Capacitor in sidechain release, bipolar or polarized?

[Flundran]

In a compressor sidechain circuit, where a capacitor in parallell with a resistor is the release component... Does it matter what type of capacitor you use? Would a non-polarized polyester work as good as an electrolytic?

/F

 

[JohnRoberts]

In a compressor sidechain circuit, where a capacitor in parallell with a resistor is the release component... Does it matter what type of capacitor you use?

no, not typically.

Would a non-polarized polyester work as good as an electrolytic?

/F

Yes... you can usually substitute NP for polar, its the other way that requires consideration for DC effects.

JR

 

[jdbakker]

Would a non-polarized polyester work as good as an electrolytic?

All things being equal a polyester capacitor will likely have less leakage current than a electrolytic cap. This will impact release time, if the cap leakage current is comparable to the current through the release resistor (for this reason the GSSL specs tantalum capacitors, as those traditionally have less leakage current than aluminium ones). Whether this is good, bad or negligible is a matter of taste.

you can usually substitute NP for polar, its the other way that requires consideration for DC effects.

+1.

There are almost never technical reasons to pick polarized over NP. The reason we don't use NP caps for everything is that for a given value/voltage rating polarized electrolytics are smaller and cheaper.

JDB.

 

[Flundran]

Ok, thank you guys!  8)

 

[JohnRoberts]

There are almost never technical reasons to pick polarized over NP. The reason we don't use NP caps for everything is that for a given value/voltage rating polarized electrolytics are smaller and cheaper.

JDB.

OK this is a bit of a stretch, but in time constant circuits where capacitors are charged up by one impedance, and discharged by another, the dielectric absorption of that particular type of polar cap can make a difference. I once designed a CX record playback decoder, and very specifically used a tantalum polar capacitor in the playback side chain circuit, to mirror the behavior of the tantalum capacitor I knew was used in the encoder side chain circuit.

This is a rather obscure and esoteric reference.. For general side chain use that don't involve encode/decode accuracy, this is not a serious consideration. The errors caused by DA would be hard to detect even in my application.

JR

 

[dustbro]

Does it matter what type of capacitor you use?

I made the mistake of using lytics in place of the tants on the release section of my gssl once... It made the fastest release time around 5 seconds, and the slowest somewhere around 15 seconds. much much much slower than what i was expecting.
I would assume that it would do the same in a sidechain circuit?

 

[abbey road d enfer]

In fact, ignoring dielectric absorption (not because it is non-existant, but because the subject is worth its own thread), the most important aspect is the leakage current. The influence of leakage current is variable, depending on the actual circuitry involved. Most "peak" rectifiers, as used in compressors, use an RC circuit where both the cap and the res are connected to ground. In that case, the leakage current will act as a resistor in parallels with the physical res, which will speed up the initial discharge, but as the voltage decreases, the leakage current decreases too, and at some point will be negligible. So the discharge is not perfectly exponential. The trouble with this is that the actual release TC is faster for high levels than for low levels, which increases LF distortion.
There are other configurations of the RC circuit where the cap is submitted to a less variable voltage and as a consequence the effects of leakage current are less noticeable.
A particular case is RMS detectors, where the voltage is quite low, because it is logarithmically converted (there is as much variation of detected voltage between 1-10V than between 0.01-0.1V). In these, the effect of leakage current is almost non-existant. Due to the log conversion, the actual release time is slow for moderate level VARIATIONS, and fast for large variations.

 

[JohnRoberts]

Agreed.. I suspect the difference noted by Flundran is indeed due to lower leakage in Tantalum caps.

In any halfway conservative design, even using aluminum, the designer should scale impedances and factor for leakage so that it is not a factor impacting time constants. That said, they don't use electrolytics in sample and hold circuts for a reason...

Data sheets used to give more useful characterizations about leakage current. I suspect there are variations in the newer, lower impedance, smaller package, etc technology. I ran into one application where the current leakage of a new cap series caused unacceptable noise in a phantom Voltage blocking application.

Trust but verify...

JR