RF Filtering - Troubleshooting Tonal Inconsistencies

[Bunkey]

Ok so,

Having learned a little about how a common emitter amplifier works (great resource here: https://www.pearl-hifi.com/06_Lit_Archive/14_Books_Tech_Papers/Horowitz-Hill/The_Art_of_Electronics.pdf )

I think I can draw some educated conclusions between what I'm hearing and what the circuit is actually doing; and there are a few things going on here:


1) It seems like, as may have been pointed out already, the original Bazz Fuss circuit is inherently unstable. It provides voltage to the base via the diode, to switch it on, which also forms a feedback loop preventing thermal runaway on account of the relationship between collector current and base through this diode network.
Unfortunately as the emitter goes straight to ground (as per the original schematic), or is often used with relatively small resistors <7k to reduce the AC signal gain, the actual DC bias point is completely dependent on the hFE value of the transistor - which varies wildly with temperature during use - as opposed to being set by the values of the supporting components in the circuit (rather like a voltage divider being dominated by one 'variable' value instead of another fixed value). Thus the bias point fluctuates and sounds great some of the time, discordant and one-dimensional at others. There is an example of a 'bad' emitter-grounded amplifier exactly like this on page 83 of the material linked up top.

By increasing the collector resistor value to match the transistor used as I did previously, I brought this point further from the 'undesirable zone' so the circuit sounded better more of the time with that particular transistor - but it was still unstable in the same way and still tips into sounding ugly when the hFE value goes beyond a range I consider musically acceptible, though not as much as it did before.

I seem to have solved this problem by greatly increasing the emitter resistor value and bypassing it so that the circuit gain is still as high as it was for the AC signal, but now the DC bias point is a product of the relationship between the emitter and collector resistors more than it is a product of transistor hFE. The bias point has also been raised by increasing the value of Re and adjusting Rc accordingly. I'm currently at a 2:1 Rc:Re ratio, which is what the DC bias sees, and the circuit seems much more stable - notes are much better defined whilst still having the same fuzz, which is feeling more touch sensitive now (as far as the issue presented in point 4 below permits) and it's also helped attenuate the very peaky note attack and smooth things out a bit. The values I'm using are still a work in progress.

That's the inconsistency problem solved.


2) Another problem I'm noticing, which might not be so apparent in the clips I've posted, is the lower notes seem to be losing amplitude when struck hard. By measuring the voltage difference between the collector and emitter whilst playing I was able to observe the value dipping below 200mV during these instances and I think the transistor C-E junction is beginning to go into saturation at this point: grounding the collector somewhat and therefore attenuating the output signal.

I need to raise the voltage difference between C and E. The value sits at 500-and-something mV when there's no signal, regardless of other parameters, so I imagine this voltage between C and E (the Vce, hurrah!) is a product of the voltage drop across the diode? This might require a bit of thinking considering I don't really want to change the diode arrangement...

.. Which brings me onto point:


3) Something I like about this fuzz - its 'phaser' undertones and 'harmonic percolator' effect - I think might be a product of the difference in phase between the signal generated through the transistor and the signal bleeding through the diode, as they combine at the collector. That's maybe why the diode choice is a key factor in the sound and simply swapping out the Ge diode for a higher Vf Si - replicating the Ge-sound with the miller-effect of a larger C-B cap value - is not a viable option here, as the Si diode doesn't allow as much 'dry' signal to bleed through to collector and create the phaser effect.

Pure conjecture at this stage, but educated conjecture nonetheless.


4) Finally, the gating issue is still present (albeit a lot less of a problem now other areas have been addressed). I realise now how this is the voltage difference between base and emitter being < ideal to put the transistor into it's 'on' state and it still requires a bit of oomf from the incoming signal to tip it into operation. I appreciate the transistors 'on' voltage is an obvious concept and something explicitly stated everywhere, I even understood this myself - but you don't really know until you know, you know?

This gating could either be because the pickups aren't able to supply the demanding higher freqencies with the extra power the input needs to switch on; or it could be a product - considering it occurs more at the higher end notes - of frequency-dependent gain in the amplifier stage related to dynamic source impedance presented by said frequencies; similar to what Rob was saying earlier about the gain of the circuit being dependent on the source impedance of passive vs. active pickups.

- Or a combination of the two as the neck pickup does fare better than the bridge.


So once I've explored the stable biasing some more (point #1) and come up with some good Rc and Re values, I'll turn my attention to the rest.


This has all become a lot less foggy given a bit of studying and practical exploration 

I almost understand what people here are saying now 

[PRR]

http://home-wrecker.com/bazz.html

[Bunkey]

http://home-wrecker.com/bazz.html

Good resource, I started here back in July, I recall you were quite helpful with a few things around that time! - Is there a point contained that you're referring to explicitly?


Whilst I have your attention I don't suppose you could explain how I factor the diode into the input impedance calculation, assuming a fixed temperature?

Could I simply measure the resistance of the diode in reverse polarity (at room temp) and begin there - aware the value is subject to change?


I also don't have a datasheet value for the transistor's Vbe ON (only Vbe SAT). Does the 'on' voltage = the forward voltage of the B-E junction when measured as a diode?

Edit: "The active region of transistor is one in which emitter-base junction is forward biased.." - I'm taking this as a 'yes'.

So based on my B-E Vf reading I need to raise the base voltage to >840mV greater than the emitter voltage to cure the gating issue - it's ~540mV right now and I assumed that was on account of the diode Vf - am I misinterpreting something here; is that 540mV the B-E forward voltage 'in circuit' (as opposed to whatever my DMM tests at) and I need to raise the base current to bring the transistor up to it's linear region instead of worrying about the diode?

Edit 2: I'm just thinking out loud here, I can investigate this for myself. Though I would appreciate some guidance on accounting for the diode when calculating impedance as it's not a very well documented arrangement.


Cheers

[PRR]

> when calculating impedance

Typical pickup levels will swing the diode from ON to OF. I'm not sure "impedance" is a useful concept.

[Rob Strand]

Congratulations you have done really well getting up to speed on this one.

Re diodes:  the diodes do have an effect on the impedance *no doubt about it*.

The difficulty is how to deal with it, especially through calculations.

Intuitively the overall effect depends on how long the diode is on and how hard. If the diode is on 50% of the time (clipping on on polarity) then when the diode is off there is no loading and when the diode is on there is strong loading.     However, that doesn't end-up meaning the overall loading effect which is twice the effective diode impedance.   It's more like you are adding the unloaded signal to the loaded signal; ie averaging the two cases.

A simple example is to add a raw signal to a low-pass filtered signal.  You don't get halve the cut-off frequency.  What you end-up with a shelving EQ which only has a -6dB step.     In other words a lot of highs still get through.

https://wiki.jriver.com/index.php/File:Shelf_positive_vs_negative.png

(here filtered = loaded, clean/raw = unloaded)

[Bunkey]

I'm not sure "impedance" is a useful concept.

It was for determining cap values and to satisfy my own curiosity really

Congratulations you have done really well getting up to speed on this one.

Re diodes:  the diodes do have an effect on the impedance *no doubt about it*.

The difficulty is how to deal with it, especially through calculations.

Thanks man,

I was thinking when the diode is reverse-biased the resistance presented to the base will be the diode's reverse resistance (1.5M as measured); then when the diode is forward biased it presents an even greater resistance to the base, considering it's orientation? (Sorry, the loading terminology is lost on me)

therefore;

So long as hfe x re (presented to the AC signal) is much less than the diode reverse resistance (by an order of 10 +), the total input impedance seen by the guitar is going to be largely determined by hfe x re anyway; and like you did earlier - the diode can be discounted in the equation and still give a reasonable ballpark figure when it comes to impedance matching and working out audible corner frequencies etc?

[Rob Strand]

I was thinking when the diode is reverse-biased the resistance presented to the base will be the diode's reverse resistance (1.5M as measured); then when the diode is forward biased it presents an even greater resistance to the base, considering it's orientation? (Sorry, the loading terminology is lost on me)

therefore;

So long as hfe x Re (presented to the AC signal) is much less than the diode reverse resistance, the total input impedance seen by the guitar is going to be largely determined by hfe x Re anyway; and like you did earlier - the diode can be discounted in the equation and still give a reasonable ballpark figure when it comes to impedance matching and working out audible corner frequencies etc?

For the reverse-biased diode what you have said is correct.

For the forward biased case the diode impedance is quite low.  If you look up dynamic resistance or diode resistance  in Art of Electronics you will find something for sure.   The diode resistance get lowers as you put more current down it.  Think of the diode characteristic past the knee point, when the current goes up the voltage across the diode can only change a little bit.  A large change in current with a small change in voltage is the same as a small resistance.  The other thing to note is the diode resistance is not really like a resistor since the resistance changes with current.

The fine details of the circuit are tricky to understand.   When it's' just sitting there and the diode is forward biased if Rc = 470k there will be roughly 17uA going down the collector, if the gain is 200 then the base current will be  0.085uA.  The base current and the diode current are the same so the diode current is Id = 0.085uA.  If you calculate the diode resistance rd = 1.9 * 26 mV / Id = 580k ohm.  The 1.9 is a rough value for the diodes "ideality factor"; see if it is in Art of Electronics. 

When you connect a resistor across the input and output of an inverting amplifier it can reduce the impedance.  So we would expect some reduction in the real input impedance compared to calculated value for the diode;  see Miller Effect in Art of Electronics.   However  because the diode clips the output, the amplifier gain is reduced and Miller effect is probably a lot less than just assuming a high gain amplifier.  Off hand in the input impedance is probably in the 100k to 500k zone.  So basically your high collector resistance, which raises the input impedance without the diode, also raises the impedance with diode.

Don't worry if you get a bit lost here.    The thing is you have little control over any of this.   When you raised the collector resistor it fixed everything and that's the main thing you have control of to set the impedance.   Noticed you worked that out without  going though any long winded arguments or calculations.  A lot of times all you need is a rough idea of cause and effect then just try it!
     

[Bunkey]

For the forward biased case the diode impedance is quite low.  If you look up dynamic resistance or diode resistance  in Art of Electronics you will find something for sure.   The diode resistance get lowers as you put more current down it.  Think of the diode characteristic past the knee point, when the current goes up the voltage across the diode can only change a little bit.  A large change in current with a small change in voltage is the same as a small resistance.  The other thing to note is the diode resistance is not really like a resistor since the resistance changes with current.

Ah that makes sense now thinking of it in terms of ohms law 

The fine details of the circuit are tricky to understand.   When it's' just sitting there and the diode is forward biased... 

Yeah I measured the Vcb earlier and was pretty stumped to see a forward bias of ~20mV at idle and a reverse bias of ~200mV with input applied - It totally threw my perception of how the current is flowing!

Gonna have to pick up a hard copy of that and take a look at the missing first chapter. As intuitive as things seem, I think there's a lot of fundamental stuff I don't really have a knowledge of which is making these heavier concepts difficult to tie together - It's like having a conversation in pidgeon english 

I really appreciate you taking the time to explain this stuff.

[Bunkey]

As an aside to this fuzz pedal; the linear boost is coming along alright -
I was just going to throw something together to test a workaround for the fuzz gating issue but it's got pretty ridiculous:





It's just a straight up LBP-1 clone - The circuit is going to hang from the jacks.

I've swapped the emitter resistor for a trimpot and I'm adding the option of diode clipping on the output using these locking PCB interconnects pulled from a butchered Marantz.

This way I can use it as a clean boost or to audition different diodes for future builds; adjusting the gain with Re to clip whatever's loaded in the sockets appropriately.
The socket is wired with a jumper so it'll take up to 3 diodes in symmetrical or asymmetrical formation - They're just running to ground on the output, between the cap and volume pot, which I hope is an acceptible way of doing things?

- No idea how it's gonna sound but it seems like a nice idea at least.

Inspired by Midwayfair's Listening Notes on Diodes: http://music.codydeschenes.com/?p=1615
... and my growing collection of Ge's. These things are like sweets 

[Bunkey]

Completely derailling the topic here but I powered through the night to get the boost finished...



Man, building this thing was ridiculous.

I can't believe how good it sounds with the right diode combination though! It wasn't even supposed to be a proper pedal 




I'll get back on task and put some progress clips together soon - I guess this has turned into a loose Q&A development blog now.

I've decided I'm going to split the pickup on the ESP next so I can do some proper single coil testing of the fuzz - It's destined for a friend's pedalboard and I don't fancy sending him a paperweight...

[Rob Strand]

Man, building this thing was ridiculous.

Very cool, I like it.   The LED and the label adds to the charm.

[Bunkey]

Back to the fuzz..

Still not tried it with a split coil yet as I'm in the middle of taking on a new property with a bit more space to work; I did manage to pick up the LP earlier and record this clip though, now I've addressed the points in the last update.

The amp wasn't quite as loud this time.

It really is something else when you hear it in person - there's so much musical saturation and harmonic content that I don't have the production talent to capture - I'm really rather pleased with it.
Provided it works with SC's I'm settled on this version.

Once I've relocated I'll try splitting the pickup in the ESP for a final test and post up the finished piece here. Exciting times!

[Bunkey]

Ooft I dunno actually - Listening to that again today it sounds a bit flat like there's a resonance to the fuzz that's not quite in tune with the notes I'm playing. It sounds a bit like when you go from 440hz to 432hz tuning, before your brain has got used to the change.

Is anyone else hearing that?

I'm not sure if it's the guitar, a product of the compression applied to the video or from the pedal itself . Maybe the harmonic content isn't sitting right with my disposition today

I'm wondering if I just get excited about progress I've made and end up hearing what I think I hear instead of what's actually there 

It would be great if someone else who isn't tone deaf could give me their opinion?

[Bunkey]

It seems I may be conversing with myself here but hopefully someone might find this of interest.


I received a parcel last week that I'd been eagerly awaiting for a while...



Here we have a nice selection of D9- germanium diodes!


I've been auditioning them in the linear boost clipping jar as asymmetric pairs.

What's interesting is their unique harmonic signatures; some of these are concordant and some are discordant. The popular types - E's, K's, J's etc - seem to be popular for a reason, they do sound pleasant. Then there are types like the D9B which in the arrangement I'm using sound a bit jarring.

Some of them have a dominant harmonic resonance above the fundamental and some of them have a harmonic resonance below - similar to what I queried in my last post about the fuzz.

Even individual diodes of the same type each sound slightly unique.

It seems the issue was a characteristic of the diodes I was using - Things get a little strange when you start to combine them, the resulting sound is a product of their signatures (as well as being more compressed) - if one drifts a little more than the other this product alters, think of it like a combination of phase shift though I guess that's not strictly correct.

It's not a simple cut-and-paste affair either because the combination which at times sounded discordant in the fuzz (as a feedback network) actually sounds very nice in the boost (as a clipping diode to ground). Likewise, the D9B mentioned above might sound unpleasant as I tested it in the boost but that's not to say it won't work in a different circuit.

This is the circuit I've been using to audition the diodes, the one I built into the chutney jar:




...and here are my obvservations:

321/285 D9E: Musicality 5/5 Brightness 5/5 Clarity 5/5 Dynamic 3/5- Awesome mid harmonics & distortion sig.
366/324 D9V: Musicality 5/5 Brightness 2/5 Clarity 5/5 Dynamic 4/5 - Dark with subtle warm harmonics. Dynamic dist.
334/306 D9D: Musicality 3/5 Brightness 5/5 Clarity 4/5 Dynamic 4/5 - Quite brash & abrasive
306/294 D9B: Musicality 1/5 Brightness 5/5 Clarity 4/5 Dynamic 2/5
322/309 D9G: Musicality 4/5 Brightness 4/5 Clarity 4/5 Dynamic 2/5 - Smooth inoffensive distortion.
329/298 D9L: Musicality 3/5 Brightness 4/5 Clarity 4/5 Dynamic 3/5 - Straightforward Neutral distortion
295/269 D9k: Musicality 4/5 Brightness 4/5 Clarity 5/5 Dynamic 5/5 - Dynamic & characterful Harmonics
356/324 D9J: Musicality 5/5 Brightness 3/5 Clarity 4/5 Dynamic 4/5 - Very smooth. Good twang that you can really lay into.

mV @ Temp = 10°C

Note: The scores are an indication of 'least to most' but not necessarily what is better.

- 'Musicality' relates to concordance of the harmonic content
- 'Dynamic' refers to useful gradation of the distortion or harmonic content between soft and hard picking; some diodes cleaned up well but were a bit on/off without much nuance (or some just clipped all the time), so these would score low.
- 'Clarity' refers to how much playing detail is being communicated, though this is not to be mistaken for;
- 'Brightness' which can be synonymous in this case with presence


The go-to 3, based on how long I spent mindlessly riffing away on each of them out of the box, seem to be the E, J & K types; they were the most attention grabbing with their presence & musical character especially when soloing at the top end and ringing out single notes; the E's & K's share a kind of sparkle that's pretty damn magical. The V & G types are also worth mentioning: The V's in particular have been toted as 'a D9E which breaks up sooner' but I actually found the V to be less granularly distorted than the E and much darker - It lacked the same clarity but had a pleasant loungy warmth and seems like a good option where others are too bright or abrasive, it's definitely better in some applications and whilst it doesn't have the presence or top end of the others I find I keep going back it. The G had an especially smooth distortion which could be an attractive asset in it's own right.  The D & L types were also useable but they just didn't have as much charm, which may be a good thing depending where they're used; the L was particularly neutral sounding with no strong harmonics so it might be useful where you need the clipping but don't want to add too much colour to the rest of the circuit.

A simplified way of grouping these:

E, B & D: Bright Pop Rocky
G, V & J: Smooth
E, K & J: Psychadelic
G & L: Neutral

Having never back-to-backed diodes before I was surprised how similar they all sounded out of the amp - I suppose this demonstrates the overall tone is more a function of the complete electrical circuit than of the diodes used, so a diode itself isn't going to completely alter the sound - however depending on where & how they're used they will add their own distinct flavour which could make or break the effect. It's like the icing on the cake.



Anyways, I've found the sound.

A similar transistor with a lower Vbe-on helped a great deal too.
I feel I'm ready to assemble a 'finished' pedal for 3rd-party evaluation, so I'll call time on this thread.



Thank you all for your help and support,

Ben

[Rob Strand]

It seems I may be conversing with myself here but hopefully someone might find this of interest.

People do read this stuff!!   It's a matter of not being able to add anything useful.

[Slowpoke101]

I will agree with Rob here. We all do read what people post to the forum but sometimes we don't have anything relevant to add. Your journey and your documentation has been fascinating. You have added to my knowledge and for that I thank you, but still I don't have anything relevant to add yet.

[Bunkey]

Its funny, I always feel like I know nothing and everyone around me has a far better understanding than I do.

..so I try to learn and I document the process; and I suppose in doing so I have become the tutor.


I guess we're all on the same page really.

[PRR]

I guess we're all on the same page really.

Same book, different pages.

Just cuz nobody else is currently on "your" page don't mean we are ignoring you.

[Bunkey]

Thanks guys.

It was just a passing quip poking fun at myself for replying to my own thread so much. I'm not on the brink of a crisis or anything  I do appreciate the comments though.

[duck_arse]

Bunkey! excellent fellow - you are a hard one to track down. for in case it happens again, I'll include my search terms .....

it's like an lpb or an electra or a simple booster, built into a glass jar, with a metal screwtop lid.



do you mind if I possibly, sometime, shamelessly rip-off your idea? for to maybe, one day, build a modular many oscillator noise maker?