Help with Gyrator circuit equivalence | opamp vs bjt

[Carlos Best]

Guys
Im planning to clone a stripped version of the EHX Metal Muff leavin just one pot and the first 4 stages only
Doing that theres a gyrator circuit made with half an opamp left
My intention is to replace that with a BJT for PCB real state matters
So

[Carlos Best]

Anyone?

[PBE6]

http://www.muzique.com/lab/gyrator.htm

[Transmogrifox]

Yes the BJT and OP AMP implementation are approximately similar.  The BJT implementation by nature has a higher equivalent output impedance at the emitter (lower gain), so the Q will be slightly lower, resonant frequency might be slightly lower.

My opinion is the difference will be inaudible.

[bool]

If you use spice, you can compare both and tweak. Note that both circuits are your basic sallen-key hpf when you look at it from a certain angle so there's a way to easy tweak.

If you just want to avoid the hassle nad a "quick fix", use something like a MPSA13 and a lower emitter resistor value (R15) to optimize the impedances (both In and Out) and the Q.


Btw. why is it "good" to post inverted color schemes on forums? Is that "fancy"? It is a PITA to read. Initially I was reluctant to even bother looking at it. While it is true that this has to to with the way I have my monitor set up (low-key), most likely I'm not the only one who is just a little too reluctant to waste time "trying to read" something.

Moral of the story: It is you who is interested in finding an answer, not "the rest of us". Make it easy for the "rest of us" to help you.

[ggedamed]

Another resource: Active Filters Using Gyrators - Characteristics, and Examples - Simplified Version [Elliott Sound Products]

And this is how it looks simulated in Tina-TI:

[Carlos Best]

Great Guys!
I just needed someone to point me at the right direction.
I found several pages talking about gyrators but none of them told me that.

Looking at the plots Ggedamed posted there is a -10db difference between circuits, and the BJT has a slightly lower frequency
Just as Transmogrifox pointed out,

I managed to make a square 1 1/2 inch side layout, i'll check it and post it here when it is verified.

BTW Sorry about the inverted color scheme, thats how my Eagle CAD is set up.
Should be a sticky post with netiquete to post schems.

[Transmogrifox]

BTW Sorry about the inverted color scheme, thats how my Eagle CAD is set up.
Should be a sticky post with netiquete to post schems.

I think this can be a matter of taste.  I personally prefer the black background because it hurts my eyes less.  I often work my circuit simulations with black background so this is a format that is familiar to me, thus easy to read.

Opposite to bool, I was one on the other hand more likely to look at the schematic because it didn't glare at me in a painful way

I don't know what proper netiquette is in this case.  For some clearly it is difficult to read, and then for others it is easier to read...

[PRR]

> inverted color schemes on forums

I try to bite my tongue.

I absolutely do not understand why schematics need every part a different color. I am only mildly interested in non-circuit bits (notes, arrows) being different color. Having lived through real blueprints, I sure am glad it is possible to do black ink on white paper.

This is a rant against schem-draw programmers, not the people who use them.

Interestingly, *this* schematic has no white, only half-white. When I do a Negative, the half-white text becomes half-black. It appears they do not use half the dynamic range from 100% to 0% brightness. Or like their pen is running out of ink. Cranking the Gamma to 2 (should be 1) gives much better text, though the bold red becomes a washed-out pink.

Getting back to the electrons......

The one-BJT job has at least 30 Ohms emitter resistance, which is not-small compared to the quite small 100 Ohm "coil resistance", so the Q and F are shifted. The '072 at unit gain has near 1 Ohm output resistance. Higher BJT current would get the emitter resistance down.

I am wondering about LARGE signal performance. Ultimately the "tuned circuit" comes down to a 100 Ohm series impedance. But if this has to be driven by a BJT with 10K bias loading, or a '072 with 2K minimum happy load, large signals may distort. In this case, maybe no, since the simulated source is a 10K resistor, and if source voltage is less than supply voltages, it won't run out of current. However I would generally prefer to use >1K here (as in Carlos' 2.2K), or do more detailed overload studies.

[bool]

You weren't paying enough attention.


To get both gyrators to perform almost equally, you must do this:

1. Use a darlington, like a MPSA13
2. Change R14 to 120K
3. Change R15 to 4K7 (or down to 3K3 for larger signals)
4. Change R18 to 1K8


A simulation with a MPSA13 model and above values aligns both circuits responses to within a -0.35dB and almost complete phase response overlap.

More than good enough for rock'n'roll IMHO.

[duck_arse]

BTW Sorry about the inverted color scheme, thats how my Eagle CAD is set up.
Should be a sticky post with netiquete to post schems.

I think this can be a matter of taste.  I personally prefer the black background because it hurts my eyes less.  I often work my circuit simulations with black background so this is a format that is familiar to me, thus easy to read.
Opposite to bool, I was one on the other hand more likely to look at the schematic because it didn't glare at me in a painful way
I don't know what proper netiquette is in this case.  For some clearly it is difficult to read, and then for others it is easier to read...

^ +1, me too. [/topic] proper netiquette? what would be the point, no-one hardly etiquettes anything in any walk of life now, do?

I take great pains and time to hack local style-sheets such that I am not blinded by the dreaded white pages all over the innernets. (it doesn't waork, but doesn't stop me trying.) and I use colour in my schemming so I can see/differentiate the things in and on my circuit. in future, I will include a link to a white-backed circuit copy. [topic]

[Transmogrifox]

More than good enough for rock'n'roll IMHO.

That about sums it up. 

I also did a simulation using a BC846B BJT and an LT1022 op amp (reasonable performance class as TL072 and similar opamps we use in the stompbox circuits).  I set center frequency to 1.083 kHz (nominal based on equivalent LC network). 

Each of these circuits were put into a typical op amp feedback creating a peaking filter similar to a typical EQ configuration using exactly the same resistor value in feedback loop.
Here are the results:

BJT Gyrator Circuit
Fc = 1.085 kHz
Gain = -0.1 dB
Q = 2.7

Op Amp Gyrator Circuit
Fc = 1.084 kHz
Gain = 2.24 dB
Q = 3.75

I think when using a normal BJT the equivalent base AC resistance works in parallel with the base bias resistor to help shift the frequency up somewhat offsetting the effect of increased emitter AC resistance.

As a result my simulation shows no change in center frequency for the component values chosen, and a 3 dB reduction in peak gain (28% change in bandwidth). 

To me a direct swap-out of BJT for op amp is good enough for rock'n'roll.

[bool]

You better use a darlington. Then you can squeeze out a near-identical response .. with some tricks. It's cheap, and works exactly as advertised - in every meaning of that word.

[Carlos Best]

Trade compromises.

The darlington circuit is overkill, but the circuit has 3 other BJT gyrators, so I prefer to compromise a little of the response of the last one and use the same part number for all
Or maybe change the other 3 to MPSA13, but that would change the original response.

It is all a matter of OCD 

[thehallofshields]

Yes the BJT and OP AMP implementation are approximately similar.  The BJT implementation by nature has a higher equivalent output impedance at the emitter (lower gain), so the Q will be slightly lower, resonant frequency might be slightly lower.

Can I just add the usual Common Collector Zout (300 ohms) to R2 and call it a day?

[Rob Strand]

Can I just add the usual Common Collector Zout (300 ohms) to R2 and call it a day?

It would be close.  As R2 gets small, or R1 large,  you have to be more careful.  I tend to use spice simulations.
However, I suspect  R = 60 + R1/hfe might end-up being a closer value to add to R2.

[thehallofshields]

It would be close.  As R2 gets small, or R1 large,  you have to be more careful.  I tend to use spice simulations.
However, I suspect  R = 60 + R1/hfe might end-up being a closer value to add to R2.

I'm using the values from Geofex, and getting almost no effect.

C1 = 22n
C2 = 10n
R1 = 47k-1M
R2 = 470

I know my Differential Amplifier is working properly because I get a huge sweep when I connect a simple LPF.
The BJT getting normal voltages, oriented correctly.

Any thoughts?

[Rob Strand]

I know my Differential Amplifier is working properly because I get a huge sweep when I connect a simple LPF.
The BJT getting normal voltages, oriented correctly.

Any thoughts?

Do you have a link the rest of the circuit? 

The two values marked "Re" on this circuit are required and affect the behaviour,

[thehallofshields]

Circuit is pretty standard, and I get a huge sweep when I connect the pot wiper to 100n cap to ground.

With a BJT (MPSA13 or 2N3094) and the suggested values, I get almost no effect.

[Rob Strand]

With a BJT (MPSA13 or 2N3094) and the suggested values, I get almost no effect.

With 2.7k resistors on the opamps you will struggle to get those transistors working.
The 2N3904 has a low gain.   The MPSA13 is a Darlington which can introduce problems for this circuit.
Really you need a high gain normal transistor like MPSA18, 2N5088, 2N5089, BC549.

You might get it to work with the 2N3904 if you increase the 2.7k's to maybe 10k or 22k.   If you do that, you will have to increase R2 by a factor (10k/2.7k) or (22k/2.7k).  Then you will have to reduce C1 by a factor (2.7k/10k) or (2.7k/10k).

If you are using variable R1 going upto 1MEG then that's going to create more problems.  Probably won't work well even with the high gain normal transistor.  To be honest, with variable R1 replacing the opamps with a BJT probably isn't a good idea.  It isn't possible to tweak the circuit to the extent it needs to work well with R1 at 1MEG.