Blackfire Gain Stage & Tube Theory

[will]

Hi,

I have been pondering on why Joe Davisson’s Bipolar Transistor Blackfire gain stage sounds so good. http://analogalchemy.com/pedals/blackfire.gif and I have an additional theory on how it operates. The 1st transistor Q1 acts as an emitter follower which has very low output impedance on the top half, or positive side, of the wave form of the wave form. However with a 100K emitter resistor it has very high (poor) output impedance on the lower portion of the wave form. The input impedance of the gain transistor Q2 puts a load on the previous stage. The Q1 can faithfully reproduce the top side of the wave but is strained and will roll off some of bottom portion. Soft asymmetrical clipping just like tube circuits perform. This is before Q2 starts amplifying the signal.

Jay Doyle created a circuit that optimized this effect in a Touch of Sweetness. http://www.diystompboxes.com/pedals/schems/Sweetness.pdf  I built that circuit using 2N7000 Mosfet’s last weekend. Mosfet are much easier in this circuit than fussing around with Fet’s. No value change is necessary from the diagram. The circuit is quite trebly so I increased C2 to 1 uF and C4 to 4.7 uF. Also the bias circuits should be bypassed with a decoupling cap to minimize noise. Also Bias 3 needs the cap to function properly 10uF works great. The circuit has very low noise and little gain; however it does improve the sound of my guitar. Coupled with other distortion gain effects it seems to boost the distortion in a very positive way. It creates the allusion of greater gain coupled with noticeably less noise. Basically the circuit is magnifying a tube distortion artifact. It’s a low noise 2nd harmonic distortion generator. I do think the values for R6 and R7 are a bit on the low side, I don’t think batteries will last very long.

A fellow by the name of Gabe wrote an article about making transistors sound like tubes. He was in the pursuit of high end audio reproduction. http://gabevee.tripod.com/sstubepre.html  He discusses how the harmonic distortion is created by altering the shape of the waveform. This allows the tube to make the sound seem both louder and closer to the ear (punchier). He reasons this matches our ear as there are pressure differences on either side of our ear drum, we hear asymmetrically.

He also discusses the warmth of tube circuits as partially due to the high input impedance coupled with a shunt capacitance causes the tubes to naturally roll of the high frequencies.  This can be easily simulated using a Fet or Mosfet. But he claims that an amp made with Fet’s still don’t sound as good as tubes.

Bipolar transistors have low internal impedance so there is no interaction with shunt capacitance this makes them sound harsh. They also suffer from varying base currents in the order of .5% to 10% of the output current. As the signal increases the current increases and decreases nonlinearly with low input level. He reasons that bipolar gain stages are compressing the signal, not tubes as many claim. Clean and sterile with fatigue as compared to a good tube circuit.

He proposes to connect a Mosfet input stage to a bipolar transistor has some interesting advantages.  The Mosfet input stage will mimic the tube input for high input impedance and high frequency rounding. Connect the base of the bipolar to the source of the Fet. This causes the phase of the voltage and current to be in phase canceling out the base current feedback which was compressing the signal dynamics. Also the miller capacitance of the Fet is avoided by this configuration. The bipolar is amplifying the characteristics of the Fet. The sound is warmer and punchier with this arrangement which looks a bit like the Blackfire arrangement. His proposed “Solid State Tube” schematic is in the article above. He also limited the gain to 45 times to closer match a tube circuit max gain.  

I played around with the circuit minus the feedback and the source follower. It sounds pretty good. The sound is fantastic when not clipping or soft clipping but only ok when clipping hard. The circuit was originally designed for high headroom 70 volt B+ and minimal clipping. It seems that you don’t want to overload Q1 Mosfet. The clipping should occur in Q2. For serious guitar we want some seriously good overdrive clipping.

I am building a prototype design that takes elements of the Blackfire, A touch of Sweetness, Highway 89 and the Solid State Tube. I’ve got my Sweetened Black Tube Highway Draft Schematic scanned but I don’t have a site to post it to yet. So if you would like a copy email me at cobcastle@yahoo.com and I will send you a copy by return email.

Caveat other bipolar transistors will likely require different bias arrangements; I am building the prototype with MPSA18 for the bipolar the fixed bias values wont work. I am going to replace the fixed bias for Q1 & Q4 with tripots 200K â€" 500K and 10 uF bypass caps for Q1 connected thru a 1M resistor removing R2 & R3. Also replace the bias for Q4 with same arrangement only thru a 5K resistor removing R13 & R14 from the schematic.    

I’m hoping to tweak it to go from a sweetened clean boost to singing sustain to pulsating overdrive. With a few voicing tweaks I think it will sound great.  

I would be very interested in your feedback, comments, suggestions and great ideas.  :idea:

Regards,
Will

[petemoore]

Makes perfect sense!
 Well written, interesting lingo!!
 I would think it could be quite a desirable item.
 As the front end of a clipping stage might really enjoy this concept.

[WGTP]

When Joe first released the Black Fire, he linked to an article that sounded similar, but I don't remember if it is the same or not.  I'm sure he remembers.  

I was just spanking it awhile ago and shore nuff smokes like the volcano of Mordor.   I think that is where he is forging these things.  

[R.G.]

However, from reading my other pages, you will note that I do not subscribe to the thought that an active device like this "makes" any kind of anything. A device has to be an oscillator to make any kind of wave.

Sadly, this reflects a fundamental misunderstanding of the structure of sound and music.

I wouldn't count too much on the theory in the page. There's a bunch more, but read more widely.

This doesn't mean that the circuits don't sound good - I don't know, haven't tried them; maybe good, maybe not. But the explanation is - in my opinion - flatly wrong. It rejects a lot of math and science that has described electromagnetic phenomena perfectly for a long time.

[zachary vex]

very interesting read, will.

[aron]

The Blackfire is simply amazing, but what is even more amazing to me is how well the simple diode compression scheme works. I am getting amazing sounds out of this fundamental building block. Thank you Joe.

[Doug H]

Yeah, that's the article that Joe developed the Blackfire stage from.

Just as a side note: IIRC, Miller capacitance is gain-dependent. So using a shunt cap alone is "close, but no cigar" in this respect, IMO.

Doug

[Jay Doyle]

Will,

That was an interesting post to read, I'll have to go over it again to really dig into it.

Jay Doyle created a circuit that optimized this effect in a Touch of Sweetness. http://www.diystompboxes.com/pedals/schems/Sweetness.pdf  I built that circuit using 2N7000 Mosfet’s last weekend. Mosfet are much easier in this circuit than fussing around with Fet’s.

Wow! I'm glad someone built it! Though I don't see MOSFETs being any easier, but it would work the same except for being able to get a snot load of gain from that first stage.

No value change is necessary from the diagram. The circuit is quite trebly so I increased C2 to 1 uF and C4 to 4.7 uF.

Good mods for a full range booster.

Also the bias circuits should be bypassed with a decoupling cap to minimize noise. Also Bias 3 needs the cap to function properly 10uF works great.

The circuit has very low noise and little gain;

Yeah, bypassing the Bias supplies would help lower the noise by putting them on the first two. But as you state below you get little gain. Putting a 10uF cap across the third bias pot creates a high pass filter (from the 1k resistor and the add 10uF bypass cap) with a low frequency roll off at about 16 Hz, so any gain you got from the first stage through the second stage is rolled off by this HPF. I imagine that with the bypass cap in you get a unity gain circuit that sounds a little louder due to the added 2nd harmonic. I think you will get a lot more gain if you get rid of that bypass cap. Also JFETs inject less noise than MOSFETs so that may help you there.

I do think the values for R6 and R7 are a bit on the low side, I don’t think batteries will last very long.

It may look that way but a low value R7 is essential to the operation of the circuit and because there isn't a path to ground on its "circuit" side, it doesn't drain the battery all that much other than what leaks through the last stage FET, which isn't much. It's the bias pots that drain the battery. R6 is the "standard" value for a source follower and works in concert with R7 to create the "squishing" of the wave form. I'm pretty sure that a battery will last a good long while. I use the circuit for, if I remember right, three, three-hour shows and I didn't have any issues.

Will, these aren't criticisms, just what I think are important clarifications. I'm glad you liked it and thanks for building it.

I'll go back and read your post again, thanks!

Jay Doyle

[will]

Hi Jay,

Thanks for your reply. I found your effect really interesting; I enjoy the effect. Nice sound by itself and it really improved the sound of the subsequent effects. I like it!

Wow! I'm glad someone built it! Though I don't see MOSFETs being any easier, but it would work the same except for being able to get a snot load of gain from that first stage.

Well, I don't have any Fets currently available and I figured since the source followers are voltage controlled the Mosfets should do the trick. It worked 1st time. I found it sounded best with not a lot of gain from the 1st stage.

Putting a 10uF cap across the third bias pot creates a high pass filter (from the 1k resistor and the add 10uF bypass cap) with a low frequency roll off at about 16 Hz, so any gain you got from the first stage through the second stage is rolled off by this HPF. I imagine that with the bypass cap in you get a unity gain circuit that sounds a little louder due to the added 2nd harmonic. I think you will get a lot more gain if you get rid of that bypass cap. Also JFETs inject less noise than MOSFETs so that may help you there.

First I want to say I didn’t find the effect noisy. I'm confused here. I put a 10uF cap to decouple the bias on the third stage not for noise but to lower the impedance of the reference bias voltage. My reasoning was the 1K bias impedance will strain and override the 10K source resistor. I also tried a 500 ohm resistor there as well but it cut the really cut the gain. I believe I have an 825 ohm resistor there right now. My assumption was the 47K bias pot when biased would contribute depending on voltage trim 15 â€" 20 K swamping the 1K bias resistor. (also I cheated a bit as I used a 100K trim pot)

I do think the values for R6 and R7 are a bit on the low side, I don’t think batteries will last very long.

It may look that way but a low value R7 is essential to the operation of the circuit and because there isn't a path to ground on its "circuit" side, it doesn't drain the battery all that much other than what leaks through the last stage FET, which isn't much. It's the bias pots that drain the battery. R6 is the "standard" value for a source follower and works in concert with R7 to create the "squishing" of the wave form. I'm pretty sure that a battery will last a good long while. I use the circuit for, if I remember right, three, three-hour shows and I didn't have any issues.

I was thinking that increasing that Q2 is driving a 1K load was a bit of a strain. I was thinking of increasing R6 & R7 proportionally say 100K & 10K. I guess I was off base here. Thanks for clearing this up.

Got to go now. Off for the weekend.

Regards,
Will

PS: Jay I thought your challenge post awhile back was awesome. Thanks.

[Gilles C]

Hey Will, I'm glad to know I'm not the only one to have experimented with the SSTubeSPreamp...

http://members.tripod.com/gillcar/id24.htm

I also built and like Jay's Touch Of Sweetness, but I never thought of combining all that together.

I sent you an email anyway...

Gilles

[Gilles C]

Forgot the part about the tubes...

http://members.tripod.com/gillcar/id29.htm

[Gilles C]

[WGTP]

I love this type of discussion even though I don't understand it.  I need to go to transistors 101.

Aron, what are your preliminary Vulcan results?  Have you tried it in the Shaka Bruddrah?

In looking at the Vulcan and Obsidian, it likes like a Mosfet Vulcan with a switch to by-pass the diodes would be pretty easy to build.  The Romulan.
Unless the bias changed with the diodes in and out.  You could use a single switch on each or a 3p.  

[Joe Davisson]

The main thing (I think) is the slowing down of Q2's input charge/discharge time. A tube or MOSFET does this automagically. Sounds interesting though Will, I'll try to check it out some year when I can afford to get a scope. Too busy these days for much of this stuff

-Joe

[aron]

Aron, what are your preliminary Vulcan results? Have you tried it in the Shaka Bruddrah?

I love the Vulcan stage. I'm currently running a modified Vulcan stage - my collectors are around 2.3V!!!!!! What is the collector voltage of the "real" Vulcan?

I think it might make a fabulous Shaka Braddah and I will try that, but right now I am enjoying fooling around with it.

You guys, try and remove the emitter bypass caps from some of the stages!!!!!

The Vulcan (to me) needs:

Hi pass roll off at the end. Try doing the "current fad":

->15K->.0015 (to ground) and string them over and over at the end until you get rid of the hash that may bother you.

Then:

Try using smaller emitter bypass caps.
Try removing the bypass cap from the last stage only.
Try removing the bypass cap from the first stage only.
Try rolling off highs after a stage - i.e. the 2nd stage.

Take note on the coupling/input caps. The first stage is critical. Try a .01uF for humbuckers and the .1uF or .047uF for single coils for the input cap. Play with the coupling cap after the first stage. Try reducing the .1uF.

Finally, I think you have to put series resistance in with the final volume pot to knock the level down quite a bit.

THANK YOU JOE!

[Joe Davisson]

My rule of thumb is to put the collectors near 1/2 supply voltage (4.5v). Don't forget about Vulcanizing those J201's. JFETs should be biased with 2 resistors anyhow. (P.S. Another JFET can be the Vulcanode!)

-Joe

[WGTP]

Thanks Aron.  With all that gain reduction, do you really need 3 stages?  Would GE diodes limit the gain?  If you were doing a PNP version so you could use BS250's do you flop the diodes around too along with the battery and electos?  

[aron]

Thanks Aron. With all that gain reduction, do you really need 3 stages? Would GE diodes limit the gain? If you were doing a PNP version so you could use BS250's do you flop the diodes around too along with the battery and electos?

I like the 3 gain stages although I suppose you could get by with 2 but I haven't tried it.

No, as far as I know, the diodes don't really affect the gain too much - but that could be because there's so much! I'm tried JFETs for diodes but IMO you don't want them for every position. The 1n4148's I used have a certain "sound" it is appropriate in certain positions.

The diodes seem to affect the feel and the "fuzzy-ness" of the tone (to me).

Good question about the diodes.... I guess so, but I haven't tried it at all.

The project is cool because I had just bought a bunch of 2N5089 transistors.

As for gain reduction, that's what's happening when you take out an emitter capacitor, but there's more than that. Try it on the last stage and tell me how much it affects the "feel" and tone of the pedal. That's why I do it.

Aron

[Gilles C]

It seems the other site (Finitesite.com) have problems right now.  :cry:

I uploaded the schematic on another site as a backup, and as a test.



I told Will I would make his schematic available, and available it should be  :twisted:

Gilles