My crosspoint switch is in dire need of a major design rework. Any suggestions on design would be greatly appreciated. This is my first major overhaul.
The noise I'm trying to eliminate is hiss. I replaced the orig. 78xx series regs. with three pin, fixed value low noise regulators. This was a significant improvement, but the noise is still objectionable.
About the Ckt.
The 16 channel switcher uses (8) 5532 op amps, each in an inverting config to drive unbalanced input signals from an electric guitar and effects pedals (1.8 - 2 vpp with high impedance inputs). The signal path is from input jack to input op amp to crossspoint chip to output op amp to output jack. Schematic here
Op amp supply rails are +/- 5 VDC for the input amps and +/- 18 VDC for the output amps. The crosspoint chip can only handle signals from 0 to 10 volts, so bumping the input amp supply voltage and signal levels higher aren't options.
Originally, the input signals were cut by a factor of 4 and boosted by a factor of 4 at the output. I recently changed the gain config. to unity gain (in and output) and increased the input Z from 40K to 1M with fdbk resistors.
This improved the signal transparency, but the noise is unacceptable (as might be expected with this high of an input Z).
Design Considerations
The input impedance to each channel needs to be 1M for tone transparency
The DC supply on the input amp side can't change (input signals can go as high as 8VDC with boost type effects pedals)
The existing ckt. board doesn't have a ground plane
Use a non-inverting voltage follower design with unity gain
Questions:
Should I replace the 5532s with OPA2134s? The 5532s only have an input Z of 300K, which is too low.
What other chips might work better?
What is the best rework method to use on the existing ckt. board? I was planning to use chip sockets to facilitate a dead bug style rework.
Any other suggestions or things to keep in mind?
Can I expect a significant reduction in noise given the changes I'm planning - enough to make it worth doing?
Thanks for sharing your expertise!
The noise I'm trying to eliminate is hiss. I replaced the orig. 78xx series regs. with three pin, fixed value low noise regulators. This was a significant improvement, but the noise is still objectionable.
About the Ckt.
The 16 channel switcher uses (8) 5532 op amps, each in an inverting config to drive unbalanced input signals from an electric guitar and effects pedals (1.8 - 2 vpp with high impedance inputs). The signal path is from input jack to input op amp to crossspoint chip to output op amp to output jack. Schematic here
Op amp supply rails are +/- 5 VDC for the input amps and +/- 18 VDC for the output amps. The crosspoint chip can only handle signals from 0 to 10 volts, so bumping the input amp supply voltage and signal levels higher aren't options.
Originally, the input signals were cut by a factor of 4 and boosted by a factor of 4 at the output. I recently changed the gain config. to unity gain (in and output) and increased the input Z from 40K to 1M with fdbk resistors.
This improved the signal transparency, but the noise is unacceptable (as might be expected with this high of an input Z).Design Considerations
The input impedance to each channel needs to be 1M for tone transparency
The DC supply on the input amp side can't change (input signals can go as high as 8VDC with boost type effects pedals)
The existing ckt. board doesn't have a ground plane
Use a non-inverting voltage follower design with unity gain
Questions:
Should I replace the 5532s with OPA2134s? The 5532s only have an input Z of 300K, which is too low.
What other chips might work better?
What is the best rework method to use on the existing ckt. board? I was planning to use chip sockets to facilitate a dead bug style rework.
Any other suggestions or things to keep in mind?
Can I expect a significant reduction in noise given the changes I'm planning - enough to make it worth doing?
Thanks for sharing your expertise!
Last edited:
The basic issues have already been discussed in your old thread, I see - including input Z.
You will have to make the input circuit non-inverting. In addition, the high source impedance (easily in the 10s of kOhms) mandates an opamp with low input impedance nonlinearity. You'll need to consult Samuel Groner's opamp measurements. However, I can tell you right away that the venerable NE5532 already is among the better choices in this regard. Since you're not working with high gains here and presumably have a coupling cap somewhere, offset caused by input bias current shouldn't be much of an issue either. I presume your NE5532s will be the classic Signetics parts, which are considered better than some other variations.
Overall, don't worry about opamps too much. You main problem right now is that you have a noise resistance of half a kOhm (translating to 90 nV/sqrt(Hz) input-referred) with a noise gain of 2.
Make sure you have decent supply decoupling and install some extra caps near the opamp's supply pins if necessary.
If changes to the power supply made a big difference, possibly it also is the switching IC or surrounding circuitry that's the problem. One invariably has to shift DC levels from 0 V to +5 V here, and there are smarter and less smart ways of doing that. Can you reverse-engineer the circuit? (Including any components between input opamps and IC inputs.)
Have you already replaced the electrolytics used for +10V supply decoupling (those after the regulators, at least)?
You will have to make the input circuit non-inverting. In addition, the high source impedance (easily in the 10s of kOhms) mandates an opamp with low input impedance nonlinearity. You'll need to consult Samuel Groner's opamp measurements. However, I can tell you right away that the venerable NE5532 already is among the better choices in this regard. Since you're not working with high gains here and presumably have a coupling cap somewhere, offset caused by input bias current shouldn't be much of an issue either. I presume your NE5532s will be the classic Signetics parts, which are considered better than some other variations.
Overall, don't worry about opamps too much. You main problem right now is that you have a noise resistance of half a kOhm (translating to 90 nV/sqrt(Hz) input-referred) with a noise gain of 2.
Make sure you have decent supply decoupling and install some extra caps near the opamp's supply pins if necessary.
If changes to the power supply made a big difference, possibly it also is the switching IC or surrounding circuitry that's the problem. One invariably has to shift DC levels from 0 V to +5 V here, and there are smarter and less smart ways of doing that. Can you reverse-engineer the circuit? (Including any components between input opamps and IC inputs.)
Have you already replaced the electrolytics used for +10V supply decoupling (those after the regulators, at least)?
Thanks for the comments, guys. The op amps are Signetics parts from the late 80s. The source Z (guitar pedals) varies from 470 ohms on newer pedals to ~10K on older ones. Not sure what the source Z of a guitar pickup is.
The crosspoint switch p/n is a CD74HC22106E. Also, I replaced all caps recently - even decoupling caps.
sgrossklass -
I'm not sure what you mean by reverse eng. the ckt. The posted schematic shows the current design. The is nothing between the crossbar inputs/outputs and the op amps.
I asked about the op amp choice because I have a different switching unit that has a similar buffer ckt. on its input but imparts no noise. The design is very simple and what I would use here since I have to switch to a non-inverting config. anyway - .1uf coupling cap on non-inverting input of OPA2134 with 470K resistor from input to gnd. The output drives a small load resistor and the inverting input for fdbk. The 2134 has a very high input Z - so the input Z is determined by the input resistor. I would use a 1M.
The crosspoint switch p/n is a CD74HC22106E. Also, I replaced all caps recently - even decoupling caps.
sgrossklass -
I'm not sure what you mean by reverse eng. the ckt. The posted schematic shows the current design. The is nothing between the crossbar inputs/outputs and the op amps.
I asked about the op amp choice because I have a different switching unit that has a similar buffer ckt. on its input but imparts no noise. The design is very simple and what I would use here since I have to switch to a non-inverting config. anyway - .1uf coupling cap on non-inverting input of OPA2134 with 470K resistor from input to gnd. The output drives a small load resistor and the inverting input for fdbk. The 2134 has a very high input Z - so the input Z is determined by the input resistor. I would use a 1M.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.