Conjunctive Filters, UL design and DR Z amps

[Ed_Chambley]

I took HBP's advice and emailed Mike Z, or Dr Z about his preference in 4 ohm taps.  I got a response from Mike (Dr Z).  He mentioned the use of Conjunctive Filtering which I have read about but never tried.  Anyway, his preference to a 4 ohm tap has to do with the design of Conjunctive Filtering in his UL amps instead of using negative feedback.  I phoned him and while I did not understand everything he told me what I took away from the conversation is simple preference of limiting highs when using UL with EL84.

He simply said he preferred the response from jazz players and their use of full chords when the conjunctive filter is designed and used with the 4 ohm tap.  This is where he lost me, but he had already spend so much time speaking with me I did not want to wear out my welcome.

I tried the Conjunctive Filter of .047/1000v ceramic and a 5.6k 10watt in series on the primary of a UL design.  The idea is to equalize the frequencies.

What I still cannot explain is running a 4 ohm there is a greater equalization of frequencies.  Simply stated the high e and the low E have a much closer to the same volume output level while this changes some when changing to the 8 ohm tap and even more at the 16 ohm tap.

I have not tried it yet, but I have a few high voltage pots and am wondering if the change in resistance would cause this to shift to the other taps or if it is cap responsible or even if it is both.

I have not tried it on a regular OT, Plate to Plate and would like to hear from anyone who has.

[Willabe]

I have not tried it on a regular OT, Plate to Plate and would like to hear from anyone who has.

So you placed the C/R from UL tap to UL tap?

[Ed_Chambley]

I have not tried it on a regular OT, Plate to Plate and would like to hear from anyone who has.

So you placed the C/R from UL tap to UL tap?

Yes, Plate to Plate.  Screens were untouched.  That is how I read about doing it.  I did not leave it as it is a very expensive build.  You know, the one with KT88's.  That is why I posted.  If I am doing something wrong I hoping someone will tell me as I would like to experiment more with this.  It is a very nice improvement, especially in the 9th to 14th fret chordal range.  Also running triads up and down the neck were never harsh, even in the upper registers.

[Willabe]

Yes, Plate to Plate.  Screens were untouched.

Ok.

You know, the one with KT88's.

 

Yes, I remember. 

It is a very nice improvement, especially in the 9th to 14th fret chordal range.  Also running triads up and down the neck were never harsh, even in the upper registers.

Cool. 

[shooter]

I found this from ampgrage ;

IMO placing it directly across speaker coil is preferred as it allows you to dimension it correctly for that particular speaker. You want the amp to see a mostly resistive load. What tubes are in the amp is not relevant here. The electrons flowing from cathode to anode have no knowledge of metaphysical meaning of their travel, they just move obeying laws of physics.

I've seem, too many years ago to remember in what, RC, or just C, networks across speakers, I think you've given me something to tinker with
I'm thinking this "test" doesn't care SE vs PP?

[sluckey]

I wonder if Dr. Z's use of a conjunctive filter was influenced by the fact that the AO-35 amp uses one?

[HotBluePlates]

Very well could be, although in a very early interview he referenced RCA's term of "corrective filter" and accurately restated RCA's infromation that it is used to smooth high frequency response when negative feedback is not applied.

A speaker has inductive response over part of its range, causing the impedance to rise with rising frequency. The cap and resistor in series, placed in parallel across the OT primary, present a falling impedance with rising frequency. You can experimentally find the best balance for a given OT and speaker.

But you've pointed out enough similarities with the AO-35 over time that it seems a likely source of inspiration.

[jazbo8]

Anyway, his preference to a 4 ohm tap has to do with the design of Conjunctive Filtering in his UL amps instead of using negative feedback. 

In the Carmen Ghia, which is a direct knock-off of the Hammond AO-35 as sluckey has mentioned, there is some feedback from the speaker back to the PI or are my eyes tricking me again... if so, what did the Dr. mean by "instead of using negative feedback"?

[clyde]

The RCA reference refers to one of the RCA tube manuals.  I have it here and will try and find it.  As I recall, the conjunctive filter was used as a remedy for under-specced output transformers.  I have used it on several occasions with mixed results. 

[tubenit]

My experience with conjunctive filters is that they work well in reducing unwanted hash ............. however they seem to kill the tone and touch sensitivity/articulation on my amps at the same time.  I found the "enhance cap" and smoothing caps to be a more useful solution for me.  The conjunctive filters sounded like a thin blanket was draped over the amp.

I have tried them on 3 different amps and removed them almost immediately. 

With respect, Tubenit

[John]

Tubenit, I think you're the one who's used the small pf values across the PI plates? With good results, I believe. (thinking of doing that on mine, is why I ask)

[Ed_Chambley]

I wonder if Dr. Z's use of a conjunctive filter was influenced by the fact that the AO-35 amp uses one?

The original was built using the AO-35 and he continued using them for a while.  I do not know how many were built using the Ao-35, but he does not hide the fact.  I believe Richard Goodsell did the same with his model 17.  I saw one of them at a Guitar Center in Marietta GA.

[Ed_Chambley]

Anyway, his preference to a 4 ohm tap has to do with the design of Conjunctive Filtering in his UL amps instead of using negative feedback. 

In the Carmen Ghia, which is a direct knock-off of the Hammond AO-35 as sluckey has mentioned, there is some feedback from the speaker back to the PI or are my eyes tricking me again... if so, what did the Dr. mean by "instead of using negative feedback"?

I have not seen the schematic, but I have seen and repaired the Maserati, the Route 66 and the KT45 and have the Schematic for each and I see no NFB in these. 

What was said is the conjunctive filter properly configured will smooth and reduce noise and still maintain harmonic content.  I am thinking it is to attain some of the raw qualities an AC-30 has but without the harshness that EL84's can have.  Just guessing of course.

He did mention he was not a player and was having to rely on players feedback.

Do you have a Carmen Ghia Schematic?

[jazbo8]

I have not seen the schematic, but I have seen and repaired the Maserati, the Route 66 and the KT45 and have the Schematic for each and I see no NFB in these. 

What was said is the conjunctive filter properly configured will smooth and reduce noise and still maintain harmonic content.  I am thinking it is to attain some of the raw qualities an AC-30 has but without the harshness that EL84's can have.  Just guessing of course.

He did mention he was not a player and was having to rely on players feedback.

Do you have a Carmen Ghia Schematic?

Here is the link to the CG's schematic. Interestingly, the Maserati and the Route 66 do not have the conjunctive filter across the OPT, nor do they have NFB, how about the KT45? With the values shown, I don't see how the conjunctive filter in the CG can "reduce noise and still maintain harmonic content", as the highs get rolled off fairly early, depeding on the actual load, somewhere below 1kHz. But hey, if the players like it, what do I know...

[HotBluePlates]

What was said is the conjunctive filter properly configured will smooth and reduce noise and still maintain harmonic content.  I am thinking it is to attain some of the raw qualities an AC-30 has but without the harshness that EL84's can have.  Just guessing of course.

Link to the 1999 ToneQuest interview I was talking about earlier.

TDPRI put the relevant section of the 1999 interview next to the relevant section of the RCA Receiving Tube Manual.

I explained what it does earlier. The implied statement is that if the output at the speaker jack is a constant-voltage, then the rising impedance means rising power output with rising frequency; that would be more power to the highest frequencies. An alternative view if the output is not constant voltage is the fact that a higher-impedance load would cut below the knee of the gridlines and possibly distort earlier than a lower-impedance load; if the highs see a higher-impedance load than the lows, they might distort first and sound harsh.

Either way, the filter was a recommendation to flatten the frequency response, and Dr Z.'s comment in the interview is limited to that (beware of restatements of his statements echoed around the web; they could be adding extra claims).

[jazbo8]

An alternative view if the output is not constant voltage is the fact that a higher-impedance load would cut below the knee of the gridlines and possibly distort earlier than a lower-impedance load; if the highs see a higher-impedance load than the lows, they might distort first and sound harsh.

That view makes more sense if the output tubes run in pentode not ultralinear mode.

Either way, the filter was a recommendation to flatten the frequency response, and Dr Z.'s comment in the interview is limited to that (beware of restatements of his statements echoed around the web; they could be adding extra claims).

I think you are right after reading the post at TDPRI, the bit about noise vs. harmonics might be added by someone else.

[SoundmasterG]

The Bogen CHB100 that I gutted and modded used a conjuctive filter plate to plate, and so did a Conn organ amp that I have around here. I haven't tried them on any amps yet though as I figured it was a hi-fi thing and likely something I wouldn't want. :)

Greg

[jazbo8]

I wonder if Dr. Z's use of a conjunctive filter was influenced by the fact that the AO-35 amp uses one?

So I went digging around TDPRI after seeing HPB's post, and found this interesting thread. So despite Dr. Z's reference to the RCA Manual, the CG was a basically a modded AO-35, and to me, the conjunctive filter isn't even that important in this design. Instead take a look at the PI and the feedback network and the changes that Dr. Z made compared to the stock AO-35. I don't think there is another guitar amplifier that use the same funky PI arrangement like it, if you know of one, please let me know.

[Ed_Chambley]

Thanks for posting the Tone Quest article HBP.  Reading it I understand more of what he said. In reading the Tone Quest article he states the conjunctive filter is related to the touch dynamics.  I am sure everything he mentions along with circuit changes are a large portion of his amps tone, but he gives a lot of credit to the corrective filter.

There are a lot of mixed up info about his amps and this subject, but I have sense tried it on another amp.  A KT66 50 Watt Plexi and I quickly ran across what the "blanket" comments meant.  His reference to speakers got me thinking more so I plugged into a 2, 10's.  A Celestion Gold and a Vintage 30 as he mentions his preference to speakers.  Wired to 4 ohm.  Better, but not what I prefer, but I am not a huge fan of the lower power Z amps.  Now I know why.

Just experimenting I did not like the dull tone, but increasing the value of the slope resistor seemed to open the amp more.  I will say the leveling of response from string to string and fret to fret is not subtle and for my fingerstyle playing I have to adjust or the higher strings are overpowering.  That is why every time I have tried the Carmin Ghia, which players I know rave about, I felt I would have to over-compress to get a tone I like.  It may take a little getting used to, but a lighter finger touch for me is necessary.

It may be like many other things I have tried and dis-guarded, but I do find it interesting.

[HotBluePlates]

... Instead take a look at the PI and the feedback network and the changes that Dr. Z made compared to the stock AO-35. ...

Looks like Dr Z keep most of what the AO-35 had as far as phase inverter and output section. There are slight changes to power supply resistors to adjust supply node voltages, but as far as that goes the phase inverter grid & cathode voltages are essentially the same in the Carmen Ghia.

There is a change to the AO-35's 2.2MΩ and 3.3MΩ feedback resistors down to 220kΩ & 3.3MΩ in the Carmen Ghia is provide additional gain. The input coupling cap to the phase inverter is made smaller in the Carmen Ghia. Of course, the preamp has guitar-friendly input circuitry and a tone control circuit added.

... I don't think there is another guitar amplifier that use the same funky PI arrangement like it, if you know of one, please let me know.

Look at it as a fixed-bias long-tail inverter.

The power supply is made a bit over-complex to create a fixed-bias arrangement for your standard long-tail inverter, in order to derive set cathode and grid voltages to bias the inverter. The AO-35's 24v and 27v pretty nearly match the voltages indicated for the Carmen Ghia.

The grid reference resistor(s) for the long-tail have to assume a different form because there's no cathode bias resistor under which you can anchor them. Imagine you had a self-biased long-tail with grid reference resistors of 235kΩ each, which go from a grid to the non-cathode end of the cathode bias resistor. Ignore the tail resistor for a moment and assume you have a differential stage where the cathode resistor is connected directly to ground... If both resistors are the same value, and they were disconnected from "ground" under the cathode bias resistor, you'd still measure 0v from their junction to ground (because the circuit effectively has equal and opposite grid signals). As a result, you could avoid connecting them to ground altogether, and since they are now 2 resistors in series, you can replace them with a single resistor of 2x the value of each individual resistor. This is actually quite common in hi-fi and in instrumentation.

You probably won't see a fixed-bias long-tail in the average guitar amp, but that's probably due to the economics of "bunch of extra resistors and caps" just to switch from self-bias to fixed bias. I don't think it's a back-of-napkin calculation to prove, but the fixed bias version should allow for higher output swing and possibly better balance than is possible from the self-bias version when overall supply voltage is low. Guitar amp economics might have instead suggested, "use a long-tail when a lot of supply voltage is available."

Now that the tail resistor is hooked to a solid power supply voltage instead of the top of a shunt feedback resistor, feedback can't be injected at the cathode. So negative feedback is instead applied at the grid of the phase inverter. The AB165 Bassman does that as well.

[Backwoods Joe]

Mesa Boogie 20/20 power amp uses a fixed bias Long tail pair if my memory is correct.

[HotBluePlates]

Mesa Boogie 20/20 power amp uses a fixed bias Long tail pair if my memory is correct.

I'd almost look at the 20/20 power amp as having a fixed bias paraphase inverter.

The drive for the 2nd half of the inverter appears to come from the OT secondary through a 56kΩ/1.5kΩ voltage divider, and does include a parallel path which reduces voltage division for highs (along with an associated control) to boost Presence.

There is no shared tail resistor to the inverter so it is not a differential or long-tail type of inverter. A rare-bird, but functionally like any other paraphase where one side amplifies, then output signal is returned to the other side through a divider to invert polarity.

[sluckey]

There is no shared tail resistor to the inverter so it is not a differential or long-tail type of inverter.

I think the shared resistor is the 15K (shown on the power supply page) connected between point D and the negative bias supply. Notice there is no filter cap at point D.

It's an oddball. I've seen a couple other weird Mesa schematics too.

[MFowler]

I've gone over the Carmen Ghia with Mike Zaite in the past he freely gives up information such as our discussion on cap values in the Zwreck his version of the Trainwreck Rocket.


Here is a note a fellow amp builder shared on Ampgarage.



From the Doc in a PM I got a while back about this question: "OK , well that is a good question, one not easily answered with a simple response. As you know the original 50 or so Ghia's were built on modified Hammond organ chassis, these units varied from piece to piece do to fact that each chassis varied as a base. In 1994 I released the ghia II if you will, it was my own chassis design and duplicated transformers from the Hammond. From then on that Ghia's have stayed as close to same as humanly possable with no MAJOR design change, why change a good simple thing. But through the years I have adjusted a few values. Now I have built over 2000 Ghias, each handmade so expect some variance just by natural touch. But these are the things I have adjusted through the years: I decreased the Global feedback resistor from 2.2 meg to 3.3 meg. 2.2 a bit smoother and slightly less output, 3.3 more edge and more output. On the 100K plate resistor of V1 , second stage , recovery stage after the tone control, originally there was a .001 cap across that 100K. It would roll of some highs and warm up the signal some, again less edge, smoother tone, and less output with it in. I removed because I heard may critic it for not enough high end and less VOXY sound. Those are my only circuit tweaks. I have a company ( Orange Drop SBE ) manufacture all my coupling caps , my name is on each one there consistency is increadable. I use the same Carbon Film caps, and the same CTS tone pot. My Output Transformers have been change due to companies going out of business. A problem with building antiquated technology suppliers dry up in the course of 20 plus years. I now use Heyboer and I feel they are nicest, and most expensive OT I have used in the ghia's. Tubes yes tubes are a problem as well. In the Ghia I use two NOS tubes always have, a 5751 in V2 and a 5Y3 GT. The supply for 5Y3's is still strong, but now everyone has heard of the 5751 and supplies have dried up. I swear before I started using 5751's in the Ghia's no other manf would touch them and many didn't know what they were. So I have been
forced to use NOS 12AX7 in V2, and this does increase mid drive some and output as well. So there you go the long winded Saturday morning answer to your question. Z"

[HotBluePlates]

There is no shared tail resistor to the inverter so it is not a differential or long-tail type of inverter.

I think the shared resistor is the 15K (shown on the power supply page) connected between point D and the negative bias supply. Notice there is no filter cap at point D.

It's an oddball. I've seen a couple other weird Mesa schematics too.

Good call on the tail resistor.

[jazbo8]

The disucssion on the PI and the quote from Dr. Z are very informative, thanks guys!

[PRR]

> weird Mesa schematics

+1

Agree there is a 15K tail; connection from output is NFB.

Tail is too small, leading to the 68K:120K unbalance in PI plate resistors.

[shooter]

Tail is too small

would this be a case of an engineering error creating a happy sound

[MFowler]

Crowded layout because the component values were added to layout.

[DummyLoad]

> weird Mesa schematics

+1

Agree there is a 15K tail; connection from output is NFB.

Tail is too small, leading to the 68K:120K unbalance in PI plate resistors.

seems way unbalanced. tube cad says 75k for Ra2 if not for the wonky NFB/presence ckt.

--pete

[jazbo8]

Looks like Dr Z keep most of what the AO-35 had as far as phase inverter and output section. There are slight changes to power supply resistors to adjust supply node voltages, but as far as that goes the phase inverter grid & cathode voltages are essentially the same in the Carmen Ghia.

WRT Mesa's 20/20 schematic, how do you get both positive and negative bias voltages from the same half wave rectifier?

[sluckey]

WRT Mesa's 20/20 schematic, how do you get both positive and negative bias voltages from the same half wave rectifier?

You don't. You have to look at the power supply schematic AND the PI schematic to see what's happening. You have a voltage divider between point B (+326V) and the negative side of the bias cap. Begin at point B and follow the path of the voltage divider... 68K plate load, tube (plate to cathode), point D, 15K, finally negative bias cap. The two halves side of the tube and plate loads are parallel to each other, but just following the divider path thru one tube should clear it up.