Development of a Low-Cost Programmable Microphone Preamp ...

1

Development of a Low-Cost Programmable Mic Preamp Gain Controller ICGreater Northwest AES Section, June2016

Copyright © 2016, THAT Corporation

Development of a Low -Cost Programmable Microphone Preamp

Gain Control IC for Pro Audio Applications

Gary Hebert, Chief Technology Officer

THAT Corporation

2



Tonight’s Presentation

• Introduction• Professional Microphone Preamplifiers• Digital Mic Preamp Gain Controllers• Earlier Products• Cost Reduction Measures• Cost-Performance Tradeoffs• Measured Performance• Conclusions

3



Who’s THAT?

• Founded in 1989– 2014 was our 25th anniversary!

• Spin-off from dbx Inc.• Founders were dbx engineers

– Paul Travaline, Gary Hebert, And Les Tyler• Once made complete pro-audio

products• Now focused on Pro Audio ICs and

Licensing

4



Professional Microphone Preamps

• Balanced (Differential) Input• Low input noise required

– On the order of 150Ω thermal noise– (-130.8 dBu in 20 Hz – 20 kHz BW)

• Wide gain range required– Mic sensitivities vary over at least 37 dB– Sound levels vary with application

• Max input level should be ≥ +16 dBu for the highest-output condenser microphones

5



Digital Control of Professional Microphone Preamps

• Many preamps are now front ends for A/D converters in digital audio products.

• Digital control of the gain gives a uniform user interface for these systems.

• It also allows enhanced automation features such as setup recall and automatic gain reduction in response to clipping.

6



Typical Preamp Front End

• Differential gain = 1+ (2RF/RG)

• CC capacitors block dc inputs and phantom power

• RG low valued at high gains to minimize noise

7



What’s a Programmable Gain Controller?

• Digitally controlled feedback network for a low-noise differential amplifier

8



Programmable Gain Controller

• RG gets small at high gains

• Small RG implies low RON switches

• RF/RG is large at high gains

9



THAT’s First Controllers

• 5171 - 1 dB/step– 13.6 dB – 68.6 dB– <.0008% THD, +24 dBu out, any gain

• 5173 - 3 dB/step– 0 dB – 60 dB– <.001% THD,+24 dBu out, any gain

• Accurate gains - +/-.5 dB max, +/-.15 dB typical

• DC servos

10



5171/5173 Topology

• Both RF and RG are varied using a combination of a tapped resistor string and a set of switched paralleling resistors.

11



5171/5173 Topology

• Tapped resistor string is used for large steps

• Tapped string switches don’t effect gain or THD, but do add noise

12



5171/5173 Topology

• Minimum RG (in red) is 5.6Ω in the 5171.

• This resistor is very wide and short.

• W/L≈109 for 610 Ω/sq. poly

13



5171/5173 Topology

• Paralleling resistors are used for small steps

• Paralleling switches are in series with high-value resistors

14



5171/5173 Topology

• Maximum parallel RF (in red) is 47 kΩ in the 5171.

• This resistor is narrow and long.

• W/L≈.013 for 610 Ω/sq. poly

15



Show Me the Money

• The 5171 and 5173 have proven to be too expensive for many applications, particularly those at the entry level where some of the possible automation features might be most useful.

• So, what makes them expensive, and what do we trade off to make a less costly part?

16



Resistors – Bigger is Better

• Ratio accuracy increases with resistor area.

• Distortion due to self heating is proportional to: ∗

.

17



Switches – Bigger is Better

• RON is inversely proportional to device width

• Low RON minimizes noise from the tap-string switches

• High voltage capability also increases area

18



Cost Reduction – the Easy Stuff

• 3 dB per step• Dual channel part

– Saves some package cost– Small savings in SPI interface area

• Eliminate servo– Reduces die area– Reduces power– Requires large external capacitor

19



New Topology

• Variable RG

• Fixed RF

• Switch RON added to RG

• ∆RON adds THD• RON variation

adds gain error

20



New Topology

• Dynamic gate drive minimizes THD due to ∆RON

• Reduced max gain (51 dB) saves die area

21



How Much Gain?

85

95

105

115

125

-20 -10 0 10 20 30 40 50

Dy

na

mic

Ra

ng

e (

dB

)D

yn

am

ic R

an

ge

(d

B)

Dy

na

mic

Ra

ng

e (

dB

)D

yn

am

ic R

an

ge

(d

B)

Gain (dB)Gain (dB)Gain (dB)Gain (dB)

Dynamic Range vs. Gain Dynamic Range vs. Gain Dynamic Range vs. Gain Dynamic Range vs. Gain –––– 150150150150ΩΩΩΩ Source, Ideal Source, Ideal Source, Ideal Source, Ideal PreampPreampPreampPreamp

114 DB A/D DynamicRange (dB)

120 dB A/D DynamicRange (dB)

22



“Bent Binary” R G Scheme

• Binary resistances for RG leads to gain error at low gains

• Bending a few of the LSBs gives a good fit (+/-.2 dB nominal error)

-1.00-1.00-1.00-1.00

-0.80-0.80-0.80-0.80

-0.60-0.60-0.60-0.60

-0.40-0.40-0.40-0.40

-0.20-0.20-0.20-0.20

0.000.000.000.00

0.200.200.200.20

0.400.400.400.40

0.600.600.600.60

0.800.800.800.80

1.001.001.001.00

0000 3333 6666 9999 12121212 15151515 18181818 21212121 24242424 27272727 30303030 33333333 36363636 39393939 42424242 45454545 48484848 51515151

Ga

in E

rro

r (d

B)

Ga

in E

rro

r (d

B)

Ga

in E

rro

r (d

B)

Ga

in E

rro

r (d

B)

Nominal Gain (dB)Nominal Gain (dB)Nominal Gain (dB)Nominal Gain (dB)

Gain Error vs. Nominal GainGain Error vs. Nominal GainGain Error vs. Nominal GainGain Error vs. Nominal Gain

Binary R's

Bent R's

23



“Bent Binary” R G Scheme

• 0 – 51 dB gain range with 10 switches

• Actual gain accuracy will vary since RONdoesn’t track poly resistors -1.00-1.00-1.00-1.00

-0.80-0.80-0.80-0.80

-0.60-0.60-0.60-0.60

-0.40-0.40-0.40-0.40

-0.20-0.20-0.20-0.20

0.000.000.000.00

0.200.200.200.20

0.400.400.400.40

0.600.600.600.60

0.800.800.800.80

1.001.001.001.00

0000 3333 6666 9999 12121212 15151515 18181818 21212121 24242424 27272727 30303030 33333333 36363636 39393939 42424242 45454545 48484848 51515151

Ga

in E

rro

r (d

B)

Ga

in E

rro

r (d

B)

Ga

in E

rro

r (d

B)

Ga

in E

rro

r (d

B)

Nominal Gain (dB)Nominal Gain (dB)Nominal Gain (dB)Nominal Gain (dB)

Gain Error vs. Nominal GainGain Error vs. Nominal GainGain Error vs. Nominal GainGain Error vs. Nominal Gain

Binary R's

Bent R's

24



Resistor Area Reduction

• Resistors scaled down to meet the target die area

• These become the dominant distortion mechanism

• THD due to resistor self heating is almost pure 3rd harmonic

25



Performance

• EIN with THAT 1580 = -128.3 dBu with 150Ω RS, 20 Hz – 20 kHz BW, 51 dB gain

• Gain Accuracy– +/-.5 dB 0 - 39 dB– +/-1 dB 42 – 51 dB

26



Performance

0.00010

0.00100

0.01000

0.10000

1.00000

0 3 6 9 12 15 18 21 24 27 30 33 36 39 42 45 48 51

THD+N vs Gain at 24dBu Out, 1 kHz

27



Performance

28



Performance

-1

-0.8

-0.6

-0.4

-0.2

0

0.2

0.4

0.6

0.8

1

0 3 6 9 12 15 18 21 24 27 30 33 36 39 42 45 48 51

Typical vs. Theoretical Gain ErrorTypical vs. Theoretical Gain ErrorTypical vs. Theoretical Gain ErrorTypical vs. Theoretical Gain Error

Typical Gain Error Theoretical Gain Error

29



Additional Features

• Zero-crossing detectors (ZCDs) for each channel

• Internal time-out clock counter for ZCD• 1 general purpose logic output (GPO) per

channel• GPOs can be sync’d to the ZCD• Independent connections for RF resistors for

discrete preamp designs that require this

30



Conclusions

• We achieved a 55% cost reduction per channel compared to our 5173

• THD performance was compromised in a manner that seems acceptable to most

• Noise performance is actually slightly better than the previous designs at most gains

31



Acknowledgements

Thanks to Fred Floru of THAT Corporation for his help in reviewing this presentation.

Thanks to Rene Jaeger for inviting me to speak tonight.

32

Interfacing Digitally-Controlled Microphone Preamplifiers to A/D Converters133RD AES Convention, Oct 2012


Questions?

Development of a Low-Cost Programmable Microphone Preamp ...

Documents

Transcript of Development of a Low-Cost Programmable Microphone Preamp ...