On Tue, Jun 7, 2022 at 11:19 AM, thet wrote:

I just noticed two conceptual errors in this rough schem.

1. opamp power consumption is not constant - it is 1.8ma per opamp
quiescent, but it will increase with signal. So the opamp power needs to
be regulated by a zener, it cannot be controlled purely resistively or
it will sag at high signal levels.

2. At C5 the voltage is labelled as 38v but I've put a 35v zener there -
that is obviously wrong.

I don't think the 'real world' voltages will be anything like those you've listed in your 'sim'..
Assuming a total current drain of 4mA , then each of the 4 'legs' of the feed will need to supply 1 mA.
Even if your phantom power supply does supply a full 48V under load (which it probably won't) you are going to drop 6.8V in each phantom feed resistor, plus another 10v across each 10k
So that's a max of 31.2v available at the 'summing' point. In reality it will almost certainly be less.

Assuming the signal requires 'significant' extra power ... then I can't see how you expect a Zener to 'regulate' such a high impedance power feed?
Without any regulator, the op-amp supply cannot exceed its specified maximum.?
As the power supply 'sags' under the extra power the signal requires (which I personally can't see being significant - but I stand be corrected on that?) the lower supply voltage will simply reduce the headroom.
I really can't see the need for a regulator in this circuit, when the absolute max voltage available does not exceed the op-amp specs.?

I just noticed two conceptual errors in this rough schem.

1. opamp power consumption is not constant - it is 1.8ma per opamp quiescent, but it will increase with signal. So the opamp power needs to be regulated by a zener, it cannot be controlled purely resistively or it will sag at high signal levels.

2. At C5 the voltage is labelled as 38v but I've put a 35v zener there - that is obviously wrong.

I will attempt a corrected version.

Still the schem serves to show the idea of the power supply stacking.

On Mon, Jun 6, 2022 at 01:44 PM, <rhalfik@...> wrote:

?

I just browsed PUI catalog and found some <10mm cardioid capsules with good SNR. They measured bigger rolloff in the bass though. Does anyone know a good place to learn how to equalize these capsules? I can design a filter for a loudspeaker, but I'm not sure what's the best way to go about it with a microphone.

I've found that many attempts at internal frequency response corrections within the microphones? themselves can often tend to be too simplistic to be really useful.
Passive filters are often the 'wrong' shape - and can introduce noise into the signal path.
For recorded signals, it's often useful to utilise the comprehensive parametric EQ? features found in most DAWs, and correct as required in post.

For 'real time' correction I've found introducing custom filters into the signal path - at line level - within the mic channel can work well.
You have the option of making the filters as simple - or complex - as you like, and working with the signal at line level makes the addition of any extra noise much less of a problem..

I made some notes ??on how I went about trying out that idea...?

Yes, I mean low shelf along with a high frequency notch that is needed with some cheap capsules. There are some interesting cardioids out there that don't have a very steep rolloff. Only 3dB per octave.

For example:

Also some 25mm Chinese capsules. They need just a touch more bottom, which I imagine can be used as a switchable proximity filter as well.

--

Newbie. Ordered this capsule from?
Unsure about what OPA to build.?
Please advice.?
Thank you.

here's a? schem to show the idea of what I mean about stacking the power supply. To use the same current twice like a schoeps.



current is assumed to be 2ma per channel, 4ma total, for sake of this schem, supplied by two XLR phantom channels.

the idea is the zeners never kick in in normal operation so an extra filter on the 9v shouldn't be needed.

I'd tweak the resistor sizes so that there's a bit more leeway between the zener voltage and the actual operating voltage but this is just to show the idea.

R26 is a nod to the need to make sure that the 9v load? draws enough current to match the opamp, it could be done with a 9v zener but then we'd need another filter and caps are bulky and there's more caps than I'd like already.

maybe the capsule source resistors should be trimmers? but that's also bulky. I'm trying to get the physical size of the circuit down.

so in the next step I'm asking what large components ie caps (if any) can I eliminate and how?

On 06/06/2022 14:57, Jerry Lee Marcel wrote:

Do you have any suggestions about what values would work best for the bal and vol pots?
I would use 20k for balance and 50 or 100k for volume, assuming there is an opamp after.

why those values? I'm not challenging it - I'm just not sure what your criteria are.

If the bal and vol pots are the same size then with the balance pot all the way over there is a 50% reduction in level on one side and none on the other. So 6dB difference. That seems like a reasonable maximum adjustment to me which is why I made them the same size.

On reflection maybe 3dB would be enough.

I'm not sure what the output impedance of the capsule fets is in source follower configuration, so I don't know how small a load they can drive, but I assumed 10k should be ok.

additionally to reduce power demand the power could be delivered with the opamp power in series with the capsule fet power since there is excessive voltage and low current - this would require a more complex power supply setup though.

Indeed, something like paralleling several Alice circuits could be made to work.

I'm not sure what you mean here. So many things have now been called Alice. Do you mean Scott's original schoeps-like circuit? If so how does it help?

Circuits that use the capsule fet as a phase splitter make balance and volume control more difficult it seems to me.

I will draw what I meant and post it when I get time.

On 06/06/2022 12:12, Arjay 1949 wrote:

I'm intrigued by the presence of some of the components.......
? Why include the 10v zener D0 ? ... It draws current, could (potentially) add noise, and restricts headroom. The derived 'half line' voltage supply will automatically bias the op-amp outputs to 'swing' symmetrically. The absolute supply voltage value is not critical (a long as it doesn't exceed the specified maximum supply - which it won't here )

firstly, this is a quick draft - component values are not optimised and the power supply isn't the bit that's new or interesting.

However the zener is there to protect any 25v or less caps in the event of switch-on surges or any other unforeseen higher voltage situations that might exist during troubleshooting or experimentation. The 10v was a quick and dirty interim value entered to allow the capsules fets to be run from the same V+ rail as the opamps. However I will probably do it a better way.

I will certainly refine the power supply. If I can make the operating voltage lower than the zener thus making it inactive for normal operation and perhaps dispense with C5 as a result then I will...

? I cannot see the point of R5, R6, R16 and R17 ?....

as Jerry said - they are there just in case a gain greater than 1 is needed or desired, if only unity gain is needed then it is trivial to omit R6 and replace R5 with a wire link. I'm trying to make a PCB that is as versatile as possible so it should allow the possibility of changing the opamp gain.

The bit I'm actually unsure of is the mixer and volume pot and how the various impedances there might interact, and what tradeoffs there might be.

开云体育

thet wrote:

Can a wire be threaded through the bellows to use one pcb with two outputs (one for each hand)? Or would it get tangled up?

Yes, the typical accordion pickup system does this -- the left-hand cable threads through the bellows (they snake the wire in an "S" shape and tape it to the tips of the bellow creases to allow elongation without tangling or interference with the reed banks -- ).

I am planning to do that and to have the jack(s) at the bottom of the accordion, so the cables can dangle between my legs (when sitting) or hang straight down (when standing).

JLM wrote:

I realized that this list has helped me before on this question: is a level control on Simplest P48 as simple ?

The 120k resistor should be halved in order to make sure the capsules receive enough voltage for pentode operation. Pentode operation is mandatory for correct mixing operation.
However the balance between both capsules depends on good matching of the capsules, so some kind of comparison/matching is necessary.

Sorry, that "120k" was a leftover from an older SP48 schematic. I will do matching for level as well, thanks.

In terms of choosing R for an arbitrary capsule in this kind of arrangement, the idea is just to choose it such that the voltage is on the higher end of the range? I.e. I would refer to the tested values from the Simonelli et al. PDFs (and/or test voltages myself), and err towards lower R to ensure higher voltage?

When you say "Pentode operation is mandatory for correct mixing operation" you are referring to the "parallel capsules" method of mixing as shown in the schematic, right?

Putting all these requirements within the constraints of phantom power is a little challenging and requires some brains agility, but I reckon it is feasible.
However, integrating that in an accordion seems hardly doable, so it would be an external box with 3 or 4 cables going to the accordion.
When the definition of the product(s) is settled, I can draw a schematic, even layout a PCB. It's not too difficult.

That's very generous, thank you! Given that I think I will get what I need from the above-discussed schematic (Parallel SP48 with a simple level knob), I will think about these other options and when/if I can be more specific about the goals I'll ask again. (And of course I will follow the ongoing discussions re: balancing mixer with interest.)

-c

On Mon, Jun 6, 2022 at 03:08 PM, Jerry Lee Marcel wrote:

?

I agree, partially. I'm not worried about zener noise, since it is conveniently filtered by R7/C4. But I agree with increasing the zener voltage, unless the opamp maximum rail voltage is exceeded. Several rail-to-rail opamps are spec'd for 11V maximum.

I agree that in this case the filtering should eliminate anyzener noise. I just can't see the point of introducing a potentially noisy regulator where it's not needed.
In the worst case here - assuming an infinitely low impedance 52v phantom power source -? then, with each op-amp drawing 1.8mA (and ignoring the tiny? bias voltage current) there is never going to be more than 22v DC available to power each op-amp. (Each leg - 1.8mA through 6k8 + 10k = 30v.....52v- 30v = 22v )
The OPA 164* series op-amp has a max supply of 36v, and a rail to rail output. I can see no reason for a zener in this case?...

? I cannot see the point of R5, R6, R16 and R17 ?....
Adding resistors into the signal path will simply increase noise (albeit marginally). The extra gain added is probably better applied by the mic preamp (unless it's a really bad mic preamp!).
Simply connecting the op amp outputs to the inverting inputs will configure the op-amps as? minimum component non inverting buffers.? R22 and R23 will apply the required DC 'half rail' bias to the op-amps.

I believe it is convenient to incorporate these resistors in case the extra gain is necessary. There's low level here, considering losses in the balance and volume pots

Add any required gain in the preamp.

Having a hefty signal there takes care of noisy input stages in some cheap mixers.

I would expect a half decent mic pre-amp to have a noise floor 20 or 30dB lower than this circuit..... Surely better to add the gain in that pre-amp?
If the pre-amp really is bad enough to have around the same noise floor as this circuit, then of course it makes little difference where you add the gain.
But even 'cheapies' like Behringer have noise floors down in the -120dB region these days.......

Le 06/06/2022 à 14:44, rhalfik@... a écrit?:

Plenty of low frequency on that graph!

I just browsed PUI catalog and found some <10mm cardioid capsules with good SNR. They measured bigger rolloff in the bass though.

One probably explains the other...

Does anyone know a good place to learn how to equalize these capsules?

What do you mean by equalize?
If you mean compensating LF loss with some kind of boost EQ, you need to realize that the drop reaches 12dB/octave, which is quite difficult to implment.

开云体育

[Edited Message Follows]

On Mon, Jun 6, 2022 at 10:45 AM, thet wrote:

Here's my incomplete single board 4 mic 2 output draft circuit.

I'm intrigued by the presence of some of the components.......
? Why include the 10v zener D0 ? ... It draws current, could (potentially) add noise, and restricts headroom. The derived 'half line' voltage supply will automatically bias the op-amp outputs to 'swing' symmetrically. The absolute supply voltage value is not critical (a long as it doesn't exceed the specified maximum supply - which it won't here )

I agree, partially. I'm not worried about zener noise, since it is conveniently filtered by R7/C4. But I agree with increasing the zener voltage, unless the opamp maximum rail voltage is exceeded. Several rail-to-rail opamps are spec'd for 11V maximum.

? I cannot see the point of R5, R6, R16 and R17 ?....
Adding resistors into the signal path will simply increase noise (albeit marginally). The extra gain added is probably better applied by the mic preamp (unless it's a really bad mic preamp!).
Simply connecting the op amp outputs to the inverting inputs will configure the op-amps as? minimum component non inverting buffers.? R22 and R23 will apply the required DC 'half rail' bias to the op-amps.

I believe it is convenient to incorporate these resistors in case the extra gain is necessary. There's low level here, considering losses in the balance and volume pots

Add any required gain in the preamp.

Having a hefty signal there takes care of noisy input stages in some cheap mixers.

Le 06/06/2022 à 11:34, thet a écrit?:

On 06/06/2022 09:59, Jerry Lee Marcel wrote:

?

That would work, probably some of the values may be tweaked.

Do you have any suggestions about what values would work best for the bal and vol pots?

I would use 20k for balance and 50 or 100k for volume, assuming there is an opamp after.

that is my concern.

It's something you need to experiment with. With large chromatics, you can run the wires along the left and right straps, but for smaller diatonics and bandoneon, very often players don't use straps, but I think a convenient and safe solution is to route the wires along the player's arms up to their neck.

1.8ma per opamp, plus <1ma per capsule total max 7.6ma - but if there are two outputs then there are two sources of phantom power

That's correct. A 4-capsule system would be feasible with some attention.

additionally to reduce power demand the power could be delivered with the opamp power in series with the capsule fet power since there is excessive voltage and low current - this would require a more complex power supply setup though.

Indeed, something like paralleling several Alice circuits could be made to work.

Plenty of low frequency on that graph!

I just browsed PUI catalog and found some <10mm cardioid capsules with good SNR. They measured bigger rolloff in the bass though. Does anyone know a good place to learn how to equalize these capsules? I can design a filter for a loudspeaker, but I'm not sure what's the best way to go about it with a microphone.

On Mon, Jun 6, 2022 at 10:45 AM, thet wrote:

Here's my incomplete single board 4 mic 2 output draft circuit.

I'm intrigued by the presence of some of the components.......
? Why include the 10v zener D0 ? ... It draws current, could (potentially) add noise, and restricts headroom. The derived 'half line' voltage supply will automatically bias the op-amp outputs to 'swing' symmetrically. The absolute supply voltage value is not critical (a long as it doesn't exceed the specified maximum supply - which it won't here )

? I cannot see the point of R5, R6, R16 and R17 ?....
Adding resistors into the signal path will simply increase noise (albeit marginally). The extra gain added is probably better applied by the mic preamp (unless it's a really bad mic preamp!).
Simply connecting the op amp outputs to the inverting inputs will configure the op-amps as? minimum component non inverting buffers.? R22 and R23 will apply the required DC 'half rail' bias to the op-amps. Add any required gain in the preamp.

Here's my incomplete single board 4 mic 2 output draft circuit.



the capsule power is not connected yet. it could share the opamp V+ or it could use a series power configuration.

this version the power supply is shared between the left and right hand channels and a dual opamp is used

However if two separate boards are required for left and right hands then the circuit could be halved - a single opamp could be used on each board with it's own power supply.

putting it all on? one board is less components overall - but requires a wire through the bellows.

separate boards might be more applicable to other instruments with two mics.

On 06/06/2022 09:59, Jerry Lee Marcel wrote:

?

That would work, probably some of the values may be tweaked.

Do you have any suggestions about what values would work best for the bal and vol pots?

Note that the upper stage needs to be grounded too.

yes - just a rushed omission

Although it's powered from a 9V battery, there is no doubt it can be adapted for P48.

that was just a shorthand for 9v derived from P48, so yes

Can a wire be threaded through the bellows to use one pcb with two outputs (one for each hand)? Or would it get tangled up?

I would do that only as long as the wires can be adjusted in a way that does not interfere with playing.

that is my concern.

Or is it easier to have separate preamps and outputs on each side?

It's more versatile.

yes, but it results in an XLR cable hanging off each side of the accordion , rather than two cables on one side. I'm not an accordion player and I'm not sure which would be preferable. I'm assuming separate signal outputs for the two sides in either case.

The former allows power supply sharing and fewer overall components.

For a full system with 4+ capsules, it may exceed the capabilities of phantom powering (<10 mA)

1.8ma per opamp, plus <1ma per capsule total max 7.6ma - but if there are two outputs then there are two sources of phantom power

additionally to reduce power demand the power could be delivered with the opamp power in series with the capsule fet power since there is excessive voltage and low current - this would require a more complex power supply setup though.

Le 06/06/2022 à 09:05, thet a écrit?:

Shouldn't the resistor be halved whether the volume control is there or not?

I have also been assuming that left and right hands have separate outputs - they are essentially separate instruments, sharing only the players brain and an air source.

My attempts to work out a mixer are purely to mix 2 (or more) capsules on one side of the accordion at a time.

Does anyone have any feedback on whether something like balance arrangement will work satisfactorily?

?

That would work, probably some of the values may be tweaked. Note that the upper stage needs to be grounded too.
Although it's powered from a 9V battery, there is no doubt it can be adapted for P48.

I've mostly drawn up a full schematic with opamp outputs, I will link when done. The parts count is way up of course, but it would be easy enough to make a pcb.

Can a wire be threaded through the bellows to use one pcb with two outputs (one for each hand)? Or would it get tangled up?

I would do that only as long as the wires can be adjusted in a way that does not interfere with playing.

Or is it easier to have separate preamps and outputs on each side?

It's more versatile.

The former allows power supply sharing and fewer overall components.

For a full system with 4+ capsules, it may exceed the capabilities of phantom powering (<10 mA)

On 06/06/2022 03:51, Jerry Lee Marcel wrote:

The 120k resistor should be halved in order to make sure the capsules receive enough voltage for pentode operation. Pentode operation is mandatory for correct mixing operation.
However the balance between both capsules depends on good matching of the capsules, so some kind of comparison/matching is necessary.

Shouldn't the resistor be halved whether the volume control is there or not?

I have also been assuming that left and right hands have separate outputs - they are essentially separate instruments, sharing only the players brain and an air source.

My attempts to work out a mixer are purely to mix 2 (or more) capsules on one side of the accordion at a time.

Does anyone have any feedback on whether something like balance arrangement will work satisfactorily?

?

I've mostly drawn up a full schematic with opamp outputs, I will link when done. The parts count is way up of course, but it would be easy enough to make a pcb.

Can a wire be threaded through the bellows to use one pcb with two outputs (one for each hand)? Or would it get tangled up?

Or is it easier to have separate preamps and outputs on each side? The former allows power supply sharing and fewer overall components.