1
\$\begingroup\$

I've been working on this project to sum and scale DC voltages from 2 sets of four 0-5V variable voltage dividers, into two separate mixes with identical circuits. In my previous version, I fed the voltages straight off the dividers without buffers and ended up having lots of problems with voltage droop, as the mixer was forming a parallel path to ground for each divider.

I added voltage follower configured op-amps to buffer each stage and that solved my issue. Now I get the correct maximum voltage when all stages are turned up fully, (about 5.09V) but I've run into another perplexing problem. Each summing mixer also has another input from an external source that is clamped to a voltage range of +/-3V, and is not scaled. When the signal coming into the mixer from the dividers is at it's maximum, (5.09V), and I apply a 3V external signal to this non-scaled input, the voltage at the mixer output doesn't respond as expected, only increasing a total of roughly 25mV from the 5.09V instead of the full 3V. If I attenuate the voltages at the dividers to offer more headroom then the circuit reacts normally to external voltage, adjusting the output as expected in response to the 3V input.

My question is: Why is this circuit preventing the output from going higher than 5.35V?

It feels like I'm reaching the limit of the op-amp swing, but the supply is +/-12V (minus 0.3V Schottky diode drop on each for reverse voltage protection). The op-amp is an LM324, so unlike the TL074, it should be capable of going to the negative rail, and I can't imagine the upper voltage limit being 5.5V below the positive rail. Is there something I'm missing with regards to this op-amp's output range?

Here is a picture of the schematic, is there anything else clearly amiss?

schematic

\$\endgroup\$
5
  • \$\begingroup\$ Can you show the desired expression? You say 'sum and scale' but never mention a scaling factor. \$\endgroup\$
    – Reinderien
    Apr 23 at 14:59
  • \$\begingroup\$ Or maybe the scaling is always going to be varied by potentiometer. \$\endgroup\$
    – Reinderien
    Apr 23 at 15:10
  • \$\begingroup\$ The summing is the sum of 4 voltage dividers, each with a possible range of 0-5V, thus the input range is 0-20V and then I am scaling that by about 1/4 (47.5Kinput resistor/ 12.1K feedback resistor) The second input branch of the summing circuit is not scaled (12.1K input / 12.1K feedback resistors) but it is intended to add a precision DC offset to the mix between -3V and +3V and is clamped with diodes to a ref voltage (on another page of the circuit made with resistor dividers into op-amps in a voltage follower configuration set to 3V and -3V). Scaling works, output just won't pass 5.3V max. \$\endgroup\$
    – Jesse Jo
    Apr 23 at 23:24
  • \$\begingroup\$ expression: Vout=((V1+V2+V3+V4)/4)+V5) (Where 0v<=V1::4<=5V, -3V<=V5<=3V) Set of possible output voltages intended to be possible for Vout is -3V to +8V) \$\endgroup\$
    – Jesse Jo
    Apr 23 at 23:31
  • \$\begingroup\$ R39-44 form a passive adder that intrinsically divides by four already. Have you taken that into account when further attenuating the signal? \$\endgroup\$
    – Reinderien
    Apr 24 at 15:11
0
\$\begingroup\$

I don't see why you need the stage at U7.2 at all, nor why U7.1 is set to invert. Also the ratio of R50 to R24,41,43,45 is "about" right (you can get closer to 4.00 than 3.93 with a different selection of resistors). Not showing the input followers, this should work with something that looks like

schematic

For your op-amps consider the onsemi NCV2007, which comes in four-channel devices that support your +/- 12V supply and are explicitly rail-to-rail.

If you want to stick to the 324 family, you might be able to get away with the ST LM324APT which will drop anywhere between 1.5V-3V from the upper rail. With Vpp=12V and a maximum expected 8V out, maybe you're safe.

\$\endgroup\$
1
  • \$\begingroup\$ Thank you for the op-amp recommendation, I really appreciate it. I see what you mean about not needing the second stage, I'll give this configuration a go and see if it solves my issues. Thanks again. \$\endgroup\$
    – Jesse Jo
    Apr 24 at 21:58
1
\$\begingroup\$

Your mixers divide by 4 which limits your output signal from each input. If you wanted unity gain, you simply change all the 47.5k to 12.1k. Of course if each input were maximum then the sum of all inputs may exceed the supply range at some point if the peaks are in sync, so you need specs for “ headroom “ to limit the maximum input.

The Schottky Diodes clamp the input at ~ 350mV @ 1mA above 3V and below -3V with unity gain after this.

  • the LM324 may get close to Vee but has a 3V drop from Vcc (full temp range with 10k load) due it’s high side Darlington driver and bias. See [datasheet1 “Electrical Charactersistics for Voh=27V typ. and Note(1) Vcc=30V

  • therefore you have designed it for a +/-3 V + diode clamp Vf. If you wanted 8V then you would need to change the clamp diode reference voltage or R input ratios or output gain.

\$\endgroup\$
1
  • \$\begingroup\$ if the max at each of the 4 stages is 5V the sum is 20V, the scaling of the first inverting mixer stage divides those be 4 down to the Max of 5V, but the +/-3V external signal isn't scaled (notice it's got 12.1K input resistance) so the Maximum Voltage expected at the output is intended to be 8V, but I can only seem to get 5.3V. It shouldn't be a problem if I input 20+V if I am using a feedback resistor to scale the opamp output to something that is within it's voltage range, right? My issue here is that it seems like it should have plenty of headroom to get up to 8V, but it doesn't. \$\endgroup\$
    – Jesse Jo
    Apr 23 at 14:58

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.