Jack Orman's MOS Face, why does the BS170 MOSFET work in place of a standard BJT

[bushidov]

Hi Guys,

I was reading Electrosmash's explanation for the standard Fuzz Face, which I know is Germanium PNP based. I get the conversion from PNP to NPN. https://www.electrosmash.com/fuzz-face#link4

So, looking at Jack Orman's MOS Face http://www.muzique.com/schem/projects.htm, how does the differences in the output stage translate? Like, how is the output impedance calculated or understood? Voltage gain? Clipping? Etc?

Just curious and trying to understand the audio differences a MOSFET brings to the table of fuzz and distortion as opposed to BJT transisters and such.

Thanks for your time,
Erik

[antonis]

Don't get me wrong but, if you've understood enough FF "architecture", you shouldn't have any issue by replacing second BJT with a MosFet.. 
(just consider MosFet as a BJT of much higher input impedance and much lower specific configuration gain..)

In circuit analysis, ignore any current going from Q1 collector to Q2 BE junction (for a pure BJT FF)..

[PRR]

> looking at Jack Orman's MOS Face

"Vbe" of a MOSFET is significantly higher than a BJT. Like 1V-2V instead of 0.6V.

This forces the collector of Q1 to sit a little higher. But a 1V shift under a 9V supply makes very little difference in Q1 current through R1.

The MOSFET has "infinite hFE" so does not divert current from Q1, but for any not-low-hFE BJT at Q2 this difference is very small.

A MOSFET might typically have higher "plate resistance" than a jellybean BJT, but in either case the 6.6k dominates the collector impedance (~~1K at the output tap to C3, 0-125k at the jack).

I don't see any functional difference. What does your ear say?

[antonis]

A MOSFET might typically have higher "plate resistance" than a jellybean BJT

<off-topic ON>
I'll take your word for that, Paul..
(although my limited experience on MosFets have steered me to the opposite notion..)
e.g. less than 10μmhos output admittance for BJTs instead of more than 50μmhos output conductunce for MosFEts on both "typical" working conditions..
(both the above either extracted from datasheets or estimated by Early voltage / Collector quiescent current and reciprocal of Drain quiescent current x channel-length modulation parameter..)

<off-topic OFF>

Of course, in either case, Collector/Drain much lower resistor value dominates transistor output impedance..

[bushidov]

Thanks for the info Antonis and Paul.

I don't see any functional difference. What does your ear say?

My ears say it sounds great. I'm just trying to understand it. Also, I haven't done much testing against it in regards to pedal board mixing and cable lengths. But from just a Fender Strat to Pedal to Amp with 10 feet-or-less cables, it sounds pretty good.

I guess my more specific questions would be as follows:
1. Output Impedance:
I know that in typical FF architecture, you can get a rough idea by taking the max volume pot resistance, in this case 500K and taking the value of R2, which is 1K and getting the parallel resistance of those, 500K || 1K, and that will give you the base impedance of roughly 998 ohms. However, I also know that due to the feedback network of the FF architecture, that raises the value to something like 7-15K, which I know from an output impedance perspective, is not too good (trying to keep it under 10K), as without a buffer after the pedal in a signal chain, issues may occur.

I am aware that MOSFETs tend to do better when it comes to handling output impedance, so I was wondering how the math of getting the output impedance has changed, if at all?

2. Total Voltage Gain:
I know that if it were a BJT, if the 1K Fuzz pot is dimed (1K), voltage gain looks like (R2 + R3) / Rfuzz or (1K + 5.6K) / 1K or 6.6K (as Paul pointed out), so long as I ignore the feedback network. So, the voltage gain can go from 6.6 (16 dB) to as high as the BJT's internal gain, assuming the Fuzz pot is dimed/maxed.

But taking into consideration the feedback network, Q2 source is around 19.5dB, so I take that minus the attenuation of (R2 / (R2 + R3)) = 1000 / (1000 + 5600) = 0.1515 (-16.4dB) giving me 19.5dB - 16.4dB = 3.1dB

Going off what Paul and Antonis is saying, I am guessing the Total Voltage Gain maths don't really change in regards to any of this, so long as Q1 is a BJT and Q2 is a MOSFET. Am I correct on this assumption?

I ask this, because if I wanted to increase the gain of this pedal, I can just treat it like a standard BJT when trying to figure that stuff out, as it appears to make little difference in this particular use case of Q2 becoming a MOSFET?

Sorry for asking a lot of technical questions and sounding like a noob (probably because, as far as pedal electronics go, I am a noob)

Thanks,
Erik

[antonis]

I am guessing the Total Voltage Gain maths don't really change in regards to any of this, so long as Q1 is a BJT and Q2 is a MOSFET. Am I correct on this assumption?

Quite Yes..
(and Quite No..) 

For a CE/CS single stage amp, total voltage gain is [-g_m (R_C/D//r_o//R_load)] / (1 + g_m R_E/S) times Rin / (Rin + Rss), where where R_E/S is the unbypassed Emitter/Source resistance, r_o is transistor output resistance, Rin is equivalent input impedance & Rss is signal source internal resistance..
So, for same working conditions, it depends more on g_m differences between BJTs & MosFets (and less on r_o differences..)
But, here NFB comes to take over the above issue so raising total gain should be mainly obtained by modifying the amount of feedback..

[PRR]

> Output Impedance:... roughly 998 ohms.

Uh, not unless the output pot is maxed. At mid-turn (which may be likely if the FF full output is more than the amp can swallow nicely) the output impedance can't be less than pot/4, or 125k for a 500k pot. Actually we can take the 1k and the 500k together and figure 501k/4, which is still 125k FAPP.

[Rob Strand]

One difference between Q2 being MOSFET and BJT is the MOSFET will roll off the highs a lot more.  Especially when the collector resistor of the first stage is high (100k) like in the MOSFACE.  For my version of the MOSFACE I used a lower valued resistor for the first stage, like 27k.   To be honest (IIRC) I initially chose that value so I could swap MOSFETs and BJTs without rebiasing 

[MaxPower]

Breadboarded this and I like it. Put a 250k pot in place of the 100k fb? resistor. Now it goes from clean to fuzzy goodness.