Jason Butcher and I am working on a single-axis rotator board based on the Cadet X multiplier.
Since we want everything to be four quadrant, we are centering the 1V p-to-p ramps at 0V (so -0.5V to +0.5V). Therefore we would like the offsets in the CV inputs to be bipolar as well (trimmer between +2.5V and -2.5V).
If there are four multipliers on each board (plus several op amps), can we buffer a global 2.5V using a single op amp for the entire board?
(I suspect yes, since there is no loading apart from the trimmers, but maybe they need to be isolated from each other)
The more complicated question: can we simply op-amp invert the +2.5V from the TL431 to get -2.5V, or should we make a separate TL431 “flipped upside down” to get -2.5V to be buffered by one (or more) op amps?
Not sure exactly what the requirements for precision and stability for the 2.5V references are in the design (and the consequences of deviating from them) but the only sources of error introduced in the inverter as opposed to a non-inverting op-amp buffer are from the resistor divider (relative precision and temperature dependency).
If i’m understanding you correctly, the negative reference is only used for the trimmers, and as such already dependent on the trimmer resistance in the same way.
On the other hand, the offset signal is effectively inverted twice along the signal chain, so you should also be able to shift the trim signal down by adding a positive reference into the negative terminal of U3.2 (and possibly modifying the R21/R22 divider to increase the range of the trimmer)
The idea is to be able to manually trim any small voltage offsets introduced by any of the op-amps along the way in each rotator module, since offsets in the signals will shift the center of rotation, and because there will be three rotators in series with any given axis passing through two of them. See block diagram. It seems to me the best way to do this is to make the trimmers bipolar, then we don’t need to fuss with other values in the multiplier.
Either way, if I’m reading the LT1251 datasheet right, the answer to 1 seems to be yes, and the answer to 2 is that you could get the same result with fewer opamps by doing as suggested above, but i think just inverting the 2V5 for the other end of the offset would also work.
If you do that, you can also swap the inverting buffer for a non-inverting one, since you now have a full range bipolar input to begin with.
Is “Y_IN” connected directly to a jack? I do not recommend connecting the input jack to the negative input of an op amp. Every video module will have a 470-499 ohm series resistance which will be added to your 10K Series input resistor. The gain will not equal -1 exactly. Use a 100K resistor from your jack to ground and connect it to the positive input of the input buffer.
You may want to, yes. Buffering a single 2.5V reference will grant the benefit of trimming it at one point.
I suggest using one half of a TL072 to buffer the 2.5V signal and the second half to invert it. No need for a second TL431. You may add a trimmer on the feedback pin of the TL431 and another trimmer on the feedback pin of the inverting op amp.
An alternative reference solution was shared by Lars with claims of improved noise.
Thanks - yes, some of those are known issues (opamp in R’s missing, mistakes on input routing) as I was trying to get as much on the page as possible in a short time frame. I do appreciate the tips re: ref voltage and opamp configuration and will post an updated schematic shortly. Thanks again!
Use a 100K resistor from your jack to ground and connect it to the positive input of the input buffer.
If I’m understanding correctly, You’re suggesting an additional non-inverting stage ahead of what is in the drawing (as U1A/U1B) with corresponding resistor omissions - as otherwise including the offset via a non-inv / inv pair would result in an inverted signal on the multiplier input (e.g. U5A)?
I assumed they were supposed to be hand trimmed before use, not as pots. The precision appears to be very important here.
I’ll look it over more thoroughly again shortly.
Consider this, if you work with the voltage as +/-1V instead of +/-0.5V, then you can attenuate the signal at the end instead of 4-times before the LT1256’s. Four of the buffers are redundant. Sorry for the bad photoshopping: