THS4551: What's the problem of the simulation

I got it to run, but not very pretty - often, a node like that can use a 10M or so to ground to help the sim, 

Looks ok up to the input of the FDA, 

Check DC bias conditions on the THS4551, looked ok, here I slowed it down to 1MHz, still glitching, maybe the diodes are turning on? Try a series 100 that way, 

I opened up that line, no change, not sure what is going on here, 

Thanks a lot! But I still can not run it at my computer even add 10M resistor to GND.

I am using V11, it has options that help, but I did not use them yet. On V9, try switching to trapezoidal integration, 

My version is 9.3. I switch to trapezoidal. Still see below problem.

Well, you were using the 2016 original model which has some benefits, there is a newer 2020 update where they improved convergence but lost the Zout resonance, it gives better results for me, maybe it will work for you

OPA810_THS4551 app with 2020 model.TSC

Your file works fine at my computer.

But I see some problem of the simulation.

 when set source = 300V, the output shows 3V.

But it should not higher than 1.9V per basic analysis, because we use diode D1 and D3.

It doesn't have to do with the diodes:

xiaoqiang_ths4551_1.TSC

Kai

I need the diode to limit the output voltage THS4551 to +/-1.8V, because it's the adc driver input limit.

Hi Xiaoqiang,

the THS4551 input voltage should be limited to 0V and +1.8V, actually.

Have you thought about using a tranformer, as recommended in the datasheet of ADS4222? Even the ADS4222 EVM has a transfomer at the input.

Kai

Or doing it this way?

xiaoqiang_ths4551_2.TSC

Kai

1. I did not use transformer because I still need to measure DC voltage.

2. Thanks for new proposal. But why my solution doesn't limit output voltage in simulation?

Hi Xiaoqiang,

to be honest, I don't know. Maybe it has to do with your chosen Vocm and supply voltages of THS4551? Or it is just a weird Spice model issue? Eventually, Michael can tell more?

But wasn't the diode clamps at the input of THS4551 originally planned to protect the THS4551 itself? Haven't you said this in an earlier thread? I wouldn't have recommended diode clamps at this location to protect the ADS4222. 

Kai

Well you need to look at the actual swing in that resistor midpoint clamp - you have a level shift going on for the output Vocm target, This is +/-200V input. 

So lets say we want to clamp a little above, just seting those diodes biases to these max swings will clamp diode drop above or below 

No, this is not going to work as the impedance changes so much in that clamp point - move you clamp diodes to the OPA810 input. 

Hi Xiaoqiang,

the THS4551 becomes instable when its output voltage comes too close to the negative supply rail of THS4551 (Vee_1):

xiaoqiang_ths4551_3.TSC

That's the cause of ringing and very probably the reason why the simulation with the diodes failed.

Kai

Move diode in front of OPA810 will impact the input resistance as the leakage of the diode. It will reduce the measurement accuracy.

I want to achieve a circuit like a oscilloscope to measure transient, which is about 10M input resistance.

Hi Xiaoqiang,

then add a stage behind the OPA810 and in front of the THS4551 just for the limiting. Mount the protection diodes there and drive them with low impedance.

Kai

And, if you are going down that path, don't use diodes, use the OPA698 output clamping amplifier. Operate with some decent max valid swing there, then attenuate in the FDA stage. 

Here is an example using the OPA698, that part requires a lot of output headroom so I went to +/-5V supplies with clamps set to +/-2.1V right now. The input was cut down by 1/2 at the OPA810 input then gain up by 2 in the OPA698 stage. Here is +/-200V input with the expected +/-2V at the OPA698 output, if I increase the input swing from here, that output should start to clip off. 

Then, here is +/-250V input, 

Not as clean as I would have hoped, but when limiting I doubt you care - now this limited output needs to be scaled through the THS4551 correctly - I had cut that gain by 1/2, but then cut it again at the inputs So I may need to go back to a gain of 1, try that next

So yes, these are nicely limited outputs single ended at the ADC inputs - not sure if everything is set up right here, but I was just demonstrating how useful the OPA698 migh be in this signal path for hard limiting. 

Here is this last file, in V9

OPA810 OPA698 to THS4551.TSC

Thanks for your ideas! So you limit the output using the OPA698 headroom requirement.