I need a super high speed peak detector, able to accurately peak detect a 10ns to 20ns pulse. Is there such an animal, what is the schematic and what amplifier should I use? I see a TI app note sbos138 with a BB 3507J amplifier but the circuit may not be fast enough and apparently the 3507J is no longer available.
Any help please?
There are a couple ways to peak detect a high speed signal. 10ns-20ns puts you in the 50MHz-100MHz range, so you need high BW and fast slew rate. You also need an amplifier that can handle driving large capacitive loads and can settle relatively quickly (<10ns).
The most basic peak detector configuration is to use a single op amp with a diode in the feedback path and a diode at the output like in the configuration below. The main drawback of this approach is that you will be limited on speed and wouldn't be able to reach the 10-20ns you're looking for. Also note, that there will be two diode drops so your peak voltage will Vpeak - (2* VD) which may or may not be a concern depending on the amplitude of the input signal.
In order to reach higher speeds, you could use the OPA615 and BUF602 in the configuration shown on figure 48 of the OPA615 datasheet. This device has a built in OTA to source additional current and help reduce the rise and fall times of the amplifier. This allows the amplifier to more quickly charge the sampling capacitor its peak value, enabling sampling times in the ns range. In addition, there is sample and hold circuitry included in the amplifier which would allow you to tune the sampling rate of the amplifier. This configuration would also enable you to detect negative peaks as well as positive peaks by adding the BUF602 circuitry. Overall, this appears to be a very good option for high speed peak detection.
Regards,Luke LapointeHigh Speed Amplifiers
Regards,Luke LaPointeHigh Speed Data Converters
In reply to Luke LaPointe:
Thanks very much Luke. This looks very promising and also I see that an eval board is available for the opa615, so I can try it out. Great help, thanks again!
BTW, I found an EDN article using a Maxim MAX9690 comparator that looks promising too but Maxim doesn't have any info about such a circuit and also they say they don't have an eval board for the part: http://www.edn.com/file/18037-92602di.pdf
In reply to John Baker97919:
In retrospect, after more carefully reading your note, I see the output voltage "there will be two diode drops so your peak voltage will Vpeak - (2* VD)" so that may be a problem since my input voltage peak will be only 1V.
In order to detect 1V peaks as you describe, you can amplify the signal before the peak detecting stage. This will give you a little more margin for the diode drops and still maintain the signal. The downside is that increasing the gain too much will cause the signal to clip and too little gain will not forward bias the diodes to let the signal get through. You will have to decide what the minimum detectable peak is for your design and set your gain accordingly. You can also use Shottky diodes with a 0.4V drop each to help mitigate this effect.
I am suspicious that your assertion about the 2 diode drops is incorrect. I haven't yet run a SPICE program on the circuit but my reasoning is as follows. If for example, the input voltage to the resistor divider is 2V, the divider would cut it in half to 1V. Then since the opamp would drive the output until the voltage at the negative (-) opamp input is the same as at the positive input, or 1V, then the current to ground thru R4 will be 0.1mA and also the current thru R3 will be 0.1mA, resulting in 2V at the opamp output since no current enters or exits the negative input to the opamp. Thus, no 2-diode voltage drop.
I analyzed a peak detector circuit using MicroCap 8 and don't see any 2 diode drop. The circuit is not optimized for high speed but just a test of the idea. Circuit is not fast enough and getting too much droop. But it looks promising.
It appears as though you are correct. When simulating, the frequency as well as capacitance value will determine what value the output will settle too as well as how much droop is present. I ran the same circuit with 50MHz and 5MHz. See results below.
Let me know if you have any further questions.
Much thanks Luke. It looks like the OPA615 will do the job for me. I can adjust the cap on the Hold_CTRL output for overshoot, depending on the pulse width, so should work great. I have recvd a couple of eval boards and have some OPA615 SO14's on order. Attached is a pdf of my simulation of the circuit with Microcap 8.
Whoops! I meant I can adjust the cap after the diode to set the overshoot, depending on the pulse width.
Can you please load the microcap file, i want to open it using microcap
All content and materials on this site are provided "as is". TI and its respective suppliers and providers of content make no representations about the suitability of these materials for any purpose and disclaim all warranties and conditions with regard to these materials, including but not limited to all implied warranties and conditions of merchantability, fitness for a particular purpose, title and non-infringement of any third party intellectual property right. TI and its respective suppliers and providers of content make no representations about the suitability of these materials for any purpose and disclaim all warranties and conditions with respect to these materials. No license, either express or implied, by estoppel or otherwise, is granted by TI. Use of the information on this site may require a license from a third party, or a license from TI.
TI is a global semiconductor design and manufacturing company. Innovate with 100,000+ analog ICs andembedded processors, along with software, tools and the industry’s largest sales/support staff.