B1 Active Crossover

[wintermute]

Thanks Manfred I've been having some fun with ltspice (which I've only just started to learn), not getting particularly useful results, but I suspect that is more the user than the program I'll see how I go with the one you have posted, I suspect I really need to just build some stuff and measure it!

Tony.

[wintermute]

Well I've done a lot of reading, brushed up on my algebra, played with ltspice a lot, and finally modeled something that I think might be workable. My planned crossover freq is 200Hz. (which seems to be right as the curve is down just over 6db at 200Hz).

Screen shot of the results of comparing an opa134 based gyrator to a passive circuit with a capacitor and real inductor. both curves are almost identical so I think the circuit is working as intended (note the phase droops at the low end due to the effective series resistance of 100 Ohms in the simulated coil, something that I added as a parameter to the "real" coil to get a like for like comparison)

Now I guess I just have to get round to actually building one!!! but at least this is some progress....
I've also attached the ltspice file of this model in case anyone is interested

Tony.

[AndrewT]

Hi,
did you say you added 100r as the resistance of the real inductor to make it match the gyrator?

Am I also correct in assuming that since this is a high pass 200Hz that the error that becomes measurable <8Hz will be completely inaudible?

Is it also correct to assume that since the 134 is in the simulated inductor path, it has virtually no audible effect on the passband signal?

All we "hear" in the upper frequencies are the effect of two B1 follower stages.

And finally,
thanks Winter.

[wintermute]

Hi Andrew, I'm just learning so I can't answer with authority, but here goes:

yes I added 100 Ohms as the series resistance in the "real" coil's parameters in spice. Attached is an expanded plot with the "real" coil having a series resistance of 1 Ohm (which is probably lower than possible for a 35H coil!)

Am I also correct in assuming that since this is a high pass 200Hz that the error that becomes measurable <8Hz will be completely inaudible?

The error actually starts quite a bit earlier than 8 Hz but at 10Hz it is only about 3 degrees... at 20Hz it looks like about 1 degree, so I guess effectively yes the error should be inaudible

Is it also correct to assume that since the 134 is in the simulated inductor path, it has virtually no audible effect on the passband signal?

I've read a fair bit about FDNR (this gyrator is like one) circuits in the past, and whilst some claim that they aren't in the signal path, others disagree, but I suspect that the effect will be lesser than if the signal was actually passing through the opamp.

All we "hear" in the upper frequencies are the effect of two B1 follower stages.

that's the plan! I didn't like the idea of using a sallen key circuit and had been planning on trying something different, I was considering doing something with fdnr circuits, but on page three of this thread, D to the G posted a link to SY's acheron circuit, and this is the above post is the result of me reading enough to be able to understand the various bits and work out how to use LT spice to simulate it. Needless to say it has been a valuable learning experience!!

But until I actually build one and measure it, it is still speculation as to whether it will work...

Please don't hold your breath on that one, I tend to take a while!! but I have at least got the fets, and have roughly sorted them into ranges. I just have to get over the obsessing over other components and order the rest of the bits

And finally,
thanks Winter.

Your welcome!!

Tony.

[JimT]

wintermute,

Please post your opa134.sub and ?.sym files.

Thanks
Jim

Originally Posted by wintermute

Well I've done a lot of reading, brushed up on my algebra, played with ltspice a lot, and finally modeled something that I think might be workable. My planned crossover freq is 200Hz. (which seems to be right as the curve is down just over 6db at 200Hz).

Screen shot of the results of comparing an opa134 based gyrator to a passive circuit with a capacitor and real inductor. both curves are almost identical so I think the circuit is working as intended (note the phase droops at the low end due to the effective series resistance of 100 Ohms in the simulated coil, something that I added as a parameter to the "real" coil to get a like for like comparison)

Now I guess I just have to get round to actually building one!!! but at least this is some progress....
I've also attached the ltspice file of this model in case anyone is interested

Tony.

[wintermute]

Hi Jim,

I've added opa134.sub and opa134.asy to the original zip file and re uploaded here (should have thought about that in the first place!), note that it is an asy file in ltspice but I suspect it is what you are after as It is under the sym directory.

The .sub was the pspice .mod file I downloaded from TI (hopefully accurate), the .asy was modified (simply by changing the SYMATTR Value, and SYMATTR Description lines) from the generic opamp2.asy file under sym/opamps

I'm a real newbie with ltspice so if I've done anything really dumb please let me know!!!

Thanks,

Tony.

[HiFiNutNut]

Originally Posted by AndrewT

the filter should see near zero ohms feeding the input and near infinity ohms as load on the output.

A standard B1 gets close if you don't hang anything else on the B1's input/output.

Unfortunately B1 needs some resistance on it's output to be stable. some have reduced the 1k0 to 220r and find that works. Maybe a bit of experimentation on that to find what works. But whatever, the next R, of a low pass filter, must take the source resistance into account when assembling the active filter.

Andrew,

What is the imput impedance and output impedance of the standard B1 using 2sk170 JFETs?

Thanks,
Bill

[AndrewT]

the input impedance of a B1 is set by the input resistor to ground. If this is omitted the DC input resistance is very much higher.
There will be some capacitance at the jFET gate but this too will result in a very high input impedance at audio frequencies. The B1 with nothing attached to it's input quite effectively meets the infinite input impedance that filters require to see.

The output impedance is the sum of the Stage output impedance and the output resistor. The standard B1 uses 1k0, giving a total output impedance of ~1050ohms.

The 1k0 output resistor can be reduced. As in the DCB1 220r is used giving a total output impedance of ~270ohms.

This does not meet the zero ohm source impedance that filters require to see.

An active low pass filter starts with a resistor feeding the remainder of the circuit. All that is required is to measure the actual output impedance of the B1 and subtract this from the theoretical first filtering resistor, so that the filter sees a total R as the feed resistor.

Not as easy for the high pass which starts with a series capacitor as the first component of the filter.

[HiFiNutNut]

Thanks, Andrew. So I presume the jFET has an output impedance of 50R.

[AndrewT]

Originally Posted by HiFiNutNut

Thanks, Andrew. So I presume the jFET has an output impedance of 50R.

very approximately.
I think it varies significantly with Idss of the chosen jFETs.
Higher Idss should give lower output impedance.