dual Vs single opamp for stereo mic preamp

[pstamler]

For a Zobel I'd go with Yamaha's recommendation; they generally knew what they were doing. The 1.6uF cap means that over most of the audio spectrum (everything above 62Hz) the transformer's load is 1.6k || 20k, or 1481 ohms. That means your mics are seeing an input Z of about 300 ohms, way too low for most modern mics.  Really, go back to the Yamaha Zobel.

About opamps: the LME49710 has an ideal source impedance of 1562 ohms; the LME49990 has an ideal source impedance of 338 ohms. Since, as I recall, your transformer's secondary is 750 ohms it's about midway between those two ideals, and should give fairly quiet results; however the LT1128; ideal source impedance is 850 ohms which is closer than either of the others. Forget FET imput opamps like the OPA604 and OPA627; they're only quiet with high-turns-ratio transformers, which yours aren't.

Figure out the input Z of your recorder's line input; unless it's 600 ohms the opamp should be happy driving it without the booster transistors; without them you'll decrease both current draw and stability problems.

The sparks, however, don't sound like oscillation to me; they sound like two things that are too close to each other.

Peace,
Paul

 

[spaceludwig]

Paul,

for the sake of clarity, here is the schematic of Yamaha's input module:

Questions: The two 47pF caps to ground at pins 2 & 3 of the XLR input are to flush high frequency garbage? The 220R/470pF going into the opamp's non-inverting input is a 1.5 MHz lowpass filter?

About opamps...

Thanks for the info regarding the FET input opamps. My observations of the OPA627 was that is seemed to have a very effective distortion/interference/noise cancellation circuit going on as on the breadboard the output (of the circuit on page 1) exhibited a lot less of what seemed like externally generated noise (rf, hum, etc) than most of the other opamps I tried. Though I'm very much looking forward to trying the others mentioned, not as an exercise in finding characteristics that match the beyond-ridiculous rhetoric used on HiFi audio forums but too see which has the least audible noise when gain is cranked to 11.

unless it's 600 ohms the opamp should be happy driving it without the booster transistors

That would be sweet indeed if I could keep the design to a bare minimum - basically throw a transformer in front and after an opamp with volume pot and voila! I think the circuit would consume less than 10mA notwithstanding phantom.

 

[ricardo]

for the sake of clarity, here is the schematic of Yamaha's input module:

Err..rh!  Yamaha don't use a Zobel.  The BP 1/16 is just the input coupling cap.  I second Paul's advice.  Use Yamaha's Zobel.  ie none.  Removing your 1u + 1k6 will give mikes a much better time.

My observations of the OPA627 was that is seemed to have a very effective distortion/interference/noise cancellation circuit going on as on the breadboard the output (of the circuit on page 1) exhibited a lot less of what seemed like externally generated noise (rf, hum, etc) than most of the other opamps I tried.

That means your breadboard is oscillating.  Make sure you have some 10u electrolytics decoupling the power rails AT the OPA.

That would be sweet indeed if I could keep the design to a bare minimum - basically throw a transformer in front and after an opamp with volume pot and voila!

You could even throw out the output transformer  and run everything on a single 9V battery as in my MAX410 circuit.  ;D

LME49710 / LM4562 would work in that circuit with more current, very slightly more noise and a tendency to latch on switch on.

 

[pstamler]

Questions: The two 47pF caps to ground at pins 2 & 3 of the XLR input are to flush high frequency garbage?

Yep.

The 220R/470pF going into the opamp's non-inverting input is a 1.5 MHz lowpass filter?

Actually, since the mic's impedance, transformed up through the transformer, is in series with the 220R resistor, it's a lowpass filter at a lower frequency. With a 150 ohm mic (assuming it's resistive), about 395kHz.

Thanks for the info regarding the FET input opamps. My observations of the OPA627 was that is seemed to have a very effective distortion/interference/noise cancellation circuit going on as on the breadboard the output (of the circuit on page 1) exhibited a lot less of what seemed like externally generated noise (rf, hum, etc) than most of the other opamps I tried.

That's not surprising. FET-input opamps are much better at ignoring RFI crud than bipolar-input opamps. I remember building a preamp for my father, who lives a few blocks from the antenna farm on top of the Hanc**k Building in Chicago; only FET-input opamps would work.

In your case, it shouldn't be a big issue once the circuit goes into a shielded box; the transformer's inherent bandlimiting, the little bypass caps on the XLR jack, and the lowpass filter ought to keep RFI thoroughly at bay, even with a bipolar-input opamp (which I'm sure is what Yamaha used). Looking at that schematic, though, I can see why Yamahas had a reputation for being exceptionally RFI-proof.

Though I'm very much looking forward to trying the others mentioned, not as an exercise in finding characteristics that match the beyond-ridiculous rhetoric used on HiFi audio forums but to see which has the least audible noise when gain is cranked to 11.

Once it's inside that shielded box, the LT1128 should be the quietest. See earlier posts.

unless it's 600 ohms the opamp should be happy driving it without the booster transistors

That would be sweet indeed if I could keep the design to a bare minimum - basically throw a transformer in front and after an opamp with volume pot and voila! I think the circuit would consume less than 10mA notwithstanding phantom.

And ought to sound quite good. Don't forget to provide good decoupling capacitors on the DC rails; a couple of 1000uF 25V electrolytics, plus something like 0.1uF stacked-film polypropylenes near each IC package's supply pins, ought to do it.

And I second Ricardo (who, admittedly, was seconding me): go with Yamaha's original termination network for the transformer. It'll give you a slightly low load for the microphones (about 850 ohms). Most mics will do okay on this, and SM57s will positively thrive.

Peace,
Paul

 

[spaceludwig]

Use Yamaha's Zobel.  ie none.  Removing your 1u + 1k6 will give mikes a much better time.

Hmm, problem is without them the square wave is all over the place. Perhaps I should upload a picture to illustrate my point.

 

[ricardo]

Use Yamaha's Zobel.  ie none.  Removing your 1u + 1k6 will give mikes a much better time.

Hmm, problem is without them the square wave is all over the place. Perhaps I should upload a picture to illustrate my point.

Yamaha provides all the damping via the 4k7 across the secondary.  That value would be my starting point for a Zobel.
___________________________

But before we go on, you need to consider what you are REALLY trying to do.

Is it to make a portable very low noise preamp to be used with a portable recorder?

Or is it to make something resembling a JLM Neve copy, perhaps as a Golden Pinnae project?

Do you really need continuously variable gain up to 2000x?

If you look at LNPrimer.doc in my MicBuilders directory, you'll see how my MAX410 design starts with the requirements; the soundcard it is feeding and the microphone to be used.  These decide the gain you need for good 16b digital recording.  The requirement for single 9V battery operation decide the MAX410 and so on and so forth ...
__________________

But to answer just one of the questions you need to ask, assuming you want a "standard" mike input impedance of 1k5 - 2k with P48V.

Then using the Yamaha 600:3K; a 1:5 impedance step up, making your R2 = 11k will show the mike 1k89 once you factor in the 2x6k8 P48V resistors.  This gives about 1/2 the damping that Yamaha have.

I would start with a Zobel of 10n + 11k across it.  The two 11k in parallel gives 5k5 which is near Yamaha's 4k7

Feed the transformer from a square wave generator with source resistance of 150R cos this is worse case for damping.  Look at the output of the whole preamp.  Adjust  R1 = 11k until the overshoot ju...ust goes away.  Do it by soldering in new resistors instead of twiddling a pot.  You can now REDUCE C1 = 10n by soldering in new capacitors until the overshoot returns.  Then increase R1 again and so forth until you can't increase R1 or reduce C1 any more.

Alternatively, just use the EXACT Yamaha circuit.  4k7, 1u, 47k, 220R, 470p exactly as they show it.  I don't think it's optimum but its likely to be OK.

But you need to be clear what you are trying to achieve first.

 

[spaceludwig]

Hi Ricardo,

Thanks for the thorough answer and sharing the methodology. I will try that next weekend when i have some spare time.

To answer your qeustion: I would like a clean and quite portable stereo preamp with Phantom power to use with a stereo microphone I just bought

http://www.superlux.com.tw/productInfo.do?pdctid=cd52cb8a-ec05-441f-a00d-96924cbf5122&pdkid=0d123a9c-ae09-4a09-b5c8-482af0209800&level=2&lv0=1#skill_zone_li

I would like to run each channel on a 9V battery. The amplified signal would be going directly into the line-in of a portable stereo recorder.

 

[spaceludwig]

I have etched a new board minus the current boosting transistors at the output. The circuit works in so much as it amplifies but i do not get the signal at the output of the transformer.

Could someone confirm that the output of the opamp should go to pin 1 of the LL1517, and pin 4 grounded (with the rest wired as per the diagram, see attached picture below)?

 

[PRR]

> on a 9V battery. ...directly into the line-in of a portable stereo recorder.

Why do you need output transformers? Or buffers? Short wire, no common ground: drive the line unbalanced or balanced-impedance.

> -36 dBV/pa(15.8 mV)

Why do you need heaps of gain? Or Low-Low noise?

With mikes like that, in a fixed installation calibrated to 0dBsf=2.8V peak, I had 40dB gain in the chain but normally potted-back 10dB to 30dB (small string quartet to large orchestral); 30dB to 10dB working gain. In a location rig with one recorder you would turn-up the recorder as-needed, and adjustable line inputs should surely go well under 1V peak, for another 10dB gain in chain. So IMHO your preamp needs gain of 20dB max. You already have 6+dB in transformer. You need gains of 14dB and say unity/0dB in the chip.

It would not be hard to build a superlative preamp with two transistors eating a shade over 1mA per channel.

Unless your design is really bad, the noise figure is entirely in the head-amp inside the microphone, not the preamp. The SuperLux specs work out to 2.5 microvolts of output hiss.... a selected TL072 will go there, a 5532 will give perfect reproduction of the head-amp's hiss.

I would trust Yama on the transformer loading, a bit over 4K.... exCept the SuperLux is rated minimum load 1K. That may only apply to max SPL, but that's not amazingly generous. So if the iron is rated 600:3K, and you need to show 1K to the mike, the secondary load should be 5K or higher.

BTW: Zobeling and other tricks tame high-ratio transformers. Low-impedance and low-ratio windings give MUCH less treble-trouble and at much higher freqs. Any reputable pro-audio 1:2.2 winding should be flat far above your ADC's input filter (probably over 70KHz).

RFI on the breadboard should not guide your development unless you need to work under high-power radio sources; that's what the box (and transformer) are supposed to prevent.

 

[spaceludwig]

Hi PRR,

Why do you need output transformers?

I probably don't. I have some output transformers and thought this might be a good time to integrate them as it gives me some flexibility with the preamp if ever I want to use in other applications. Admittedly, the unit would be smaller and lighter without them (stereo=2)

drive the line unbalanced or balanced-impedance.

I do not understand what this means.

Why do you need heaps of gain? Or Low-Low noise?

Ambient sounds a very very low, much lower than someone speaking softly into a microphone so I wanted to make sure i had adequate gain to boost and also in the event I am using a quiter mic. Low-low noise for the same reason. 2000x is probably too much for this particular application but it's not like I have to use all the gain I have available.

It would not be hard to build a superlative preamp with two transistors eating a shade over 1mA per channel.

That is beyond my skills at the moment. Though I certainly wouldn't be opposed to having a pre with 30~40db run from a few transistors.

Upon consideration I would be okay losing the transformer but I still want to understand why they are not working for future reference. I have obviously looked them up wrong but I don't understand where.

 

[ricardo]

I have some output transformers and thought this might be a good time to integrate them as it gives me some flexibility with the preamp if ever I want to use in other applications. Admittedly, the unit would be smaller and lighter without them (stereo=2)

drive the line unbalanced or balanced-impedance.

I do not understand what this means.

Have a look at the circuit I posted.  There is both an XLR output (balanced-impedance) and an RCA phono output (unbalanced).  The XLR output has better distortion & drive than your output Lundahl and at least as good balance if you match the 47u electrolytics.  This is not a cheaper or less flexible alternative.  It is a better one.

Ambient sounds a very very low, much lower than someone speaking softly into a microphone so I wanted to make sure i had adequate gain to boost and also in the event I am using a quiter mic. Low-low noise for the same reason. 2000x is probably too much for this particular application but it's not like I have to use all the gain I have available.

Too much gain on one knob makes it more difficult to adjust accurately.  It also compromises the noise performance at one end or other.

The MAX410 circuit has from 12 - 42dB gain.  With your Superlux, the minimum gain is at the maximum that PRR recommends.  With the 7dB from the i/p transformer, this is more than enough gain for any mike unless you are using an STC4038 with original 30R nominal trasformer.  MAX410 takes 1/2 the current / OPA of LME49710 etc.

Once your analogue noise exceeds, the 16b dither level by a sensible amount, there is no advantage to more analogue gain.  I explain this at MicBuilders

Do you need continuously variable gain?

This is where guru PRR comes up with his 2 transistors & a bent pin circuit which is 30dB better in all respects than anything I could dream up 

 

[spaceludwig]

Have a look at the circuit I posted.  There is both an XLR output (balanced-impedance) and an RCA phono output (unbalanced).

Ah, ok, offer both balanced and unbalanced. I don't have a problem with that and don't even mind using IC driver such as DRV134 to accomplish the task. However, I would still like to understand why the transformer is not working in my circuit. Have I hooked it up incorrectly?

This is where guru PRR comes up with his 2 transistors & a bent pin circuit which is 30dB better in all respects than anything I could dream up 

Is there a link to this circuit?
[/quote]

 

[PRR]

> Ambient sounds a very very low, much lower than someone speaking softly

Of course. But soft-speech usually does not have to hit the wax at MAX level. The real goal in soft work is to get the "unavoidable noise" (room or microphone) up above wax scratch (tape hiss, ADC bottom bit). Taking soft speech as 65dB SPL and that mike (most good mikes) hiss near 15dB SPL, there's 50dB from "unavoidable noise" to desired sound. In days of Dolby B tape, that means I could capture the entire acoustic spread from room/mike to soft or medium speech; but for loud speech I'd have to turn-down and let room/mike hiss fall below tape hiss. Now with 90+dB dynamic range, we can set 15dB SPL just over digital garbage floor, let 105dB SPL fall near 0dBfs. There's a LOT of acts that don't peak at 105. When you face a clavichord in a large chapel, 70dB SPL, it may peak at -35dBfs. It is entirely reasonable for much soft music to never twitch the -20dB mark on tape VU meters. As long as your room/mike hiss is significantly above your digital dither, it's fine, and you can normalize later.

But maybe I'm getting ahead of you.

Or maybe it's just April. (April was when all the cellists recited to close out the semester.... I still have nightmares of running out of tape and yet another Bach Suite to go....) (Insert rant about you young guys can put a year of music on a thumb-card and never ran around with boxes of tape just to get through the evening....)

> would be smaller and lighter without them

I confess to heavy bias. I did live recording, both fixed and location, for some decades. After 5 recitals in 3 different rooms, every day for weeks on end, the LAST thing you want is ANY extra weight. Yes, a hobbyist going-out a few times a year may not mind heft.

Quality is important also; but often less is more. As Ricardo mentions, good transformers are excellent but not perfect. Better than ground-loop hum across a large facility, but not as good as a short wire.

> I have obviously looked them up wrong

Another reason to K.I.S.S. early and often. I admit, if I recall the right post, it looked right to me. But I've been fooled many times. Worst was a red and black power pair, and it took months to realize the Red was _negative_. Little miscommunication between designer assumptions and user assumptions trip you up. (Clip-leads are your best friends.)

> offer both balanced and unbalanced.

No; any good "balanced" input will take an un-balanced signal. You may get more cross-talk and interference in long lines.... but here your lines may be less than a foot long and not near much other electrical crap. An intermediate case between true balanced/floating and simple unbalanced is "Balanced Impedance"; instead of hard-grounding one side, you return through impedance equal to your one live source pin. I wouldn't think of it for runs under 10 feet, and have run much longer pure-unbalanced; but ricardo showed the plan and it isn't like expensive.

> This is where guru PRR comes up with his 2 transistors & a bent pin circuit

I could freshen-up some well-proven preamps; but in fact it is not a trivial design chore (the first 90% takes 10% design time and the last 10% of validation takes 90% of tedium mixed with panic re-design) and as often as not I went with op-amps. The extra $5 (less today) and few-mA current difference does not compensate for the simplicity of op-amps.

 

[spaceludwig]

Thank you PRR-ofessor 

I'm certainly ready to move on past the balanced output as per the suggestions proffered and concentrate on the power supply (probably another headache that I've naively assumed would not be too hard to make...) but I am really bothered by the fact that the transformer is not working and would love to at least understand why or where I went wrong. I did a search but found no circuits using the LL1517 at the output, though I imagine the principal is the same for any 600:600 output xfmr?

Can anyone enlighten me on where I went wrong?

 

[ricardo]

I did a search but found no circuits using the LL1517 at the output, though I imagine the principal is the same for any 600:600 output xfmr?

Can anyone enlighten me on where I went wrong?

With reference to the bottom pic on the Lundahl datasheet.

- Do you measure 2x9R2=18R4 across p1 & 4 and 19R across p7 & 11?  If not, you may have a broken winding or not connected it correctly.
- Have you tried another transformer?
- If you have 1V 1kHz sine on the output of the OPA, do you see this across p1 & 4?
- Do you see this same voltage across p7 & 11?

You need to connect the earth on your scope to the pins on the transformer for these last 2 tests.

 

[spaceludwig]

- If you have 1V 1kHz sine on the output of the OPA, do you see this across p1 & 4?
- Do you see this same voltage across p7 & 11?

Ricardo,

Thank you kindly for the help. I will try your recommendations. I have no signal at output 7 & 11. Are pins 1 and 4 supposed to be tied to the output of the opamp or should pin 4 be grounded?

I see that my revised schematic did not upload in the post above with the LL1517 datasheet. Here it is, this is how the PCB is wired together:

*note:The space between R5 and GND is an error

 

[ricardo]

I see that my revised schematic did not upload in the post above with the LL1517 datasheet. Here it is, this is how the PCB is wired together:

My 4 questions remain the same.  Measure directly at the transformer pins for both resistance and voltage.

 

[spaceludwig]

You need to connect the earth on your scope to the pins on the transformer for these last 2 tests.

Which pins? 4 & 9?

 

[ricardo]

Which pins? 4 & 9?

p4 for the 1st sine wave test and p11 for the 2nd

 

[spaceludwig]

Do you measure 2x9R2=18R4 across p1 & 4 and 19R across p7 & 11?  If not, you may have a broken winding or not connected it correctly.

Wired correctly but resistance way off: 25k pins 1 & 4 and dead (according to DMV) across pin 7 & 11

Have you tried another transformer?

Have another that measures correctly; will desolder problem one and replace then do rest of measurements.