MOSIS Testing Rapport

E. Aldrete, J. Altet, D. Mateo Electronic Engineering Department Universitat Politècnica de Catalunya

Barcelona - Spain

This testing report includes results from a Low Noise Amplifier (LNA) and a built-in differential temperature sensor, both placed in the same silicon die. Testing results from low frequency and high frequency scenarios for each circuit are merged in this report. Other active and passive devices, i.e. resistors and MOSFETs, acting as a heat sources are included in order to characterize the temperature sensor. At the end of the document the original Proposal submitted is attached, where details of the measurements performed are explained. This is a preliminary document, as we have not finished the measurements. New versions of the report will be sent as new data is available.

Design Identification

• Design ID Number: 75589. • Fab-ID: T65VAA. • Design name: LNA_TEMP. • Technology: TSMC25. • Requested fabrication options: NON_EPI, THICK_TOP_METAL. • Samples received: 3 packaged in OCP_LQFP44A and 37 unpackaged. • Layout size: 1250 x 1250 microns; area: 1.563 sq millimetres. • Parts tested: 2 packaged, 3 unpackaged. • Functional Parts: 1 packaged, 2 unpackaged. We have 2 samples with the

temperature sensor non-functional (1 packaged and 1 unpackaged). We attribute the malfunctioning to the fact that we did not include ESD protection for the adjusting lines of the sensor circuit.

Circuit Description

LNA The single-ended Low Noise Amplifier employs a cascade topology with inductive degeneration and on-chip spiral inductor. The figure shows the schematic and simulated figures of merit.

VBIAS VDD

VRF_in

VRF_out

RB1

RB2

CB

Lo

Ls

Lg CEX

MNB MNC

MNA

39.6mWPower

Consumption

5.2dBNF

-35.54dBInput ImpedanceMatching

8dBMaximum PowerGain

2.4GHzFrequency

PerformanceParameter

Simulated Figures of Merits

Schematic of the Low Noise Amplifier and its simulated Figures of Merits.

Differential Temperature Sensor

The differential temperature sensor consists of a variation of a CMOS Operational Transconductance Amplifier (OTA). Its schematic is the following.

MpOff

MnOff

Vout

VBias

MP1 MP2 MP3 MP4

MN2MN1

QS1 QS2

Q1 Q2

VDD

GND Schematic of the differential temperature sensor

When the circuit is properly biased, ∆Vout = St·(TQS1-TQS2), where TQS1 is the

temperature of the bipolar transistor QS1 (in red in the figure above), TQS2 is the temperature of transistor QS2 and St is the differential thermal sensitivity of the sensor. Detailed information about this sensor can be found in [1]. From simulation St is 2.5 V/ºC. Heat Sources

There are a total of 6 devices acting as a heat sources (HS): 4 resistors, R1, R2, R3 and R4, and 2 MOS transistors connected in diode configuration, M1 and M2. The goal of these devices is to generate temperature gradients in the IC layout to test the performances of the differential temperature sensor

Layout of the circuit

The figure below shows the placement of the different circuits on the IC.

IC layout

The pictures on the left side above show a detailed view of the placement of the

bipolar transistors QS1 and QS2 and the devices acting as heat sources.

The following is a photograph of the chip fabricated for testing.

chip photograph

PCBs for measurement

Two different PCBs have been designed to perform the measurements. The

photograph on the left side below is the PCB used to test the packaged circuits, whereas the photograph on the right side is the PCB used to test the unpackaged circuits. Chip on board (COB) has been used for the last one.

LNA

SENSOR HS

25µm

45µm 27.5µm

QS2

MNC

R1 R3

M1

R2 R4 QS1

M2

Photograph of PCBs for testing IC prototypes.

Equipment used to perform the measurements

Brand and Model Description Specification

Agilent E3632A Power Supply 0-15V, 7A; 0-30V, 4A 105/120W Tektronix AWG2021 Arbitrary Waveform Generators 125 MHz Agilent E4443A PSA Spectrum Analyzer 3 Hz – 6.7 GHz Agilent E4438C ESG Vector Signal Generator 250 kHz – 6 GHz, +10 to -136 dBm SR830 DSP Lock-In Amplifier 1 mHz – 102 kHz Agilent 33220A Arbitrary Waveform Generators 20 MHz

Report of the measurements 1. Heat sources.

The figures below show the I-V characteristics of the devices used as heat source. Left is for resistors and right is for MOS transistors in diode configuration.

I-V characteristics of the heat sources devices. Left: Resistors, Right: MOS in diode.

2. Temperature sensor.

Similar to the work reported in [1,3], we have done static and AC measurements.

Static measurements: The figure on the left side below shows the output voltage variation as a function of the DC power dissipated by each resistor. Four different curves are superimposed. Depending on the resistor that dissipates power, a temperature increase is generated at the bipolar transistor location QS1 and QS2.

Static response of the temperature sensor.

As it can be seen, resistors R1 and R2 have stronger thermal coupling to the bipolar

transistor QS2 and QS1 respectively than resistors R3 and R4, as they are placed closer to them. The figure (right) shows the sensor’s output voltage variation as a function of the power dissipated by each MOS transistor. In this case, the curve for MOS1 and MOS2 are superimposed in the graph. Dynamic measurements: In this case we activate a heat source with a sinusoidal function plus a DC bias voltage and we measure the amplitude and phase of the sensor’s output voltage at the same frequency of the AC signal applied to the heat source. Measurements are done with a lock-in amplifier. The figure shows amplitude of the sensor’s output voltage as a function of the signal’s frequency applied to the heat source (in this case, the resistor R1). Four different curves are superimposed, depending on the value of the sensor’s load resistance. As it can be seen, the bandwidth of the sensor is about 1 kHz.

Dynamic response of the temperature sensor.

3. Thermal coupling characterization

Bode (amplitude) of the thermal coupling impedance between the point named hot spot and the resistors R1 and R3

Using a laser reflectometer set up [2,3] we have obtained the bode of the thermal

coupling impedance between the resistances that act as heat source (R1 and R3) and the point marked in the figure as hot spot (left figure). The figure (right) shows the evolution of the bode amplitude as a function of the frequency. The vertical axis is proportional to ºC/W. 4. Electrical characterization of the LNA

The figures show the gain of the LNA for a specific bias: VDD=VBIAS=3.3V. The figure (left) shows the gain of the LNA for different input power levels obtained in the packaged circuit, whereas the right figure shows the gain obtained in the chip on board set up.

Gain of the LNA. Left: packaged. Right: Unpackaged.

The following figure shows the output power delivered by the LNA at 900 MHz as

a function of the input power applied at the same frequency.

5. Characterization of the LNA through temperature measurements

As indicated in the proposal, the main goal of the IC is to obtain the central frequency and bandwidth of a LNA by measuring temperature [4]. We are still working on this point (we will re-send the testing rapport once all the measurements are done). As example of results, the figure compares: output power delivered by the LNA as a function of the frequency of the input signal (input power level: 2 dBm). Thermal measurement performed at 3.012 kHz as a function of the central frequency of two tones spaced 3.012 kHz applied to the input of the LNA (input power level of each tone: 2 dBm). Measurements have been done with a stabilized Michaelson Interferometer devoted to dynamic temperature monitoring in ICs [2,3]. The temperature has been measured at 10 microns from the centre of the cascode transistor MNC.

Left vertical axis: Output voltaje provided by the interferometer. Right vertical axis:

Output power delivered by the LNA (input power: 2 dBm).

Michaelson interferometer: Measurement set-up

REFERENCES

[1] Josep Altet, Antonio Rubio, Emmanuel Schaub, Stefan Dilhaire i Wilfrid Claeys. (2001). THERMA COUPLING IN INTEGRATED CIRCUITS: APPLICATION TO THERMAL TESTING. IEEE JOURNAL OF SOLID-STATE CIRCUITS , 36 (1) : 81-91. ISSN: 0018-9200

[2] Josep Altet, Stefan Dilhaire, Sabastian Volz, Jean-Michel Rampnous, Antonio Rubio, Stephane Grauby, Luis D. Patino, Wilfrid Claeys i Jean-Bernard Saulnier. (2002). FOUR DIFFERENT APPROACHES FOR THE MEASUREMENT OF IC SURFACE TEMPERATURE: APPLICATION TO THERMAL TESTING. MICROELECTRONICS JOURNAL , (33) : 689-696. ISSN: 0026-2692

[3] Josep Altet, Wilfrid Claeys, Stefan Dilhaire i Antonio Rubio. (2006). DYNAMIC SURFACE TEMPERATURE MEASUREMENTS IN ICs. PROCEEDINGS OF THE IEEE , 94 (8) : 1519-1533. ISSN: 0018-9219

[4] Diego Mateo, Josep Altet, Eduardo Aldrete-Vidrio i José Luis González. FREQUENCY CHARACTERIZATION OF A 2.4 GHz CMOS LNA BY THERMAL MEASUREMENTS. A: RFIC - 2006 IEEE RADIO FREQUENCY INTEGRATED CIRCUITS SYMPOSIUM. IEEE Service Center, 2006, p. 1-4.

Design for MOSIS Educational Program (Research)

Project Title: ”Electrical characterization of a 2.4 GHz Low Noise Amplifier

through temperature measurements” Prepared by: Eduardo Aldrete-Vidrio (PhD student), Josep Altet (associate professor) and Diego Mateo (associate professor). Institution: Technical University of Catalonia, North Campus, Electronic Engineering Department (UPC/DEE). Date of Submission: April 28th 2006.

Electrical characterization of a 2.4 GHz Low Noise Amplifier through temperature measurements

Project description: A 2.4GHz Low Noise Amplifier has been design in TSMC025 technology in order to be electrically characterized by thermal measurements, for what a thermal sensor has been integrated together with the LNA. A recent new technique has been presented, which proposes the electrical characterization of high frequency analog circuits by low frequency thermal (non invasive) measurements [Altet06][Mateo06]. It has been proved theoretically and by means of simulations, and the present project proposal is aimed to prove it experimentally. The LNA that has been designed is a single input, single ended LNA, with inductive source degeneration and inductive matching network at the input, with an inductive choke inductor as load. All inductors are on-chip inductors. The LNA works at 2.4GHz as central frequency and the simulated voltage gain obtained is 17 dB, with a noise figure of 5 dB and 2.7 mw of power consumption. The thermal sensor designed is a differential temperature sensor. It uses npn2 bipolar transistor in the input differential pair in order to sense temperature. To calibrate the sensor different heat sources (resistors) have been also integrated. The objective of the project is to obtain the gain, central frequency and bandwidth of the LNA through thermal measurements by using the previously mentioned temperature sensor. The experimental results will be correlated through thermal measurements carried out by using laser reflectometer setup and a Michaelson interferometer [Altet02]. Fabrication process The circuits has been implemented in the TSMC 0.25 micron mixed-signal technology (1P5M+SILICIDE 2.5V/3.3V). The technology code for this process is CR025 (CM025). Packaging requirements No packaging will be required: from previous experiences in RF design Chip on Board measurements are preferred (better control of input and output parasitics), and it allows at the same time external silicon surface temperature measurement (see Test and Characterization Plans). Estimated project size (length and width) The total estimated area of the design is 2 mm^2. The total estimated length and width are 1.41mm and 1.41mm respectively. These data include the 20 pads required (the design is core limited).

Simulation plans First, the circuit level simulation of the temperature sensor is carried out by using Cadence and HSPICE environment in order to find the proper transistor dimensions to get the greatest temperature sensitivity suitable for thermal testing in RF applications. Therefore, the static and dynamic electrical responses of the sensor, as well as its performance as thermal sensor have been analyzed. Also, the circuit level simulation is carried out by using SpectreRF in the Analog Design Environment from Cadence (version 5.033) in order to estimate the common parameters which are important in the design verification of Low Noise Amplifiers. Various analyses to determine the compromise among gain, noise, power, linearity, stability and matching performances are completed into the test-bench where the single-ended LNA is placed. An appropriate modeling of the parasitic capacitance and bondwire effects are taken into account. The expected behavior of the temperature increase on each transistor of the LNA, due to its operation, and the viability of the proposed technique we want to prove, have been carried out by using Dynamic Link between ADS and Cadence. S-parameters and Harmonic Balance analysis have been performed for this purpose. Also, both circuits, thermal sensor and Low Noise Amplifier, are simulated with normal and corners manufacturing process variations in order to verify the performance under these process conditions. The layout of the circuits is done by using Virtuoso Layout Editor from Cadence. Test and characterization plans The goal of the measurements are: first, to characterize the built-in temperature sensor; second, to extract the electrical characteristics of the LNA by using classical approaches (using Network and Spectrum Analyzer connected to the input and output of the LNA); third, to obtain its electrical characteristics (specifically, central frequency and bandwidth) by measuring temperature close to the LNA by means of the integrated differential temperature sensor; and fourth, to correlate the electrical and the thermal measurements obtained. The final goal is then to demonstrate that electrical characteristics of RF amplifiers can be obtained by measuring temperature. In addition to the specific objectives of the project, we also characterize the thermal coupling through silicon substrate of the TSMC 0.25 microns technology used. Specifically: - Characterization of the built-in differential temperature sensor: Temperature gradients will be generated by using integrated resistors as heat sources. We will perform DC characterization and AC characterization activating the heat source with a periodic voltage source and measuring the sensor’s output with a lock-in amplifier. These are similar measurements to what is reported in [Altet02]. - Electrical characterization of the 2.4 GHz LNA: the following figures of merit will be extracted: S parameters in the ISM band, voltage gain, bandwidth, current consumption and linearity. - Following the procedure established in [Altet06][Mateo06], extraction of the bandwidth and the central frequency of the amplifier by measuring temperature using the built-in differential temperature sensor.

- Following the procedure established in [Altet06][Mateo06], extraction of the bandwidth and the central frequency of the amplifier by measuring temperature using off-chip temperature measuring systems: Laser interferometry and a laser reflectometer set-up [Altet02]. Using off-chip temperature sensors allows detecting the effect of noise coupling in the measurements performed with the built-in temperature sensor. - Correlation of the thermal and electrical measurements and the simulation.

[Altet02] J. Altet, S. Dilhaire, S. Volz, J.-M. Rampnoux, A. Rubio, S. Grauby, L. David Patino-Lopez, W. Claeys, and J.-B. Saulnier, “Four different approaches for the measurement of IC surface temperature: application to thermal testing”, Microelectronics Journal, vol. 33, no. 9, September 2002, pp 689-696.

[Altet06] J. Altet, D. Mateo, J. L. González, E. Aldrete-Vidrio, “Observation of High-Frequency Analog/RF Electrical Circuit Characteristics by on-Chip Thermal Measurements”, IEEE International Symposium on Circuits and Systems (ISCAS 2006), Island of Kos, Greece. To be published.

[Mateo06] D. Mateo, J. Altet, E. Aldrete-Vidrio, J. L. González, “Frequency Characterization of a 2.4 GHz CMOS LNA by Thermal Measurements”, IEEE RFIC Symposium 2006, San Francisco, CA. To be published.