THS3491: THS3491 Burning Up

Hi Martin,

This is a high output current device, so it is capable of driving a large amount of current, however, this is not something that we have experienced in the lab with this device. Could you please share a schematic and PCB that is being used with this design. It is capable of driving large capacitive loads, but it is surprising it is causing this to your traces. Does the device get damaged when running these tests?

Best Regards,

Ignacio

Thank you for the response. I can only describe the schematic and layout. It is setup as a non-inverting amplifier with a Feedback resistor RF = 576 Ohms and Rg = 294 Ohms, the output has a 50 Ohm (and 1000uF in parallel to the 50 ohms) in series. Our V+ = 15V and V- = 15V. We also have PD tied to V+
I did not read anything regarding what the max voltage applied to PD can be? Is it lower than +15V? Now that I look at the schematic our decoupling caps 6.8uF/16V are probably a little too low for our +15V, might see a spike of voltage on that line that might have caused that cap to short to ground and cause the rush of current to the traces. Maybe a possibility of what happened. The datasheet shows the decoupling caps but not the voltage rating. Any recommendations?  

The traces on the V+ and V- are burned up. There are no burn traces on the Output traces. Which tells me that this was not driving at a high current rate?

That seems to be the one that you are talking about that has Maximum Current Output per the datasheet reading 550mA. What are the maximum ratings for the VIN+ and VIN-, perhaps I am driving this too much? Causing the V+ and V- to draw more current and stressing out the amplifier?

This under our standard testing does not burn up, the end users after what seems to be running continuously for 1-2 months experiences this issue. Our PCB has enough cooling, so I do not assume this is a heating issue. I am not aware of much of the setup that they cannot share. The only thing I do know is that this output (from the amplifier) is connected to a well-protected antenna. In any case I wanted to ask what can cause a V+ and V- to draw an exceeding amount of current which caused the traces to burn up.

Hi Martin,

This is good information. PD can be tied to 15V so that should not be an issue assuming REF is within the expected range. Typically, we ground REF. The choice of Rf might be relatively low but if you do not see any sort of oscillation at the output then it should be ok. For the choice of decoupling we do suggest a larger capacitor closer to the supplies which 6.8uF is ideal but we also suggest adding additional decoupling capacitors as close as possible to the device supply pins. The values we suggest are around 100nF for these capacitors. Do you have this scheme on your PCB? We have seen issues on customer boards when excluding these smaller capacitors. However, I am not certain this would cause the device to burn up unless there is an oscillation causing excess current to be drawn. For voltage ratings anything above 15V should work we typically use 50V. The common-mode voltage range of this device is around +/-11V which could lead to see issues if overdriven. Is there any information on how the device fails? Or are the supply traces too small for the amount of current through them? 

Best Regards,

Ignacio

We are using RGT Package. In the 8-Pin HSOIC package there is a pin for Vref but what about RGT Package? I am not seeing any oscillations. Those decoupling capacitors (6.8uF and 100nF) are installed. The 6.8uF cap we use AVX: TAJA685K016R looks like it has a surge voltage of 20V. The traces width is wide enough to handle 3A. Investigating further we have a AD8337 driving the input of the THS3491. Under our standard ABSOLUTE MAX testing I cannot get this to break even if I run it for 1 week straight! Only customer shared information I have is that the device works for about 1-2 months before failure. I will let you know if I find anything else. 

Hi Martin,

The reference for the PD for the RGT package is the GND pin which is pin 5. Is there any information on how the device is turned on. We have had customers damage devices by not powering on the device correctly, either by having an input signal prior to the device being on or turning on one supply while the other supply is still off which could cause the device to fail. 

Best Regards,

Ignacio

Hi Ignacio,

From what I know the device in test has been on continuously for months before failing. I do not think it is being injected some signal before it turns on, but it could be the case. The Chain is a AD8337->THS3491. If the customer is indeed injecting a signal the AD8337 would be the first thing it hits.

Hi Martin,

Seeing the traces burning up on the supplies and nowhere else could point to maybe the supplies being powered incorrectly where one supply is on before the other. Floating supplies can cause damage to the amplifier as the current has no clear path and can cause excessive current draw and damage. However, this is just a guess as it is hard to give any definitive reason. Does the system violate any input/output ranges when operating? Or do the supply pins get any transient spikes that exceed the recommended supply?

One other consideration that was mentioned to me which I did not consider is potentially some edge triggered ESD cells being damaged when the supplies are turned on too fast. We have this spec in the abs max table (second row). This could also be a point of failure worth considered.

Best Regards,

Ignacio

The board does not seem to violate the VIN- or VIN+ (the most this pin can get is +5V/-5V) 

Since we have a non-inverting configuration. 

9.4.1 Wideband Noninverting Operation The THS3491 is a 900-MHz current-feedback operational amplifier that is designed to operate from a power supply of ±7 V to ±16 V. (THS3491 - SBOS875C – AUGUST 2017 – REVISED FEBRUARY 2023 Page 27) - 

So +16V and -16V on the power supply ranges. 

We have a dual channel. So, we have another THS3491 that is not being used (no input injected) and that one is not being burned up. So, I do not think the voltage rails are being violated there since the only one that gets burned up is the one receiving some kind of input. 

I am thinking perhaps the load at the output might contribute to some unforeseen oscillation that our standard testing does not see. 

Hi Martin,

That is good to know it might not be the supplies if the same supplies are powering on the unused THS3491. I know you have mentioned an antenna type load, is there any type of isolation resistance in your circuit to help isolate your complex load.

Best Regards,

Ignacio

Hello Ignacio,

A selectable 50/75 Ohm in parallel with a 1000uF/16V Capacitor are in line with the output of the amplifier that connect directly to the complex load. Note: Input and Output traces that connect to the THS3491 are not burned up. 

Hi Martin, 

Would it be possible to get the schematic for the circuit in a private message? I am not sure what else we can try without digging into the schematic to see if anything sticks out. 

Best Regards,

Ignacio

Hi Ignacio, 

I got the green light to share via private message. 

Hi Ignacio, 

If it is okay with you. I would like to leave this opened in case someone else is interested or experiences the same thing. If we can't figure anything out in our private messages, then we can close this thread.