Questions concerning the THS4524EVM

Hello Ben,

Here are my responses to your questions:

1) What kind of signal amplification do you need? There is no maximum recommended gain, per se. There are tradeoffs to be made as the gain gets higher, in terms of bandwidth, distortion, noise, etc. The THS4524 will be limited by its gain-bandwidth product, meaning that an increase in gain will result in a reduction in bandwidth (see the section on small-signal bandwidth on the top of page 5 of the datasheet and compare the different bandwidths at different gains).

The higher the gain, too, the lower the loop gain of the amplifier. Loop gain is important for distortion performance, among other things (see p.8 of Designing for low distortion with high-speed op amps - slyt133). You can see in Figure 35 in the datasheet that as the gain increases, the distortion performance degrades. Also, the input referred noise will be multiplied to the output by the same gain as the signal, so higher gain means higher output noise.

2) I do not completely follow what you are saying here. For a typical amplifier circuit, the output noise will be higher with higher gain, and lower with lower gain. By referring the noise measurement to the input, the gain-dependence of the noise is removed. In this way, different op amps can be directly compared by their input-referred noise, and the output noise can be calculated through analysis (see Noise Analysis for High Speed Op Amps - sboa066)

3) From a practical standpoint, one downside with using single-supply versus split-supplies on any op amp is having to pay extra attention to interfaces. By shifting the supplies from split-supply (dual polarity) to single-supply,  the input/output range of the amplifier is also shifted.

For example, if your signal is bipolar and swings around your ground reference, you have to make sure that the resulting signal to the amplifier does not violate its input range. Sometimes it is necessary to AC-couple the signal and rebias the signal at a different DC level so that it is within the amplifier input range. The THS4524 has an input range that includes the negative rail (GND in single supply case), which mitigates voltage level issues.

4) I am not sure I understand what you mean with the 15ohms to ground on each side. Do you mean that your magnetic-field sensing coil will essentially act as a differential current source with 15ohm (30ohm total) effective output impedance? Alternatively, it could be modeled as differential voltage source (swinging around ground) with 15ohm series resistance on each side. Can you please share a quick schematic/drawing showing how your sensor will interface with the THS4524 and what is meant by these 15ohm resistors?

5) I would recommend using the same value suggested by Figure 66 for your 750pF load, which looks to be about 9-10ohms. The recommended value will not change significantly as the load resistance increases beyond 1kohm.

I suggest using TINA-TI to simulate your circuit. You can download a TINA-TI reference design circuit on the product page under Tools and Software at this link.

Hello Kristoffer,

I am working with Ben on the same project.  Thanks for your help.  

We are using the circuit depicted in Figure 60.  We have a single polarity power supply of 3.3V.  We are using Rf = 1Kohm (the part that came on the evaluation board) and Rg = 30ohms.  The frequency range of data is 800Hz-4.5KHz, so the gain-bandwidth product is not a limitation.  The amplitudes of the input signals are < 1V.

In general, the gain has not followed the Rf/Rg rule. Given our setup, we expect a gain of about 30, but instead we see a gain of 6. The gain changes when we change the supply voltage to a dual supply of +/- 2V.  Also, we changed Rg to 92 Ohms and saw no change in gain.

1) Why is the gain not following the Rf/Rg equation?

2) Why is the gain not changing when we change Rf?

Thanks for your help.

--Alex

Just to clarify one issue, the outputs are not saturated.  The output voltages are always  below 1 volt.

Thanks.  Ben

Hi Ben, Alex,

I have some questions about your setup:

For your testing, what is your signal source? Since you refer to Figure 60, I assume you have removed the termination resistors populated on the evaluation board and your signal source goes in differentially into the EVM. For example, the EVM has 52.3ohm resistors from each input to ground before the Rg resistors.

The output impedance of your signal source will also play into the gain. Rf/Rg would be the gain if the differential source impedance is 0 ohm. If your source is not low impedance, then the source impedance should be added to Rg when calculating the gain. Changing Rg should reflect a change in the gain though, and it is interesting that it does not...

How are you measuring the gain? Do you use the output transformer circuit on the EVM? The default output circuit that converts the differential output back to single-ended will have an attenuation of 31.8dB (into 50ohm terminated equipment).

Oscilloscope screen captures and schematic with all modifications made to the default EVM configuration may help me understand the issue.

Hello Kristoffer,

We are using differential input and output.  The output transformers have been removed.  The resistor across the output lines has also been removed.  We removed the resistors to ground on each input line, and replaced the 1 KOhm Rg with 30 Ohms.  Here is a link to our circuit diagram:

http://i.imgur.com/LrTtk.png

One important change to this circuit diagram is that there is a center tap on this coil that goes to the amplifier ground.

Each channel is being driven by an inductive magnetic field coil, used to measure changing magnetic fields.  Each coil has finite resistance (about 32 Ohms across the coil, 16 ohms from each side to a center tap, which is also the amplifier ground).

To measure the gain, we have done two things. 1) We used a controlled DAC to input a known signal amplitude into the amplifier and measured the output value (with an ADC).  2) We measured changes in gain by sending out a signal on a transmitter coil and picking it up on a receiver coil--the inductor in the figure.  We changed Rg from 30 ohms to 92 ohms, and we saw no change in the output waveform.

Given the above schematic and setup, can you tell us how the gain should be calculated?

We work in the field on Tues/Thurs so any assistance by tomorrow (Thursday) morning would be greatly appreciated.

Thanks,

Ben and  Alex

Hello Kristoffer,

One more point about our circuit diagram.  The output of the THS4524 amplifier feeds a twisted-pair shielded cable.  The other end of the twisted-pair shielded cable feeds an Ithaco 1201 istrumentation amplifier with a +input a -input and a ground.

Ben

The THS4524 data sheet refers only to gains of 1, 2, 5, and 10.  Can we use this amplifier for a gain of 50 or 100? Note, we only need frequencies up to about 20kHz.

What values of resistors (Reference Table 1) should we use for a gain of 50 or 100?

We need a differential input and single ended output.  Can we just leave one of the SMA connectors open on the amplifier output for single ended output?

Ben Sternberg

University of Arizona

bkslasi@email.arizona.edu