain to throw away inside the feedback loop (and phase margin with the OT/speaker, but that's another story).

I have soldered a Zobel network ...
1. It takes away highs as expected but it's pretty much like putting a blanket over the amp. A cut control works better IMHO, and it can be dialed out.
2. It has a more pronounced effect on amps without global negative feedback. On amps with global negative feedback it only makes a discernible difference if the presence control is turned up.

Exactly. This is for amps who don't already have feedback wrapped around the output stage to smooth out the frequency response.

And this isn't a magic-bullet for "tone!" If you already like the sound of the amp, it's silly to tack on something that will change it. Also, you, by definition, have to tinker the values to perform properly with any specific amp, so copying someone else's values won't work too well... especially if you're using different speakers or OT's.

[TIMBO]

Hi guy's,Just doing some topic surfing to do some SMOOOOOOOTHING out of the Plexi Gain 20 as it is producing some unwanted ringing of struct notes (more at the beginning than at the end of the note).I looked at Tubenits amp smoothing options and conjunctive filtering was on the list  but it seems to be a subjective mod and requires a lot more technical equipment and knowledge than i have,so if any one can offer any direction that would be great.Thanks

[HotBluePlates]

If your amp rings, you'll want to address the cause of the ringing. A conjunctive filter probably won't do the job.

Could be a microphonic tube, too much treble response, poor wiring layout, or some other source of oscillation. But definitely a conjunctive filter won't solve oscillation or a preamp EQ issue.

EDIT: I felt the need to clarify "ringing."

Adding a resistor and cap to a PT as a snubber was suggested to avoid "ringing." This is a form of oscillation that occurs in a power supply, and is related to diode switching. A cap and resistor network (usually on the secondary side, possible across a choke) is effective at reducing this form of ringing, assuming you correctly tune the resistor and cap values to the inductance present and the frequencies involved. This form of ringing might be supersonic.

If however we hear "ringing" as in a microphonic preamp tube, we are probably dealing with a situation that a conjunctive filter in the output section will not address.

[tubenit]

Here are some examples of what I have found helps me for smoothness.

I have NOT used 22uf or more on a 12A_7 cathode for yrs now. They add too much grit for my taste. I usually am using a 2.2uf or 4.7uf and rarely a 10uf.  Exception would be reverb circuits.

NFB and local negative feedback (like on some Dumbles) can add some clean tone but not really smoothness per say, IMO.

Plate voltages are important.  I think some types of caps in the "wrong" place can give a harsh or brittle tone. Others can smother the tone for some reason. And it depends on the amp also.

On my TOS and TBM, both needed a lower value cathode cap on the power tubes to get a smooth tone like 10-22uf.
Of course, they are cathode biased.

I've learned to do alot of A/B testing on smoothing ideas. And I suggest doing one at a time. If you normally use an effects pedal, plug it in as you check for tone.

With respect, Tubenit

 

[TIMBO]

Thanks HBP,I can say that it has a bit to much treble and maybe i could tone it down a bit and see if that helps.I was wanting to use this amp with strat and LP Jr. Thanks

[Deric]

The amp pictured above is mine.  I've used the "Ruby Mod" on every "18 Watt" (4 or 5 maybe) I've built.  Never build it with the "mod" - just A/B it after everything else is running and always noticed a slight smoothing of the high end distortion when cranked - no noticeable difference when the power amp/PI isn't distorted.  Some amps more than others - several different OT's (Heyboer, Mojo, Trinity, MPS.....).

I've used a Zobel on 2 different amps with great success - a BrownNote Dirty 18 and a Daisycutter.  Both are pretty high gain with no negative feedback.  The Zobel was very good at smoothing out the high end when the gain is cranked and not changing the tone at lower gain settings.  I first learned about the Zobel as part of Randy Fay's Daisycutter schematic.

To me, they do 2 different things.  The "Ruby Mod" is for "buzz" when the power amp is cranked and the Zobel works well with "high gain" pre-amp distortion "fizz".



I have a friend who has a Vox Night Train and was complaining about some high end fizz when cranked.  We made up a Zobel on some alligator clips to try out - my thinking is that it will sound like a "blanket".  Haven't heard back from him yet....  8>)

[The_Gaz]

I've used the zobel and corrective filter, and ripped it out as soon as I put it in because of the 'blanketing' effect.

I much prefer the Vox style cut control, and playing with the cap value/pot combo to get the right cut-off frequecy and sweep for the particular amp. Usually I prefer a 250k linear pot and a cap from 1000pf-.0047uf.

And TIMBO, is that your design? Pretty cool, but I image it's hard to mange there preamp modes with only one EQ if you're swithing on the fly. The cut control may work really nice with that amp if you have room for it, or simply just some treble attenuation here and there.

[TIMBO]

Hi Gaz, Tone Junkie presented this one awhile back http://www.el34world.com/Forum/index.php?topic=11645.0 and i couldn't resist a build being an old headbanger from wayback  i ment  no i really mean  This one has been great for the learning and at the moment it is sounding great but i sure i can get it sounding a hole lot better,so i'll try some of T's tweaks and if you could throw your 2c in that would be great.

Its just lacking a little smoothness.Thanks

[kagliostro]

I was reading with interest this tread

Before I remembered only I've seen Zobel Networks in a Chambonino amp used to avoid instability

today I noticed that someone has put a trimmer cap in the same position of the conjunctive filter

is the purpose of that the same of the conjunctive filter ?

here the schematic of a GEC 1957 amp using conjunctive filter

the Chambonino and the Hiraga

Thanks

Kagliostro

Look to the Hiraga Schematic

[HotBluePlates]

The Hiraga doesn't look like a conjunctive filter.

First, there's overall negative feedback around the OT, and a conjunctive filter is used in place of such feedback. Second, it would make more sense that the cap to ground is being used to alter the phase response of the OT and speaker, to keep the negative feedback stable.

The only point of the conjunctive filter is to reduce the impedance seen at high frequency, so that it is consistent with the impedance seen at low frequency.

[kagliostro]

Many Thanks HotBluePlate

Kagliostro

[whoops]

Hi,
I've been reading on Conjuncting Filters and found this thread.
Thank you all for the great info.

isn't 2 conjunctive filters that the Watkins Dominator has at the Output transformer primaries?

[HotBluePlates]

... isn't 2 conjunctive filters that the Watkins Dominator has at the Output transformer primaries?

Yes, you could say that.  Or you could say it has one conjunctive filter, split into 2 parts (one for each half-primary).

Why would I mention the "silly way" of talking about it?  I suspect Watkins was thinking about hard Class AB operation.  When the output stage pushes beyond the Class A range, one tube shuts off and that side of the OT is effectively out of the circuit.  Splitting the plate-to-plate form of the filter in two parts (plate-to-CT) probably ensures the filter is still doing something in the Class AB range.

[whoops]

Well it seems Marshall forgot this part of the circuit when the dominator was copied.
Maybe the high end fizz some 18watts users seem to complain when the amp is driven hard has to do with the 18watts not having this corrective filter.

What do you think?

[12AX7]

I tried this a few years ago to tame highs and i came to the same conclusion.....blanket effect.....useless. For taming high end it felt like a band aid.

I have soldered a Zobel network (basically the same thing as a conjunctive filter but on the speaker side of the OT) into a small stompbox size case across the input jack. I can simply add that to any amp with more than one speaker output jack. My findings:

1. It takes away highs as expected but it's pretty much like putting a blanket over the amp. A cut control works better IMHO, and it can be dialed out.
2. It has a more pronounced effect on amps without global negative feedback. On amps with global negative feedback it only makes a discernible difference if the presence control is turned up.

To be honest, to me it is pretty useless.

Cheers Stephan

[HotBluePlates]

... Maybe the high end fizz some 18watts users seem to complain when the amp is driven hard has to do with the 18watts not having this corrective filter. ...

Maybe.

I tried this a few years ago to tame highs and i came to the same conclusion.....blanket effect.....useless. For taming high end it felt like a band aid.  ...

If you got "the blanket" then the R was too small and/or the C was too big.  The circuit is no different than a guitar's tone pot, except that the source of the rising response is the speaker used.

If you were reasonably happy with the amp's sound before adding the filter, then you messed up right there... This is not a way to "add TONE!" but a way to fix a problem.  You can tell some Dr Z amps which used the conjunctive filter that they would be way to bright and harsh without it, so it makes sense they'd use it.

So if you applied it to an amp which didn't really have a problem (because the speaker's rising impedance was offset by its falling treble response), or it runs counter to your tastes (guitar often benefits from some bite at the speaker), the filter was misapplied.

[12AX7]

If you were reasonably happy with the amp's sound before adding the filter, then you messed up right there... This is not a way to "add TONE!" but a way to fix a problem.  You can tell some Dr Z amps which used the conjunctive filter that they would be way to bright and harsh without it, so it makes sense they'd use it.

Well, no. i wasn't and thats why i tried it. The amp was too bright and i'd spent months trying all the usual  things and felt none of them reduced the brightness without dulling the top to a point i wasn't happy with. I tried this because it sounded liek it was the perfect solution but it wasn't. I did try different values but i can't recall which. In any case the amp just wasn't for me i guess and today i have a similar version all the way thru the tone stack but tweaked further and with a totally different PA which i think makes it work for me. This one is not at all bright and sounds like what i was going for with the other amp.

What i have come to think, at least for me, is that when designing an amp i find that cutting treble is never a good thing, but that it;s best to design it so that there is not excessive treble in the firat place so you don't HAVE to cut. It sounds much more alive and "right" to me like that. I'm the same with guitar tone controls. don't like them. I'd rather have the amp tame it in the way it;s configured. I don't even know how, i just experiment till it does and this one does. It has no top end taming at all aside fro the tone stack. No snubbers of any of that, and it sounds full and meaty, and if anything im often have to ADD top end via treble and presence controls. It's balanced tonally just right. Thats why i didn't like the conjunctive filter or whatever you call it. It like other methods of taming top end always seem to leave me feeling the result also loses the amps liveliness. And in trying to retain that liveliness i came to believe to do so you need to design the circuit so that itls not too bright in the first place. I'm no tech and you are infinitely more knowledgeable than I, but thats the only way i could get what i wanted. But while no tech, i do know what i hear and that worked for me and really well leaving me with the first amp i built that does what i wanted.

[whoops]

I'm with HBP:

"If you got "the blanket" then the R was too small and/or the C was too big."

[HotBluePlates]

... The amp was too bright and i'd spent months trying all the usual  things and felt none of them reduced the brightness without dulling the top to a point i wasn't happy with. ...

What i have come to think, at least for me, is that when designing an amp i find that cutting treble is never a good thing, but that it;s best to design it so that there is not excessive treble in the firat place so you don't HAVE to cut. ...

There's a lot of merit to this idea.

The trick really is (if you're not designing from a blank sheet of paper) to know what's causing "too much treble" to start with.  Often (especially in medium- to high-gain amps), "too much treble" is really the result of many places where bass is being cut to keep the distortion clear-sounding.  But then you play cleanly and it sounds harsh & gutless.

If you were using the conjunctive filter to attempt to offset the effect of bright caps across the volume pot or voltage dividers, it would likely not have good results.

[PRR]

What's the Frequency, Kenneth?

3,300 and 0.005uFd is 10KHz. This isn't guitar tone, and not a good fit to loudspeaker >2KHz rise. It smells like the OT goes squirrelly above the audio band, which screws with the NFB, so they loaded it down to damp the >10KHz response and soothe the NFB phase margin.

If you have a 6.6K pp load and it's "too bright", think each side 3.3K with more like 0.025uFd. 5 times bigger than what gives 10KHz, so you get ~~2KHz. That may be "blanket", but now you have the problem boxed-in.

On tube amps these will have to be HIGH voltage caps. 1,000V may not be over-kill. The idle voltage is small, but the signal voltage is large, and in overdrive the voltage and current may be very large for typical caps.

[shooter]

fwiw, if your eye turns pink, call your Doc

[BetterOffShred]

I was waiting for the "Conjunctivitis" joke ..   
-Brett

[shooter]

38yrs working in hospitals does that to ya

[BetterOffShred]

I have a 4 year old son, so I got it twice in a month.  It was the worst!