BSIAB 2 analysis..

[dschwartz]

Hi there..
i was just simulating BSIAB 2 ckt (from GGG) with LTspice.. and found 2 things:

- First mu-amp stage is pure treble boost, very trebly, it has 45db at 10kHz, and rising up..and -6db at 100hz...look exaggerated to me, but i can understand it..

- After both mu-amp stages, wave look nicely compressed and with round edges..

- but then the last fet adds some serious square clipping, and clips a lot of the nicely compressed wave.. that part confused me...

is the overdrive coming from the last fet?, looks that it kills all the compression gained in previous stages..

at last..the output frequency response is pretty trebly too, 60db peak at 10kHz, and 10db at 100Hz..but  in real life i think this pedal is not specially piercing

can someone explain taht to me?

[flo]

About that last FET: Did you use a J201 model or a 2N5457?
People have stated that the J201 has too much gain, 2N5457 sounds better.
Perhaps a J201 lets it clip to much like your analysis shows?

[dschwartz]

i tried both of them..
both clipped the signal a lot, almost like putting  diodes there...

[John Lyons]

Once a signal is compressed it can not be uncompressed (at least with a guitar signal).
The last stage may be clipping the rounded peaks of previous stages but that will even
further compress the signal while squaring the waveforms.

Take a look at the Minibooster, Booster 2.5 and BSIAB v1 to see the progression to BSIAB2

[dschwartz]

i just can´t find a schem for bsiab 1..dop anyone has a link?

i found a "blues booster" from ed guidry, that looks just like a bsiab, without the last fet..
i´m gessing most of the clipping happens at that stage..its a very interesnting ckt...but i have the feeling that most of the work done by the mu amps is eaten by that fet

[stm]

i just can´t find a schem for bsiab 1..dop anyone has a link?

Daniel, I have the BSIAB I schem at home.  Will send it tonight.

Regarding the JFET models that come with circuit simulation software, they are usually not very representative of the units one actually has.  For instance, in micro-cap the J201 and J202 models are identical with VGS(off) around 2 volts or more.  Same goes to 2N5457, 5458 and 5459, where all three JFETs are treated with the very same parameters.

Based on the measurement of several JFETs, I have tweaked the SPICE models for VGS(off) and IDSS representative of the devices I *actually* have, so I get far more accurate and representative results in the simulation.  Nevertheless, the simulations are not perfect regarding clipping and saturation behavior, but at least with some adjustments gain and biasing prediction is much more accurate.

As usual, YMMV.

[dschwartz]

hi Sebastian..thanks for your input..
this is the spice model of j201 i use:

.MODEL J201 NJF(Beta=1.621m Betatce=-500m Rd=1 Rs=1 Lambda=2.236m
Vto=-600m Vtotc=-2.5m Is=114.5f
+ Isr=1.091p N=1 Nr=2 Xti=3 Alpha=506.8u Vk=251.7 Cgd=4.667p M=227.1m
+ Pb=500m Fc=500m Cgs=2.992p Kf=0.6042f Af=1)

may you share with me your j201 parameters for simulation please? i use LTSpice, but i think spice parameters are the same for both softwares...
thanks in advance!!!

[stm]

may you share with me your j201 parameters for simulation please? i use LTSpice, but i think spice parameters are the same for both softwares...

Sure!

The three JFET models I use are:

* J201 with VGSoff=-0.8V and IDSS=0.6mA
.MODEL J201 NJF (VTO=-0.8 BETA=0.94M LAMBDA=2M IS=114.5F RD=1 RS=1
+ CGD=4.667P CGS=2.992P M=.2271 PB=.5 FC=.5 VTOTC=-2.5M BETATCE=-.5
+ KF=604.2E-18)


* 2N5457 with VGSoff=-1.6V and IDSS=3.3mA
.MODEL 2N5457 NJF (VTO=-1.6 BETA=1.29M LAMBDA=2M RD=1 RS=1 CGD=6E-12
+ CGS=2.25E-12 KF=6.5E-17 AF=0.5)


* MPF102 with VGSoff=-2.5V and IDSS=6mA
.MODEL MPF102 NJF (VTO=-2.5 BETA=0.96M LAMBDA=5M RD=1 RS=1 CGD=1.54248P
+ CGS=2.567P PB=1.49 KF=7.90591F AF=499.953M)


Your J201 model is a little on the high-gain side, but perfectly reasonable, as it has VGSoff=-0.6V and IDSS=0.58mA.
My model has a little less gain as it is representative of the average of a lot of units I measured.

Just in case you are wondering where did I pull the 0.58mA:  IDSS= BETA * (VTO^2).
So, if you want a JFET with a certain value of VGSoff and IDSS, just change the model so:

VTO = VGSoff
BETA = IDSS / (VGSoff^2)

Other parameters are less relevant.  For instance, LAMBDA accounts for the Vds effect on the drain current.  As it is a very small value, it doesn't produce any significant effect unless you are powering your circuit with 50 volts or more.
Other parameters model parasitic capacitances (which affect behaviour at high frequencies), or model temperature dependency.

[dschwartz]

great thanks!!!!!

[mojotron]

Just a few thoughts here....

Also - look at what you are using for an input signal; a 1vp-p to 2vp-p wave is good for frequency analysis, but a wave form analysis might be better done using smaller signals: 1vp-p or less as well as larger signals. The clipping may just be a factor of how big the signal is wrt Vds or Vgs.

Also - what settings are you using for the gain control? I generally try to sim things with several different settings if a mid-setting on each pot does not look right.

I've never simmed this circuit, but that's what I would look at.

BTW, stm, those extra models are great to have - THANKS!

[dschwartz]

i use 100V p2p signal, 1kHz sinusoidal, i tried different gain settings and the clipping is there almost allways..
i´ll try STM´s models now..

[mojotron]

i use 100V p2p signal, 1kHz sinusoidal, i tried different gain settings and the clipping is there almost allways..
i´ll try STM´s models now..

A guitar is generally going to give you something like a 1Vp-p signal, maybe a single coil will be 100mV-400mV, so 100Vp-p is way too big to stuff into a 9VDC circuit - it's clamped by the supply voltage and Vds is lower than that. That's why Its clipping...

I would try signals like .1v/1kHz, .5v/1kHz, .01v.... Then you may see that there is a lot more interesting stuff going on in the transfer from stage to stage in the time domain as the knob settings are changes.

Also, the settings for the trim-pot R16 may need to be tweaked as well so you will want to compare test points on the gate of Q5 and the drain of Q5 for values of R16 from 25k-80k...

[dschwartz]

i meant 100mV!!!..hahaha
i´m not that newbie..
i think 1V p2p is a lot larger than a guitar output, line levels are about that voltage..i´m pretty sure guitars go around 100-200mV

[Ed G.]

- but then the last fet adds some serious square clipping, and clips a lot of the nicely compressed wave.. that part confused me...

is the overdrive coming from the last fet?, looks that it kills all the compression gained in previous stages..

I got around to breadboarding the circuit, it's been some years ...
I found that Q5 made the circuit a bit too compressed sounding and too 'cluttered' and 'congested' sounding for lack of a better term. I think I was hearing what you are seeing in the sim.
I'm trying some different values right now, I increase the value of R14 to 100K and R15 to 4.7K. So far, I like what I hear, but I want to rework that whole stage, I think it's the weak link in the circuit. Possibly use a source-biasing scheme. Also, I want to tune the filters a bit more. They are dark sounding to most people.

[dschwartz]

Hey ED!!
nice to hear from you...

well, i´m kind of confused with that last stage, too.. is heavily overdriven and looks sqare and fuzzy in the simulator....but the real thing sounds great!!.
I believe that the very heavy filtering at the first stages (leaving almost only harmonics to distort) gives that crunchy raw sound we like when hard clipped..

i dont think the bsiab is dark.. but maybe is too dependable of input impedance of the next stage..as most of the fet distortions around..

is the blues booster sound too different from the bsiab?

[Harold]

I'm trying some different values right now, I increase the value of R14 to 100K and R15 to 4.7K. So far, I like what I hear, but I want to rework that whole stage, I think it's the weak link in the circuit. Possibly use a source-biasing scheme. Also, I want to tune the filters a bit more. They are dark sounding to most people.

Hey Ed, dit you ever find the time to work on it? I built a BSIAB II with the "countour" pot today and was wondering what tweaks are available. Can you tell me what the 4k7/100k combination around Q5 does?

[J0K3RX]

I'm trying some different values right now, I increase the value of R14 to 100K and R15 to 4.7K. So far, I like what I hear, but I want to rework that whole stage, I think it's the weak link in the circuit. Possibly use a source-biasing scheme. Also, I want to tune the filters a bit more. They are dark sounding to most people.

Hey Ed, dit you ever find the time to work on it? I built a BSIAB II with the "countour" pot today and was wondering what tweaks are available. Can you tell me what the 4k7/100k combination around Q5 does?

Harold, this post (BSIAB 2 analysis) is from way back in October of 2008! Ed's last post in this forum was October 28, 2012... Good luck 

[Harold]

Harold, this post (BSIAB 2 analysis) is from way back in October of 2008! Ed's last post in this forum was October 28, 2012... Good luck 

That should have given him more than enough time to come up with a solution!

[tca]

The three JFET models I use are:

* J201 with VGSoff=-0.8V and IDSS=0.6mA
.MODEL J201 NJF (VTO=-0.8 BETA=0.94M LAMBDA=2M IS=114.5F RD=1 RS=1
+ CGD=4.667P CGS=2.992P M=.2271 PB=.5 FC=.5 VTOTC=-2.5M BETATCE=-.5
+ KF=604.2E-18)


* 2N5457 with VGSoff=-1.6V and IDSS=3.3mA
.MODEL 2N5457 NJF (VTO=-1.6 BETA=1.29M LAMBDA=2M RD=1 RS=1 CGD=6E-12
+ CGS=2.25E-12 KF=6.5E-17 AF=0.5)


* MPF102 with VGSoff=-2.5V and IDSS=6mA
.MODEL MPF102 NJF (VTO=-2.5 BETA=0.96M LAMBDA=5M RD=1 RS=1 CGD=1.54248P
+ CGS=2.567P PB=1.49 KF=7.90591F AF=499.953M)


Your J201 model is a little on the high-gain side, but perfectly reasonable, as it has VGSoff=-0.6V and IDSS=0.58mA.
My model has a little less gain as it is representative of the average of a lot of units I measured.

Just in case you are wondering where did I pull the 0.58mA:  IDSS= BETA * (VTO^2).
So, if you want a JFET with a certain value of VGSoff and IDSS, just change the model so:

VTO = VGSoff
BETA = IDSS / (VGSoff^2)

Other parameters are less relevant.  For instance, LAMBDA accounts for the Vds effect on the drain current.  As it is a very small value, it doesn't produce any significant effect unless you are powering your circuit with 50 volts or more.
Other parameters model parasitic capacitances (which affect behaviour at high frequencies), or model temperature dependency.

Happy finding!

Thanks Harold.

[amz-fx]

3 gain stages is a lot...  more than enough.  That's why I cut the original Mini-Tubes back to 2 stages, and then made the Booster 2.5 from the same general configuration. The schematic is still on Aron's page:



Best regards, Jack