Controlled feedback

Players,
I'm looking for that magic moment of melodic feedback, which I always struggle finding.  And I'm not talking about the squeal of unpotted pickups, I want a note related to the one I'm playing.

For gigging I use the GM, and most of time I use an Line6 HD500X as midi controller and input effects like wah, OD, EQ, etc.  I play in a 7 piece so cranking my amp to 10 is not an answer for me - although I can play loud(ly).  I'm open to any channel at any setting.  I am using an SG with moderate gain SD pickups, and sometimes use an LP with stock P-90s.

On a side note, at band practice I use a Digitech GNX3000 into a solid state power amp, and I can get that to feedback rather easily.  And that's at a fairly low volume.  Is it a natural compression in either the Digitech or the power amp?

Help me Mr. Wizard?

Natural Feedback frequency is determined by your distance to the amplifier. You can force it to be (almost) anything you want with a sustainer or an Ebow.

A couple of years ago I performed with Steve Vai, and he had all these spots on the stage marked with tape. I asked Thomas Nordegg what they were and he said they were where he had to stand at certain points in certain songs to get the feedback frequency for whatever song was at hand.

I've heard that about Steve Vai. I've also seen Adrian Belew and the way he gets different notes by changing guitar orientation.
But neither of those explain why with two different rigs in the same location, one rig feeds back much easier than the other. For dress rehearsals at practice I use the GM36 and it just doesn't feedback like my crappy rig.

I find the same thing Namklak, my GM doesn't give too much in the way of sustained feedback except at extreme gains which I don't normally use. I use a Les Paul Traditional, that's the version with the unchambered body, so it's a pretty heavy beast. It's not a dead guitar in normal playing but I had always wondered if it was the guitar which was unresponsive in some way, but I do get feedback from it with some other amps.

I've heard the same thing about Gary Moore, the crosses on the stage for the long sustained notes. It won't be purely the distance to the amp, (time of sound from speaker to guitar), which denotes the frequency, it will also be transit time of signal through the amp and the additional path effect of soundwave reflections which are added to that.

While distance is certainly not the sole factor, it is the one variable that can be changed at any moment (you move. That`s it). Whether you get your 5th harmonic feeding back at 11 feet or 13.266 feet, moving away from your amp will affect the pitch the same way.

As far as why one rig feeds back "easier" than another, it`s a matter of harmonic oscillation and how sensitive that particular device is to it. Of course a tube amp with a failing tube going microphonic will feed back even at bedroom volumes and be nearly uncontrollable at rehearsal levels.

As I mentioned, the Ebow has a setting to coax the harmonics of whatever note you`re on (feedback is ultimately a harmonic of the tone causing it). Joe Satriani uses a sustainiac to get feedback style harmonics without having to be so specific about where he is on the stage (effectively being able to illicit any feedback pitch he wants at any time).

Okay, so I saw The Aristocrats about 2 months ago. Guthrie had a 1x12 combo, I was behind the guy leaning against the stage and almost couldn't hear GG's amp, never changed his distance from his amp (although he did change orientation occasionally), and he got copious amounts for glorious melodic feedback! What is his secret??? I mean he is a god, but even god's have tools, eh?

namklak wrote:Okay, so I saw The Aristocrats about 2 months ago.  Guthrie had a 1x12 combo, I was behind the guy leaning against the stage and almost couldn't hear GG's amp, never changed his distance from his amp (although he did change orientation occasionally), and he got copious amounts for glorious melodic feedback!  What is his secret???  I mean he is a god, but even god's have tools, eh?

He`s very secretive about his guitars, so I didn`t really notice if he was using a sustain device but he could have been. Other than the obvious of using the wang bar or otherwise altering the origin tone causing the feedback you can`t organically alter it without a device of some sort. I`ll ask him next time I see him (a joke. I don`t actually know the guy on a friend level, we just happen to cross paths once in a while).

Great pic VoodooJeff!
I'm not necessarily ruling out a catalyst device like a Fernandez Sustainer, or more convenient,a pedal...

Um, bump?
Compressor? Switch to AX8?

I've always taken the view that, just as Jeff says, feedback of our sort is mainly an acoustic issue.  You have the loop:- guitar -> pedals/effects -> preamp -> effects loop -> power amp -> speakers -> air volume -> guitar.  The different stages through this sequence all add in their own time delays to the signal, mostly swamped by the speakers -> guitar area again just as Jeff says.  This time delay is a major factor in the frequency of feedback but it isn't the sole effect.   They also each impose their own frequency curve with unique peaks and troughs on the signal as it goes through so that when it comes back to the guitar it has been shaped to produce an overall curve which favours certain specific frequencies.

There is another effect which is not often mentioned, reflection from surfaces.  Consider this setup.  An open backed cabinet set a few feet in front of a brick back wall.  The guitarist is standing in front of the amp with some space to play around with.  The sound from the amp will travel forward to his guitar, taking up some time to do this, and the rear sound will travel backwards to the wall and bounce forwards again until once again it hits his guitar, taking more time to travel its path.  Obviously the difference in the path length that they travel is "2 x distance from back wall".  The rear radiation of a speaker is 180deg out of phase with the front so would normally cancel as we all know but if the delay in covering that distance is enough to bring the wall reflection close to being in phase with the front it will then add to it and increase the feedback effect, if it maintains it near 180deg out of phase it will prevent feedback.

Don't know about you but I usually find that I get feedback around G14 (A), that's 440Hz and it works out to a wavelength of about 2.6ft.  That means that at distances where the front back difference is "1.3ft + a number of full wavelengths" the waves will add and increase acoustic feedback.   That's 1.3ft, 3.9ft, 6.5ft, 9.1ft, 11.7ft etc.  Try this experiment.  The next time you are in a position to create feedback, (maybe home doesn't allow that much volume), set your amp up a few feet in front of a solid reflective wall, that's brick or plaster or glass rather than curtained.  Set yourself in front of your amp and generate just the right amount of gain to achieve feedback on any note that occurs and note your distance from the amp.  Now move in a semicircle in front of your amp keeping the distance from it the same.  If it's only the front direct radiation that counts then it should feedback at any point on that path, allowing for the fact that the volume drops off a little as you go off axis.  As you move round the semicircle the front wave stays the same distance from your guitar but the rear reflected wave changes.  This means that, at that frequency , it is going out of phase and will cancel out rather than strengthen.  In a corner, a few feet from the back wall and a different distance from the side wall, it gets even more complex, and glass windows say will improve things again.  My bet is that, if you have the guitar just on the verge of feedback you will find points on the path where it will feedback and points where it won't, and that's those other room reflections coming in to play.

Like others I have watched the Gods move to specific points on the stage for certain notes.  They aren't just moving to a distance from their sound source, that's a very rough first approximation, they are moving to specific points where all of the sources, including the multitude of reflections, add up to the amount needed to feedback.  The whole environment has a focusing effect and produces points where there is a genuine sharp peak in sound pressure.  And it's different for all frequencies as they all have different wavelengths and the path differences have a different effect.  We usually consider the front radiation to be the creator of feedback but in the real world it isn't.  It's a combination of all of the reflected soundwaves, major and minor.

And there is another thing to think about.  The loading on your guitar is a crucial factor in tone that I have bleated on about before.  The most basic factor affecting the tone of a guitar is its fundamental electrical resonance.  No amount of "good wood" or high quality fittings can make up for a guitar with the wrong resonant frequency.  This electrical resonance is set by one factor and one factor alone, (remember it's electrical), and that is the resonant circuit set up by the pickup inductance and the load capacitance.  High load capacitance does not "bleed off treble"!  It shifts the resonant frequency down.  Too far down gives the muddy sound which is often cited, too high up gives the guitar a shrill and thin quality.  Higher inductance pickups need to be matched with higher load capacitance to sound good.  Resistance of a pickup is the most useless measurement to quote, it gives absolutely no hint as to how the pickup will sound in use.

I know this because my own LP suffered from it when I started to use my active buffered cables.  It needed the capacitance at the input of the head amp increased to compensate for the now 12" cable to the head amp.  Once that was addressed it became much sweeter and beefier.  Load resistance (note that is specifically resistance) of the combined guitar controls and the amp input then dictates how large the resonant peak will be but it doesn't affect its frequency at all.  Now all of this will also have consequences for our feedback because the pickups are already leaning towards a specific frequency with a peak in their output, and higher signal output at a certain frequency will make that frequency more susceptible to feedback.


EDIT:  Whoops, idiot alert!  When I said "Higher inductance pickups need to be matched with higher load capacitance to sound good." I really meant that the other way around.  

Code:

                      ______
Fres = 1 / ( 2 x Π x √(L x C) )

If L goes up C must COME DOWN to keep the same centre frequency.

I've noticed this too namklak, I love my GM36 to bits but I definitely notice a difference, it is harder to coax natural feedback compared to any other rig that I've owned. Same room, same guitar, same volume.

I dig the sciencey explanations here about feedback (seriously, it's fascinating!) but I'd love to see a post that explains why the GM seems to be a little prudish with it

Thanks!!!

Bordonbert strikes again. At practice where I get the desirable feedback my amp is about 4 from the back wall, and I'm using an 18' George L. Note this setup does not use the GM36.  All other times I'm using a line 6 wireless with the cable compensation set to zero feet, and big stages with no close back wall, or outdoor stages.
Well, with my rig,I can put my amp back as far as I want, like 3.9.feet from the back wall, because I use my power soak to drive a separate small 1x12 cab monitor on the front of the stage. Then I just need to find that 2.6'*X spot to stand on stage to stand.  Sigh.... it could happen...

Well, now I disagree with bordonbert on something.

First let's take the circuit of an ideal generator and parallel RC load (shunt to ground). In this case, changing the R or C does not change the freq response of the circuit. But no generator is ideal - real generators have finite non-zero output impedance. In the real world we then have the equation
f = 1 / (2 * Pi * R * C). Changing R or C changes the freq response.
Guitar pickups are very far from ideal generators. They have a complex (resistive and reactive) and significant output impedance, so every load change will make a difference in freq resp and resonant peak. I've seen on the 'net very well documented (meaning the test setup was well documented) graphs showing that just changing the tone or volume pot resistance can change the resonance peak and freq of a pickup. The tone pot circuit is an RC shunt circuit on a generator, just like the circuit above. So I disagree with the quote from bordonbert above "Load resistance (note that is specifically resistance) of the combined guitar controls and the amp input then dictates how large the resonant peak will be but it doesn't affect its frequency at all." Load R does change the freq.

But all this does support bordonbert's point that a cable can change the resonance of the guitar/pickup combo. Now I just need to find a 3 ft cable that loads like my 18 ft George L.

f = 1 / (2 * Pi * R * C) is correct Namklak but does not apply to our case.  You're absolutely right, changing either R or C in that equation will alter the frequency but that is for RC filters not a LC tuned resonant circuit.  You should be using:

f = 1 / (2 * Pi * sqrt (L * C) )

and as you can see that is independent of R with both L and C altering frequency.  The R comes in only as a damping factor to the resonant peak.

In truth I simplified the circuit a little but it doesn't make too much difference.  Real pickups are a little more complex than a straight L + R in series.  They also have their own winding capacitance in parallel too, but this gets swallowed up in the parallel capacitance we attach as loading.  Humbuckers are of course two coils in series each with their own LC circuit so again it gets more complex but for our simple purposes the straight inductance of a pickup is the parameter we should be looking for, resistance tells us absolutely nothing about how it will sound except how much wire was used in making it, and even then it also depends on what gauge wire was used.  But the myth of "resistance" of pickups being a measure of anything useful continues.  And then there is the never considered question of Eddy currents....  

I think I've posted this before but this is a great read for anyone interested in this sort of stuff.  All the info you could want to get a clear picture is there, and a few debunkings on the way!  The Secrets of Electric Guitar Pickups

EDIT:  Make sure to have a look right at the bottom for Fig 14 and Fig 15 which show the curves of a Strat pickup with just the load resistance varied and just the load capacitance varied.  That demonstrates it clearly in real life.

Harmonic feedback? A real moment when it happens, whether it comes off is another matter, but good fun trying! For me it's getting to know your guitar, which pickup is more likely to work on, use your volume control ie back it off a bit and use your ear and coax it with the pot. I find strings important. B string does it for me more often than G or E. A string pinching the string usually produces results.

I find my Les Paul Standard does it best, bridge pickup for the sustain into the 5th, or middle position for the endless note. Boogie Mark iii does it better than any other amp followed by my Marshall rackmount JMP1 and 20/20 power amp. Unfortunately I've pretty much given up on the GM, nice concept but I can get the tone I want and I'm on the verge of giving up and selling. Just got a Boogie Mark 5:25 head and waiting to gig it in a couple of weeks.

It's a shame to hear you're giving up on the GM Stapletongas, I'm still learning how to get the best out of it and I have to say, I still find it the most versatile amp I've ever played.  It is a complex setup but when you can play one feature off against another it gives an amazingly wide set of sounds.  That said, there are some days when I can't get to grips with it and end up with far too much of a modern sound for what I need.  I just seem to lose track of the balancing of each aspect and it ends up far too strident.  Then I get it right and it really sings and all is right with the world.  I do now feel that if I had to make a criticism it would be that the power soak options do not give me enough variation in control.  I would like a couple more power values there between 5W and 36W and it would be better to have a couple based on the full 4 x EL84s.  People talk about output valve saturation but I feel like it also really needs to drive that output transformer as an overall part of the saturated sound of the power amp.  Just using 2 x EL84s may drive the valves but the transformer is pretty much sitting there idling.  (Didn't someone else once mention this was how he approached getting his preferred sound?   )

We seem to be looking at this whole feedback thing from a narrow perspective.  Some link it to the amp they use and others link it to the guitar.  In fact it has to be seen as a whole loop system, that's the way it works and that's the only way it works!  The theory of feedback as in 'Control Systems' is a wide ranging one in the engineering world and comes down to well established general principles which are readily available on line.  How to display the response of a system and what criteria are useful to measure and test whether feedback is likely is more involved but the basic idea is simple.  Our acoustic feedback in a guitar system is a classic case for the theory.

Any form of self generating feedback in any field relies on one concept, a loop gain of >1 with a feedback signal which is in phase.  Our loop, (as I said earlier), is something like:  guitar -> pedals/effects -> preamp -> effects loop -> power amp -> speakers -> air volume -> guitar.  If you break that loop at any point it will have an input coming to it (from the left) and an output from it (to the right).  These may be in the form of a voltage or a sound pressure or even other forms in other feedback systems.  The first half up to the speaker input is electrical and the remaining portion back to the guitar is physical.  Let's say it is a voltage we are dealing with as in the power amp node.

For self generating feedback to occur we need to have a setup where at some frequency, when the system is triggered by the slightest signal and feeds it into its output and then around the loop, it generates at least the same signal back at the input to that node.  If we put 100mV on the input of the power amp, it will work its way around the loop until it finds itself back at the input again.  For feedback to occur the signal after passing through the entire loop must be greater (or equal to technically) than the 100mV we applied to create it.  Then the signal grows and becomes self generating.  It will continue to increase each time it passes around the loop until some part of the system limits it from any further growth, then it will stabilise at that level.  If the returned signal ends up lower than the signal we applied to cause it, it will die away and cannot be self sustaining.

You can't really disassociate the guitar contribution from the amp, they are all a part of that loop and play their part in passing the signal through, perhaps converting it from one form to another, either amplifying it or attenuating it.  So what makes for easy feedback?

Well, the loop gain which dictates whether we will have feedback or not (>=1) is a combination of the gains from every stage of the system multiplied together.  Looking at it overall there are areas where the local gain will be low naturally.  The speakers are terribly inefficient and the speaker to guitar interface via the air will also be incredibly lossy, a loud output from the speaker making the guitar vibrate in sympathy will only induce a very small voltage in the guitar output.  If this means an effective gain of only 0.01x then the gain in the other parts of the system will have to more than make up for this loss with at least 100x to restore the overall gain to 0.01 x 100 = 1 (or more).  

So as per the discussion so far, finding that spot where the acoustic transfer is focussed and hence more efficient is crucial to making it easier to get feedback overall as it lessens the need for higher gain in the rest of the system.  I reckon the H&K is a difficult amp for this because of its design aim to sound good for modern high gain metal sounds.  I don't have any problem getting acoustic feedback with my really heavy LP Traditional in the Ultra channel, but the tone by my standards is dreadfully sharp and coarse.  It suits some, just not me. It will work on the Lead channel with reasonable gain as long as I am in a location where I can up the power soak and MV levels a bit without bringing the environmental officers round with a writ.  I also think the GM may have a less compressed sound than most other amps, which is great for some sounds but bad for others. I find this is really noticeable at 18W power level and below and may be made worse by the halved output transformer drive and the purely resistive power soak.  This again means that settings may have to be made a tad lower than say a Marshall for the same perceived loudness.  I think this makes some of us deliberately try to dial these aspects out in our search for smoother more classic tones and in doing so we reduce the gain in the electrical stages.

It's a weird area and up for discussion of course.  Personally, I don't miss this aspect on the good days when I can get the amp on song and it sounds smooth and velvety.  When I lose track of this and I start panicking and dialling in perhaps too much gain or guitar volume then things do go a bit awry, and that is when my feedback gets easier!

Just noticed your mention of your George L cable Namklak.  Their 0.155 and 0.225 are both listed by "other sources" as 67pF/m.  At 18', (which is almost exactly 5.5m), that gives you about 370pF.  All you need is to load up any other cable with enough capacitance to bring it up to that level and Robert is your mummy's male sibling as we say!

When you say "All other times I'm using a line 6 wireless with the cable compensation set to zero feet", does this mean the guitar load capacitance is set to the equivalent of zero feet of cable?  If so you should really be setting it to close to that full 370pF of your GL.  If necessary a 330pF or 390pF (standard values) inside the guitar plug across the contacts would load it up from nothing and may sound more like your sooper dooper one. I trim my own buffered long cable in exactly the same way. Without it it sounds a bit thin and sharp, (high resonance).

I'll go backwards in time responding to posts.

The Line6 G30 wireless has a "Cable length simulation" switch - with the word simulation I don't think it adds capacitance.  Most users leave it set at 0feet.  I probably have a 330ishpF cap laying around, I could maybe mod a short cable like I use with my wireless.

After reading all of this (bordonbert, don't forget to mention phase in your control systems dialog) (also, I just started the excellent write-up on secrets of pickups), I'm thinking my Digitech GNX-3000 must have a built-in feedback catalyst algorithm,  because I can easily get feedback at basement practice volume using my 6lb SG (which has almost no natural sustain, it's all SG snap).  So maybe I'm chasing a false god...  Because as brodonbert pointed out in his control systems post (of which I has two semesters of circa '83), loop gain must be greater than 1.  And the phase must be correct, but that is a whole other beast as every device in your chain skews the phase response the signal.  Hmmm, maybe my HD500x has a 0/180 setting...

Stapletongas, I play in an Allman Bros tribute band, so there is another lead guitarist. He uses the MB Five-90 Combo (6L6/EL34 instead of EL84 the 5-25 uses).  He only uses two channels, he doesn't get feedback any easier than me, and his tones are really no better than mine at twice the price.  When our CD comes out in early Nov, I'll post a track or two.  Please report your findings on the 5-25 vs GM36.  Why don't you try the JP-IIC and let use know about that?  

Update on the Boogie Mark V 25. Buying this amp has made me realise that you shouldn't need to spend much time getting the tone you want and getting it to do what you want it to do. I think in retrospect I got sucked into really liking the idea of the GM and wanting it to meet my requirements for tone, because I liked the concept so much and I did that to the point of all the tune swapping, trying different guitars and cabs, when the truth is the concept was great, but the tone of the amp simply is too flawed for my liking.

Gigging the 5:25 the other weekend was an absolute joy playing classic rock covers. I'd spent little time with it pre gig, added a chorus and delay through the loop and it performed and sounded fantastic.

Getting back to the subject though...Do Anything You Want To - Thin Lizzy...that bit in the middle where Moore and Gorham sustain in harmony, I got that effortlessly, picked up the harmonic feedback and infinite sustain.

So it's bye bye GM. On eBay coming soon.

Thanks for the community though people.

Sad to see you go Stapletongas, though I undeerstand exactly what you are thinking. I did recently realise that I'm spending more time thinking how to mod this amp than playing it now, and that gives rise to some very uncomfortable questions. It's an immensely controllable beast but it has to be said that it has an underlying high gain bias in its sound which seems to come through whatever you do with it. If that doesn't suit you then yours is the only answer. And for myself, I'm missing some of that old Marshall chestiness for my classic rock stuff which I can't really seem to find anywhere on the GM. Maybe I'll go the whole hog and build one myself from scratch. At least that way I'll get some of the tone I want, if not the flexibility.

Stay in touch and do let us know how you get on. It's been great talking with you.

Father, forgive me for I have sinned - I have lusted after another amp.

To me, the GM is the best modeler there is - ya can't beat modeling with tubes.

But...

I've been in this AllmanBros tribute band for 3+ years now, and the band is getting stronger and getting better gigs all the time.  Instead of modeling a Marshall, maybe I should just ...   I've been lusting at the JVM205H - there I said it!  I'll practice an extra hour tonight for penance!

This next gig is a bar gig (yuk), so I'll experiment by putting my main 1x12 on a tilt-back stand right behind me like at practice (instead in more of a backline setting like most gigs) - that close proximity seems to help elicit the harmonic feedback I'm looking for...  That make sense - I'm increasing the coupling between guitar and speaker, increasing the loop gain.  Maybe I don't need no stinkin Marshall after all! Besides, I've got the GM Ultra channel sounding something like the Warren Soldano tone,

And I absolve you of your sins, in the name of Greg, and of Duane and of the Holy Grandmeister. Amen.

I understand.  I sympathise.  I forgive.  I'm kind of going through exactly the same thing as you Namklak.  It's noticeable that we seem to have pretty similar demands in terms of the tonal limitiations of the GM36.  I spent a couple of hours last night browsing other low power valve amps in the same quest, could I find one that was cheap but with a killer classic tone?  I'm looking for a creaminess to replace the sharp top end which we have all always said is a characteristic of the amp.  It can be tamed but is always there as a tribute to the modern sound they aimed it at.  And I still want those tone controls sorted out!

In addition to my LP Trad I now use a Line6 Variax JTV59 which I got at a reasonable price.  Once again I'm stunned at how good modelling technology is despite the purists who will always reckon it stinks.  We play numbers with slide in so I need to have access to Open D for some and in particular to a ZZTop EADGBD hybrid tuning I use.  I also like having the variety of guitar types on a knob.  We do numbers like Tom Petty's Running Down a Dream and Refugee and his crisp Rickenbacker tone is there in the JTV, as is some really good acoustic modelling for Frampton's Show Me The Way.  The JTV has all of this at the turn of a knob and on top of that the GM's Clean/Crunch side is spectacular for this sort of sound.  No more retuning or carrying a guitar for everything I would like to play.  I still prefer the LP and use it for most numbers in the set but now I only need a second guitar and can set up for the next number in seconds.  And it has to be said there are times that the GM36 let's the side down and is the weak link.

It's sad to see that H&K have nailed their colours to the mast with the arrival of the GM40.  They had an opportunity to address this and at least include the ability to set it up with softer classic tones.  They plain ignored it as far as I can see form the vids online and went along their same chosen path despite having the feedback to know that a lot of owners of GMs were looking for this.

That said, I've set myself a goal and have been working on a project which I think may bear some fruit for The Faithful.  With data that I now have 'at my disposal' I've been doing some calculations and modelling relevant areas of the amp in Spice.  The results are pretty interesting.  I've restricted myself to the SS stages at the moment.

The input buffer/overdrive pedal is an area I can work on.  Looking at it has already suggested that there is a very simple mod to be tried.  The feedback around the opamp is pretty standard apart from the diodes which we have already talked about.  The feedback in the Clean channel has in it a 3k3/470nF time constant as the Clean low end roll off.  That's 3dB down at 103Hz.  We don't want an extended bass for guitar of course but that could be a tad bass light considering the E string at 82Hz.  That could be increased a little. This also shows a high end shelving network which is pulled in for the Lead and Ultra which has its lift break point at 2k2/1uF which is 72Hz. This will also take away some of the lower end. And the clipping diodes could easily be changed to limit the signal to just under the supply voltage on both sides.  You would then have a clean Buffer/Boost and your own choice of whether to use a pedal or not. And you would still keep real overvoltage limiting to outside the opamp which we all know is a good thing to do.  Personally I think there is room to set up a better more gradual diode feedback network rather than just removing it.  A diode clipping a signal will have a portion of its range where it gradually slides in, becomes more severe and eventually locks pretty solid.  The proportion of the signal where this all takes place is significant. If you amplify the original signal and lift the clipping level as with a zener I reckon that the clipping range in relation to the overall signal then becomes much less significant, more abrupt.  That sounds harsh and I don't like the idea of it. I would want to have the same gradually reducing slope after the chosen clipping point, maybe even a tailored slide into limiting.  As an aside, I'm designing a very simple pedal to allow me much more control of this.  It's kind of like a 'Parametric Clipping' setup as opposed to two diodes butchering the signal with little control. I have control of the 'onset of bending' level, the sharpness of clipping from that point up, and the ability to control these for both sides of the signal independently. It's easy if you bother!

My main area of interest is the tonestack.  It is a fairly classic piece of design - with H&K voicing tweaks.  Those tweaks are significant and can't be modelled in anything like the Duncan tonestack calculator, good though that is.  Spice gives a much better, more accurate if not so easy to manipulate set of results.  And once you have it set up in there you can get a picture of where the tonestack can benefit from being modded.  It will take a bit more work but I will have a comprehensive set of pictures of the curves for the various settings available soon.  Unfortunately there are a host of htem as the tonestack is changed for the different voices.  I'll post them here of course.

Anyway, whether this would maybe convince you to mod the GM36 is another thing.  I'm hoping that I can get it to sound a bit less extended in the upper end, with a little less upper mid and more lower mid, and with a bit more weight at the bottom without becoming flubby.  This may interfere with the modern high gain sound that the amp is set to give but that isn't an issue for me.  It's time to break out the soldering station and get to work.  I'm not going to do anything that would not be reversible, and I'm not planning on touching much of the valve circuitry other than a few really simple things, and the digital and midi circuitry is staying exactly as is.  But I'm convinced there is potential to sort this amp out or at least improve things vastly.

If you can stay on the path for a while maybe I will have something to offer to help your own 'doubts'.  But you're right, that Marshall JVM205H looks very tempting.  And they are going for about the same price as my TM36 would fetch!!!!!

And then I played the GM tonight in my bedroom real low while the wife was on the phone and was reminded why I have this amp.... plus the 18 lb head concept when I'm loading out at a gig.

A guy sat in with us last week with a Divided By 13 2x12 combo - lot's of low end thump - and a little muddy actually.  He was pretty proud of his very warm tone when he set up, but was confused when his solos couldn't cut thru a (by this time) 9 piece band.  Point being...

I realized my GM doesn't have a good thump.  I'm not helping by using a 1x12 cab.  But I just recently realized that is what the Resonance knob is for.  So on all my lead patches, I now petty much have it cranked.  Again, some of that might be alleviated by a 4x12, but I like your idea of a little more kick in the bottom without flub.  My GM is leaving warranty soon, so I'll be interested as you move forward.  I'm not afraid of standard soldering, as long as there's no SMD work involved.

I've grown to accept the clipping diode issue - so none of my patches overdrive, or do so by only 3-6 dB with a clean overdrive.  I might still change the lone-in-that-direction zener to be a little bigger value someday...

Oh yea, Variax - congrats and I'm glad to read about another person who likes it.  My "pedal board" consists of a Line6 HD500x and that's it.  It does great control of the GM, and if/when I end up in a general cover band (or back with my buddy the singer songwriter) where I need lot's of flexibility, a Variax, the HD (which can be highly integrated with the Variax), and the GM would be perfect.  In fact in that case, I'd sell the proposed Marshall and go back to the GM.

Beho-wuld Father, and I have done a thing worthy in thy si-yught.  I have kept my Bu-rother from straying from the Path of Right-e-ousness!  And I say-eth unto Mar-ushall, "Gu-reat though thou art, Let-uh My-uh Pe-uh-ple Go-uh!"  And they-uh releaseth my pe-uh-ple to come for-uth from bondage.  (Ok Father, where's my $25?)     That explains the babbling I often do on here, it's just me speaking in tongues.  


At the moment I really believe that what is holding this amp back is the number of places that distortion is introduced.  In a modern amp this is the sound that is wanted for high gain work, but we are experiencing the fact that we can't remove it.  In a classic amp you have distortion from the preamp valve stages and, well, the power amp valve stages.  That's not too many sources to work with so it doesn't become "distortion of distortion of distortion of...", well I'm sure you get what I mean.  In the GM36 we have passive zener diodes limiting the input voltage, an input buffer/Boost stage which has an assymetric diode clipping network in its feedback, a Gain stage with a pair of mysterious diodes within a complex feedback network which may cause a second stage of solid state distortion under some conditions, the preamp valve stages themselves, an active Volume control stage with a pair of zener diodes in its feedback, a Master Volume stage with a zener diode network in its feedback, the power amp valves in the Phase Splitter and output stages (with no feedback).  Now I am curious as to what levels each of these stages will allow before these mechanisms clip the signal, if they ever do in normal working.  I don't like the idea of distortion gradually being added by such a large a number of stages on top of each other, I would prefer a more classic setup with first the valves overdriving, then maybe a single SS stage kicking in.  All other stages should have protection to prevent internal clipping inside the opamps, but this should be the minimum amount that is necessary to protect without affecting signals.  That may just be what we have of course, the H&K designers are thorough and know their product infinitely better than I do, but I want to prove it to myself otherwise I will want to adjust some of those values to make some of the stages clean.

So I'm putting together a bit of data regarding the way some of the stages behave using Spice simulations.  Spice is not perfect but it's very good at this simple sort of stuff and close to what happens in the real amp as long as you don't try to model too much in one bite.  I'm setting up each stage with all of the factors in place which may affect it, like additional attached passive tone shaping filters, to get as true a picture of how it works as possible.  This will involve a lot of curve plotting and I would like to post as much as I can that is useful.  As this may be a bit dry for most members I'm going to start a new thread just to cover this work and post everything there.  That way this thread can continue on its proper track and those who are not interested can ignore it.

Cool work BB. A have a couple of brainstorm, and we should maybe start a mod thread?

The op amps can handle +/- 18v rails. I should check the whole schematic, but potentially the 7815 & 7915 could be changed  to 7818 &7918. Easy if they are TO-220 package. Then we can make the protection zeners real big.
Kind of unexpected, the el84s are run off a rail that has no pi filter, just a BIG cap. I was wondering if the lack of thump was due to a resistor that could be swapped to a choke -'but no need. I'm kind of surprised the GM36 doesn't him a little....

EDIT: Never mind the +/- 18 - the 4053 can't handle it...