Hi,
When I debug THS3091 according to following configration(Power supply is +_15V,G=5,and the input is +_2V bipolar signal),when the input signal's frequency is 1MHz,the output is right(+_10V,20Vp-p),but when I increase the frequency of input signal,THS3091's output amplitude will decrease,as per my test, the output(Voutp-p) with 2MHz input signal will start to decrease,but it's normal and right for below 2MHz input signal. After I check,when the output decreases,the input signal's amplitude decreases(Vinp-p) too.In THS3091's datasheet,the BW is about 190MHz with G=5,should be enough,and its SR is enough high too. I also tested on THS3091EVM,the phenomena was same.I don't understand why this input signal decreases along with input signal's frequency increase,because of THS3091's Noninverting input capacitance? Would anybody like to help me?Thanks in advance.
Best Rearge
Li
Hi Li,
Since your input is on a non-inverting input and it is not terminated, you input impedacne seen by the source is, from the datasheet (1.3Mohm || 0.1pF) since the board will always add some parasitic, say 2pF, you have in effect an input impedance that will decrease with frequency and has a pole at 58.3kHz. So after 58kHz, the input impedance will decrease. To eliminate this phenomenon, you will have to set a much lower impedance at the input. The datasheet bandwidth was measured with a 50ohm matched impedance. The test instrument have a 50ohm termination that is matched at the non-inverting node.
Figure 59 in the datasheet shows the configuration used.
Best regards,
Xavier
Best Regards,
Hi Xavier,
Thanks.But I added the terminall resistance and I also tested this chip with 49.9 ohm terminall resistance on THS3091EVM(Figure 74 THS3091 circuit configration),the result was same and the input decreased.
Appreciated for your more help.
Best regards
How are you measuring the decrease in the signal amplitude? If you are using a probe, the additional probe capacitance may be the cause of the phenomenon you are observing. It is possible to eliminate the capacitance interaction with the previous stage by isolating the probe with a 100 ohm series resistance.
What is the signal source you are using? What is its output impedance?
Could you provide a description or a schematic of the signal chain you are evaluating as well as the function you are trying to design?
I use oscillograph to observe signal amplitude's decrease,it's obvious with input frequency's increase.If I use a series resistance,should I still add a 50 ohm terminal resistance?
I use Agilent's 33250A as input signal source.
I want to use THS3091 in a wave generator circuit,and max output is a 20Vp-p(+_10V) output with G=5/+_2V(4Vp-p) input.
Thanks&Best regards
Thanks for the details. You should still use the termination resistance. 100ohm isolation resistance between the probe and the test point should be sufficient.
I just tested according to your advice(adding a 100 ohm series resistance and 50 ohm termination resistance) on THS3091EVM board,the phenomenon is same.I also try to change these two resistances' position,the result is same.Following is my test data(G=5,input is fixed at +_2V/4Vp-p sine wave from Agilent 33250A) and I only change the input signal frequency:
Input Frequency/ input from oscillograph observation output from oscillograph observation (V)
10KHz / 4Vp-p 20Vp-p
100KHz / 4Vp-p 20Vp-p
1MHz / 4Vp-p 20Vp-p
2MHz / 3.84Vp-p 19Vp-p
3MHz / 3.6Vp-p 18.4Vp-p
5MHz / 3.04Vp-p 16.4Vp-p
10MHz / 2.04Vp-p 11.4Vp-p
15MHz / 1.46Vp-p 8.56Vp-p
20MHz / 1.11Vp-p 7Vp-p
Looking at the ratio of output to input, it appears that the THS3091 is peaking at higher frequencies in your setup, but the overall gain is close to 5V/V as you would expect at low frequency and increasing with frequency. To simplify the problem, look at the output of the generator directly using a 50ohm termination. The 50ohm termination is required by the signal generator for calibration purpose. If the 50ohm resistance varies (by adding a second path in parallel for example), you would have the roll-off that you are seeing.
From the number that you are giving at 20MHz, it appears that the capacitance in parallel with the 50ohm termination resistor is from 100pF to 150pF. This pretty large capacitance would explain the frequency roll-off of the signal generator.
You are right,after I tested the output of generator directly only using a 50ohm termination resistance,it decreased with the signal's frequency increase,and same result when I only changed the termination resistance with fixed frequency.
1: You mean,is the reason the output impedance(larger capacitance) of front-end circuit? Agilent 33250A's output for test purpose or front-end circuits's output in my actual design. Or is the reason the influence of oscillograph's probe when I look at the signal's wave? It shouldn't be the influenceof PCB's parasitic or distribution capacitance because I tested on THS3091EVM too,is right?
2: Do you have any advice to resolve this issue?
Appreciated for your response.
In our another design project,I found the input didn't decrease with frequency's increase,but the output still decreased.I didn't know why.
For above test,it should be influenced by signal generator(I will check Agilent's generator soon).
Thanks.
1- I believe it to be the probe. If you can, connect the output of the generator directly to the oscilloscope. by bypassing the probe, you should have the adequate performance for the signal generator alone. The THS3091 EVM should have a few pF (no more than 5pF) parasitic capacitancc on the input.
2- Try not using the probe and connecting the board directly to the oscilloscope. If the input impedance of the scope is 1Mohm, you will want to have the termination on the EVM. If the scope termination is 50ohm, all you need on the EVM is the matching resistor.
What is the bandwidth of the oscilloscope?
I will test it according to your advice.The oscilloscope I use is Tektronix's TPS2024 200M oscilloscope.Thanks a lot.
You are right,it's probe.Thanks a lot for your great support.