Hi Ian,
I agree. I also recall the crumb about resistors wandering off to ground as well. However, that being said these are the only resistors to ground that are shown in the patent. Claim 8 specifically covers their use. They are signficant. They are not wandering.
Your idea of shunting an active input node sounds good (no pun). However, that would mean connecting the shunt directly to the actual signal which NP says is not the case. The SuSy line amps are single stage designs AFAIK. Not many places in a single stage design to find a node that's at virtual ground and not directly connected to the signal. I still think that 42 and 43 could be the charm.
Note that NP claims the use of resistors 42 and 43 in claim 8 as a possible part of a SuSy circuit but does not specifically state that he is claiming this as a gain adjusting element. Very interesting.
The idea of using bipolars for switches is not that farfetched. Back in the late '60's and early '70's (before 4053's) the electronic organ makers used BJT keyers for switching low level audio signals. An organ contained hundreds these in parallel so they had to be designed well enough to provide serious isolation when off but have almost no resistance when on. The circuit used a BJT, a diode, a couple of resistors for bias and a small cap to eliminate clicks. I have an example in a text book somewhere. All things considered a BJT switch done right is a better choice in this application. Linearity does not have to be an issue. On the other hand if I was switching the signal itself directly I would definitely use FET's.
Cheers,
Graeme
I agree. I also recall the crumb about resistors wandering off to ground as well. However, that being said these are the only resistors to ground that are shown in the patent. Claim 8 specifically covers their use. They are signficant. They are not wandering.
Your idea of shunting an active input node sounds good (no pun). However, that would mean connecting the shunt directly to the actual signal which NP says is not the case. The SuSy line amps are single stage designs AFAIK. Not many places in a single stage design to find a node that's at virtual ground and not directly connected to the signal. I still think that 42 and 43 could be the charm.
Note that NP claims the use of resistors 42 and 43 in claim 8 as a possible part of a SuSy circuit but does not specifically state that he is claiming this as a gain adjusting element. Very interesting.
The idea of using bipolars for switches is not that farfetched. Back in the late '60's and early '70's (before 4053's) the electronic organ makers used BJT keyers for switching low level audio signals. An organ contained hundreds these in parallel so they had to be designed well enough to provide serious isolation when off but have almost no resistance when on. The circuit used a BJT, a diode, a couple of resistors for bias and a small cap to eliminate clicks. I have an example in a text book somewhere. All things considered a BJT switch done right is a better choice in this application. Linearity does not have to be an issue. On the other hand if I was switching the signal itself directly I would definitely use FET's.
Cheers,
Graeme
The resistors may not be wandering but they are difficult to implement in practice as they mess up the DC conditions and also reduce the CMRR. Their purpose is to reduce common mode noise on the output but in practice it is better to solve this in other ways.
Are you sure that NP says that Waynes volume control does not connect to the signal? That sounds very curious. Did he perhaps say that it does not appear in series with the signal? Controlling voume without connecting the signal would be difficult (impossible?) even it were to act by changing the gain of an active stage.
I understand that bipolars can be used as switches but I was wondering what it is about their application in this context that makes them the preferred device. In general FETs are easier to implement when dealing with a bidirectional (+ve and -ve) signal. Bipolars do not work well in reverse.
I guess if we want to take this forward we will need to come up with a specific circuit that others on this forum can pick to pieces and hopefully improve. I'll give this some more thought.
Are you sure that NP says that Waynes volume control does not connect to the signal? That sounds very curious. Did he perhaps say that it does not appear in series with the signal? Controlling voume without connecting the signal would be difficult (impossible?) even it were to act by changing the gain of an active stage.
I understand that bipolars can be used as switches but I was wondering what it is about their application in this context that makes them the preferred device. In general FETs are easier to implement when dealing with a bidirectional (+ve and -ve) signal. Bipolars do not work well in reverse.
I guess if we want to take this forward we will need to come up with a specific circuit that others on this forum can pick to pieces and hopefully improve. I'll give this some more thought.
Why would you not use some of the mosfet switches with isolated photovoltaic gate drive instead of bipolars? I'm assuming because of the capacitance?
How do you use bipolars on a circuit that has an alternating polarity? Is it biased above ground?
Someone throw me a crumb here... this is interesting stuff.
How do you use bipolars on a circuit that has an alternating polarity? Is it biased above ground?
Someone throw me a crumb here... this is interesting stuff.
Commercially available analog switches are also fairly well used today... even JFET types I think... They are available in arrays which would be very conveneant, I would think... When your impeadance is 10K or something. 1 ohm or less tends not to have a great effect
Any signal dumped to ground better be dividing the original signal perfectly, or the resultant signal path will now contain extra signal that was not divided (distortion)!
Just a quick few peices of info, I think we should be pondering.
Any signal dumped to ground better be dividing the original signal perfectly, or the resultant signal path will now contain extra signal that was not divided (distortion)!
Just a quick few peices of info, I think we should be pondering.
Good Morning Folks
BINGO
I personnally consider a device between gnd and a shunt not in
the signal path.
All those switch arrays are of poor audio quality AND they do not
support a lo enough on resistance. The series resistor is not the
criteria, the shunt is very much lower.
Example:
Series resistor is 10k. Shunt is eg 101R wich is ~40dB, a not so
unusual attenuation in a pre. when you have a switch with
a resistance of about 50 - 100R, this leads to gain errors up to 6 dB!
Uli
HBarske said:
BINGO
I personnally consider a device between gnd and a shunt not in
the signal path.
All those switch arrays are of poor audio quality AND they do not
support a lo enough on resistance. The series resistor is not the
criteria, the shunt is very much lower.
Example:
Series resistor is 10k. Shunt is eg 101R wich is ~40dB, a not so
unusual attenuation in a pre. when you have a switch with
a resistance of about 50 - 100R, this leads to gain errors up to 6 dB!
Uli
Okay, let's clarify this for dummies like me:
When the transistor is open, there's only VCEsat left between collector and emitter. This could be made rather low, if IB (high positive UBE I assume) is high enough. But: There's still an AC signal present. The BJT won't have problems with a negative IC?
When the transistor is open, there's only VCEsat left between collector and emitter. This could be made rather low, if IB (high positive UBE I assume) is high enough. But: There's still an AC signal present. The BJT won't have problems with a negative IC?
There are 2 ways to implement such a switch, normal (emitter
connected to gnd, collector to shunt) and inverse mode (emitter
connected to shunt, collector to gnd). The difference is that in
inverse mode the CE offset voltage is in the range of 1 - 3 mV
whereas in normal mode it is about 10 - 20mV.
Because of the Uce - Ic "Kennlinie" around Zero you can use
a BJT as a switch assuming Uce is lower than Uce-sat, below
50mV or so. In our thought experiment the signal voltage across
CE should be significantly lower than that.
(P. Skritek, "Handbuch der Audio Schaltungstechnik" pp 185 ff.)
Uli
connected to gnd, collector to shunt) and inverse mode (emitter
connected to shunt, collector to gnd). The difference is that in
inverse mode the CE offset voltage is in the range of 1 - 3 mV
whereas in normal mode it is about 10 - 20mV.
Because of the Uce - Ic "Kennlinie" around Zero you can use
a BJT as a switch assuming Uce is lower than Uce-sat, below
50mV or so. In our thought experiment the signal voltage across
CE should be significantly lower than that.
(P. Skritek, "Handbuch der Audio Schaltungstechnik" pp 185 ff.)
Uli
I suspect that the problem has as much to do with keeping the BJT turned off in the presence of a large signal as it does with handling the on case where the signal across the BJT is quite small. One can only reverse bias a junction so far without it breaking down. I'm guessing that this is the reason for preferring an implementation at a virtual earth point, i.e. where the signal level is always low.
Indeed and this is one particular conundrum that others have pondered in another thread. Wayne's volume control claims to have a factor of 10 improvement in the area of maximum signal level so presumbly has a novel solution to this particular problem. Presumably this must involve additional complexity (hence the SMD solution) but so far nobody has even speculated as to what this might be. Any ideas?
Hi Ian,
it might be that Wayne´s claim is that always at least one BJT is
turned on. In contrary to the before shown circuit it is also possible
to use not say 10 shunt resistors alternately but connect
them in series and switch the in between connections to gnd. So
there is always some attenuation present thus leading to more
headroom for the input signal of the pre.
Uli
it might be that Wayne´s claim is that always at least one BJT is
turned on. In contrary to the before shown circuit it is also possible
to use not say 10 shunt resistors alternately but connect
them in series and switch the in between connections to gnd. So
there is always some attenuation present thus leading to more
headroom for the input signal of the pre.
Uli
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.