• TI Thinks Resolved

THS4524: Implementing a single ended input using THS4524 for a data logging application

Part Number: THS4524

Hello forum,

We're currently working on an environmental data logging application. We're building the analog front end using the THS4524 FDA and a ADS1274 ADC, which seems to be a common pairing.

However we've hit a few blockers that are halting our progress, and would appreciate any assistance from any experts in the field of analog, specifically FDA implementation.

The input signal

Output topology: Single ended or differential (Manufacture selectable, we want to use single ended)

Output impedance: 50Ohm (single ended), 100Ohm (on each line in differential)

Output bias: Either DC coupled (to Vsupply/2) or AC coupled (Manufacture selectable, we want to use AC coupled)

Output voltage: 0V to Vsupply. (in our case 0-5V)

Signal frequency: 2 - 20kHz.

The ADC

Accepts a differential input voltage between 0 and 5V, DC coupled to 2.5V.

 

Q1) Differential or single ended?

Since we have the option to select differential output from our source, is there any advantage to using the THS4521 in differential mode? I have read that this simplifies the process of selecting termination resistors, is that true, or can we achieve the same performance with single ended (albeit with a slightly more involved termination resistor calculation?).

Q2) Can we remove the capacitors between the input and output of the THS4521?

We purchased the ADS1274EVM-PDK, which has THS4521 circuits on-board. On the board, it uses 1.5nF capacitors in parallel with 1KOhm feedback resistors on the THS4521 (see image below). 

These capacitors form a low pass filter with a cutoff of 103kHz on the output which is undesirable for us.I removed them manually from the board. However this resulted in the signal shown below.

Yellow = input, blue = output.

The input was single ended from a 600Ohm source. When the capacitor is present, the FDA works fine, when it is not present, we see the signal in blue above. Is there a reason for this behavior, are these capacitors necessary for stable function?

Q3) Reliability of the 8802.FDA Design tool for calculating gain, feedback and termination resistances.

I have been using the 8802.FDA Design tool to assist in selecting values for RT, RG, RF and RM based on the single ended circuit below.

When I run this calculator with RF, RG and RIT from the datasheet (page 24, table 5, shown below) for a gain of 1 V/V, the calculator shows a full-swing ADC voltage of 0.25V for a 0.5V input (suggesting a gain of 0.5), am I missing something?

Calculator result

3) Using three FDAs in parallel on the same input signal

We intend to use the THS4524 to apply gains of 1x, 8x and 64x to our input signal. We will placing all three FDA inputs in parallel. Will this cause issues with termination, and if so is it best we use a buffer or isolation transformer on each input?

Apologies for the length of this post, but it will be an incredible exciting application for TI's products if we can get this to work.

Thanks in advance,

Josh

  • DiffAmpCookbookTemplate.docxSingleToDiffAmpCookbookTemplate.docx

    Hi Josh,

    You have covered a lot of subjects so please ask followup questions if necessary.

    1. The main benefit of a differential signal is that if noise couples into both traces, the differential signal is unaffected. If all that I am measuring is the voltage between two wires, my measurement unaffected if I add 1 volt to both sides. However, for a single ended signal referenced to ground, adding 1V will appear in my measurement. In addition to using differential signals, the main benefit of an FDA is that it allows the output to go negative for a single supply, which doubles the effective output swing. Whereas a single ended output can only swing between 0V and 5V with respect to ground, the positive differential output can swing 0-5V with respect to the negative differential output, which can also swing 0-5V. That means the output rails are 5V-0V= +5V and 0V-5V=  -5V.  As far as terminating resistors, its just the difference between one 50 ohm resistor and 2, 25 ohm resistors.

    2. The capacitor is not necessary here for stability, since the FDA is in a gain of 0dB (it is internally 2 single ended amps in a gain of -1/+2) . It might be helpful for noise filtering. Are you familiar with the subject of op amp stability? Sometimes a feedback capacitor is needed to prevent the parasitic input capacitance impedance from making the internal single ended gain =0dB over frequency, since they are typically minimum gain of 2 stable.

    3. I have attached some cookbook circuits to help. I don't like relying on tools too heavily.

    3 cont. Yes this will cause issues since the effective gain impedance will be all those resistors in parallel. You will need a buffer.

    Best regards,

    Sean

  • In reply to Sean Cashin:

    Post Script: I'm not sure how you connected the 600 ohm signal, but please follow my single ended to differential cookbook to match the input impedance correctly. maybe just attach it to the op amp input without a resistor and put 600 ohms to ground on the other side?
  • In reply to Sean Cashin:

    1. Understood, thank you.

    2. Yes I’m familiar with that concept, but I can’t say I’ve seen the behaviour shown when a capacitor isn’t present between the input and output of an amp (oscillations yes, but this looks like rectification!). What are your thoughts on that?

    3. These cookbooks are a great resource, I’ll go through them using our values and let you know how I get on.

    4. The cookbooks refer to the THS4561, do you happen to know when the release date of that IC is?

  • In reply to Josh Elijah:

    2. This amplifier is technically unstable, but the capacitor filters out the resonant frequency. You can stabilize it and therefore eliminate the capacitor if you put a 1pF capacitor directly parallel to the two feedback resistors for 100MHz of bandwidth.

    DiffAmpStabilityBode.TSC

    4. That may be a long time. However, you should be able to use the THS4521 for this.

  • In reply to Sean Cashin:

    Thanks Sean,

    What software do I need to open that .tsc file?

  • In reply to Josh Elijah:

    TINA-TI, Texas Instrument's circuit simulation software found here http://www.ti.com/tool/TINA-TI.

    Best regards,

    Sean