Why not noiseless-bias BSIAB2 and Pinnacle style circuits?

[darron]

Hey all. Here's something I've wondered for years. The BSIAB2 and Wampler Pinnacle (and many similar) circuits use Mu amp circuit biased by a network of 1M+1M from V+ and ground. Here's a reference schematic for the BSIAB2 https://generalguitargadgets.com/pdf/ggg_bsiab2_sc.pdf

MY question is; can we use noiseless biasing with say a 10K+10K voltage divider connected by a 470K/510K resistor? Example of this here: http://www.geofex.com/Article_Folders/modmuamp/modmuamp.htm

I've seen reports of people building the circuit this way, but not successfully. Only in debugging threads.

I do this for clean boosts. But maybe there's a good reason NOT to do this for distortions? If Brian Wampler doesn't... Maybe the distorting transistor state affects the bias and then the clipping nature? Some magic like that?

Thought it would be better to just ask rather than design a new experimental board layout. Is this a design consideration, or an oversight? Thanks.

[Rob Strand]

MY question is; can we use noiseless biasing with say a 10K+10K voltage divider connected by a 470K/510K resistor?

Absolutely fine doing that.  From the amps perspective it's identical - except lower resistor noise and some more current from the supply rail through the 2x10k's.

In many circuits the bias resistor is applied to the input of the circuit.  In the case of the mu-amp the bias resistor is applied after the gain stage, it's kind of a buffer.   There's also a 100nF or 1uF across the gate-source.   What I'm getting at here is the bias network probably contributes less noise for the mu-amp than some other circuits.    I haven't checked exactly how much but that's what it looks like to me.

[antonis]

From the amps perspective it's identical - except lower resistor noise and some more current from the supply rail through the 2x10k's.

IMHO, there is another important (frequently overlooked) factor, which is the divider filter cap..
In the abscence of that cap, power supply rejection is only 6dB (half of Vcc ripple) severely dominating stage's stability, especially when the amp must supply large currents into a load..

[amz-fx]

What I'm getting at here is the bias network probably contributes less noise for the mu-amp than some other circuits.    I haven't checked exactly how much but that's what it looks like to me.

I should measure that. I believe that I have some proto boards that could be adapted for that purpose.

Best regards, Jack

[darron]

Thanks everyone. I have a little more confidence now, and will do a new board design with the filtered divider. If I don't report back here down the track then we'll assume it was a success

I'm planning to make the boards publicly available. It's on the long to-do list.

Still wondering why this isn't applied more often. It's not like it adds more parts or complexity where there are multiple stages.

[GGBB]

I did the noiseless bias thing on the BSIAB2/Pinnacle a number of years ago - works fine.

[darron]

I did the noiseless bias thing on the BSIAB2/Pinnacle a number of years ago - works fine.

Perfect! That's a really cool schematic, too. I haven't actually tried the Pinnacle or the Brownacle. Brownacle mods look interest, I should give that one a shot... So double thanks : )

[Clint Eastwood]

I did the noiseless bias thing on the BSIAB2/Pinnacle a number of years ago - works fine.

C15 surely doesn't have to be 100uF?   10uF would already give a roll-off frequency less than 1 Hz.

[antonis]

I did the noiseless bias thing on the BSIAB2/Pinnacle a number of years ago - works fine.

Gord, I'd move LED & respective circuit before R19..

P.S.
We continually argue about RLED1 & RLED2 values but this particucalar configuration of yours does credit to me..

[GGBB]

Gord, I'd move LED & respective circuit before R19..

P.S.
We continually argue about RLED1 & RLED2 values but this particucalar configuration of yours does credit to me..

Old schematic - from before I knew better.

[MrStab]

It's like a milky coffee if you keep the Nyquist noise of the original biasing: Brown Sound with a hint of white noise

[PRR]

C15 surely doesn't have to be 100uF?   10uF would already give a roll-off frequency less than 1 Hz.

For frequency response, we want "small" cut at 20Hz or 82Hz.

For power crap filtering we want LARGE cut at 50/60/100/120Hz. As much as we can afford or fit.

100uFd against 23.5k is like 0.07Hz. Compared to 60Hz this is 857:1 attenuation. Volts of ripple come out milliVolts. In bulk the cost of 100uFd 10V is 10 cents or less, not a severe dent on profit.

Of course if the power is clean you do not need that much filtering. A 10uFd is pennies cheaper. If it buzzes in the customer's rig, blame his power supply.

In DIY I just use whatever medium big cap is handy.

[darron]

Thanks everybody. I did the noiseless bias method and it worked great.

Schematic here: https://www.dazatronyx.com/docs/Brown-Sound-schematic.pdf

If anybody wants to grab a PCB off me for the project hit me up. It's partially SMD populated, since the transistors are going to be SMD FETs anyway.

Don't mean to make it too much about advertising, but in the spirit of DIY, I also have the project files or a complete kit for sale on my site as a result of this conversation: https://www.dazatronyx.com/#brownsound

[antonis]

Did you intentionally place C9 in series with only R7 (and not before both R7 & C8)..??

[GGBB]

Perfect! That's a really cool schematic, too. I haven't actually tried the Pinnacle or the Brownacle. Brownacle mods look interest, I should give that one a shot... So double thanks : )

I still have a few Brownacle PCBs left over in case anyone is interested. PM me for more info.

[printer2]

So used to seeing the 470k/470p combination in series with a large pot. Would there be less noise if a follower were after the gain stages and scaled down part values were used, say scale the parts by a 10X? (47k/4.7n/50k)

[darron]

Did you intentionally place C9 in series with only R7 (and not before both R7 & C8)..??

Yes. I think that way makes a little more sense, and shouldn't have any change tone-wise. But I could be wrong. I've been building them like that for a while.

[darron]

So used to seeing the 470k/470p combination in series with a large pot. Would there be less noise if a follower were after the gain stages and scaled down part values were used, say scale the parts by a 10X? (47k/4.7n/50k)

That's a good point. I always wondered if the 470K heavy signal series loading was an important part of the distortion/interaction at the next stage. Same with the 500K gain pot. At least it will only add noise on the lower frequency shelf, not the upper one.

ZVEX Box of Rock has this same magic 470K/470p section.

Geofex has an article on modding the MU amp to be like a buffer, but without adding a third follower transistor. Just adding a 1K resistor: http://www.geofex.com/Article_Folders/modmuamp/modmuamp.htm

[POTL]

So used to seeing the 470k/470p combination in series with a large pot. Would there be less noise if a follower were after the gain stages and scaled down part values were used, say scale the parts by a 10X? (47k/4.7n/50k)

the input impedance and the internal capacitance of the amplification stage must be taken into account. Unlike operational amplifiers, discrete amplifiers can be dependent on these values. I will be a bore (haha) but once again I will speak out against the idea of ​​​​Mu Amp. This design is not as well filtered as single transistors, and wave clipping is not like the class A amplification of single transistors or triodes. My verdict, Mu Amp and Jfet should have retired a long time ago (especially since they are getting harder to buy). Instead of Mu Amp, it is better to use Baja-style OpAmp, it is more predictable and flexible. If you need the behavior of a triode, look to Mosfet, it is also more predictable and flexible.

[printer2]

the input impedance and the internal capacitance of the amplification stage must be taken into account. Unlike operational amplifiers, discrete amplifiers can be dependent on these values. I will be a bore (haha) but once again I will speak out against the idea of ​​​​Mu Amp. This design is not as well filtered as single transistors, and wave clipping is not like the class A amplification of single transistors or triodes. My verdict, Mu Amp and Jfet should have retired a long time ago (especially since they are getting harder to buy). Instead of Mu Amp, it is better to use Baja-style OpAmp, it is more predictable and flexible. If you need the behavior of a triode, look to Mosfet, it is also more predictable and flexible.

Irregardless of the type of amplification stage, is there any reason to use high impedance circuits that are lifted right out of a Marshall tube amp or would there be a noise advantage to scale the parts by (as an example) 10X?