Filter capacitors and ESR

[shattered]

I'm looking for snap-in capacitors measuring 25mm diameter and 35mm height for some Laney GH series amplifiers. One of the original parts is Samwha FL and the other one is Panasonic UP, both 100µF 500V.

I can't find the Samwha FL, and Panasonic UP (discontinued) datasheet does not specify ESR, so I suppose they are General Purpose.

I'm considering the EPCOS B43548A6107M060 as a reliable alternative, but the datasheet specifies it's very low ESR (datasheet in PDF format). Is it viable to install a small resistor in series to make it behave as a General Purpose capacitor? If so, what value and wattage would you suggest?

In the past, I had trouble using Panasonic FR caps in a circuit with linear regulators, so I learned that using low ESR capacitors with linear power supplies may not be a good idea.

-- edit. I found a possible alternative, Nichicon LGN2H101MELA30. Do you think this one would be ok? datasheet (PDF format).

[Snicksound]

I've never heard about low ESR being an issue in linear power supplies. Low ESR is a good thing, runs cooler.

Where did you hear that?

[R.G.]

I'm looking for snap-in capacitors measuring 25mm diameter and 35mm height for some Laney GH series amplifiers. One of the original parts is Samwha FL and the other one is Panasonic UP, both 100µF 500V.

I can't find the Samwha FL, and Panasonic UP (discontinued) datasheet does not specify ESR, so I suppose they are General Purpose.

Many datasheets specify a limit on ESR, but do it in the most obscure possible way. For instance, it may be specified as a dissipation factor or as "tan theta". Both relate to ESR and the ESR can be back computed as noted here: https://en.wikipedia.org/wiki/Dissipation_factor

I'm considering the EPCOS B43548A6107M060 as a reliable alternative, but the datasheet specifies it's very low ESR (datasheet). Is it viable to install a small resistor in series to make it behave as a General Purpose capacitor? If so, what value and wattage would you suggest?

In the past, I had trouble using Panasonic FR caps in a circuit with linear regulators, so I learned that using low ESR capacitors with linear power supplies may not be a good idea.

> it is viable to install a small resistance to fake a cap's ESR.
> value has to be guessed/estimated; probably best to guess-timate based on the ones you took out.
> difficulties with capacitor ESR and linear regs came up with the change from old-school bipolar linear regs and LDO types. Earlier LDOs used PNP or P-channel MOSFETs as pass transistors, and this added incremental voltage gain to the regulator's forward loop. That made the regulator's stability depend on load esoterica, including the ESR of any output capacitors. Many recent LDOs have been designed to make ESR not so sensitive an issue. A number I've used recently simply say to use a low ESR ceramic cap on the output. All of that is to say: it depends on the regulator you're supplying.

It might be good to look at the part number for the linear regs in the amplifier and consult their specs on what ESR will and won't work OK. Maybe you don't have to do anything special.

[shattered]

That's interesting information, thank you so much. The device in which the low ESR capacitors caused problems had a pair of 7805 linear regulators.

So low ESR would be fine for a tube amp filter caps, isn't it? Anyway I have more trust in the Nichicon brand and will probably use the LGN2H101MELA30 instead of the EPCOS. The amps are Laney GH50L and Laney GH100TI.

There are a pair of linear regulators in the circuit (7805 and 7905), but there I stick to Panasonic NHG (General Purpose and Long Life, AEC-Q200 certified), so it should be fine.

[R.G.]

That's interesting information, thank you so much. The device in which the low ESR capacitors caused problems had a pair of 7805 linear regulators.

That's a bit odd. The 7800 series is one of what I thought of as old-school without lot of ESR issues. I did once run into a situation where the self-inductance of the power distribution net was making a linear reg prone to oscillation, so yeah, there is a world of things that can conspire. But if it happened, it happened. A small MOX or wirewound should fix things.

So low ESR would be fine for a tube amp filter caps, isn't it? Anyway I have more trust in the Nichicon brand and will probably use the LGN2H101MELA30 instead of the EPCOS. The amps are Laney GH50L and Laney GH100TI.

Yeah, low ESR for tube amp supplies is fine. In fact, it's what we tried to get for everything back when I was designing power supplies for a living. We could not get ESR low enough. Switching power supply caps really pointed up the need for low ESR in a big way.

Is there a lot of low voltage logic stuff in those amps? DSP core, something like that? Our more difficult cases were with big high speed switching stuff that had to be remote from the DC power supply.

As an aside, I only got into the baby-sized supplies, things like 5V at under 50A or so. The big iron machines would sometimes run 2-3V at 200A for a logic supply.

There are a pair of linear regulators in the circuit (7805 and 7905), but there I stick to Panasonic NHG (General Purpose and Long Life, AEC-Q200 certified), so it should be fine.

Again, I'm a little surprised at the circuits even being sensitive to filter cap ESR. Is this on the first rectifier filter caps, a bypass before the regulator, or an output filter cap after the regulator? Most of the time, power supplies don't care much in terms of stability about the ESR of the first filter cap and not much about bypass before the regulator. It's the output filter that hangs on the outer end of the feedback loop that can get you issues.

One issue you can run into is just putting too much capacitance on the output of a regulator, maybe coupled with too little bypass out at the user loads. The regulator itself may be fine for suppressing switching noise at its output. A bigger output cap won't help much if the actually logic/switching load is some inches away over wires or copper traces. A much smaller local filter out at the load itself will have better suppressing effect, as the noise loop currents stay local, not running over the power and ground wires. I wonder if the ESR would matter as much if you cut the capacitance in half or a quarter.

[nuke]

I was looking for it, but came up empty handed, but TI used to have a really good application note article on the 7800 series regulators.

The older 7800/7900-series had a tendency to oscillate when the current limit function was triggered. TI also does not recommend a large filter capacitor on the output terminal, only a small 0.1uf ceramic.

I can't find the specific app note at the moment, but the TI data sheet for the 7800/7900 series regulators is pretty good and interesting reading, and applicable to other brands. I would also say, the datasheet's recommendations are largely ignored in practice. I've seen lots of large filters on the outputs in any number of equipment, and I don't think I've ever seen the recommended capacitor discharge diode to prevent reverse voltage on shutdown actually implemented.

For straight up power filtering, low ESR is a good thing, tube amps, transistor amps, so on. Long life, less ripple heat, etc.

[shattered]

Yeah, low ESR for tube amp supplies is fine. In fact, it's what we tried to get for everything back when I was designing power supplies for a living. We could not get ESR low enough. Switching power supply caps really pointed up the need for low ESR in a big way.

Is there a lot of low voltage logic stuff in those amps? DSP core, something like that? Our more difficult cases were with big high speed switching stuff that had to be remote from the DC power supply.

Great! By the way, I found a better alternative, Rubycon 500VXG100MEFCSN25X35, with longer life and more ripple current rating. It's 5mm taller than the Nichicon, but there is enough margin inside the chassis.

Fortunately it's a relatively simple circuit, single channel, a pair of TL072 for effects loop and slave out, a J175 and a BC236 transistors, no digital bullshit - otherwise it wouldn't be in my stable . I don't know how the circuit works, the regulators are near to a delay turn on circuitry (to increase tube life according to the manual), the HT supply to the tubes gets switched on only after they have warmed up and are ready to go - approx 30 seconds.

Again, I'm a little surprised at the circuits even being sensitive to filter cap ESR. Is this on the first rectifier filter caps, a bypass before the regulator, or an output filter cap after the regulator? Most of the time, power supplies don't care much in terms of stability about the ESR of the first filter cap and not much about bypass before the regulator. It's the output filter that hangs on the outer end of the feedback loop that can get you issues.

It was a Sega Mega Drive I (also known as Sega Genesis), can't find the schematic right now. If I recall correctly it was an output filter cap after the regulator, but I'm not sure. It didn't like the Panasonic FR caps, the console produced video artifacts and ran hot as hell, once I replaced them with General Purpose it worked fine again.

[R.G.]

[...] I would also say, the datasheet's recommendations are largely ignored in practice. I've seen lots of large filters on the outputs in any number of equipment, and I don't think I've ever seen the recommended capacitor discharge diode to prevent reverse voltage on shutdown actually implemented.

For straight up power filtering, low ESR is a good thing, tube amps, transistor amps, so on. Long life, less ripple heat, etc.

Spot on. Designers would find out that mostly they worked, so the fine points from the application notes got ignored. It's a tribute to the original designers that they were so resistant to abuse. I'm digging through my in-head archives for the original manufacturer; it wasn't TI, but TI bought them out.

I lived through the mass adoption of three-terminal regs. When I did my first regulator designs, the power-in-charge would not let me use the then-current uA723 IC. It had to be discrete, and provably worst-case-designed in front of a review committee. We longed for the simplicity of just plugging in a three-terminal. Of course, they didn't let us use plastic packaged power devices of any kind for my first few years; too unreliable.

The original datasheets and app notes on three terminals didn't include specialized cap recommendations or that reverse protection diode. I remember being surprised at the inclusion of the reverse-discharge diode after using 7800s for a while. I had to defend the reverse diode's inclusion a time or two, and I got in the habit of including it in my designs, but it's always been rare in consumer-grade stuff.

You bring up a great point for the OP: running any power circuit near a current limit can cause it to oscillate back and forth across the limiting edge. It's an issue with current limiting in high power solid state amps where the output stage is adequately protected against low-resistance output loads.

Fortunately it's a relatively simple circuit, single channel, a pair of TL072 for effects loop and slave out, a J175 and a BC236 transistors, no digital bullshit - otherwise it wouldn't be in my stable . I don't know how the circuit works, the regulators are near to a delay turn on circuitry (to increase tube life according to the manual), the HT supply to the tubes gets switched on only after they have warmed up and are ready to go - approx 30 seconds.

Hmmm. Without a lot of switching on the regulator loads, 78xx devices are pretty solid. As nuke said, even pro designers regularly ignored the datasheet recommendations on caps and everything else. That is curious enough to make me think that there might be some oscillation funnies going on in the circuit itself, or possibly that the circuit is running near the edge of current limit or voltage overhead if you're seeing this in an actual real-world setup. I'd love to see scope pics of the regulator input voltage and output voltage when a too-low ESR cap on the output made it oscillate. If it does, it's worth examining what other than ESR is making things go south.

Again, I'm a little surprised at the circuits even being sensitive to filter cap ESR. Is this on the first rectifier filter caps, a bypass before the regulator, or an output filter cap after the regulator? Most of the time, power supplies don't care much in terms of stability about the ESR of the first filter cap and not much about bypass before the regulator. It's the output filter that hangs on the outer end of the feedback loop that can get you issues.

It was a Sega Mega Drive I (also known as Sega Genesis), can't find the schematic right now. If I recall correctly it was an output filter cap after the regulator, but I'm not sure. It didn't like the Panasonic FR caps, the console produced video artifacts and ran hot as hell, once I replaced them with General Purpose it worked fine again.

Output filter after the regulator would be my guess. Quite a few newer LDO regulators have special instructions on how low or high an ESR the output cap needs, limits on the output cap values, even charts of output cap values that show stable and unstable regions for cap versus ESR. But the 78xx series didn't bother with that annotation.

Zooming back out a little, nearly all voltage follower stages, even simple BJT emitter followers, can go into oscillation from capacitive loading. It's a quirk of how the feedback reacts to a pure(ish) capacitor. You'll find the recommendation in many opamp datasheets that capacitive loads need a small resistance between the opamp output pin and the capacitance for stability.