Deliver the lowest distortion and noise in a low-power, wideband, ADC interface (Part 4 of 4)

[Editor's note : we are presenting this lengthy, insightful article in four parts:

  • Part 1 looks at driving high-performance, high-speed ADCs, and benefits to using an input transformer in a differential inverting amplifier design
  • Part  2 looks at adding an interstage filter between the amplifier and the ADC
  • Part 3 looks at managing SFDR degradation in a low-power interface, and provides and example of broadband, low-power design with tested performance
  • Part 4 provides measured results for the test circuit

Also, for convenience , the extensive references for all four parts are repeated as a group at the end of each part.]

Measured results for the test circuit

The actual interface implements the Rp resistor as an attenuator into a 1:1 measurement path transformer that allows a direct measurement of the frequency response to the ADC inputs. In this example, that measurement path introduces a -25.8dB loss in the measured midband gain (as shown in Figure 8 ) but the shape is very close to the desired flat through 120MHz interface. It is showing a bit lower bandwidth than expected (180MHz) and this can be attributed to the input transformer slightly rolling off at the upper frequencies.

Figure 8. Measured small signal response to the ADC input.

Finally, the FFT for this interface and what we are assuming is a typical example of the ISLA112P50 for test purposes is shown in Figure 9 .

Figure 9. Measured FFT for a 105MHz -1dBFS input signal

The combined SNR at -1dBFS is 63.6dB, slightly lower than the predicted 64.3dB but still only 1dB below the typical ADC itself (and we may have had a slightly lower than typical ADC on SNR). This could be a slightly lower than nominal ADC SNR, higher clock jitter than in the original data sheet work, etc. If we assume the ADC has typical HD2 and HD3 (-91/-86 dBc) we can use Figure 6 to estimate the HD2 and HD3 coming into the ADC from the amplifier path.

Actually, HD2 is more than 6dB lower than the ADC but we can use Equation 11 to calculate HD2 into the ADC was in the -88dBc region. HD3 is less than 6dB lower (2dB actually) so Figure 6 suggests we came in with an HD3 at least 12dB lower than the ADC typical of -86dBc or about -98dBc. This HD3 makes sense from both the superior output HD3 vs HD2 for the ISL55210 but also that this tone it is falling well into the interstage filter cutoff (315MHz) where we should get an added -16dB of HD3 attenuation looking at Figure 8.

The HD2 at the output pins of the ISL55210 will get only a -6dB attenuation in the filter (Figure 6, at 210MHz). This suggests it was in the -82dBc region at the amplifier output pins to explain the inferred -88dBc at the ADC inputs to match the measured combination of input +ADC of -83dBc. Figure 8 in Reference 2 is showing approximately -80dBc HD2 at 100MhHz and 2Vpp for loads in the 500Ω region.


This work has shown one way to combine a very wideband, high-dynamic-range FDA with an input wideband transformer to deliver a very low-power, low-output-noise and distortion last-stage gain solution for ADC driving. The approach is very flexible and can use a wide range of transformers and gain settings for the VFA FDA shown here. The input transformer provides a simple single-to-differential conversion and offers both a noise and loop gain benefit to the overall solution. This can be used to improve an already very high performance FDA by 2dB to 5dB on both HD performance and Noise Figure.

Once we have this low noise and HD at the amplifier output pins, the benefits of an interstage 2nd order filter were shown. These include a much lower integrated noise for a given target frequency-response flatness region and some HD suppression for the out of band HD2 and HD3 terms. While not shown, the IM2 terms also get filter help for the F1 + F2 terms, but the IM3 terms are normally in-band requiring exceptional native IM3 for the amplifier. The example here was focused on a very broadband, 1st Nyquist-zone solution. Narrowband and/or higher Nyquist-zone solutions can still use the input transformer to good benefit but will require a different output-stage filter.

Sidebar: Transformer modeling

Wideband Transformer background : The particular type of transformer used in these examples is a transmission-line transformer in that the bandwidth comes from winding twisted-pair wires around a core (Reference 7 ). The model described in Reference 7 is a lumped-element representation of the separate pieces of the transformer characteristic. But the device itself is really pair of coupled inductors with a mutual-coupling coefficient very close to 1.00.

In the configuration used here, it is sometimes called a “Flux Coupled Transformer” – but aren’t all transformers flux coupled? Turning this transformer on its side and making it a 1:1 turns ratio yields what is sometimes called a common-mode choke – or, even more confusing,  sometimes called a  “transmission line transformer”.

In any case, for the input side of this circuit we are looking for a bit of step up to get the noise and loop-gain advantages described here. As the turns ratio goes up, we get more of “free” voltage gain at the cost of lower and lower min to max bandwidth span. For the broadband first-Nyquist zone applications considered here, a turns ratio of 1:2 (ohms ratio of 4) seems like a good maximum. Narrowband applications at higher frequencies can take advantage of higher turns ratios.

All transformers are “bandpass” circuit elements with a minimum-to-maximum passband region. These AC limits are often specified in terms of -1dB, -2dB and -3dB frequency spans. While those spans will be specified for a certain source and (presumably) matched load, transformers will “work” with any source and load impedance. So a 75Ω specified transformer can be used in a 50Ω environment and vice versa.

What will change are the minimum and maximum frequencies as the source and load impedances are changed. Given the specifications of a transformer, a simple two-inductor model with a mutual-coupling coefficient can be derived that will correctly predict the changes in min-to-max bandwidth with different source and load impedances (Reference 8 ).

For example, the ADT4-1WT model is shown in Figure A (Reference 9 ) where P1 = 2µH, S1 = 8µH and the coupling coefficient was 0.99488. This is the correct model to hit the specified 2MHz to 775MHz F-3dB span when driven from 50Ω and terminated in 200Ω.

Figure A. Example simulation circuit to validate the ADT4-1WT model.

Running this simulation with a 50Ω source and 200Ω load gives the expected 2MHz to 775MHz F-3 dB bandwidths. Running the same transformer model with a 75Ω source and 300Ω termination simply shifts the passband up in frequency as shown in Figure B .

This is showing a 0dB passband gain as the source is considered on the other side of the source impedance – so this model takes a -6dB attenuation into the input of the transformer (load reflected through as a match) then a 6dB gain (1:2 turn ratio gives a 2× V/V gain to the output side) for a net 0dB simulated gain. The green curve is 75Ω while the red is the specified 50Ω for this transformer. This model is not picking up the midband insertion loss.

Figure B. Simulated transformer response with matched source and load.

While the Mini-Circuits ADT4-1WT was used in the examples here, Table A shows it plus some alternate 1:2 turns-ratio devices.

Table A. Representative broadband 1:2 turns ratio transformers.

One other option is the presence of a secondary centertap. The input-side application discussed here does not use that centertap, but when used on the output side as a final stage to the ADC, the centertap is often used to bring in the Vcm bias for the ADC. The centertap is not used in these examples since the output Vcm of the FDA will bias the input DC operating point to the same level.

Connecting the centertap to any AC or DC ground or bias runs the risk of introducing signal imbalance due to gain and phase imbalance in the transformer itself. Those issues are rendered moot if the centertap is either not there or not used.


1. “Simple Circuit Techniques Yield High-Dynamic Range Amplifier”, Electronic Design Analog Applications Issue, June22, 1998 pp22-33.
2. ISL55210 data sheet.
3. Contact the author for the NF derivation using two op amps in the input transformer coupled differential inverting design.
4. Intersil “Active Filter Designer”
5. “Noise Analysis for High Speed Op Amps” TI application note SBOA066A, 1996, Michael Steffes
6. ISL112P50 data sheet.
7. “How RF Transformers Work” Mini-Circuits application note
8. “Spice model simulates broadband transformer”, Michael Steffes EDN Design Ideas, May 11, 1995 pp135-136.
9. Contact author for a copy of the ADT4-1WT model  that can be used with Intersil’s iSim PE simulator.

About the author

Michael Steffes is Senior Applications Manager, Intersil Corp.with more than 25 years of experience in high-speed amplifier design, applications, and marketing.Previously, he wasthe Market Development Manager for High-Speed Signal Conditioning, and a Distinguished Member of the Technical Staff, at Texas Instruments Inc. With more than 25 years of experience in high-speed amplifier design, applications, and marketing, Michael currently provides product definition and customer design-in support.

Michael earned a BSEE from the University of Kansas and an MBA from Colorado State University. He shares several basic patents in high-speed op amp designs and has written more than 85 product data sheets, scores of contributed articles, applications notes and conference papers.

0 comments on “Deliver the lowest distortion and noise in a low-power, wideband, ADC interface (Part 4 of 4)

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.