Moog Ladder Filter - Research Paper

[andy-cytomic]

So .. this kept bothering me, so I just put a simplified thing into that online circuit sim...
ext.png

With all defaults in the sim (except bumped the beta a bit just in case), this thing actually gives: 32uA and 31uA as the collector currents, with 63uA total over the 150k, which is basically balanced with bases at -6.4mV and -7.2mV. With "spice-default" instead of "default" for the transistors we have about 33uA vs. 28uA which is still pretty close to balanced.

I don't know. The fact that the two transistor models in the sim disagree to a certain extent suggests it could be a case of "doesn't seem like the math is works out here, let's tweak.. oh right now it's closer" and the design is actually nowhere near as imbalanced as it might seem. Or it could just be that this poor sim is not realistic enough.

Grab LTSpice, it's free and as detailed as you want to get, win and macos

[mystran]

Grab LTSpice, it's free and as detailed as you want to get, win and macos

I've had LTSpice on desktop and old laptop since forever (and that's what I usually use to validate designs before I actually try building stuff) ... the editor is horrible to use on a touchpad so I've never got around to installing it on this thing.

[urosh]

Oh, you're probably right... I must have had a brainfart thinking there was some sort of a voltage divider... though that still doesn't answer why the odd values for what seems like just feedback.

Any idea though why the asymmetric bias though? For gain?

Likely did not care or omission. 1:2 is not terrible imbalance, relevant for high fidelity mic preamp or poweramp, not here.

So .. this kept bothering me, so I just put a simplified thing into that online circuit sim...
ext.png

With all defaults in the sim (except bumped the beta a bit just in case), this thing actually gives: 32uA and 31uA as the collector currents, with 63uA total over the 150k, which is basically balanced with bases at -6.4mV and -7.2mV. With "spice-default" instead of "default" for the transistors we have about 33uA vs. 28uA which is still pretty close to balanced.

I don't know. The fact that the two transistor models in the sim disagree to a certain extent suggests it could be a case of "doesn't seem like the math is works out here, let's tweak.. oh right now it's closer" and the design is actually nowhere near as imbalanced as it might seem. Or it could just be that this poor sim is not realistic enough.

Hi, i'll chek on Monday in sim when i'm back home, but you could try something, change resistor at q27 base from 100k to 200k, and see how ltp balances

[mystran]

Hi, i'll chek on Monday in sim when i'm back home, but you could try something, change resistor at q27 base from 100k to 200k, and see how ltp balances

Makes no difference. LTSpice suggests what actually makes the most difference is the PNP and depending on the choice the asymmetry can be a bit more or a bit less than the web simulator. The NPNs also seem to matter a bit, but I'd say probably less.

I think there's basically two things to note. Using 2n3904 and 2n3906 models, we have 653.1mV or so over the 27k resistor (so slightly higher Vbe for the pnp than the 0.6V estimated) and (perhaps more significantly) the PNP base current adds another 4.5uA or so on top of the ~24uA over the resistor.

These choices give about 29uA and 34uA for the LTP, but simply swapping other transistors can make it a bit more balanced or unbalanced. I feel like when we're looking at a couple of uA this or that way, the design might actually be about as close to balanced as you can reasonably expect without matching or a trimpot.