OPA354: OPA354

Hi Sima,

thanks a lot for your help but I have problems to debug this simple circuit and I'd like to simulate, of course using a dual power supply +/-5Vdc is out of range of the limit +Vmax, but I am using the OPA354 with a single rail voltage of 3.3Vdc and GND. As you can see from my simulation the RF is 2.7Kohm and Ri is 330 ohm then the gain G = 1 + RF/Ri = 9Vdc raffly. My input signal is unipolar, as visible, is a positive pulse coming from a Silicon Photumultiplier it will go to the input of the the first amplifier its output then enter a Peak Detector circuit with an analog switch TMUX1101 as reset for the hold capacitor 10 nF. If I can't simulate the single non inverting amplifier with OPA354 I can't simulate the Peak Detector circuit. The reset circuit is coming from an integrated comparator inside an MCU, when the Peak is more positive of the reference the comparator send the reset signal 0-3.3V to the analog switch. I am interested only to the counter rate of the input signal it is for a rate dosimeter for gamma and beta particles. Could you help me to simulate this part of the project is the analog part tha does not work (Ampli + Peak Dector). We used all TI components and we encounter problems for a total of 10 board produced by another group (Turkish Ankara) from our collaboration that designed the boards. I am attaching my simulation and the scheme of the analog part of the project. I send my best regards and thanks a lot for your help in this task.

Andrea Papi  

Hello Andrea,

   For simulation, when using a 3.3V single supply, the amplifier will accept inputs from -0.1V to 3.4V and can output from 0.1V to 3V. For your values of 10mV, the input is fine, but output can not go below 100mV. This is why simulation and amplifier will not work for this type of input. You would need to add a DC offset/bias at the input as shown in the previous reply.

  Thank you for the all details! Since this is being used in a TIA type application, is the issue with the boards directly at the output of the TIA? If so, are you running into oscillation type issue, or just incorrect type of output? If a TIA issue: What is your PMT's input capacitance? Usually PMTs require large reverse bias voltages which tend to have higher input capacitance.

  Or is the issue at 2nd Stage 0 and Stage 1 which looks to be a gain stage after AMP0 and AMP1, or is the issue in PH0/PH1?

Best Regards,
Sima 

Dear Sima,

Thank you very much for your valuable help. However, I am having some difficulties simulating and understanding the different blocks of the project. As I attached in my previous emails, we have a gain stage with the OPA354 and a peak detector stage with the OPA2354, using AC coupling to avoid the 100 mV pedestal from the single-supply op-amps, as you mentioned regarding the common-mode voltage (CMM) issue.

At the input of the OPA354 we have a Mux TIA (LMH32404IRHFR from TI). On the four channels, as you can see, we have inputs from ON Semiconductor Silicon Photomultipliers at 50 ohms, and these are AC coupled to the TIA. You will see the scheme of the TIA and the details, we also trasform the differential signal of the TIA LMH32404 into single end signals by Balum.

I would like to run a TINA9 simulation, but I am having trouble importing the PSpice model from TI into TINA9. Could you give me some hints and tips to make this process easier? Or a tuturial or application note to understand how to import model into TINA_TI_9.

Thank you for your help.

Best Regards.

Andrea Papi

Hello Andrea,

  I apologize for the long delay on a reply again. Your block diagram makes sense for your application, the only concern I see is the AC coupling to avoid the 100mV clipping due to the single supply operation + input/output voltage violation of the amplifier. The amplifier will still have the issue even when covered by the AC coupling, and amplifier may not work in a linear operation + there could be reliability concerns in the long-term operation. If unable to offset the voltage via the resistor divider from the supplies, I would suggest instead to add a low noise negative bias generator. This generator works with single supply positive amplifiers and you do not need to provide a negative supply. This generator usually outputs -100mV to -300mV which is enough to avoid the amplifier saturating at VS- (VEE). 

  Here is a circuit example using the LM7705: https://www.ti.com/lit/an/sbaa373a/sbaa373a.pdf?ts=1753313982117. 

  For the second question on importing PSpice components into Tina-TI, below is a guide:

Extract the zip folders until you reach either a .lib, .cir, or .mod file. Insert the file into TINA-TI by selecting Tools, New Macro Wizard, then browse and select the .MOD file. Press Next, load shape from library, and select the correct template symbol/shape with the right amount of pins. Select next, assign the pins to the right areas of the symbol based on the pin assignment in the netlist, select save and now it is ready to be inserted anywhere in the open schematic. Here is a more in-depth guide of the process: http://www.ti.com/lit/an/slva527/slva527.pdf and in video format: https://training.ti.com/importing-spice-models-tina-titm-simulator. Start at Figure 3 in this case. 

Thank you,
Sima

Dear Sima,

Thank you very much for your help. Based on your advice, if I use the LM7705 to power the gnd pin of OP354 pin 2 and pin 4 of OPA2354 I can avoid placing AC coupling at the input and output of the Peak and Hold circuit, as shown in my attached schematic.

The output of the Peak and Hold circuit feeds into a comparator with a programmable threshold. This same comparator controls the TMUX1101 to reset the hold capacitor C82, and the STM32 microcontroller increments its count each time the threshold is exceeded.

During my debugging process, I also identified issues with the LMH32404 TIA.

Our colleague from Turkey used the Evaluation Board schematic directly for the TIA, but the SiPM should be connected so that the input current enters the TIA, not voltage. Moreover, when simulating with a pulse generator at the 50 Ω resistor at the TIA input, I observe an oscillation at the output of the Balun Amp0 and Amp1 input of the OPA354 on the falling edge.

I have attached the links to the Evaluation Board schematic used by our colleague. I would also appreciate it if you could verify whether the connections to the Balun are correct.

www.ti.com/.../LMH32404RHFEVM

I would like to connect the SiPM as indicated in this block diagram:

However, we only have a positive BIAS available, and I need to connect it to the Cathode. Could you please confirm if my modification is correct and if this could solve the low amplification issue of the TIA?

In my opinion reading the data, the TIA should provide a gain of approximately 20 mV per 1 µA input current.

Thank you for your help.

Best regards

Andrea Papi.

Dear Sima,

I am attaching the scope plot of the oscillation as indicated above:

Oscillation Issue

• Oscillations are observed both at the input and output of the TI transformer (Balun).
• The oscillation manifests sharply on the falling edge of the signal.
• Despite efforts, the origin of this oscillation remains unclear.
• The bandwidth and gain of the OPA354 filter out this oscillation at its output.
• Relevant diagnostic materials include the schematic and oscilloscope screenshot labeled Probe_1_Out_TIA.

I am also searching for a reference design for Lidar using LM32404 but I am not able to find any reference design using this chip for automotive Lidare example with all the analog part detailed before to enter in the digital domaind like FPGA or ARMMCU.

Thank you very much for your help we have to solve this kind of problems because we have to replicate other 20 boards.

Best Regards.
Andrea Papi.

Hello Andrea,

  Yes, I would fully recommend to use the LM7705 to power the ground pin of OPA354 (pin 2) and OPA2354 (pin 4). This will avoid clipping the negative rail which is causing the offset. You do not have to remove AC coupling if you do not need DC information of your signal, but you can remove them if you do want to track DC signals. Otherwise, having a small negative voltage at VS- for your devices is most critical to keep amplifier working in a non-saturate/clamping operation.

Issue with LMH32404 TIA

Yes, you are correct, SiPM connected should be an input current source to the TIA, and not converted to a voltage prior to input of TIA. The TIA will be the current to voltage conversion while keeping the system stable. In the LMH32404 schematic, please

• Remove R24, R28
• Short C52, C56 with a 0 ohm resistor
• Short R22. R26 with a 0 ohm resistor
• Remove C54, C58 and would recommend placing the SiPM as close as possible to the input of the LMH32404 on the layout

Above bullet points follow for correct SiPM connection is done with the block diagram you shared. I would though double check if the SiPM you are using is voltage or current output. Some SiPMs have an integrated resistor for current to voltage conversion. If current source, please proceed with above bullet points and your block diagram.

Since you are using a positive bias (cathode is biased to a positive voltage and anode is tied to the amplifier input), you can still use the LMH32404. The features that won’t be available are:

1. Integrated ambient light cancellation will not function. This means you are correct about asking about IDC_EN, since IDC_EN will have to be left logic low, and we recommend driving this digital pin rather than leaving the pin floating because fast-moving transients can couple into the pin and inadvertently change the logic level.
2. When the photodiode sources the photocurrent (cathode is biased to a positive voltage and anode is tied to the amplifier input) a soft clamp activates when the amplifier input is overloaded (high input current saturating the amplifier). When the soft clamp activates, the amplifier takes longer to recover which causes increase in pulse stretching/extension. The recovery time depends on the level of input overload. Therefore, the soft clamp is the active clamp when the photodiode is sourcing a photocurrent rather than the fast recovery clamp. 

Thanks for the scope shots! That does look like ringing on the pulse signal. Would you be able to probe at R32 and R33? If you still see the same issue, please follow above bullet points after checking if SiPM is current output rather than voltage. Also, what is your SiPM internal capacitance at your set reverse voltage bias?

Thank you,
Sima

Also, here are some references for automotive Lidar that includes analog + digital portion:

1. Multi-channel optical front-end reference design overview using the LMH32404
2. App note on optical front-end reference design overview using the LMH32404
3. LMH32404-Q1 Datasheet has application information on section 8 starting on page 23
4. E2E thread that includes the board design files
5. Application brief overview: Time of Flight and LIDAR - Optical Front End Design
6. Reference Design: LIDAR-Pulsed Time-of-Flight Reference Design Using High-Speed Data Converters using OPA857 which is similar 
7. Reference design: maximizing transimpedance bandwidth for LIDAR and time-of-flight (ToF) applications using OPA858 
8. Overall LIDAR module

Hello Sima,

thanks a lot for your replay I am working to modify the scheme as you indicated for the SiPM input to TIA. SiPM is the Micro FJ 30020 from Onsemi is a current output. The capacitance at the anode is raffly  is 1nF at Vbias of 30Vdc (Cathode).

Yes on R32 and R33 I still have the some ringing on the pulse fall.

I am sending the data of the SiPM used .

Thank for your help.

Best regards.

Andrea.

MICROJ-SERIES-D.pdf

Hi Andrea,

I was able to look over the conversation regarding the TIA portion of the circuit and I was wondering if the changes in the circuit Sima suggested were implemented? You mentioned you still see these issue but wanted to confirm it is with the changes. Do you see any ringing behavior at the output of the LMH32404 when no signal is sent? Or are the outputs just a dc signal at 1.1V? Does this behavior change with changing the transimpedance gain?

Best Regards,

Ignacio

Hi Ignacio,

I will replay each items:

1) I still see this ringing but the modifications have not been made at this time. The circuit is very dense and complex. It is a portable dosimeter for medical physics for measuring gamma, beta, and neutrons, and the TIA functions both as an amplifier and a mux for the three measurements at different times.

2) I am sending simulated Silicon Photomultiplier Signal semigaussian and as you can see at the output of the second stage amplifier OPA354 is still visible the ringing.

3) In any case I can test the TIA Output when no signal is sent at the input.

The TIA gain may be slightly higher if the output load resistance is increased (not 100 ohm), due to the internal configuration, but the nominal/typical design value is 20kΩ and is not intended to be user-adjustable, it is fixed by internal resistor values and architecture I believe.

Thanks a lot for your help.

I send my best regards.

Andrea.

Hi Andrea,

Thank you for the information. I would like to see if maybe this ringing is seen at the output of the LMH32404. If the customer can send the same signal and probe prior to the 2nd stage, that would be good to see if it is the TIA stage or the 2nd stage. Is there a way to decrease the input signal to see if you see this issue when the output is decreased? Another potential issue is the setup of the device. If the device is sinking current, the current cancellation circuitry must be disabled. This means the IDC_EN must be set high. It is currently set low, which I am not sure how it affects the response of the device. To rule this out I would recommend cutting this trace to ground and setting this pin high with some kind of external wire if possible. One other recommendation would be to remove the output capacitors (C67 and C70) as these will affect your phase margin. I am not sure how much it affects the circuit but it is something else I would remove for debug purposes.

Best Regards,

Ignacio

Hi Ignacio,

Yes, I’m attaching oscilloscope photos taken at the probe point at the output of the T1 transformer balun, at the input of the OPA354 with a gain of about 10 V/V. The oscillation occurs at the output of the TIA and is also directly present at the Pair1+/Pair1- and Pait2+/Pair2- (R32-33-34-35) ringing is present also changing the peak of the input signals from 50mV to 25mV.

At the moment, the configuration is incorrect as per the schematic, but the system, even though wrongly connected with the Silicon Photomultiplier, is in a sourcing current configuration. The same sourcing configuration will be present when I'll make the modification as suggested and shown in the attached schematic at the TIA input. I suppose that, in this case, we are in a sourcing—not sinking—configuration, we would need to have the anode at a negative bias and the cathode connected to the TIA Input in order to achieve a sinking configuration but it is not possible with minor modifications. Then if my assumptions are correct it means that IDC_EN should be connected to GND.

For phase margin I can remove C67 and C70 in an easy way and try the ringing will be reduced.

I send my best regards, I sincerely thank you for your help in solving these problems on a board that has already undergone pre-production and is very compact and difficult to rework.

Andrea

Hi Andrea,

It does get very confusing with how the configuration affects the device. Just to confirm, when the anode is negative biased and the cathode is connected to the input of the TIA, the device is sourcing current (photodiode is sinking current) and in this configuration the cancellation feature can be used. If the photodiode cathode is positive biased and the anode is connected to the input the device sinks current (photodiode sources current), and the cancellation feature must be disabled. I would like to add, depending on the current range for this SiPM, you could consider using a resistor to convert the current to a voltage and use a high bandwidth device to gain that voltage signal to the desired value. You could also use a discrete TIA design with a smaller feedback resistor to gain the current. With the LMH32404, the linear current range for this device is around 70uA because it has a high internal gain. I believe your current range is higher than this which would lead to unideal behavior even if you do make the adjustments in the input stage. 

Best Regards,

Ignacio

Hi Ignacio,

you are right, I got confused between the SiPM and the TIA input. In the configuration I need to implement, the SiPM sources current on the anode, while the TIA input sinks current; therefore, the IDC_EN pin should be set high. This is very difficult, since the PCB has 8 layers and the LMH32404 package is very compact, so I don’t know how to perform the rework to pull the IDC_EN pin high. 

We must use the LMH32404 because it allows us to multiplex different channels using pins M1-2-5 as indicated in the schemes. We are just follow your  suggested approach: as you can see from the schematic on the SiPM anode we have resistor R21-25-29-31-24-28 that converts current into voltage. At minimum, we have about 40 µA across a 50 Ω resistor, giving about 2 mV, which is then fed to the TIA through a 2 kΩ resistor. That corresponds to about 1 µA, which multiplied by the 20 kΩ gain of the TIA gives approximately 20 mV output. With an additional gain of 10 from the OPA354, we reach around 200 mV at the input of the Peak detector. However, we are encountering issues with oscillation at the TIA output and offset in the OPA354 output, and we will likely need to use, as suggested by Sima, the LM7705 chip to bias gnd for Opa354 and Opa2354 (Peak_Detector). We have to solve both these problems: place IDC_EN pin high (for the ringing) and using LM7705 for input common mode error in OPA, we could also increase the conversion resistor R from 50 Ω to 470 Ω to see if we gain more input dynamic range for the TIA, while keeping this configuration even if it is not fully optimal. The voltage on the 50 ohm is sourcing current and TIA sink this current. Waiting for your useful comments and hints to solve these problems on the 10 board assembled, I send my best regards.

Andrea.