Bearfoot - Mint Green Mini Vibe - Trace and Build

phi1

Well-known member
This was a satisfying project for me. I don't often trace pedals, and I don't consider it my strong point, but I took my time on this one.

I've been interested in trying a MGMV for years. Like many BJF pedals, the reviews online are glowing and seem magical. And for my part I'm pretty much bought in, I love both the Honey Bee OD and the Deep Blue Delay clones I've made. So I finally found a used MGMV to try, without really intending at first to trace it. It happened to be a hand-painted octopus enclosure... that's all I could find available for sale at the time.

I noticed right away that it sounded like a harmonic tremolo, albeit a uniqe sounding one. There's an FSB thread (some pictures and component labels, no schematic), but I couldn't figure how a 6 Jfets and 1 opamp circuit would be configured for H trem. So I started poking around and couldn't stop. The board was so compact with all the standing resistors so I couldn't see the traces without removing the parts, and I eneded up removing every single component. (A better tracer probably could have kept it in tact).

I drew out the circuit and still couldn't really make sense of how it worked. I planned to try my own build to verify the trace but put it aside for a while. Eventually I realized that it's following the Fender Brownface harmonic tremolo schematic (ex 6G5-A), replacing tubes with JFets. But there's 2 harmonic tremolo blocks instead of 1 in the amps. This seemed pretty cool and got me motivated to get pcbs made and build it.

At first go, it sounded really trebley, to the point I thought something must be wrong with my trace or my build. Then I started swapping around my jfets and with the right configuration it sounds perfect, so I guess it's sensitive to matching the jfets. The original was all 2n5952, so far I used all 2n5457 (the pinout is reversed, which I applied). I may try to pull out the original jfets and install to see if there's much noticeable difference.

It really is a great sounding pedal, more vibe-y than my beloved Cardinal V2 harmonic trem (design by jon patton), probably because of the 2 stages. I will be boxing it up soon.

R44 is appears to be a fake-out resistor to thwart would-be tracers such as myself... I hope at this point, long after the pedal is out of production, that BJF would take my endeavor as a compliment.
 

Attachments

  • MGMV_trace_schem_page1.png
    MGMV_trace_schem_page1.png
    366.1 KB · Views: 38
  • new pcb build.jpg
    new pcb build.jpg
    628.5 KB · Views: 38
  • octopus pic.jpg
    octopus pic.jpg
    476.3 KB · Views: 34
  • gut shot.png
    gut shot.png
    1.2 MB · Views: 38
Really outstanding work! The circuit is way more interesting than I would have thought, especially that it uses two stages.

For the JFETs, I've found the dual 2SK2145 that shares a common source pin are decently matched, even though they're not listed as matched in the data sheet.

Did you ever run into issues with LFO thump or ticks?
 
Nice idea for the 2sk. The odd thing is, I had pre-sorted my 2n5457's by Vp, and these were all in the -1.2~1.3V range, so I expected them to be decently matched. I could revisit the measurements with more of an emphasis on Idss I guess. I know Vp is the key for phase90 matching, not sure about this circuit.

Maybe another idea could be to put a trimmer for the Drain resistor for each of the 4 fets in the audio path? Or, separate out the Source R/C and put a trimmer there?



I've mostly played this with Headphones so far while the rest of the house is sleeping (Spark Neo). I'll try more with the amp soon, but I didn't really notice LFO noise so far. I was mostly focused on the JFET configuration to get the best EQ.
The "SW" wire does ground out the LFO when the pedal is bypassed (which also turns off the LED), I assume that is to prevent any LFO noise from leaking out when not in use.
 
Sure I got 5 made, so send me a DM and we can figure out how to get one to you.

The original fit in a 1590B with a battery. I'm slightly embarrassed that my pcb design fits in 1590BB (with side jacks). It could certainly be brought down to 125B or 1590B with a re-design. I just didn't feel like working on optimizing it at the time. Lmk if you want to explore that more.

Also, I'd consider if you build it:
1. socket the JFETs, since I found swapping them around made a big difference in EQ balance.
2. replace the 14.7k drain resistors (R7, R8, R20, R21 - I used 15k) with trimmers, so you can adjust the balance of the paths without relying on the perfect combination of Jfets. (I think this would work, maybe @jubal81 will give some more feedback about that idea). The pedalpcb Trimmit! could help implement this cleanly.
3. Either socket, timmit!, or make a pot of R36, and socket C23. R36 and C23 make a lowpass filter right at the end of the circuit. I was experimenting with this before swapping the jfets, and changing the C23 was a powerful tone shaping tool, but in the so far I've left it as is.
 
2. replace the 14.7k drain resistors (R7, R8, R20, R21 - I used 15k) with trimmers, so you can adjust the balance of the paths without relying on the perfect combination of Jfets. (I think this would work, maybe @jubal81 will give some more feedback about that idea). The pedalpcb Trimmit! could help implement this cleanly.

The use of 14K7 on all the JFETs indicates the value isn't critical and the gain pairs were selected to match.

Consider the JFETs as being set up as phase splitters: equal drain and source resistors to give equal outputs that are out of phase. They are biased at the gate to ground. Maximizing headroom wasn't a design goal, as there's likely enough for the effect.

The gain stage pairs share a source resistor, so it's about half the value of the drain resistor. A bypass cap is added to the source to make gain at the drain output, which is what moves on to the next stage.

The trick is using a matched pair for the ones that share a source connection. I'd bet the Toshiba dual JFETs would work fine, but there's manufactured pairs designated as matched in the data sheet, which are very expensive.
 
Can you measure the Vp and Idss of the transistors from the traced unit?
Traced Bearfoot unit 2n5952s:

Q1: Vp = -2.43v IDSS = 6.60mA
Q2: Vp = -2.49v IDSS = 6.83mA
Q3: Vp = -2.42v IDSS = 6.49mA
Q4: Vp = -2.42v IDSS = 6.67mA
Q5: Vp = -2.45v IDSS = 6.87mA
Q6: Vp = -2.48v IDSS = 6.99mA

These were measured on my own pcb design, which copies the bottom, simpler circuit from here. It's from the runoffgroove - 'a closer look at the fetzer valve' article.

These are definitely well-matched, and it was worth measuring, they seem closer to what I expect from 2n5458s rather than 2n5457s typically. I measured the 3 2n5458s I have and they're all over the place.
1: Vp = -3.38v IDSS = 6.20mA
2. Vp = -2.84v IDSS = 5.24mA
3. Vp = -1.63v IDSS = 3.53mA
 

Attachments

  • IMG_1097.jpeg
    IMG_1097.jpeg
    918.7 KB · Views: 1
Back
Top