FMOD high pass filters?

[DHT 845]

A LC filter is second order (12 dB/oct slope), and is more desirable IMHO for this application.. However, a purely LC filter will have a very high resonance peak (high Q) at the cutoff frequency, and adding a resistor into the circuit (see below, source: Wikipedia) will let you adjust the Q.
The problem is, when you plug in the numbers, a 70 Hz cutoff high pass filter will require a huge inductor, e.g. when C = 10 μF, the required L = 0.52 H, which is going to be pretty big.

Thanks
so I choose LC filter, keeping coil moderate 50mH and inexpensive (ab. 6 USD) and C=100uF polypropylen cap. The coil is without core but it is made with pretty thin wire
making it quite resistive (9 Ohms). Do you think that coil resistance could "damp" the resonanse or additional damping resistor is needed. If needed I have no idea how to calculate it's value for optimal damping. Also This resistor, coil resistance and load resistance will affect effective R that preamp will see, right?

 

[DHT 845]

Most of the cost is putting the components in an enclosure which connects to RCA male plugs on each side.

I suspect they use cheap small parts to fit inside, some reviews say that these FMOD add distortions (probably due to individual impedance issues?)...
I would try to split RCA cables and mount the filters in the middle in some "soap box".

BTW.
Danny Richie from GR-Research ecourages to use single cap as hi-pass filters in the line-level before the amp. That is minimalistic approach!

Hi-pass filters | GR-Research

gr-research.com

 

[Ron Texas]

I suspect they use cheap small parts to fit inside, some reviews say that these FMOD add distortions (probably due to individual impedance issues?)...
I would try to split RCA cables and mount the filters in the middle in some "soap box".

BTW.
Danny Richie from GR-Research ecourages to use single cap as hi-pass filters in the line-level before the amp. That is minimalistic approach!

Hi-pass filters | GR-Research

gr-research.com

$28 is pretty cheap. Unless you value your time at zero it isn't worth going the DIY route for this. As for the components being cheap, that's just your guess. Without tearing one down nobody will ever know.

 

[DonH56]

"Cheap" and "small" do not always go together... I would guess they are pretty inexpensive, but likely more than adequate for audio.

A single cap works for a single-pole (6 dB/oct) HPF but you'd better know the impedance it is driving to get the corner frequency right...

Here is a Wikipedia article on Q factor that has the equations for an RLC circuit: https://en.wikipedia.org/wiki/Q_factor For this I'd prefer overdamped, Q of 0.5 or so, but the Q is also affected by the source and load impedances. I'd just buy the Fmod, but if you want to DIY then 9 ohms coil resistance is probably enough given the rest of the circuit. You could always add a little series ("snubber") resistance to the capacitor to reduce the Q but not likely to be worth it. You can get component models from the manufacturers and simulate it using LTSpice (free download) or whatever but that's a steep curve for someone who has never used a simulator.

 

[restorer-john]

@escape2 If the setup is permanent and the amplifier is a normal integrated or other amplifier with a typical RC front end to the power stage, a simple single capacitor change internally would be the easiest and most cost effective.

Back in the day, we had plenty of useful filters on our gear (the L-81 was one of my first 'nice' amplifiers). The 70Hz filter was very useful with small bookshelf speakers and vinyl, preventing a lot of feedback and excess cone excursion in small woofers.

 

[NTK]

Thanks
so I choose LC filter, keeping coil moderate 50mH and inexpensive (ab. 6 USD) and C=100uF polypropylen cap. The coil is without core but it is made with pretty thin wire
making it quite resistive (9 Ohms). Do you think that coil resistance could "damp" the resonanse or additional damping resistor is needed. If needed I have no idea how to calculate it's value for optimal damping. Also This resistor, coil resistance and load resistance will affect effective R that preamp will see, right?

Here is the simulation of your filter. The frequency response looked good. But the problem is that the load of this filter to the source has a minimum impedance of 10 Ω.


Below is for a 2 RC filters cascade. The limitation of using only R & C is that the Q of the resulting filter is low (cannot exceed 0.5 for a 2nd order one), but it represents a more reasonable load (but still on the low side) to the source.

 

[DHT 845]

Here is the simulation of your filter. The frequency response looked good. But the problem is that the load of this filter to the source has a minimum impedance of 10 Ω.

1. So there is no way with LC circuit to make effective resistance high? Coils do not have huge resistance from their nature... How about adding resistance in series with the coil for. ex. 100k Ohms? Wouldn't it rise impedance to ab. 50k Ohms? Or maybe it will "overdamp" the circuit in some way?
2. Maybe simple RC circuit with 100k Ohms and 22nF would be the best in terms of impedance but still I don't get the idea how come it makes 12dB/ Oct without a coil (and not 6dB/Oct), it is like black magic to me...
3. On previous page you proposed 0,47uF + 4,7k Ohms, that makes 4,5k Ohm in parallel to the load. Isn't it better to make it 50k Ohms, like with 22nF + 100k Ohms, especially that minimum impedance is in the crossover frequency that is pretty low?
4. From what I understand till now it seems that LC circuits are better for figher crossover points where low impedance is not such a problem(?)

 

[escape2]

@escape2 If the setup is permanent and the amplifier is a normal integrated or other amplifier with a typical RC front end to the power stage, a simple single capacitor change internally would be the easiest and most cost effective.

Back in the day, we had plenty of useful filters on our gear (the L-81 was one of my first 'nice' amplifiers). The 70Hz filter was very useful with small bookshelf speakers and vinyl, preventing a lot of feedback and excess cone excursion in small woofers.

View attachment 329356

View attachment 329357

Thanks. This is an old thread. Since then, I ended up getting an SVS SB2000 sub and am using its built-in 80 Hz high pass filter.

 

[DonH56]

1. So there is no way with LC circuit to make effective resistance high? Coils do not have huge resistance from their nature... How about adding resistance in series with the coil for. ex. 100k Ohms? Wouldn't it rise impedance to ab. 50k Ohms? Or maybe it will "overdamp" the circuit in some way?

At the resonance point the reactance of the capacitor and inductor cancel leaving just the resistance of the coil in this circuit. You can add a series resistor before the capacitor or in series with the inductor. That will also affect the corner frequency.

2. Maybe simple RC circuit with 100k Ohms and 22nF would be the best in terms of impedance but still I don't get the idea how come it makes 12dB/ Oct without a coil (and not 6dB/Oct), it is like black magic to me...

The two capacitors and associated resistors comprise a two-pole response, though rather weak (low-Q) as @NTK said. Each capacitor is a reactive element, with the resistors provided the needed current-steering to provide second-order response.

3. On previous page you proposed 0,47uF + 4,7k Ohms, that makes 4,5k Ohm in parallel to the load. Isn't it better to make it 50k Ohms, like with 22nF + 100k Ohms, especially that minimum impedance is in the crossover frequency that is pretty low?

You want the impedance of the filter circuit much lower than the load if possible so that the load impedance does not affect the filter circuit. But, you also want it much higher than the source (driving) impedance to prevent loading the source. It's a trade. One reason active circuits are usually used for line-level crossovers.

4. From what I understand till now it seems that LC circuits are better for figher crossover points where low impedance is not such a problem(?)

The low impedance is because the L and C cancel each other at the corner frequency, no matter how high (or low) that frequency is. At that point the only thing raising the impedance is the parasitic resistance of the L and C components, plus whatever series resistance you add to the circuit.

Using an L and C makes it easy to produce a second-order transfer function with only two components. An RC design generally needs four components to provide some measure of isolation between stages (otherwise the two capacitors will look like just one capacitor of smaller value).

HTH - Don

 

[DHT 845]

You want the impedance of the filter circuit much lower than the load if possible so that the load impedance does not affect the filter circuit. But, you also want it much higher than the source (driving) impedance to prevent loading the source. It's a trade. One reason active circuits are usually used for line-level crossovers.

I want passive. So what is the optimal impedance of the filter in case when source has 20 Ohms and amp has 100k Ohms? The impedance that is the least probable that might cause any problem for the source or for the amp? The "dream impedance"...

 

[DonH56]

I want passive. So what is the optimal impedance of the filter in case when source has 20 Ohms and amp has 100k Ohms? The impedance that is the least probable that might cause any problem for the source or for the amp? The "dream impedance"...

Infinite ohms seen by the source, zero ohms seen by the amp. Difficult to realize with actual components.

Is there a reason you want to roll your own instead of just buying the Fmod or another passive solution (Marchand or whomever)? Versus learning the engineering required to design your own, or have someone here work it out for you? Assuming you get the component values required, have you the soldering and assembly skills to manufacture your filter? Sounds snarky, but we really need to know your skill level for some of this (and apologies for not reading the entire thread if you've already stated that). Also, and again I probably just did not read far enough back, do you know the crossover frequency and order (roll-off) needed, and (sanity check) you just need the high-pass filter and not a corresponding low-pass?

The simplest answer for now would be to just use the passive RC design @NTK provided above or some similar version. The minimum load your source should see is hopefully part of its spec sheet so you can design to that.

 

[DHT 845]

Infinite ohms seen by the source, zero ohms seen by the amp. Difficult to realize with actual components.

Is there a reason you want to roll your own instead of just buying the Fmod or another passive solution (Marchand or whomever)? Versus learning the engineering required to design your own, or have someone here work it out for you? Assuming you get the component values required, have you the soldering and assembly skills to manufacture your filter? Sounds snarky, but we really need to know your skill level for some of this (and apologies for not reading the entire thread if you've already stated that). Also, and again I probably just did not read far enough back, do you know the crossover frequency and order (roll-off) needed, and (sanity check) you just need the high-pass filter and not a corresponding low-pass?

The simplest answer for now would be to just use the passive RC design @NTK provided above or some similar version. The minimum load your source should see is hopefully part of its spec sheet so you can design to that.

From my experience with Marchand XM26 that I owned it took me several different CP and slopes to find the best sounding values with my previous speakers (required soldering), so I know that even if I buy FMOD pair it would be room for curiosity and experiments. I do want to buy several pairs of FMODs. I prefer to cut some spare RCA cables I have in two and solder some parts in between. I just want to have a good theoretical starting point from since I want to avoid mistakes like impedance mismatch, resonances etc. (I will be trying 40-70Hz region). I need HI-PASS only (Dayton APA150 for the base, 50-150hz 18dB/Oct CP). Marchand XM46 PLLXO (to be more precise, just 2 passive filter boards that I'm considering) are tempting me but I am not sure about impedance possible issues with them. However I think that it is hard to find the amp with higher impedance than mine (100k) and Pre90 also has not the worst output impedance (20 Ohms) for the preamp so for whom are passive LC crossovers like XM46? I read very positive opinions ab. Marchand XM46. Marchand have manual (with schematic) and even calculator on their site...

Marchand Manual

Marchand Electronics: Sound card Function Generator, Real Time Analyzer, Audio Oscillator, Calculator, free download

 

[Sokel]

From my experience with Marchand XM26 that I owned it took me several different CP and slopes to find the best sounding values with my previous speakers (required soldering), so I know that even if I buy FMOD pair it would be room for curiosity and experiments. I do want to buy several pairs of FMODs. I prefer to cut some spare RCA cables I have in two and solder some parts in between. I just want to have a good theoretical starting point from since I want to avoid mistakes like impedance mismatch, resonances etc. (I will be trying 40-70Hz region). I need HI-PASS only (Dayton APA150 for the base, 50-150hz 18dB/Oct CP). Marchand XM46 PLLXO (to be more precise, just 2 passive filter boards that I'm considering) are tempting me but I am not sure about impedance possible issues with them. However I think that it is hard to find the amp with higher impedance than mine (100k) and Pre90 also has not the worst output impedance (20 Ohms) for the preamp so for whom are passive LC crossovers like XM46? I read very positive opinions ab. Marchand XM46. Marchand have manual (with schematic) and even calculator on their site...

Marchand Manual

Marchand Electronics: Sound card Function Generator, Real Time Analyzer, Audio Oscillator, Calculator, free download

The old XM9 modules I have measured maxed my old interfaces ability,so the XM44,so more than 102db SINAD (they state 110db if I'm not wrong).
And apart from Marchand,even the cheap one I measured here is totally decent (read the whole front page as there is a progression)

The permanent problem with FMODS and the likes are mismatch impedance,otherwise everyone would use them because of their ease.

I would ask Marchand if I was you.

 

[DHT 845]

Sticking to diy RC FMODS-like idea (I have not a clue how to rise LC filter impedance) I figured out (correnct me please if I am wrong) that ideal RC filter impedance for me will be 1.4k Ohm. Why? Because 1.4k Ohms is 70 times higher that 20 Ohm and ab. 70 times lower than 100k Ohm.

So ideal (starting point for experiments with CP) RC hi-pass filter will be C=1.6uF, R=1.4k Ohm, that will give -3dB at 71Hz.

Changing only cap for:
1.8uF will give 63Hz at -3dB
2.0uF will give 57Hz at -3dB
2.2uF will give 51Hz at -3dB

 

[DonH56]

Very few if any consumer products are rated to drive 1.4 k-ohms. IME most are rated for 10k minimum, though I am sure there are exceptions.

If all you want is a 6 dB roll-off and know your load is 100k, then you can neglect the 20 ohm source and just calculate using fp = 1/(2piRC), or C = 1/(2*pi*fc*R). Then you just need a selection of capacitors providing the range of desired corner frequencies.

Here is the result with R=100k so C=0.027 uF for a 60 Hz corner. The source never sees less than 100k ohms.

It is almost impossible in practice to isolate the two sections of a passive RC filter, so they will interact, changing the corner frequency and roll-off compared to a simple cascade of two isolated first-order filters. Usually an active buffer between solves that problem, but if you want to stay completely passive and use RC instead of LC filters, you have to keep the R and C values about a decade or more apart and either solve the more complex circuit for poles or iterate in the simulator. I have the equations, but for this just tweaked values in the simulator to achieve the desired 60 Hz corner (-3 dB) frequency whilst keeping the input impedance above 10k ohms. Note the roll-off is rather "soft", only -9 dB instead of -12 dB at 30 Hz, due to this interaction. You can tweak C and/or R values for other corners.

 

[DHT 845]

O man! Now we're going back to the minimalistic approach

Hi-pass filters | GR-Research

gr-research.com

 

[DonH56]

O man! Now we're going back to the minimalistic approach

Hi-pass filters | GR-Research

gr-research.com

Wish I'd said that...

 

[DHT 845]

this is my first attempt to create schematic in LTSpice, I am not too good in it, cannot properly describe voltage source and measuring points
but I will try, never did anything like it, I studied Finance and Banking and I am ignorant in creating simulations in electronics

 

[NTK]

I recreated the simulation file from your schematic. Looks like the capacitor values weren't specified correctly, as I don't have the "F" and have a "μ" instead of "u" in the value. The AC voltage source looked different too. (Did you choose "voltage" when you placed the component in the schematic when creating the AC source?) The simulation command ".ac dec 100 10 1k" looks correct.


There are lots of training materials on the web and in YouTube. Below are links to a few from Analog Devices, which had acquired Linear Technology (the LT in LTspice).

https://www.analog.com/en/education/education-library/videos/video-series/ltspice-ac-noise-analysis-tutorial.html

https://www.analog.com/en/technical-articles/how-to-generate-a-bode-plot-with-ltspice.html

Getting Started with LTspice - Webinars - Webinar - EngineerZone

If you are new to LTspice, this seminar is for you. This seminar covers the fundamentals of drafting circuits in the schematic editor and how-to bring circuits to

ez.analog.com

 

[DHT 845]

There are lots of training materials on the web and in YouTube.

Thank you, I need a training and little more time.