TMUX8212: PSpice Simulation Results & Clamping on Output of Analog Switch

Hi Mitchell

Thanks for explaining your simulation and what i would like to start of saying is that this multiplexer isn't designed to handle what our team calls overvoltage events.

In simulation and in real life devices like the TMUX8212 will show incorrect operation when signals higher or lower than there supply rail are placed on the io pins.

This FAQ will give you a general idea of what is going on here.

However TI does have a family of multiplexers with a feature called overvoltage protection that will isolate  the drain pins of the multiplexer from the source pins  when  a overvoltage event happens

Here is a FAQ about overvoltage protection multiplexers

With the TMUX7411F,TMUX7412F or TMUX7413F you will likely not need TVS diodes in your application!( There are also other configurations available if you are interested)

Please let me know what you think and I am happy to assist

Regards,

Kameron

Hey Kameron!

Thanks for the information. Those devices you mentioned look great - but it seems like from what you talked about they'll isolate the output pins and not clamp them? Are there devices that would let me clamp the inputs to the rails instead. I'm looking for the device to behave like a Zener at the rail voltage. 

Let me know if there is any other information that you need. 

Thanks!

That's awesome!

Thanks for the additional information - I'll run some simulations and then close the issue. 

Appreciate the support!

Hey Mitchell,

Perfect I think I understand what you are asking for better now.

The TMUX7462F would actually allow you to clamp the voltage on the drain pin during a overvoltage event.

This is because this device has a drain response pin that can be set to clamp the voltage on the drain pin to the rails.

For your application I believe you would connect VFP to VDD and that way you could get the clamping behavior you are looking for.

Please let me know if I am misunderstanding anything and I can take another look.

Regards,

Kameron

Good morning, Kameron!

Do you know if there are SPICE models available somewhere other than the TI Website for this component. I'd like to be able to verify this design with simulation data but I don't think the provided IBIS models will be able to fully verify this.

If you think the IBIS model should be sufficient for verification then I can figure something out such that I'd be able to simulate that with the software I have - but I was under the impression that IBIS would be better suited for determining signal integrity issues rather than verification of a circuit design. 

If there isn't a model available then is the actually switching circuity similar to the TMUX8212 in operation? If that's the case then I don't really have much concern and can just follow application guide in the datasheet for implementation. 

Would you mind if I followed up with a schematic for verification of the design?

Thanks! 

Edit #1: Also wanted to make sure I was understanding this from the datasheet. On Pg. 25 of the TMUX7462F Datasheet in Section "8.3.2.1 Input Voltage Tolerance" it specifies that if DR if left floating an an overvoltage conditions occurs it'll be clamped by the rails. I just wanted to confirm that if DR if left floating outside of this specific situation it'll still be able to act as a switch (e.g. When there is an infinite load on the DR that they'll maintain their voltage.) Figure 9-1 (Typical Application) would suggest this is true since DR is driving an OpAmp - but just wanted to confirm since not explicitly stated.

Edit #2: From looking at the datasheet it doesn't seem like there is a method of controlling whether the switch is closed or open other than if there is an overvoltage condition. Is there a part that could combine the functionality of the TMUX8212 and the TMUX7462F - where I'd be able to select whether the lines were closed with a select line and then also have it go high impedance if there is an overvoltage condition. If not I supposed I could connect the TMUX7462F after the TMUX8212 in series. 
 

Hi Mitchell and good day to you as well.

So actually there is no PSPICE model for the TMUX7462F due to the fact that its particular overvoltage protection circuitry is difficult to represent in simulation( The actual OVP is hard to show and create in simulation)( The internal protection features in the TMUX7462F are different than the TMUX8212)

However the ibis model shows its signal path performance of the switch and during normal operation 

The ibis is good to show the signal path performance of this device in normal operation.

The best idea I think here is to follow the application guide in the datasheet.

In addition I understand your question about section 8.3.2.1( This section can be tricky but let me try to explain)

So the DR pin being set to GND allows the device to clamp during a overvoltage event( Drain pin pulled up to VFP or VFN during overvoltage event)

When the DR pin is set to Logic High or floating then the device drain pin will behave like a open circuit( Signal path will be in a high z state and the voltage on the drain pin will be  dependent on what load is connected to the device) 

This is more of a system design question and is important if you would have a infinite impedance load connected to the mux( like a op amp buffer) but this shouldn't be a problem in your application to my knowledge.

This is a app note that i think is related to this corner case that is important to remember in some signal chain applications.

Please let me know if I clarified your question and I can take another try if this didn't work

Regards,

Kameron

Good morning, Kameron!

Thanks for the information and explanation. Some of the analog switches are connecting to external measuring points that will be connected to with a oscilloscope so I will have infinite impedance loading but never in a situation where the oscilloscope is connected and those measuring points won't be connected. (e.g. assuming it doesn't cause any issues I don't think I'll need to pull-downs on those non-selected lines since I don't care about their voltage when not selected)

The application note was a good read so I appreciate that information - at some point we were looking at using buffers - but decided otherwise - but it's an issue I'm sure will come up a some point. 

I'm not sure if I edited the post before or after you replied yesterday - so I apologize if it was after you replied and I just didn't see it but let me know if you're able to find a component that can combine the functionality of the two devices without having to have two chips. I should have said this earlier but the goal of this project is to make a bus switch that will be able to limit the voltages to the rails. 

Thanks again and let me know if you need any more information!

Sincerely, 

Mitchell M. Gresham

 Hey Kameron!

Thanks for all the information. I've been working on this issue for about a week now and have had a couple ideas on how I might solve it. I'm not sure if the options that TI has in analog switches will work since all of the options presented have a larger profile than we have space for on the board. 

I've actually been thinking about using an OPAMP with a SHUTDOWN pin as a switch and at the same time solve our clamping issue by limiting the input to the rails. I'll ask a related question about this part but I'm thinking the OPA2992 might work. Since I can just turn ON/OFF whichever OPAMP depending if the signal is an input or an output and then use the rails to clamp. 

Again I'll close this issue and open another one - but just wanted to look for some initial feedback. 

Thanks!

Hi Mtichell,

Thanks for messaging me back and actually I am going to loop in TI op amp team to see if they think the device you selected could work here.

Regards,

Kameron

Hi Mitchell,

Mitchell Gresham said:

I've actually been thinking about using an OPAMP with a SHUTDOWN pin as a switch and at the same time solve our clamping issue by limiting the input to the rails.

OPA2992S op amp has output shutdown feature, and the disable to output voltage is high in GOhm range. However, the abs. max. volage in the part is 42V. 

I did not go through the previous reply in detail. If the output volage spikes are less than 42V, then output voltage may be able to survive it. Otherwise, you would still need to voltage clamp to clamp the op amp's rail below or equal to 42V in a shutdown state. 

If you have other questions, please let me know. 

Best,

Raymond

Thanks so much for the assistance Kameron!

Hey Raymond! Just wanted to run over the idea that I have real quick for clarification. 

This is the configuration that I'm suggesting. 


I'm setting the op-amps as buffers and then putting resistors in front of them such that the current going into the device if the voltage on inputs need clamped then it's limited. I'd like to set the rails to +-20V and then limit the input which could be up to +-30V  to the +-20V on the rails. The SHUTDOWN pins will be configured from an external source to be placed into an "INPUT MODE" which I'm assuming the source is coming in from the left and an "OUTPUT MODE" where the source is from the right - since I'll always know where the source is coming from. Since SHUTDOWN pins will be controlled by an external source I'm not too worried about them but let me know if you think this configuration will work. 

Thanks!

Hi Mitchell, 

Mitchell Gresham said:

then limit the input which could be up to +-30V  to the +-20V on the rails.

Per your description, this is my interpretation. Please see if it is correct. There is no feedback between one op amp from others. It looks like a summing amplifier configuration.  

OPA2992 Buffer with SD 07162025.TSC

It looks like that the over_voltage is occurred at input. If the input is protected, then output may be protected. 

In our catalog, we have OPA2206, which it may naturally clamp the overvoltage input (without the use of Zener, TVS or diode protection clamp

. 

Alternatively, we may use the following topology to protect the overvoltage input (using diode and TVS/Zener to clamp the Vcc or Vee). 

https://www.ti.com/lit/an/slvaex7a/slvaex7a.pdf?ts=1752692068671&ref_url=https%253A%252F%252Fwww.google.com%252F

https://www.planetanalog.com/op-amp-input-protection-can-be-noisy/

Please let me know if my understanding is correct. 

Best,

Raymond

Good afternoon, Raymond!

Thanks for the update - and from what I see the schematic does look how I'd expect it to. 

It's my understanding from the OPAx992 Datasheet that "Input pins are diode-clamped to the power-supply rails. Input signals that may swing more than 0.5 V beyond the supply rails must be current limited to 10 mA or less" (5.1 (Note #2)) which seems to suggest to me that there is overvoltage protection on the input of the device as long as I have a large enough resistance to limit the current through the diodes to a reasonable current (<10mA). 

I'd really like to use a model of op-amp that has a SHUTDOWN pin since I can eliminate the need for an analog switch by turning ON/OFF the op-amp itself. 

If needed then I can do as suggested and use a diode clipping circuit on the input of the op-amp but I'd like if I could minimize the amount of external parts needed for the switch to operate. I don't expect any insane voltages on the inputs of these devices but it is reasonable to assume that it could be 10-20V above what's expected. 

Just let me know what you think - I'd like to be able to protect overvoltage from input and output depending on how it's configured. 

Thanks!

Hi Mitchell, 

Mitchell Gresham said:

Input pins are diode-clamped to the power-supply rails. Input signals that may swing more than 0.5 V beyond the supply rails must be current limited to 10 mA or less" (5.1 (Note #2))

Yes, the ESD protection scheme are used in near all op amp at input terminals, which is a standard (may use different protection design topologies). The suffix -S features are only presented in handful of op amps, where OPA2992S can shutdown individual op amp' output.  

You mentioned that the input signal can potentially go up to +/-30Vpk, thus we can use ESD diode as the current path, if the input current is limited to +/-10mA. In addition, the Vcc and/or Vee rails need to be clamped as below Vs. The op amp's configuration should not violate the following abs. max. ratings. Otherwise, it is considered as EIPD (External Induced Physical Damage). 

You will need the TVS voltage clamps as shown in the image below, if the input transient is close to ±30V. 

Please let me know if you have other questions. 

Best,

Raymond

Awesome - okay this makes a lot of sense now!

So just to confirm if I'm understanding what you're saying - then as long as I'm able to current limit the input rails on the op-amp I should be good in this configuration assuming that I also have a diode on the rails connected to the op-amp that can sink the current to ground? 

So for example if I had an input that was a a 50V then I'd have a ΔV of 30V and then 30V/10kΩ = 3mA which means as long as the ΔV isn't more than 100V - which it will never be close to - then we'll be in specification for the diodes in the op-amp itself and assuming I select diodes on the rails that are also rated to such current. 

If my understanding is correct then the power from this drop will be dissipated in the resistor itself. 

Thank you so much for the assistance and let me know if I've misinterpreted something!

Edit #1: I'm kind of confused the role that the TVS Diode is playing on the rail of the op-amp. I'd like the clipping voltage for the op-amp to be adjustable and therefore I'm expecting that the current be sunk into the rail itself rather than through the TVS diode since I've designed it such that the TVS has a breakdown around 22V which is the upper-limit of the clipping voltage required. Is there a reason why the current flows into the TVS rather than directly into the LDO that's powering it and is sunk in that device or am I confused?

Hey Raymond! Thanks for the information. 

A couple things and then one question. Would you be able to confirm that the information I provided is correct in the previous response? Your previous response demonstrates a 31.4V drop across the current-limiting resistor which led me to the assumption that it will be the element that dissipates the power and the ESD Diodes won't actually dissipate much and instead are acting as a current-path. Also would it be possible to move the resistor from the input of the op-amp to on the rail itself in a pull-down configuration. In this way whenever the overvoltage situation occurs the ESD Diodes will sink the current to the rails and since the LDO can't sink it'll end up sinking through the resistor to ground? I made a diagram of this proposed configuration below. 


Let me know if this would be a good solution since I can't do a Zener on the rails since the voltage will change to clip. 

Thanks!

Hi Michell,

Mitchell Gresham said:

Also would it be possible to move the resistor from the input of the op-amp to on the rail itself in a pull-down configuration. In this way whenever the overvoltage situation occurs the ESD Diodes will sink the current to the rails and since the LDO can't sink it'll end up sinking through the resistor to ground?

I am referring to this diagram, where 10kohm or larger resistor in series with Vin+ to limit the ESD current. 

Mitchell Gresham said:

I can't do a Zener on the rails since the voltage will change to clip. 

Please configure your overvoltage protection  as shown in the image above. 

The resistors of 100kohm shown in your previous drawing are loads for +/-20Vdc, and the two 100kohm resistors do not do anything to limit the input current at Vin+. You will be using TVS to clamp the overvoltage conditions.  

Best,

Raymond

Good morning, Raymond. 

I'm unable to use the previously mentioned configuration because of the changing clipping voltage on the rails. Would it be possible to still have the 10kOhm as pictured on the input and then put another 10kOhm on the rails in place of the TVS Diode. I agree with the configuration up to that point. 

Also you mentioned that LDOs usually won't sink and source current at the same time? Is it that they won't do it because of protection in the circuit or is it that they're no supposed to - because my assumptions are based on the LDOs having some protection in them that won't let them both sink and source current only whatever they're supposed to based on output voltage. 

Also is it possible to have a call with the applications engineering team if this issue isn't resolved sometime soon? I do feel like some information is getting lost in translation here and it's quite an important part of a project our company is working on. 

Thanks!

Edit #1: I ran simulations with the previously mentioned strategy and am getting results that seem good - but still would like some perspective on them. 

Edit #2: Added rails into simulation using different op-amp just for confirmation. LM741. Clipping results were same as with previous circuit. 

Hey Raymond!

Awesome! I ran the simulations that you attached and then ran some of my own and understand what you're talking about! I'll see if over the weekend I have any questions about this once I do some analysis that isn't simulation - but I appreciate all the help! Sorry took me a little bit to see what you were saying but I understand now!

Thanks!

Hi Mitchell, 

Ok, please let us know. 

Best,

Raymond

Hey Kameron!

Spent a good while working on this problem and while I thought the op-amps would do the purpose I wanted. I'm concerned about the power dissipated in them in standby and was wondering if you had an opinions regarding the TPS16416. While I liked the previous suggestions that you gave for limiting chips I'm concerned about the high on-resistance of them. This chip seems to have all the protections that I'd like - and more like power protections and current limitations while maintaining a low on-resistance. 

Let me know if you think this would work for the application because I'm back to wanting to use the TMUX8212 with some type of protection on the front - but just want to make sure I'm not missing something. 

Edit #1: Realized right after I posted this that I did miss something and that was that you can't use the TPS16416 in an application where signals could go negative. Instead the TVS1800 as suggested by Raymond might honestly by itself solve this issue. Let me know what you think!

Thanks!

Hi Mitchell,

Because that is a TVS diode (TVS1800) I will loop in the TI team who can help more with that device 

Hi PD Team can you help here?

Regards,

Kameron