This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

THS3491: Slew rate limiting THS3491 to improve impedance match ?

Part Number: THS3491

The THS3491 datasheet says it has a slew rate of 6000 - 8000V/uS. However looking at Fig 27 I see the output impedance rises significantly with frequency. So using this part to drive a transmission line with a series matching resistor is it necessary to limit the output slew rate way below what is possible to get a good low output impedance and minimize reflections ?  For example say a 5V step at 5000V/us is1nS so assume 1nS rise, 1nS on and 1nS fall and 1nS off that,s a frequency of 250MHz. But from fig 27 Z out is around 15 Ohms at this frequency, so assuming this is inductive Z out would be sqrt (15^2 + 50^2) = 52 Ohms. So could limiting the slew rate be used to effectively lower Zout or is there a more elegant  way to get a better broadband impedance match without signal loss or degradation ? I need 50+/-0.5 Ohms.

  • Morning Steve, 

    You are convolving linear operating things with non-linear (slew rate)

    Essentially, the slew rate on the THS3491 is so high that you should not hit it (adding a non-linearity to your step) for most apps. 

    For a step output, an approximate dV/dT on the output (assuming perfect input edge) is 2.85*Vstep*F-3dB. Use that to see if you are approaching the slew limit (in fact, it is hard to find a limit on this part as its CFA nature just pushes more and more edge rate as needed, some tests suggested>9000V/usec)

    You can run a small signal sim of output impedance to see where you are including the "closed loop" Zout of the amplifier. 

    If you need to tune an apparent inductive total Zout, try a small cap across the physical 50ohm out. 

    If you have an intended schematic, send it over if you would. 

  • For example, here is a gain of 5 output impedance sim looking back at a 50ohm + amplifier

    This is dB ohms so 49.5 ohm min would be 33.9dBohms and 50.5ohm is 34.06dBohms, so it does deviate low more than that. The details would depend on your amplifier set up externally. 

  • Thanks Mike, I figured you might be the one who answered. My question is subtly different from slew rate limiting what I'm concerned about is the device output impedance on a transient. So for example the THS3491 could be well within it's slew-rate specification but it's output impedance on an edge could be high enough to cause an impedance mismatch I believe.

    Going back to my example yesterday 8000V/uS the slew rate limit is 8V/nS so for a 5V step is around 625pS. So if I limited the input slew to say 1nS the realized slew becomes 5,000V/uS well within the parts capabilities. However assuming 1nS rise, 1nS on, 1nS fall and 1nS off =4nS total period is 4nS or 250MHz. From the datasheet the Zout of the part according to fig 25 is 15 Ohms at 250MHz so the total series termination I believe will be SQRT{(50^50 + (15^15)} = 52 Ohms approx (assuming the Zout is inductive) but for our application the client ideally needs a better impedance match !

    If I limit the input slew to say 2000V/uS then a 5V step would be 5nS or a sine as before would be 10nS ie 100MHz at this rate Zout would be 3 Ohms so Z = SQRT (50^50 + 3^3) ie Z = 50.09 Ohms. The price I'd pay for this though is slower settling time.

    So my question, do you agree with this, if not please tell me any error you see. If you do agree is there any way to get the fastest settling/slew while maintaining a good output impedance match ? Bear in mind repetition rate may vary as well as amplitude as would say a precision pulse/ramp generator.

    Thanks Steve

  • Well Stieve, quite a bit going on here - 

    One way to think about the higher Zout at higher F in a pulse settling concern is that Zout is varying across the settling time - just one of many effects getting into the shape to a final final value. Not sure how to interpet that source accuracy, normally it might be more of a VSWR band over frequency - you can tune it with passives I suppose. 

    From a final load settling time perspective, yes staying out of slew limiting is a good idea, 

    Touched on that in these articles

  • Hi Michael,

    Yes good point about Zout being a dynamic over the settling time period, I was thinking just worse case because of the reflections created by any mismatch but in reality it's not until the reflected pulse from the cable impinges back onto the THS3491/Series R by which time the rise/fall will be slower due to cable group delay so now I'm coming to the conclusion that this is even more difficult to analyze than in my original post ! I agree about avoiding slew rate limiting also maybe shunt termination at the load end which if low inductance would provide a slew-rate independent termination. Anyway thanks for your thoughts definitely has helped me to crystalize what's going on a bit better. Best Steve.