OPA857EVM-978: OPA857EVM

Una buona giornata Andrea,

I have just had a look into the datasheet of R7400U. Which type are you using?

Kai

We are using the model   R7400-U03.

Thank you for your help.

Kind regards.

Andrea Papi

Hello Andrea,

For clarification on your design requirements:

• C cable RG58 (100 meters) = 8.2n you have 8.2nF input cap at input of 857
• Ci = OPA = Cdiff + Ccm =OPA857 input capacitance is 2pF
• RF = 1K /2.2K Not quite clear on this.. you are adding these two resistors in parallel in the feedback? You would like to switch between these two gains? 
• CF = Sqare Root of = Cc+ Ci / 2 pai Rf x GBP (OPA 857) ( compesating C cable). You would like to compensate the effects of the input capacitance here & bandwidth lower than 2MHz is needed? Wasn't clear on this statement, "The costant time of the CF and RF should be more than 10 times the T of the maximun pulse PMT frequency 18Mhz (I mean 500ns)."

Best,

Hasan Babiker

Dear Hasan,

I'll answer to your questions:

you have 8.2nF input cap at input of 857 = Yes the current signals from PMT come from a 100m cable RG58 Ctot = 8.2nF.

OPA857 input capacitance is 2pF = Yes In the data is only specified the input pin capacitance non Ccommon Mode 

Not quite clear on this.. you are adding these two resistors in parallel in the feedback? You would like to switch between these two gains? = No it was  my mistake the OPA857 EVM will provide two different TIA Gain 5K to 20K the I'd like to use these gain;

You would like to compensate the effects of the input capacitance here & bandwidth lower than 2MHz is needed? Wasn't clear on this statement, "The costant time of the CF and RF should be more than 10 times the T of the maximun pulse PMT frequency 18Mhz (I mean 500ns)."= The big input capacitance 8.2nF from cable should be compensated by an external CF to stabilize OPA857: We are not working in pulse mode but we'd like to measure the PMT photocurrent then the BW is not so important. In principle we will have PMT pulse input to TIA from low rate 1Mhz to high rate 20Mhz (low light presence and high light presence) and we want a DC level in the ouput. This is an application for detecting low lighr leakage in a dark box.

We want to use a TIA System to convert current in voltage coupled in DC then we send the output buffered to a comparator and set two alarm signals one is TTL and the other is LVDS signals to turn off the HV Bias of the PMT.

We'd like to know if your OPA857 Evaluation Board can fits our need for prorotype, we will have 15 PMT R7400U to monitor 15 dark box each containing 64 multianode PMT to protect.

We will need to know the tau of the TIA (Opa857) expecially if the long 100 meters of RG58 cable at the input can create problem to your chip, but I know that it is possible to use RF and CF externally expecially CF to compesate the Cin (from cable), we are not interested to the bandwidth because we are not woeking in pulse mode but we want to convert a current-to a voltage level on the TIA output we will plase a Low Pass Filter and a buffer to drive the comparator.

I attach the PMT input pulse from a simulation at different rate.

Thank you for your help.

Best regards.
Andrea PapiPMT_1Mhz.pdf20Mhz_PMT.pdfPMT_1Mhz_Zoom.pdf 

Hello Andrea,

Thanks for the explanation, using a 8.2nF cap at the input will not damage the device but as you mentioned there will need to be compensation so that the OPA857 isn't unstable. In this case it will be better to compensate for the 5k-ohm gain option and confirming that this works for the 20-kohm option as well. I've attached an TINA sim where stability can be evaluated using the AC peaking of the device. Developing a full scale stability analysis is difficult due to the integrated nature of the model. 

We get around 2.3dB of peaking with an additional 5pF of feedback capacitance (45.28 deg of phase margin) & stability can be improved further by slightly increasing the feedback cap (0.07dB peaking at 8pF feedback cap). Note that even with 2.3dB of peaking here, the bandwidth for 5k-ohm transimpedance is at 6MHZ (2MHz for 20kohm option). Therefore to achieve the 20MHz bandwidth of your signal you will either need to add a feedback resistor to lower the overall transimpedance gain or select a device with a greater GBW (such as OPA858 or OPA855).

Andrea_OPA857.TSC

Best,

Hasan Babiker

Hi Andrea,

if I had to use the OPA857, I would do it this way:

4807.andrea_opa857.TSC

The phase stability analysis shows that the phase margin is > 68°:

andrea_opa857_1.TSC

Kai

Hello Hasan,

thanks a lot for your design, I think the BandW could not be a big problem, my signal is coming from a PMT R7400-03 Hama, I am sending these kind of signals (I'am using NI Mulsim but it is the same). As you can see the 1Mhz rate of the PMT is when we have Low Light Leakage in the box to protect from light and the rate of 20Mhz is when we have to send an alarm to the High Voltage Power Supply for shutdown the VBias of all the Photomultipliers.

As you can see the rise time, fall time and Imax is -40uA in the pulsed system. I want to use the TIA as a Photocurrent Converter I mean in the TIA output I'd like to have a DC signal proportional to input rate. At 1Mhz it will be a low value at 20Mhz it will be high value then using a comparator I can set the High TH and when I have 20Mhz the comparator set a signal High TTL to send the Alarm signal.

The problem is the 100 meters of RG58 cable and CF (Fedback Capacitor) should stabilize the OPA857 or other OPAMP. Then at the gain of 20K the average current of the pulsed signals at different rate could generate a different DC signal output to send to the comparators in my opinion.

I attached the -40uA signals generated by simulation closed at 50 ohm Rload.

Could you generate this kind of signals to test your circuit if it will work as I expected in transient mode. In effect this application in a middle way between a TIA ampli and CSP ampli.

Thank a lot for your help.

Kind regards.
Andrea.20Mhz_PMT_Pulse_High_leak_Light.pdfPMT_1Mhz_1.pdfPMT_FWHM_2.3nS.pdfPMT_Pulse_Fall_Time_1ns.pdfPMT_Rise_Time_3ns.pdf 

Hello Hasan,

thanks a lot for your design, I think the BandW could not be a big problem, my signal is coming from a PMT R7400-03 Hama, I am sending these kind of signals (I'am using NI Mulsim but it is the same). As you can see the 1Mhz rate of the PMT is when we have Low Light Leakage in the box to protect from light and the rate of 20Mhz is when we have to send an alarm to the High Voltage Power Supply for shutdown the VBias of all the Photomultipliers.

As you can see the rise time, fall time and Imax is -40uA in the pulsed system. I want to use the TIA as a Photocurrent Converter I mean in the TIA output I'd like to have a DC signal proportional to input rate. At 1Mhz it will be a low value at 20Mhz it will be high value then using a comparator I can set the High TH and when I have 20Mhz the comparator set a signal High TTL to send the Alarm signal.

The problem is the 100 meters of RG58 cable and CF (Fedback Capacitor) should stabilize the OPA857 or other OPAMP. Then at the gain of 20K the average current of the pulsed signals at different rate could generate a different DC signal output to send to the comparators in my opinion.

I attached the -40uA signals generated by simulation closed at 50 ohm Rload.

Could you generate this kind of signals to test your circuit if it will work as I expected in transient mode. In effect this application in a middle way between a TIA ampli and CSP ampli.

Thank a lot for your help.

Kind regards.
Andrea.1680.20Mhz_PMT_Pulse_High_leak_Light.pdf0211.PMT_1Mhz_1.pdf0820.PMT_FWHM_2.3nS.pdf3644.PMT_Pulse_Fall_Time_1ns.pdf7357.PMT_Rise_Time_3ns.pdf

Hi Hasan,

it's difficult to find furthergoing information on the R7400U photomultiplier tube in the www. But from other photomultiplier tubes (PMT) I know that the equivalent circuit model of a PMT is a current source in parallel with a small capacitance (usually <10pF) and in parallel with a big resistance (usually >10^12Ohm). Also, many PMT can drive rather heavy loads like a 50R resistance. Timing specification are often given with a 50R load resistance. By the way, 10pF and 50R gives a very small time constant :-)

My R4 = 3M is a bit unrealistically low but has no big impact on the performance anyway.

R2 and C2 (which have been introduced by myself) shall form a low pass filter to suppress the highest frequency contents of PMT signal before going into the OPA857. A passive low pass filter is very helpful here to unburden the OPA857 from this task. And R3=100R is small enough to allow the OPA857 to still work as a TIA.

At the same time R2, C2, D1 and D2 form a protection circuit against overvoltages and ESD. R3 additionally provides some current limiting. But its main task is to isolate the -input of OPA857 from the enormeous cable capacitance (C4) and C2. This allows using active low pass filtering capacitances (C3) in the range between 10p and 1000n (tested, eventually even more) without corrupting the phase margin.

From my feel R2 and R3 should also provide some dampening of resonances caused by the cable acting as transmission line without characteristic impedance matching. But I haven't yet done the according simulations. Funnily enough, Hamamatsu says in a documentation on their PMT that the impedance matching of cabling shall be done by trial and error :-)

Kai 

Hi Hasan,

you said:

"One issue I see with running the stability analysis in this way, is that the integrated feedback resistor & cap within the OPA857 is not included in the analysis."

Oops, yes, now I understand! Yes, you are right :-)

Kai

Hi Asan,

we want to be sensible to very low light level, photon counting mode, for this reason the amplitute is the same at different rate: What is changing is the average current that we sent to the TIA for this reason on the output we don't want to follow the pulse at the input we are not working in dynamic but just to measure the average current of the pulse. This dc signal will be the variable and at 20Mhz High Leak Light I will set an alarm.

If you want i can send all the different physic algebra for counting the charge, the current and the average.

Thanks a lot for your help.

Best regards.

Andrea.

Hi Kay,

thanks a lot for your interesting simulation, but as I said to Hasan I am not so interested to

BW, I am interested to use this TIA as converter to pulse mode to average current to voltage 

converter. The big problem is to compensate the cable RG58 100 meters capacitance 8.2nF.

Which Kind of formula did you use to calculate CF?  one most popular is :

CF = SQ   Ctotal = Ccable + Cin (Op_AmP) / 6.28 (2pai) RF GBWP

but there are also different formulas to use and stabilize a TIA expecially for high frequency is very

complex, but my desing in a middle way between a TIA and Charge Sensitive Preampli.

Yesterday i sent my PMT input signal simulated you and hasan  gently could you try to use TINA TI and send thes

signals to your design Input. I'd like to see the TIA output.

bBest regards and thank you.
Andrea.

Hello Andrea,

It looks like the input signals are you've provided are voltage measurements? Is the output of your PMT a voltage or current signal? If current can you provide amplitude of current pulses to run a transient sim?

Best,

Hasan Babiker

Hello Kai,

Thanks for your explanation of the circuit, I was wondering if I had missed something in the part datasheet. Yes probably best to be safe and protect the OPA857 in this case.

Interesting note on the impedance matching needing to be done by trial and error, wonder how reliably this can be designed for.

Best,

Hasan Babiker

Hi Andrea,

I do not use a special formula to find the best component values of the circuit, but I run the according SPICE simulation. I use a trick to allow the TIA to handle large stray capacitances at its -input, the trick of isolating this stray capacitance by the help of a small resistance. The following simulations show the performance of the standard TIA topology when varying the feedback capacitance.

10pF:

1nF:

100nF:

andrea_opa857_3.TSC

From the increasing peak in the frequency response and the increasing ringing in the transient analysis you can see that the circuit becomes more and more unstable when increasing the feedback capacitance.

With the adding of the small "isolation" resistance the circuit becomes much more stable:

10pF:

1nF:

100nF:

andrea_opa857_2.TSC

So I would recommend to add R2, C2 and R3.

In the following the performance is shown with your 50ns and 1µs distanced pulses assuming an amplitude of 100µA:

50ns, 100pF:

1µs, 100pF:

50ns, 1nF:

1µs, 1nF:

50ns, 10nF:

1µs, 10nF:

Kai

Hello Hasan,

yes of course I send my simulation PWM I_Source for the pulse current at different rate (the voltage is I_pulse on 50 ohm scope)

I_Pulse @ 1Mhz Rate Low Light PWM File:

Time         Current

1us           0

1.001us    -40uA

1.004us    0

Repeat data during simulation

I_Pulse @ 20Mhz Rate High Light PWM File:

Time         Current

46ns           0

47ns           -40uA

50ns           0

Repeat data during simulation

I think you can create this pulse current with TINA TI.

Kind regards and thank you.

Andrea.

Hello Asan,

I will try using this OPA.

Thanks a lot my thanks also to Kai for his design and simulation very useful.

My best regards.
Andrea.