Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[Marcus Black]

Hi

Does anyone have experience with the Audio Upgrades mic pre cards ? Here: http://audioupgrades.com/products/products_pre_card.htm

Would I (not a tech but have good soldering skills and some basic common sense when it comes to electronics) be able to build a compact, portable, battery powered stereo mic pre out of these ?

Is it as simple as adding the 9-18V feed, mic iP and line out connectors and a gain pot ? What about +48V phantom, will it provide that although battery powered ?

Thanks.

[ssltech]

You should need plus AND minus 9-18Volts (I'd go with plus/minus 18V to avoid the significant headroom penalty), and some means of generating +48V if you plan on using phantom. -This CAN be done using a square-wave oscillator (555 timer IC) and a diode/capacitor voltage multiplier, but beware... I don't recommend doing it that way, since you're building an oscillator near the most delicate of signals; -that's asking for trouble, though kit makers like PIAA do it, I think.

If you'll NEVER use phantom, you can probably take the opportunity to bypass the input capacitors, but Jim Williams will be the person to ask. -I have no idea if the output is balanced or unbalanced... but I'd guess unbalanced.

I have a 9V-battery powered 48V microphone phantom supply, in case you need such a thing.

Keith

[Jim Williams]

There are 8 wires to it, balanced mic in, unbalanced line out, two wires to the gain pot and + - power and ground.

So it needs a gain pot, XLR inputs, + 48 if you want phantom power, + - 15~18 volts and an output jack. It draws 9 ma idling. Output current is 100 ma and it will drive 150 ohm loads.

Noise is -129.6 EIN with a 150 ohm source. That goes down with a lower input impedance so all noise is from the source. Input impedance is 4 k ohms. Output impedance is 56.2 ohms. Gain will vary from +13 to +60 db with a 2.5k reverse log gain pot.

THD+noise is .0006%. CCIF IMD is .0002%. Bandwidth is -1 db at 200k hz. Slew rate is 3000 v/us.

[John Roberts {JR}]

I hope that's fast enough...  

JR

[Jim Williams]

I do have faster stuff if interested.  I can load up the High Speed mic preamp (rack version) with TI THS3061 current feedback opamps. Those run at 300 mhz and slew at 7000 v/us.

I also have those parts designed into a custom DAC designed by Michal J. of Mytek and me for Mark Levinson Labs using a BurrBrown PCM1704 DAC chipset, it sounds wonderful and very open and clear. I used the current feedback THS3061 as a current to voltage converter and as a 2 pole Sallen Key low pass bessel filter set to 65k hz.

[John Roberts {JR}]

Jim Williams wrote on Sun, 06 February 2011 10:46

I do have faster stuff if interested.  I can load up the High Speed mic preamp (rack version) with TI THS3061 current feedback opamps. Those run at 300 mhz and slew at 7000 v/us. I also have those parts designed into a custom DAC designed by Michal J. of Mytek and me for Mark Levinson Labs using a BurrBrown PCM1704 DAC chipset, it sounds wonderful and very open and clear. I used the current feedback THS3061 as a current to voltage converter and as a 2 pole Sallen Key low pass bessel filter set to 65k hz.

Nah, I'm OK. Just kidding you....  

I recall back a few decades ago some fell into the trap of ASSuming that specialty opamps designed for S/H applications, with very high slew rate, and fast recovery time, would translate into good audio amps. However the high slew rate in those specialty opamps was attained by saturating the input stage and producing a nonlinear input stage transconductance. Nice for the data sheet max speed but not that sweet for linear operation.

In the specific case of driving A/Ds or CODECs you do need gain bandwidth well in excess of normal audio bandwidth. While you don't literally need to slew any faster, you do need to damp very high edge rate current spikes dumped out the input pins by internal S/H circuitry or whatever.  

At some point it seems there may be some value in shunting the opamp buffer output with a HF cap to ground. Do these uber-fast opamps tolerate much (any?) capacitance on their output ? I guess if they are fast enough they can do it all with active NF.

JR

[bruno putzeys]

John Roberts  {JR} wrote on Sun, 06 February 2011 18:25

In the specific case of driving A/Ds or CODECs you do need gain bandwidth well in excess of normal audio bandwidth. While you don't literally need to slew any faster, you do need to damp very high edge rate current spikes dumped out the input pins by internal S/H circuitry or whatever.   At some point it seems there may be some value in shunting the opamp buffer output with a HF cap to ground. Do these uber-fast opamps tolerate much (any?) capacitance on their output ?

Ahem.

[John Roberts {JR}]

bruno putzeys wrote on Mon, 07 February 2011 02:12

John Roberts  {JR} wrote on Sun, 06 February 2011 18:25 In the specific case of driving A/Ds or CODECs you do need gain bandwidth well in excess of normal audio bandwidth. While you don't literally need to slew any faster, you do need to damp very high edge rate current spikes dumped out the input pins by internal S/H circuitry or whatever.   At some point it seems there may be some value in shunting the opamp buffer output with a HF cap to ground. Do these uber-fast opamps tolerate much (any?) capacitance on their output ? Ahem.

Yup, that is what I was hinting at, but I was just curious to see what the several thousand V/uSec crowd thinks is optimal.

There are also topologies that allow for a real C to ground in the input path ( 2 pole MFB LPF)


I learned decades ago when trying to make effective active anti-aliasing/anti-imaging filters for bucket brigade delay lines, with the slow opamps of the day ('70s), that you couldn't just dump digital edge rate switching currents into those opamps and expect them to behave.

Finesse vs. brute force perhaps, while a combination of both doesn't suck.

JR
 

[Jim Williams]

I used Sallen Key filter topologies with the THS3061 as they don't include a capacitor in the feedback loop, a no no for transconductance or current feedback opamps.

For voltage feedback opamps I prefer the Multiple Feedback (MFB) filter topology.

As one can see in Bruno's schematic, there is a 22 ohm resistor placed into the feedback loop in series with the opamp's output. That isolates any capacitive load from the opamp's output pins and prevents ringing, or worse. It's needed to allow that 1 pole passive LPF used to feed the ADC to operate without destroying the opamp's phase margin.

The same concept was used on the Mark Levinson DAC, it has series resistors isolating the opamp from the load. Since that TI part is class A and has mondo output current, driving loads is not a problem. Transconductance opamps have a different sound than voltage feedback opamps, I can't really put my finger on it, but they do sound more like tubes than voltage feedback devices. FFT sweeps show differences, but they are so far down in the noise floor I'm not sure if they are the source of those differences.

This is a subject I've not seen explored fully yet, the application of transconductance opamps in audio. I began using them in the early 1990's as they are fun to mess with as all those old opamp rules are re-written.

[bruno putzeys]

Jim Williams wrote on Mon, 07 February 2011 18:25

It's needed to allow that 1 pole passive LPF used to feed the ADC to operate without destroying the opamp's phase margin.

It's a second order LPF. The shunt capacitor is inside the loop so you get a complex pair of poles. Output impedance is that of a damped parallel LC circuit. It's low at audio frequencies and at RF, and peaks somewhere in-between (at the corner frequency, obviously).

This makes it quite perfect for driving sigma-delta ADC's. Cut-off frequency would be lower than in the example btw.

Current-feedback op amps can be used in MFB circuits, simply add the minimum feedback resistor directly in series with the inverting input.

The nice thing about current feedback, or more properly "transimpedance amplifiers" is that you can always optimise loop bandwidth. Externally compensated op amps offer the same benefit. The trade-off between closed loop bandwidth and loop bandwidth which transimpedance amplifiers evade is only a result of having a fixed, non-user-controllable compensation capacitor and a fixed, non-user-controllable input transconductance. By making the input conductance external and part of the feedback network, transimpedance amplifiers give this control back to the designer.

Of course, since MFB circuits require unity gain stability, using transimpedance amplifiers to implement them offers no benefits since you effectively have to compensate them (using the series resistor) to be unity voltage gain stable. So the fact that you can do it doesn't imply that it's a particularly good idea.

Sallen-Key lowpass filters seem a bit impractical to me because the output impedance of the amp gets in the way of good HF suppression. MFB is vastly superior in that respect. For highpass filters, the roles reverse.

[John Roberts {JR}]

Jim Williams wrote on Mon, 07 February 2011 11:25

I used Sallen Key filter topologies with the THS3061 as they don't include a capacitor in the feedback loop, a no no for transconductance or current feedback opamps. For voltage feedback opamps I prefer the Multiple Feedback (MFB) filter topology. As one can see in Bruno's schematic, there is a 22 ohm resistor placed into the feedback loop in series with the opamp's output. That isolates any capacitive load from the opamp's output pins and prevents ringing, or worse. It's needed to allow that 1 pole passive LPF used to feed the ADC to operate without destroying the opamp's phase margin. The same concept was used on the Mark Levinson DAC, it has series resistors isolating the opamp from the load. Since that TI part is class A and has mondo output current, driving loads is not a problem. Transconductance opamps have a different sound than voltage feedback opamps, I can't really put my finger on it, but they do sound more like tubes than voltage feedback devices. FFT sweeps show differences, but they are so far down in the noise floor I'm not sure if they are the source of those differences. This is a subject I've not seen explored fully yet, the application of transconductance opamps in audio. I began using them in the early 1990's as they are fun to mess with as all those old opamp rules are re-written.

I will add another vote against the Sallen and Key for dealing with extreme HF artifacts. While it gives you, input isolation, the output is not.

There is a phase shift associated with that isolation technique that needs to be factored into your phase response if the pole is close to your passband of interest. However with modern opamps that pole can be placed high enough to not be an issue for audio frequencies.

I think I know what you mean but transconductance opamps make me think of OTAs (operational transcondiuctance amplifiers).  Like the old CA3080 and many derivative parts. Most of the descriptions I find for transimpedance sound like conventional amps optimized for virtual earth topology, or could even be describing the old Norton amp topology (hows that for obscure?).

Based on what I think you are talking about, and from Bruno's description of their features, this sounds a little like Buff's transamp topology, often credited to Cohen after his AES paper but known to many skilled in the art back then. I even discussed this in my 1980 console article (see figure 9). While not a rigorous technical investigation, in a technical journal, certainly out there for those paying attention, and Cohen's paper in the AES certainly qualifies as publication in an audio context.  

JR

[bruno putzeys]

(for those who lost the plot)

A transimpedance amplifier is basically an op-amp like circuit with that difference that only the noninverting input is the base of a transistor. The inverting input is the emitter of the same transistor (plus a bias network to correct for Vbe). So the inverting input is a buffered version of the noninverting input and the output voltage is Gain times the current drawn from the noninverting input. Normally the input is executed in a mirror-image fashion with both NPN and PNP. The Gain of the circuit is not a voltage gain but a transimpedance i.e. a current-to-voltage conversion. It is expressed in ohms.

I've noticed that several people get the term wrong and call this circuit "transconductance amp" (probably on phonetic grounds). I've even seen this in data sheets that nevertheless correctly replicate the above explanation. Transconductance amps really do the reverse. An OTA takes a voltage and produces a current in response.

The most popular term is "current feedback amplifier" but I think it should be avoided. In power electronics the term is used to describe a circuit where the variable being fed back is the output current so as to raise the output impedance. Although in a circuit built around a transimpedance amplifier the feedback signal is a current, the thing that's being sensed and fed back is still a voltage. I could live with "current input amplifier".

John Roberts  {JR} wrote on Mon, 07 February 2011 19:48

There is a phase shift associated with that isolation technique that needs to be factored into your phase response if the pole is close to your passband of interest.

It's no different from any LPF in that respect.

[Jim Williams]

I missed that 100 pf feedback cap as another pole, most ADC's used for audio use a passive 1 pole filter in front of the ADC, after the opamp buffer. In this example it's included into the feedback path to create a 2 pole response.

As to current feedback opamps for audio, there is one designed for audio, the National LME49713. It's rather linear for a CFA and as usual has low input voltage noise. The problem with using these for audio is two fold, first the inputs are not matched, the inverting input feeds the emitter of the input transistor rather than the traditional base. The other audio problem is current noise, it's through the roof.

If used with low value resistors in a non-inverting configuration it can work quite well. Bandwidth is set by the feedback resistor value instead of the usual RC ratio.

The original Audio Upgrades High Speed mic preamp design used Analog Devices AD811's, a CFA opamp. Other than that, I know of no commercial audio products using this class of amplifiers.

[John Roberts {JR}]

Jim Williams wrote on Tue, 08 February 2011 10:16

I missed that 100 pf feedback cap as another pole, most ADC's used for audio use a passive 1 pole filter in front of the ADC, after the opamp buffer. In this example it's included into the feedback path to create a 2 pole response. As to current feedback opamps for audio, there is one designed for audio, the National LME49713. It's rather linear for a CFA and as usual has low input voltage noise. The problem with using these for audio is two fold, first the inputs are not matched, the inverting input feeds the emitter of the input transistor rather than the traditional base. The other audio problem is current noise, it's through the roof. If used with low value resistors in a non-inverting configuration it can work quite well. Bandwidth is set by the feedback resistor value instead of the usual RC ratio. The original Audio Upgrades High Speed mic preamp design used Analog Devices AD811's, a CFA opamp. Other than that, I know of no commercial audio products using this class of amplifiers.

I sure miss the good old days when IC makers published schematics of their inner circuitry. I am inclined to speculate that this topology has indeed been used in professional audio products for decades. The primary difference is that the discrete transistors with opamps wrapped around them are now integrated inside a single IC. Which just wasn't practical a few decades ago. but we have enjoyed the benefits of the topology for some time.

JR

PS: While I haven't paid close attention the the early mic preamp ICs, it seems they too used variations on this circuit topology.  

[johnR]

John Roberts  {JR} wrote on Tue, 08 February 2011 17:18

I am inclined to speculate that this topology has indeed been used in professional audio products for decades.

I've got some Neve BA338 and BA340 amp modules in front of me right now. They use the emitter of the input transistor for the inverting input (although they would normally be used with an external dc blocking capacitor).