Bellum Fuzz MKI parts values

manfesto

Well-known member
Was looking at the documentation on the FuzzDog RawFuzz V1 and see that they have two BoMs


----------

The original DBA circuit apparently uses 2N5089s (probably pre-sorted for low gain?), 430K biasing resistors, and 680nF caps in every stage, which apparently very rarely works the first time.

FuzzDog's "stable" circuit uses 2N5088s, 820K biasing resistors, and 220nF caps in every stage.

@PedalPCB's Bellum on seems to be 2N5089s, 430K biasing resistors, and 100nF caps in every stage.

----------

I know PedalPCB likes to say they "don't make the weather, just report on it", so I guess I'm wondering

A) who has the original trace correct? FuzzDog with the 680nF caps, or PedalPCB with the 100nF caps?

B) would it be smart to build the Bellum MKI with FuzzDog's "stable" BoM? Or is this circuit not actually as touchy as they're making it out to be? (yes yes I know the right answer is "just socket everything and try" but still)
 
I don't know what is right, but I have my experiences. I have built both fuzzdog versions (on their boards) and the "stable" version works fine every time. I could not get a viable fuzz sound out of the 2n5089 version with unsorted gains. Either too gated to pass signal or squeals on wheels.
 
I don't know what is right, but I have my experiences. I have built both fuzzdog versions (on their boards) and the "stable" version works fine every time. I could not get a viable fuzz sound out of the 2n5089 version with unsorted gains. Either too gated to pass signal or squeals on wheels.
Did you try building with 680n caps? If so, was it indeed too crazy?
 
The preview image was outdated, the PCB has been revised since then. (I've also just updated the preview to avoid confusion)

This circuit is a complete mess.....

2N5089, 430K, and 100nF is what was used in an original. The circuit will almost certainly not work with these values. I say "an original" because there are apparently multiple versions going around with different values. The PCB is marked for 680nF coupling caps, but 100nF capacitors were installed...

Normally yes, I go with the original circuit, flawed or not. But in this case you will not get the results you expect by following the original circuit, I could not in good faith leave the circuit broken just because "thats what was in the original"...

The PCB was redesigned to use 820K bias resistors with 680nF coupling caps. You can feel free to experiment with any variation, but I suspect you won't be happy with 430K bias resistors. All I got was a gated nasty mess.


I would really like to get my hands on an original so I can measure the specs of the transistors, but the V1 is a bit hard to find.


If you happen to see a V1 for sale, or know someone who has one and might let us borrow it that would be great.
 
Last edited by a moderator:
I've also been waiting to get this one working. This is the only one I've seen traced on a forum years ago. It has the 430k (yellow-orange-yellow) resistors and 100nfs though I've also heard of some having 680nf.
Fuzzwarparts 7tran.jpg

Besides the first half, where the fuzz is happening, the rest is confusing too. There's the 680pf in the tone section where the silkscreen asks for a 4n7. Last minute change at the factory? Then there's the missing 27k which would make the germanium clipping diodes useless.

I have a friend who told me he opened up a real one long ago that had a 4n7 tone, and no clipping diodes on the last stage. No pics though. Another guy on a forum was selling clones of one he claims to have traced and told me his had a different value in that position, but never answered back with what.

I built a clone of the DBA Soundwave Breakdown. The fuzz section is like the Fuzz War's but with 910k resistors. It's a very heavy sustain fuzz and not overly gated. The different tone and bias controls get you more versatile sounds.
Others that use those reverse gain stages but with 910k instead of 430k are: Interstellar Overdrive, Harmonic Transformer, and the Supersonic Fuzz Gun (though I had problems with that one gating too much)
 
Last edited:
Then there's the missing 27k which would make the germanium clipping diodes useless.
I remember the original preview image for the Bellum Fuzz said "27K N/A" in that spot, and the new one says "*27K*" with similar markings for one of the diode pairs, I guess to say you can omit them depending on which of the (apparently NUMEROUS) versions of the Fuzz War V1 you want to build?

Also, looks like ceramic caps?
 
In an attempt to understand the variables for this circuit a tiny bit better, I measured h_FE of 35 each of NOS (Motorola) 2N5088 and (GE) 2N5089 at 2.5 mA with a DCA55. The 2N5088s range from 538 to 639. The 2N5089s range from 510 to 839. The sample size isn't really big enough to establish meaningful statistics but mean of 2N5088 is 592, stdev 33, and 2N5089 mean is 667, stdev 85. So 7 randomly selected 2N5088s have a much bigger chance of representing the lower side of h_FE than 7 random 2N5089s.

I at least managed to get 7 of each with similar "low" gain values so will see what happens on a breadboard with the different resistor/cap combos.
 
In an attempt to understand the variables for this circuit a tiny bit better, I measured h_FE of 35 each of NOS (Motorola) 2N5088 and (GE) 2N5089 at 2.5 mA with a DCA55. The 2N5088s range from 538 to 639. The 2N5089s range from 510 to 839. The sample size isn't really big enough to establish meaningful statistics but mean of 2N5088 is 592, stdev 33, and 2N5089 mean is 667, stdev 85. So 7 randomly selected 2N5088s have a much bigger chance of representing the lower side of h_FE than 7 random 2N5089s.

I at least managed to get 7 of each with similar "low" gain values so will see what happens on a breadboard with the different resistor/cap combos.
Wow, those are pretty hot transistors! My 2N5088s usually clock in around 250-300 and my 2N5089s are about 450-500.
 
Wow, those are pretty hot transistors! My 2N5088s usually clock in around 250-300 and my 2N5089s are about 450-500.
I was kinda worried about this after seeing values elsewhere and seeing that the I_c dependence of the h_FE is not very strong, thanks for sharing. I ordered some more of each from different sources to compare.
 
My guess is that there were all kinds of revisions done on the fly to the original units. Multiple cascaded changes due to hFE variations (particularly since they're in backwards), remove the shunt feedback diodes or not, all kinds of crazy stuff happening. With all due respect to the designer, he's known for being a seat of the pants experimentalist, not for making stable, consistently produceable designs. My gut says the originals probably all sounded different anyway, so personally I wouldn't chase nailing the exact authentic sound since I doubt there is one. I built one on perf a while ago FWIW, I think I used like 400nF caps, 830k resistors, and one or two of the transistors sounded better if they weren't backwards (blasphemy!)
 
Wow, those are pretty hot transistors! My 2N5088s usually clock in around 250-300 and my 2N5089s are about 450-500.
I bought some more and it starts to become a bit clearer why the part number is not that helpful in choosing components for this circuit. These are the ranges of h_FE for 2N5088/9s from different sources (number of components in parens).

2N5088 (35) 538 - 659
2N5088 (10) 463 - 522
2N5088 (10) 528 - 574
2N5088 (20) 276 - 302

2N5089 (35) 510 - 839
2N5089 (25) 744 - 816
2N5089 (10) 498 - 554

So a set of 7 2N5089s from the 3rd source would be lower gain than a set of 7 2N5088s from the 1st source, but it you got 2N5089s from the 2nd source they'd be tightly distributed around an hFE of 773 which is really high and useless for purpose. The 4th set of 2N5088s is way lower.

My goal isn't really to make an original one, it's more to get some feel for why the higher value caps/resistors are necessary and how any of the originals could have had 2N5089s at all. I have run out of energy for buying more transistors though now that I have a ton I don't need. Time to move to a breadboard I guess.
 
Last edited:
I bought some more and it starts to become a bit clearer why the part number is not that helpful in choosing components for this circuit. These are the ranges of h_FE for 2N5088/9s from different sources (number of components in parens).

2N5088 (35) 538 - 659
2N5088 (10) 463 - 522
2N5088 (10) 528 - 574
2N5088 (20) 276 - 302

2N5089 (35) 510 - 839
2N5089 (25) 744 - 816
2N5089 (10) 498 - 554

So a set of 7 2N5089s from the 3rd source would be lower gain than a set of 7 2N5088s from the 1st source, but it you got 2N5089s from the 2nd source they'd be tightly distributed around an hFE of 773 which is really high and useless for purpose. The 4th set of 2N5088s is way lower.

My goal isn't really to make an original one, it's more to get some feel for why the higher value caps/resistors are necessary and how any of the originals could have had 2N5089s at all. I have run out of energy for buying more transistors though now that I have a ton I don't need. Time to move to a breadboard I guess.
FWIW when I did my build of the V1 Fuzz War on the PCB Guitar Mania board, I found the transistors needed to be quite closely matched and in the range of around 350-370hFE. Unmatched sets didn't sound right, and higher gain sets (500 and above) just didn't work at all. I've done two builds using those boards now. Will be interested to know if thats the case with the Bellum as my build also uses 820k bias and 680nf coupling caps.

Did a demo video too if you're interested in what mine ended up sounding like;
 
FWIW when I did my build of the V1 Fuzz War on the PCB Guitar Mania board, I found the transistors needed to be quite closely matched and in the range of around 350-370hFE. Unmatched sets didn't sound right, and higher gain sets (500 and above) just didn't work at all. I've done two builds using those boards now. Will be interested to know if thats the case with the Bellum as my build also uses 820k bias and 680nf coupling caps.

Did a demo video too if you're interested in what mine ended up sounding like;
Wow. I think that sounds awesome. I hesitated ordering one of these boards because the docs were not available and the complexity issues I’ve seen. But this sounds very tasty. Nice work.
 
Though raising those resistors values helped me, it will also be worth trying some low gain MPSA06 & MPSA05's with the original resistors values. Several people have advised me to try them on DBA clones, like mentioned here:

(On an FSB thread concerning the Fuzz Gun, which also used 2n5089's)
"reports that the current production units now use MPSA06G ... and a bunch of people reporting that changing the transistors in their clones makes them work as it should" https://www.freestompboxes.org/viewtopic.php?f=7&t=12689&start=120
 
Left is the back of the same board I posted above with the 680pf tone cap and missing 27k.
It was traced on FSB. https://freestompboxes.org/viewtopic.php?f=7&t=2692&hilit=fr

The right is one I just saw on a TGP thread with an unusual red colored enclosure. The owner didn't post any pics of the top, but I noticed some differences. The second set of diodes and the 27k positions aren't present.
It also doesn't have the extrasocket ("add cap") for the other doubled up 120n+20n tone caps.
 

Attachments

  • Two PCB comparisons 01a.jpg
    Two PCB comparisons 01a.jpg
    611 KB · Views: 31
Last edited:
Left is the back of the same board I posted above with the 680pf tone cap and missing 27k.
It was traced on FSB. https://freestompboxes.org/viewtopic.php?f=7&t=2692&hilit=fr

The right is one I just saw on a TGP thread with an unusual red colored enclosure. The owner didn't post any pics of the top,
but I noticed some differences. The second set of diodes and the 27k positions aren't present. It also doesn't have the extra
socket ("add cap") for the other doubled up 120n+20n tone caps.
So at some point they officially redesigned the PCB to drop the 27K and diodes, and also decided to stop cleaning their boards!
 
FWIW when I did my build of the V1 Fuzz War on the PCB Guitar Mania board, I found the transistors needed to be quite closely matched and in the range of around 350-370hFE. Unmatched sets didn't sound right, and higher gain sets (500 and above) just didn't work at all. I've done two builds using those boards now. Will be interested to know if thats the case with the Bellum as my build also uses 820k bias and 680nf coupling caps.

Did a demo video too if you're interested in what mine ended up sounding like;
I built PCB Guitar Manis version too. I could not get it to work as anything other than a high frequency tone generator. Now here I am in the middle of building this one and I’m...concerned. LOL
 
The preview image was outdated, the PCB has been revised since then. (I've also just updated the preview to avoid confusion)

This circuit is a complete mess.....

2N5089, 430K, and 100nF is what was used in an original. The circuit will almost certainly not work with these values. I say "an original" because there are apparently multiple versions going around with different values. The PCB is marked for 680nF coupling caps, but 100nF capacitors were installed...

Normally yes, I go with the original circuit, flawed or not. But in this case you will not get the results you expect by following the original circuit, I could not in good faith leave the circuit broken just because "thats what was in the original"...

The PCB was redesigned to use 820K bias resistors with 680nF coupling caps. You can feel free to experiment with any variation, but I suspect you won't be happy with 430K bias resistors. All I got was a gated nasty mess.


I would really like to get my hands on an original so I can measure the specs of the transistors, but the V1 is a bit hard to find.


If you happen to see a V1 for sale, or know someone who has one and might let us borrow it that would be great.
As far as the "GE" germanium diodes with no value in the build for the MK 1, is there a more specific value for these? Thanks
 
XD finished my build last night using the silkscreen values.... As expected it's a farty mess . Didn't have the time to troubleshoot but I know that manfesto made build report recently and he used 100nf caps on all the 680n ones and he seems to like it :p. In my end I think that swapping the 2n5089 for 2n5088 will help but haven't tried yet ! I am also tempted to try some 2n3903 since I ordered a bunch for the blackash recently ^^
 
Back
Top