Author Topic: RF Filtering - Troubleshooting Tonal Inconsistencies  (Read 3134 times)

Bunkey

RF Filtering - Troubleshooting Tonal Inconsistencies
« on: January 12, 2020, 11:27:53 AM »
Hello all,

Playing around with a modified bazz fuss circuit (Ge diodes, Si transistor, Battery power) - I'm noticing an inconsistency in how it performs:

Sometimes it's very musical, sounds really sweet and sustains decently with a nice sparkly roll off.

Other times it's a bit dischordant, slightly more gated, there's less musical content in the guitar's highs and it's altogether not so pleasing. Kind of like having a stereo jack not quite plugged into a piece of equipment properly.


The difference is subtle yet it makes all the difference between sounding good and sounding bad.

I'm tempted to say it gets better the longer I play through it but this really is inconclusive - it could well just be coincidental at the times I've noticed a difference.

I don't think its relative to temperature either, I'm aware of the effect this has on Ge diodes.


Everything is solid; I have considered a dry joint but it's also worth noting that the circuit itself is free-formed using solid copper wire not encased in a typical metal enclosure and there is no additional filtering on the circuit - I'm wondering if it's acting like an antenna and this has something to do with interference eating into the signal?

Given the lack of shielding is it worth adding a small cap to ground on both the input and output sides of the circuit or would one at the input suffice?

Any other thoughts on this?

Thanks for your time.



« Last Edit: January 12, 2020, 11:59:10 AM by Bunkey »

iainpunk

Re: RF Filtering
« Reply #1 on: January 12, 2020, 11:47:00 AM »
hve you tried a really large resistor in parallel with the diode, something like 1M or more? it solved my Bazz Fuss problems, which sound similar

Bunkey

Re: RF Filtering
« Reply #2 on: January 12, 2020, 11:58:11 AM »
hve you tried a really large resistor in parallel with the diode, something like 1M or more? it solved my Bazz Fuss problems, which sound similar

I haven't but someone did once recommend replacing the diode altogether with a 1M resistor for a 'classic overdriven bjt amplifier sound'.

I'll give it a go and see if it retains the Ge harmonics I like - Still not sure why there'd be an inconsistency though unless the diodes were on their way out?


Maybe the problem lies in the practically incalculable realm of things such as parasitic capacitances, microphonics and electromagnetic fields?

Maybe the planets just didn't align at the time?  :icon_lol:


Thanks for your suggestion :)
« Last Edit: January 12, 2020, 12:11:06 PM by Bunkey »

GibsonGM

  • more
  • Awesome!
  • *****
  • Posts: 5726
  • Total likes: 875
  • Mike Parker (aka. Guitar Mike, Parks, etc)
Re: RF Filtering - Troubleshooting Tonal Inconsistencies
« Reply #3 on: January 12, 2020, 12:41:28 PM »
Ge diode isn't giving the (still low) level of bias the original 1N914 does.  So, you're in this crappy non-linear region of the transistor operating range.  Leads to that inconsistency...my 2 cents.  It's a very unpredictable part of the operating curve.

Bunkey

Re: RF Filtering - Troubleshooting Tonal Inconsistencies
« Reply #4 on: January 12, 2020, 12:50:53 PM »
Ge diode isn't giving the (still low) level of bias the original 1N914 does.  So, you're in this crappy non-linear region of the transistor operating range.  Leads to that inconsistency...my 2 cents.  It's a very unpredictable part of the operating curve.

Thanks. Contrary to the schematic I have a number of diodes here.

The total combined forward voltage is measuring as 640mV - This is actually a bit more than the individual voltages added up on paper for some reason but I guess that's a whole other can of worms.

Here's the datasheet for the transistor: https://www.onsemi.com/pub/Collateral/KSD471A-D.pdf

You're probably onto something but I'm probably not the guy who's gonna be able to confirm it  :icon_lol:

So why then does it still sound great some of the time? Is it a matter of unpredictable external factors tipping some part of the circuits balance one way or the other?


As you may have gathered - I am not the electrical scholar - I just try to apply lateral thought and common sense!
« Last Edit: January 12, 2020, 06:41:02 PM by Bunkey »

Bunkey

Re: RF Filtering - Troubleshooting Tonal Inconsistencies
« Reply #5 on: January 12, 2020, 05:26:48 PM »
Well I just tried a 1M resistor in parallel with the diode..

I can see how it might eleviate some of the more obvious biasing issues people have with the traditional bazz fuss circuit.
In my case it did produce a more consistent & slightly less gated sort of sound whilst being the same pedal but it also lost a lot of the sparkle and harmonic content that made this particular pedal sound so good before.

It was a bit flat by comparison tbh and I felt like I just wanted more fuzz than it was giving me to compensate for the fact.

I think the problem I have and it's solution lies elsewhere..


I don't seem to recall having this issue with the buffered signal from active pickups when I had them, so it could well be a 'transistor-on' problem but re-biasing the circuit by means of a C-B resistor isn't the way I need to go. Like I say, it's an intermittent thing and maybe the buffered signal just concealed whatever the problem is as opposed to addressing it directly.

Thinking on it now, it sounds audibly similar to an issue I have when my washing machine is on spin in the other room and the electronic noise through the poor circuitry in this house eats into my amplifier's ability to generate or reproduce harmonic content with the increased noise floor.

I'm going to try some RF filtering as originally intended and see how I get on with that - If someone could advise whether it's worth putting this both before AND after the gain stage as I explained in my OP, that would be handy.

Thanks for the input though guys, there's always benefit to be had from exploring ideas.
« Last Edit: January 12, 2020, 06:53:16 PM by Bunkey »

amptramp

Re: RF Filtering - Troubleshooting Tonal Inconsistencies
« Reply #6 on: January 12, 2020, 10:02:40 PM »
Your biasing as it stands in the schematic you have shown is totally through transistor leakage and diode leakage, neither of which are controlled parameters.  It was common in some early transistor radios to bias the base through one resistor to the rail, another to ground and a reverse-biased diode to ground with the idea that as the transistor leakage from collector to base increased with temperature, the diode leakage also increased to shunt some of this extra collector leakage and maintain a reasonably constant bias.

There is nothing going from base to ground to stabilize the bias point, so you can expect that the circuit will change behaviour with junction temperature.

Since the circuit has the high output impedance of the collector going to the relatively low collector resistor, there is almost no power supply rejection other than the voltage division through the output capacitor and level control.  Adding a resistor to the base from the collector will reduce the input impedance through the Miller effect which multiplies the current required from the input signal and this in series with the inductance of the pickup(s) reduces the high frequencies.

Rob Strand

Re: RF Filtering - Troubleshooting Tonal Inconsistencies
« Reply #7 on: January 12, 2020, 11:23:14 PM »
I think you should try replacing the 2x Ge diodes with a 1x Si diode just to see if the inconsistencies go away.
1x Ge + 1xSi might work but I'd do that experiment after the 1xSi.

*In circuit* the Ge diodes might have a smaller voltage drop than when you measure them with a DMM.

One weakish theory for loss of highs is the input impedance is low.  Perhaps promoted by the diode's low impedance and feedback lowering the impedance by the Miller effect.  At face value you would expect the Si and Ge to be similar in this respect by perhaps the Ge diodes are on more often and that lowers the impedance.

Another thing you can do is accurately measure the collector voltage when it sounds good and when it sounds bad.  Keep a running list of measurements and you might see a correlation between bad sound and voltage.  The difference is going to be small so you need to measure carefully and use the same physical measurement points each time.
   

Bunkey

Re: RF Filtering - Troubleshooting Tonal Inconsistencies
« Reply #8 on: January 13, 2020, 08:20:41 AM »
Trying to keep my head above water whilst maintaining the impression I can swim  :icon_lol:


I had an early prototype with an Si diode and I've tried all kinds of combinations with the two before arriving at this point but it was a completely different pedal as a result. You're right though, I didn't have the problems with the Si.

The Ge's came direct from russia; ex-military stock, corroded leads and all; so yes based on what I can gather from your posts I think this might just be a product of intermittent reverse leakage through the old diodes causing more of a miller effect than I intended when I added the cap in parallel.

In a similar sense that'll be why I thought the C-B resistor sounded flat too.


I guess it's just the nature of the beast then?

This configuration still sounds killer compared to everything else I've tried - Yeah I'm talking it up but I'm proud of it haha. No shame.

For now I'll chalk it down as 'character'  :icon_mrgreen:


Amptramp, I feel there's a solution in the information you posted so I'm gonna study those concepts and mull that one over.

As an aside, could you guys suggest a good way to determine the reverse leakage in the remaining diode stock I have?
I'm putting a few of these pedals together so obviously I'd like to strive for better consistency where I can.


Thank you all so much for sharing your thoughts on this - Very humbling.
« Last Edit: January 13, 2020, 08:49:37 AM by Bunkey »

Bunkey

Re: RF Filtering - Troubleshooting Tonal Inconsistencies
« Reply #9 on: January 13, 2020, 11:02:42 PM »
A final update on this.

I added an RC low pass filter with a corner frequency of 20khz between the input cap and transistor and it's sounding a little more defined than it did - there was just a bit too much low-order harmonics before and the sound got a bit muddied at times which I think is what I was first referring to in my OP.

I imagine though the improvement is more to do with the effect of raising the input impedence than it is removing interference.

I think to summarise, there are a lot of things going on here and I don't really have the understanding of circuit theory to draw valid conclusions on paper, I just know what sounds good and fumble through it until I reach a desired outcome.


I've uploaded a wee clip here if anyone is interested in hearing this circuit - This is my working prototype that's under constant revision, cobbled together with inferior salvaged parts.
The difference between this and the production version is like removing a blanket from the amp... but I'm not quite ready to reveal the finished article yet  :icon_wink:



Take it easy  :icon_cool:

Rob Strand

Re: RF Filtering - Troubleshooting Tonal Inconsistencies
« Reply #10 on: January 14, 2020, 05:00:17 PM »
Quote
The Ge's came direct from russia; ex-military stock, corroded leads and all; so yes based on what I can gather from your posts I think this might just be a product of intermittent reverse leakage through the old diodes causing more of a miller effect than I intended when I added the cap in parallel.

In a similar sense that'll be why I thought the C-B resistor sounded flat too.
It's not likely to be just leakage as the diode is forward biased but what I suspect is happening is the base current through the transistor is small and this causes the diodes to operate at low current.   Germaniums have a quite a low voltage drop at low currents.   The higher "leakage" diodes are likely to have a lower drop when forward biased.  It is possible that is both contributing to the sound you like and perhaps to the instability of the behaviour.

The voltage across silicon and germanium diodes varies with temperature by roughly the same amount, approx -2mV / deg C.  You have two ge diodes so that will effectively double.  Normally that alone is unlikely to be the cause of your problem.  What is possible is the germaniums are running at a low enough current to make this effect worse.

So going down that path of thinking perhaps something to try is to add a 1M to 10M resistors across the Base and Emitter of the transistor.   The loading of this resistor is unlikely to cause any issues and there is no Miller effect.   What it does is puts a minimum current through the diode.   I strongly recommend playing with the value.  In fact of 10M has a better sound than 1M but still isn't quite there then you might need to try 20M.     No guarantees but worth trying.

Quote
I guess it's just the nature of the beast then?

This configuration still sounds killer compared to everything else I've tried - Yeah I'm talking it up but I'm proud of it haha. No shame.

Possible but we don't know *what* aspect of the design makes it sound good and what aspect make it drift.   If you know the answer to those questions you can say it does have an inherent problem.  The question then is can you side-step the issue by changing something.   For example the mod I mentioned.   A good motivation for working through the problem is it sounds good.

It's worth trying a few things to see what affect is has.  I'll admit trying to theorize is too open ended without  understanding both of those aspects first, or at least the first one.   I'd probably measure the collector-emitter voltage when it sounded good and when it sounded bad, then removed the diodes and use resistors to bias the transistor to exactly that collector-emitter voltage.  From that you know if the bias point is important to the sound.  Then I'd play with bias point to see how close I need to stay to those voltages, then finally try to come-up with a scheme to make it stable.
« Last Edit: January 14, 2020, 08:05:48 PM by Rob Strand »

Rob Strand

Re: RF Filtering - Troubleshooting Tonal Inconsistencies
« Reply #11 on: January 14, 2020, 07:14:11 PM »
I finally got to listen to the clip.  It reminds me of a Harmonic Percolator a bit.  The Harmonic percolator biases one the transistors with a low collector-emitter voltage so maybe the bias point is what you are seeking.  (Maybe I'm on the right track with the biasing and diode current thing.)


pinkjimiphoton

Re: RF Filtering - Troubleshooting Tonal Inconsistencies
« Reply #12 on: January 15, 2020, 06:56:09 PM »
try adding an led in series with the ge diode. or if ya have one, a 4739 like i used in the stupid simple overdrive. the problem you're having i think is from the way the ge diodes react to temp/humidity/the moon/the tides etc... i know they shouldn't change due to that stuff, but man, it sure seems to me NOTHING germanium EVER SOUNDS exactly the same ;)

Bunkey

Re: RF Filtering - Troubleshooting Tonal Inconsistencies
« Reply #13 on: January 20, 2020, 10:36:25 PM »
Alrighty,

I took a few days away from this to rebuild my guitar and regain my sanity  :icon_mrgreen:




Now I'm back on it I've been playing around with a few of your suggestions..

I went back to trying some Si + Ge combos as well as adding E-B resistance but all these things took the sound further from what I like about it.
Basically I've found the bias point is exactly where I want it to be - the flare and the 'Harmonic Percolator' effect you mention is what I'm into.

The main issue I'm having here is that sometimes I have to pick the higher single notes pretty damn hard just to get it to 'switch on'. Other times they ring out even when playing softly (to a point), which is how I want it - it's still not great but it's playable.

- I've a feeling though that you're going to tell me that's down to the bias :icon_lol:


Measuring the collector voltage (to ground), it's hovering around 600mV - dropping to ~500mV when playing the thin E and ~400mV when playing the fat E. Strumming drops it down to about 300mV.
With the Si combo it hovered around 900mV, dropping by the same amount to ~800, ~700 and ~600mV but I still had the issue of having to pick the higher notes hard to switch it on - so no change there.

The base seems to hover around 540mV regardless.

I don't really know what these figures mean tbh but they seem pretty consistent, the collector voltage drifts a few tens at most, its sitting at 580mV just now and it's sounding good.


The story goes something like: Actives work fine and I can even roll the volume right back to clean it up with really nice sparkly overtones; passive humbuckers at full volume sound good most of the time but give me the symptoms I'm trying to troubleshoot here, rolling the volume back just cuts the sound out; and single coils simply don't work with the pedal - You have to play so hard just to 'switch it on' and there is no sustain whatsoever.


So.. is there a way to get this to work well with the lower output from passives and single coils without changing the bias?

The only things I haven't played with are the coupling cap values and the transistor.
« Last Edit: January 21, 2020, 08:45:44 AM by Bunkey »

Rob Strand

Re: RF Filtering - Troubleshooting Tonal Inconsistencies
« Reply #14 on: January 22, 2020, 05:25:26 PM »
Quote
- I've a feeling though that you're going to tell me that's down to the bias :icon_lol:
Well it's definitely a factor.

Quote
The story goes something like: Actives work fine and I can even roll the volume right back to clean it up with really nice sparkly overtones; passive humbuckers at full volume sound good most of the time but give me the symptoms I'm trying to troubleshoot here, rolling the volume back just cuts the sound out; and single coils simply don't work with the pedal - You have to play so hard just to 'switch it on' and there is no sustain whatsoever.

It's not clear to me if have ever got it to sound good with single coils?  (and if yes, then the problem is it no longer sounds good on Actives)

Quote
So.. is there a way to get this to work well with the lower output from passives and single coils without changing the bias?

From what I can see you have the biasing set such that it is essentially cut-off and it needs the extra output of the actives to kick it off.  What I don't understand is if that is the case I'd expect the notes to be gated off in the tail of the note even with actives.  At some point the signal level has to drop below that of the single coils.

This gating issue is common when trying to convert some of the old GE transistors circuits to Silicon transistors.   The GE transistors reduce that gating effect.   Someone on this group did a silicon version of one of the earlier Ge Tone Benders; no doubt they spent a lot of time on it and I think they concluded if you have GE transistors then you should use them.

Anyway a different plan of attack would be to put a gain state before your circuit.  That way you can dial the gain back on actives and boost it back-up for single coils.     As for preserving the sound I cannot guarantee it. 

The problem with finicky circuits is they need a lot of experimenting and development to get them to work.  You have to do a lot of work and perhaps be prepared to fail,  or not quite get there.
« Last Edit: January 23, 2020, 03:55:07 PM by Rob Strand »

Bunkey

Re: RF Filtering - Troubleshooting Tonal Inconsistencies
« Reply #15 on: January 23, 2020, 12:01:43 PM »
Quote
It's not clear to be if have ever got it to sound good with single coils?  (and if yes, then the problem is it no longer sounds good on Actives)

I've never had it working with single coils; I just tried it once with a Squire in a music shop and it was terrible.
I don't use SC's myself but a friend of mine does so I'm trying to widen the scope of the pedal to accommodate them.

I don't have a single-coil guitar to test with either but if I can get it working good with my passive HB's I'd consider that a happy medium.

Quote
From what I can see you have the biasing set such that it is essentially cut-off and it needs the extra output of the actives to kick it off.  What I don't understand is if that is the case I'd expect the notes to be gated off in the tail of the note even with actives.  At some point the signal level has to drop below that of the single coils.

The tail end does gate off with actives and it does it in a really nice way that I can only describe as the audio equivalent of a star-filter in photography  :icon_lol:
The difference is the sustain I get before reaching that point, and how long that state of sparkly decay lasts when I do reach it. It's pretty much non-existant with the low output pickups, it just cuts.

Having done a bit of research it seems this sustain issue is quite well documented but it's solution is not.
A few have mentioned the collector resistance is key so I'm going to swap this out for a trimpot and see if I can strike a balance.

If that doesn't work out I may have to look at a simple way of boosting the input like you say but I'm not sure how to do this without introducing more clipping - I guess that'll be my next question.
The preamp schematic for the EMG 81 I was using before the passive H3 might be a good place to start but the layout of the final pedal is really tight and well thought out already - it's not subject to change - everything fits like a puzzle so I'm reluctant to start tagging on more circuit at the input side.
« Last Edit: January 23, 2020, 01:28:17 PM by Bunkey »

Rob Strand

Re: RF Filtering - Troubleshooting Tonal Inconsistencies
« Reply #16 on: January 23, 2020, 04:05:19 PM »
Quote
I don't have a single-coil guitar to test with either but if I can get it working good with my passive HB's I'd consider that a happy medium.

One possible difference between passive and actives is the output impedance of the guitar.  Some actives have low output impedance at all volume settings and some don't.   Passives pretty much have high output impedance all the time.

Where I'm coming from is:  that circuit has a gain which is very much dependent on the impedance of the guitar.  With that circuit, the low impedance actives will have a much higher gain than the passive.   

I don't know if impedance is the root cause of your problem but it is something to investigate.     If you have a Boss pedal or some pedal which is *not* true bypass you can wire it before your distortion pedal and set it to bypass.  That will buffer the signal before the distortion pedal.   If it improves things then you can build a buffer into the pedal.  You could also add a series resistor between the buffer output and input of the pedal and tweak the value for best sound.

Bunkey

Re: RF Filtering - Troubleshooting Tonal Inconsistencies
« Reply #17 on: January 24, 2020, 02:22:49 AM »
Breakthrough!!

The original schematic calls for a 100k collector resistor (10k for a darlington) so I thought the sweet spot must be somewhere around the 100k mark..

I replaced the resistor with a 500k guitar pot, started at around 90k and made my way up.
I spent a lot of time in the 100k - 120k region working in small increments and it'd sound alright one minute and not so great the next.

Eventually I got to about 180k when I realised the top strings were much easier to play and picking lightly I could get into that sparkly decay sound instead of just gating out.
I worked right up to 490k and here I can even roll the guitar volume back a bit to clean up the sound like I could with actives.

It still has very much the same character as it did so that's great!


Looks like the circuit is just very particular about matching collector resistance to the transistor used -  the KSD471 has lower gain than a 2n5088 - I was surprised this difference put it so far off the original schematic.

There looks to be a correlation between resistor value and transistor gain here:

MPSA13 ~10,000hFE = 10k ohm
2n5088 ~900hFE = 100k ohm

KSD471 290hFE = ~500k ohm


Maybe as you say Rob, the impedence of the actives = higher circuit gain so I didn't need as much resistance for it to function even with the low-gain transistor.


The amp sounds like it just wants to get cranked now  :icon_mrgreen:

I'll try a 1M collector resistor tomorrow to see if that's any further improvement
Maybe record another clip too.

Thanks for keeping this thread alive! Perseverance pays off.

Rob Strand

Re: RF Filtering - Troubleshooting Tonal Inconsistencies
« Reply #18 on: January 24, 2020, 11:29:05 PM »
Quote
Breakthrough!!
...

It still has very much the same character as it did so that's great!
Awesome work!   Extremely cool result.


Quote
Looks like the circuit is just very particular about matching collector resistance to the transistor used -  the KSD471 has lower gain than a 2n5088 - I was surprised this difference put it so far off the original schematic.

There looks to be a correlation between resistor value and transistor gain here:

MPSA13 ~10,000hFE = 10k ohm
2n5088 ~900hFE = 100k ohm

KSD471 290hFE = ~500k ohm


Maybe as you say Rob, the impedence of the actives = higher circuit gain so I didn't need as much resistance for it to function even with the low-gain transistor.

Increasing the collector resistor causes the input impedance of the transistor to increase.   It's looking like all your results put the input impedance of the transistor in the 250k to 500k.   That's very much the zone where you would expect it to no longer load the pickup, which dulls the sound.   So perhaps there is something in the pickup loading theory.


Quote
I'll try a 1M collector resistor tomorrow to see if that's any further improvement

When the collector resistance beyond 220k you will start to get some overall gain loss from the pedal.   The input impedance of the *following* pedal or amp will start to load it down your pedal.   If you make the collector resistor too high you will effectively get more loading.   So you probably end-up with a happy medium somewhere around what you already have.   This is only a rough analysis as it ignores the effect of the diode in the feedback loop.  Nonetheless the ball-park is certainly making sense from the point of view of pick-up loading.   

Quote
Thanks for keeping this thread alive! Perseverance pays off.
No worries thanks for sticking at it.   That type of result doesn't come easy.

There's an interesting follow-on result from your work.   Maybe the input impedance of the first transistor of the Harmonic Percolator has an impact on the sound.   That would mean tweaking the bias current to ride the pickup loading point like you have done here.
 
« Last Edit: January 25, 2020, 06:15:09 PM by Rob Strand »

Bunkey

Re: RF Filtering - Troubleshooting Tonal Inconsistencies
« Reply #19 on: January 31, 2020, 07:13:19 PM »
Well I've made a few revisions and tried a few ideas..

I'm planning to compile the results after a little more progress and I've wrapped my head around a few concepts but for now here's an update on how it's sounding

This time with a LP Studio  :icon_mrgreen:




Comparing this to the previous clip is like night and day. The Les Paul accounts for a bit more musicality of course (even with dead strings) but the sound is obviously a lot more open and playable. I don't mind the lower gain either - goes well with the vintage Ge vibe imo.

Still a work in progress but it's getting there.


I just want to get the transistor closer to it's linear operating state but I don't want to add anything more than I already have so right now it's a case of understanding how a collector feedback amplifier is biased, accounting for the fixed diode Vf and 9v supply. It's all new to me.

Yayyy number crunching  ::)


I'm putting together a simple linear boost just now too, to see how they co-operate.
« Last Edit: January 31, 2020, 07:22:37 PM by Bunkey »