I have been using an old french 4 way active crossover for a long time - Power FEP 204 - and it works very well, except for the extra hiss which it introduces into the system. I had some very good advice on solving a similar issue with a Linsley Hood integrated amplifier - changing some old 1980s op amps did the trick. This circuit contains six RC4156DB and one R4559NBS. Does anyone have any views on these or anything else as a possible culprit?
Presumably this is between a signal source (or preamp) and several power amps in your system? What's the input sensitivity of the power amps connected downstream of the active XO?
The reason for asking is its rather tricky without a schematic to figure out whether the opamps or the resistors around them are going to dominate the noise. But if your poweramps are quite sensitive relative to the maximum unclipped output level of the AXO, you may get a lower noise level in practice by not modding your AXO but rather by running it at a higher input signal level and then attenuating its outputs.
The reason for asking is its rather tricky without a schematic to figure out whether the opamps or the resistors around them are going to dominate the noise. But if your poweramps are quite sensitive relative to the maximum unclipped output level of the AXO, you may get a lower noise level in practice by not modding your AXO but rather by running it at a higher input signal level and then attenuating its outputs.
Thank you! The power amps get swapped around sometimes, but the current setup (using a Quad 33 preamp modified with a 12v battery power supply) has two Yamaha MX-35 for upper mid and high, the power amp section of a Yamaha AX-550 for low mid, and a pair of French Power 500w mono blocks for bass. If the crossover is turned off there is virtually no hiss at all from any of the speakers - all very high sensitivity - with the gain at max on the MX-35s, while the other two have no gain controls. I have just tried turning all the crossover outputs to low and boosting the gain from the preamp and there is definitely a much lower level of hiss. I can certainly live with it - I have been used to not going beyond 3 on the Quad output, and there is loads and loads of headroom. I'm happy not to change anything on the crossover unnecessarily, but just wondered if there were any components which had been improved since the 80s
Carbon film resistors may be creating some excess noise. The opamps are 10nV/√Hz typically which is rather high these days, and the resistor values may be higher than would be used in a modern low-noise design.
Modern low noise opamps may require better high-frequency decoupling than exists in that design though, so simply replacing with lower noise devices may bring extra problems - if you have an oscilloscope you can check for this.
As for devices there are lots of candidates with more like 3nV/√Hz or so to choose from, but the quad opamp may be tricky as modern devices are often dual only.
Given the pots I'd recommend JFET opamps, OPA1656 comes to mind as a good all round performer.
I'd hate to replace all the resistors with metal film, lots of work and not clear how much improvement that might bring, no doubt some though.
Modern low noise opamps may require better high-frequency decoupling than exists in that design though, so simply replacing with lower noise devices may bring extra problems - if you have an oscilloscope you can check for this.
As for devices there are lots of candidates with more like 3nV/√Hz or so to choose from, but the quad opamp may be tricky as modern devices are often dual only.
Given the pots I'd recommend JFET opamps, OPA1656 comes to mind as a good all round performer.
I'd hate to replace all the resistors with metal film, lots of work and not clear how much improvement that might bring, no doubt some though.
Opamps and resistors have improved in noise terms over those fitted in your AXO. However chances are substituting the lowest noise compatible opamps won't lose you more than 6dB of noise as I'm seeing plenty of resistor values in the 33k region in your pics - such a value is already twice as noisy as your opamps. I'm in agreement with @Mark Tillotson that replacing resistors isn't really practicable and it may irrevocably damage your PCB. If your poweramps are 1V RMS input sensitivity and the AXO can deliver up to 10VRMS (perhaps a stretch) unclipped then in theory you could get almost 20dB noise reduction from gaining up the inputs and attenuating the outputs. For output attenuators you could do worse than transformers like these : https://www.aliexpress.com/item/4001171636439.html. Admittedly not quite 20dB attenuation but getting there. I'd not use them backwards for step-up duty, likely the frequency response won't be very flat in that configuration.
Thank you both for your help, very much appreciated. There may be circumstances where I would be prepared to replace more than 150 resistors, but not these! All three Yamaha amps are 1V input sensitivity, and I'm not sure about the big Power APK5001s, but the speakers they drive don't reproduce hiss very well anyway. Turning the pots on the crossover right down seems to be the best solution. The MX-35s have input attenuators, so I can try using them too - they are just set to maximum at the moment, and also used with very efficient drivers meant for higher SPLs than my domestic installation permits (for 'domestic installation', read 'family'....)
I agree except for the transformer attenuation idea, more expensive and woefully under performant compared to resistor divider on the output, say 2k:220R.
I agree that trafos will have poorer distortion measurements and be considerably more expensive than resistive attenuators. However in my experience the perceived noisefloor is often lowered through use of trafos between units, likely through attenuation of common-mode noise.
Regarding the resistors: excess noise in the resistors can only contribute to the noise floor if there is a DC bias voltage across them, otherwise it only causes noise sideskirts around the signal. If everything is biased at 0 V plus or minus a few millivolts of offset, like it usually is in an op-amp circuit with symmetrical supplies, I would not expect resistor excess noise to affect the noise floor substantially.
Resistor thermal noise is another story, of course, but that only depends on temperature and resistance, not on whether the resistors are carbon or metal film types.
Resistor thermal noise is another story, of course, but that only depends on temperature and resistance, not on whether the resistors are carbon or metal film types.