Phantom blocking input capacitor configurations

[leswatts]

I'd like to revisit an old subject.
I have a THAT 1512/1646 mic pre I built for the studio a good while back. In a couple channels I used the original 1512 demo board
layout where 100K input bias resistors and an input servo allowed the use of small film caps rather than large electrolytics.

I liked the ideal of not using unpolarized electrolytics when phantom power is OFF.

But in fact I have found that this configuration really doesn't work that well in practice.

So....I have a little spare time and I thought i'd rip that stuff out. I think I'll try ditching the servo and going with the now recommended
1K/22K/1K T configuration input bias network.

But I still don't like the idea of lytics running unpolarized...including half of a bipolar lytic if that is used.

I'm familiar with the technique of  doing a -++- series pair of caps where the center is connected to +48 through 100K or so resistors.
That way both caps are properly polarized with phantom on or off PROVIDING in the phantom on condition a mic is drawing some current
such that the mic side of the caps is well below +48 from IR drop across the 6.81K phantom resistors.

However with no load and phantom on the first cap would still have no bias.

I was thinking this. What about configuring the input caps +--+ with the center connected to say the -supply rail (-15 to -20V) through 100k or so.
So...

Phantom on with mic drawing some current
first cap is polarized by somewhat less than 48V + rail voltage; second cap polarized with rail voltage since the pre
amp input side is near ground due to the input bias resistors

Phantom on with no load
First cap polarized with 48+rail voltage; second cap polarized with rail voltage

Phantom off with load (like a dynamic mic)
First and second cap polarized with rail voltage

Phantom off no load
First and second cap polarized with rail voltage

In the last  two cases the 6.81K phantom resistors would have to have a path to ground with phantom OFF.

Comments?

Les
L M Watts Technology

 

[gyraf]

I see your point.

But Wima recently introduced a series of small-footprint polyesters in the MKS2-series (5mm pin spacing) - 10u does not have to be very big...

Jakob E.

 

[leswatts]

I'll check that out... we use some WIMA films in the mic circuits and they are fairly small.

I'd want to have 20-40 uF even with Tbias just to get very very good CMRR at line frequencies.

But as far as I can see the lytic balance I mention will work...I just have to imagine the behavior in every possible fault condition.
I think the only time the caps would not have bias is when the power is off.

Let's take it a step further. What if the cap middle terminal was biased by an ELECTRET? I wonder if there would be any potential
there with typical cap leakage. Then caps would be biased always...on or off.
I could try it with an  electret from a cheap EC capsule...but i'd need an electrometer to measure it. I suspect cap leakage would
result in very little potential remaining.

Les
L M Watts Technology

 

[JohnRoberts]

When phantom power is defeated, you don't need the caps at all, while you need to account for DC offset.

JR

[edit] IIRC Sam Groner has posted some studies of how electrolytic caps behave in response to bias voltage. There is little problem from leaving phantom resistors biased up at low DC voltage just to insure a positive bias on the caps. Note: I am not suggesting this is worth doing, I have never personally measured this myself or experienced audible distortion problems from running polars with no bias.. Look at Sam's measurements for some real data. Data is our friend. [/edit]

 

[leswatts]

Of course John, an option is to switch out blocking caps when phantom is disabled. Some consideration of timing
and charge/discharge rates would be in order. Certainly doable, and probably is done. I considered it.

But it's money and complexity.

I figure if modern long life low esr electrolytics have very low distortion levels, why not simply set them up as I mention,
which should keep them polarized properly in all but power off conditions.

I realize that the usual solution is to leave them in with phantom disengaged and hope they are not reverse biased more than about 1.5 volts
for any significant time...and they shouldn't be at audio frequencies where very little voltage is across them. But that doesn't cover everything.


BTW Jakob turns out I have some WIMA MKS-2 series in the lab. 2.6 u. I checked the MKS4 series where mouser stocks a 33u 63V.
It's big, but not impossibly so. I'm not liking the price. $18 each.

What i'm asking here, please, is there any fault  or other condition I haven't thought of that would cause undesirables with my +--+ with junction at negative power rail through 100K? As far as I can see it would keep forming voltage on the caps better than -++- with junction to hi Z +48. Have I missed something? It would seem economical and effective. No switching or super expensive capacitors. A slight increase in ESR,
but with modern lytics not much of an issue as far as I can see.

Les
L M Watts Technology

 

[JohnRoberts]

First thought is that you need higher voltage caps (larger and more expensive).

Second thought, leakage current in caps is a function of voltage across them so more voltage is more leakage. Significant bias even when phantom is off seem undesirable (does negative bias need to be -20V?). Noise from this leakage is generally insignificant but I have seen at least one series of caps we had to black list from phantom application due to noise. Note: in hindsight the caps I black listed may have come from the factory unformed. Not my problem to figure out why they didn't work so i just disappeared them. 

Finally, review how caps discharge should fully charged input cap network get momentarily shorted (like by a patch bay). This will probably be similar to single cap as far as voltage step the preamp electronics see.

JR

 

[leswatts]

Thanks john.

Yes the first cap would have to be 100V rather than the usual 63. I had factored that in. But second cap could be less than 63.
I'd say a cost wash.

I think a short to pin one would be similar to single cap. Some put a few ohms buildout to limit current. It's a pretty common thing especially when people build their own cables. It's gonna happen.

Obviously I don't like lytics in general and especially unpolarized. I don't trust many data sets. I tend to trust cap manufacturers more. It makes me think of the reforming procedure I have to do for the caps in my large CNC VFD if left off more than 3 months. Or the transient noise for the first few minutes in pres or some mics  that have been on the shelf a long time.

Seriously looking at button cells too for power off. Depends on leakage for an old cap worse case.
Call it belts and suspenders. But SSL at least seriously thought about it.

Les
L M Watts Technology

 

[JohnRoberts]

Thanks john.

Yes the first cap would have to be 100V rather than the usual 63. I had factored that in. But second cap could be less than 63.
I'd say a cost wash.

except two caps in series need to be 2x C to effectively deliver 1C.

I think a short to pin one would be similar to single cap. Some put a few ohms buildout to limit current. It's a pretty common thing especially when people build their own cables. It's gonna happen.

Obviously I don't like lytics in general and especially unpolarized. I don't trust many data sets. I tend to trust cap manufacturers more. It makes me think of the reforming procedure I have to do for the caps in my large CNC VFD if left off more than 3 months. Or the transient noise for the first few minutes in pres or some mics  that have been on the shelf a long time.

Be clear about non-polar technology... film is polarity insensitive but I have taken some non-polar electrolytic apart to find two polars inside.

My preference when possible is to avoid electrolytic caps in audio path, while in my lifetime I have used millions that way.  :-[

Seriously looking at button cells too for power off. Depends on leakage for an old cap worse case.
Call it belts and suspenders. But SSL at least seriously thought about it.

Les
L M Watts Technology

Batteries? Not my cup of tea for long term reliability. Too many bad memories of leaky batteries on digital boards before we had flash memory.

JR

 

[emrr]

I'm assuming you've been over all of this:

http://www.proaudiodesignforum.com/forum/php/viewtopic.php?f=6&t=14

 

[leswatts]

except two caps in series need to be 2x C to effectively deliver 1C.

Right. The T bias helps a bit there.

Batteries? Not my cup of tea for long term reliability. Too many bad memories of leaky batteries on digital boards before we had flash memory.

Yeah some very long term newer batteries are advertised now. Haven't evaluated them. One thing I did check on is the electret I mentioned.
As far as I can tell, it could supply a potential depending on cap leakage current. But I think that's not enough. Forming, reforming, or maintaining a dielectric is an electrochemical reaction that requires current...no matter how small. An electret can supply none. Oh well.

I'm assuming you've been over all of this:

http://www.proaudiodesignforum.com/forum/php/viewtopic.php?f=6&t=14

Sure have. ALL over it.

Even though this pre is just a one off for our studio, I can't help thinking of things as products. We'll probably do a pre to compliment our mic line. funny thing...due to our vintage marketing vibe it will probably be transformer /tube. Especially since we make OEM transformers for microphones. But I like to have a very clean pre in the studio, and it's hard to beat the THAT products. It's just that their old "Phantom menace"
film cap version in the 1512 demo board didn't work out too well in practice.

Les
L M Watts Technology

 

[emrr]

cool.  and these:

http://www.proaudiodesignforum.com/forum/php/viewtopic.php?f=6&t=256

http://www.proaudiodesignforum.com/forum/php/viewtopic.php?f=6&t=559

 

[leswatts]

I did look at those...thanks Doug.

I found that the offset error was pretty high at high gains with the menace configuration. Even with an OP-07 servo.
I didn't trim it though...and really didn't want to.

Les
L M Watts Technology