Portable mic preamp project

[carlmart]

Hi,

Have just joined the club (thanks, Wayne!) and I am open to tips and suggestions.

I am involved in designing a 2-mic small sized, affordable, balanced mic preamp, with 48v phantom power, single-supply (12 to 14.4v), 12-segment or more LED metering and, did I mention affordable?

Even if I should be using THAT 1510 chips, the basic design I am starting on is a mix of their suggested schematic and the one I used on an SSM2017 based preamp I built in 2000. Which was also based on AD's design for that chip by Walt Jung.

The different approach this time is that I want to use a single instead of a split-supply, which adds some restrictions and demands, like obligatory use of inter-stage cap coupling.

Before someone does suggest it, as always seems to happen, this project can't involve transformers (due to budget restrictions) or use DC-DC supply for the chips power. There will be a +48v DC-DC supply, but that's all.

The phantom booster is based on a Maxim chip that I have used on the previous product, that proved very effective. At the time I also tried some LT based boosters, but their RFI was much higher.

There will also be a headphone preamp and a 1KHz tone generator.

Undecided yet on whether to use a THAT based limiter in the second stage, as my aspiration had been to use a limiter on the first stage, with an LDR. Vactrols now seem to be banned or in the way to be, so I wonder what should I do.

Another area which is critical, considering my budget, is the VU-meter. It has to be LED, to be able to see it in the dark and from a distance. I'm looking for bars, but the longest I could find, easily drivable, was just 12-element long.

[mediatechnology]

Thanks for joining us and posting!
Sounds like a fun project.

The different approach this time is that I want to use a single instead of a split-supply, which adds some restrictions and demands, like obligatory use of inter-stage cap coupling.

You might want to look at this thread as a starter for the single-supply aspect: viewtopic.php?f=6&t=191

Glad you already got that +48V happening. Would like to see the switcher design.

[carlmart]

Here's the DC-DC switcher. It works quite well.

Presently I'm modifying my old SSM2017 design, which was fed by +9/-9v batteries into one using just one single 11 to 14.4v battery, depending on what you plug into it.

I still don't know how the different voltage will affect the output, as the 0db VU point will change as voltage goes down.

But I don't want to regulate anything, because that will dissipate as heat and wasted amps, which on a portable design is more important than wasting a few dBs. I did ask THAT for a voltage-output curve for the 1510, which they will hopefully provide.

I'm not planning on using a pot for the gain, but a three position switch, as I did on my other design. There will be a pot at the output, after the cap and before the passive bass cut, so it did cover all the levels you might need.

What I do have to be aware of is to never let power get less than 10V, so 11v will be my low battery alarm level.

The output doesn't need to be balanced, so I was planning on using a dual opamp there, which on my original design was an OP-275.

But as I stated, I need to solve the VU-metering question. Also adding a headphone amp (nothing too fancy or current hungry) and a 1KHz generator.

[JR.]

Here's the DC-DC switcher. It works quite well.

Presently I'm modifying my old SSM2017 design, which was fed by +9/-9v batteries into one using just one single 11 to 14.4v battery, depending on what you plug into it.

The audio shouldn't know much difference...just make a solid V/2

I still don't know how the different voltage will affect the output, as the 0db VU point will change as voltage goes down.

?? 0VU should be a fixed voltage level independent of the battery.

A bar graph (solid bar) will waste current unless you put them all in series. I would just light up a dot display and you might even consider driving the LEDs with a pulsed capacitor. The LEDs these days are very efficient and yiou do not need a solid diapsly

But I don't want to regulate anything, because that will dissipate as heat and wasted amps, which on a portable design is more important than wasting a few dBs. I did ask THAT for a voltage-output curve for the 1510, which they will hopefully provide.

Running from battery you should not have hum or PS noise to worry about.

I'm not planning on using a pot for the gain, but a three position switch, as I did on my other design. There will be a pot at the output, after the cap and before the passive bass cut, so it did cover all the levels you might need.

What I do have to be aware of is to never let power get less than 10V, so 11v will be my low battery alarm level.

The output doesn't need to be balanced, so I was planning on using a dual opamp there, which on my original design was an OP-275.

it cost about a penny to add a balancing impedance.

But as I stated, I need to solve the VU-metering question. Also adding a headphone amp (nothing too fancy or current hungry) and a 1KHz generator.

Headphone amp and LED meter will consume battery life. But it's your battery.

JR

[carlmart]

What exactly would it mean to make a solid V/2? Just one well filtered divider to take care of all V/2? Buffered or unbuffered?

I think the only way to keep repeatable values is using a regulator. Other than that there will be small variations. Maybe on a passive VU-meter, with diodes and needles, that wouldn't be so, but in active ones, with leds and all, I think it might. Am I wrong?

Please do go on on what you think might be practical solutions for the vu-meter.

If you balancing resistors, on what I call faked balancing, I already used that and it does work. But in this case I should be mostly interfacing with unbalanced inputs.

Battery size shouldn't be a problem on this application, within certain parameters. And I do need good monitoring for audio and level.

[JR.]

What exactly would it mean to make a solid V/2? Just one well filtered divider to take care of all V/2? Buffered or unbuffered?

Since V/2 is taking the place of ground, it needs to act like ground in the context of the application. Since I do not know the exact circuit I cannot know what might be a problem.

XLR pin 1 which is also 0V for the mic phantom current return, if connected to v/2 (fake ground). Should be able to accept typical mic current. There may be a way to connect pin 1 to actual 0V with non-polar or double polar input blocking caps.

I think the only way to keep repeatable values is using a regulator. Other than that there will be small variations. Maybe on a passive VU-meter, with diodes and needles, that wouldn't be so, but in active ones, with leds and all, I think it might. Am I wrong?

yes the meter can have an internal reference voltage, could be as simple as a couple diode drops.

Please do go on on what you think might be practical solutions for the vu-meter.

I have designed too many to list. As I said before to conserve energy I would only light 1 led at a time, and perhaps strobe it with a cap to use even less power. Or some combination of a cap and resistor, so when it first hits, it lights brightly but if steady on, the light dims.

Wayne has a whole thread somewhere with simple LED meters ,,, maybe something using a LM339 comparator.

If you balancing resistors, on what I call faked balancing, I already used that and it does work. But in this case I should be mostly interfacing with unbalanced inputs.

For single ended interfaces ground integrity is important. While battery power should keep PS noise at bay, keep in mind that V/2 Fake ground will be the hard 0V references for audio output.

Battery size shouldn't be a problem on this application, within certain parameters. And I do need good monitoring for audio and level.

If battery size is not a concern, two batteries could provide a split supply and simply using typical application notes.

JR

[mediatechnology]

What exactly would it mean to make a solid V/2? Just one well filtered divider to take care of all V/2? Buffered or unbuffered?

It's gotta be buffered. Use this for V/2 and input bias.


THAT1510 1512 Single Supply Preamp

Pin 5, the 1510 Reference connection must see a low impedance V/2 source or the CM rejection will degrade.
Note that the input capacitors are required to be bipolar or bipolar-connected since the polarity under phantom on/off conditions reverses.
The good news is the right-hand pair are continuously formed.

I think you'll find that 9V will be the minimum the 1510 will operate. 10V is the spec'd minimum. Your 10V min/11V alarm should be OK.
The MC33178 (SMT only now) pulls very little current.
You could also use a MC33078 which is still available in DIP but it pulls more current.

IIRC the 1510's p-p output level is about 4V less than the supply.
If you were to use balanced output drive you could pickup - at least with balanced loads - +6 dB more output level.
You could then run the 1510 with 6 dB less overall gain to conserve headroom with low battery voltage.

[carlmart]

I have three suggestions for implementing an analog supply on several chips: SSM2019, INA217 and THAT 1510.

What I wonder is they will work on all those chips with the mods suggested, particularly the one I called SSM2019_single2

[mediatechnology]

Based on a quick glance they all will work across the THAT1500-series, INA 217 and SSM2019.

I don't recommend the Zener clamps because under phantom fault conditions they will likely short.
I'd use the diode bridge to rails circuit I posted with 10R series resistors. (R5 and R6).
R5 and R6, at 49R9 add 100 additional Ohms of source impedance. With 10R/leg it drops to 20 Ohms.
The reason can be found here in a paper Rosalfonso Bortoni and I co-authored: http://www.thatcorp.com/datashts/AES790 ... eturns.pdf

C1 and C2 in the first two drawings may be too small.
The differential impedance in the first example is approx 2K1 making the cutoff 160 Hz.
The second drawing has about 20K Zin-diff making the cutoff 16 Hz.

In the second example, the Zeners are also already pre-biased to 1/2 the supply which may cause them to limit too early.

The first example, which has T-bias, provides a common mode input impedance on the right-hand side of C1 and C2 that is roughly 10 times higher than what it would be if the 22K were 0R.
This improves the LF common mode rejection considerably because it lessens capacitor matching requirements.

If you do switch phantom I'd put the switch after C3.
It's just a personal preference but the peak turn-on current can be quite large if the 48V supply is fairly stiff.
If the 48V supply was capable of it the peak turn-on current could be as high as about 480 mA.
You might also want to back-ground R3 and R4 when phantom is off to discharge C1 and C2.

C1 and C2 cannot be polarized in the first or second examples.
When phantom is switched off the polarity is one direction: The right-hand side is more positive.
When phantom is on the left-hand side is more positive.

"single 2,"with a few adjustments, should work with the THAT1510,THAT1512, INA217 or SSM2019.

[ricardo]

I don't recommend the Zener clamps because under phantom fault conditions they will likely short.
I'd use the diode bridge to rails circuit I posted with 10R series resistors. (R5 and R6).
R5 and R6, at 49R9 add 100 additional Ohms of source impedance. With 10R/leg it drops to 20 Ohms.
The reason can be found here in a paper Rosalfonso Bortoni and I co-authored: http://www.thatcorp.com/datashts/AES790 ... eturns.pdf

I strongly second the THAT diode bridge protection with 1n4004s even if you use 2019 or the Texas equivalents.

My experience of this was not even a P48V application but Factory Test gear with the 2017/9 and their recommended circuit. They died with depressing regularity. Switching to the THAT bridge protection made the problem go away.