OPA549: OPA549 Output Short to GND

Hi Bing,

what is mounted, R8 or R9?

Now to your question: Inductive kick-back can destroy the output of OPA549. So if you plan to connect an inductive load to the output of OPA544, please add the protection diodes shown in figure 8 of datasheet

Kai

Oh, just noticed that your snubber contains 5R and not 10R as recommended in the datasheet...

Kai

Hi Bing,

As Kai pointed out, what types of load do you have at the OAP549's output? Per your description, the OPA549's output stage seems to be damaged. 

I have some questions:

a. I did a short on output, which made the output short to GND directly and then disconnected the output immediately. 

      How long did the OPA549 goes into shutdown mode after output is shorted. Short circuit should result OPA549's internal heating up, and thermal detection circuit should be activated when the junction temperature of OPA549 is reach its threshold. 

b. Your snubber circuit is 5Ω and 10nf. From this, I assumed that you have an inductive or complex load. The cutoff frequency of 5Ω and 10nf is approx. 3.2MHz. With 10Ω and 10nf, the cut off frequency will be 1.6MHz. Do you know the inductive kick back voltage pulse looks like at the output? It is also a function of how fast you Open/short the output of inductive load in Ldi/dt = Vavg. We can estimate R and C, if you know the pulse width and Voltage spike. 

c. It is not conclusive how the part may be damaged. Please provide us more information.

Enclosed in the Safe Operating Area (SOA) plot. OPA549 should be able to handle 10A short and activate the thermal shutdown circuit (you also described the shutdown sequence). Per the SOA plot, 8A is recommended short circuit current, 10A short is pulse current limit for the part. 

It is possible that the part is damaged via inductive kick back at the output.  Please let us know the inductive load, and see if we are able to simulate it a transient event. 

OPA549 E2E SC 07202021.TSC

Best,

Raymond

Hi Raymond,

The image above shows the setup. Once OPA549 output is connected to the input of DCDC EVM, then DCDC starts to work.

The output of OPA549 is a 40V square waveform with 50Hz. After DCDC EVM was connected to OPA549, I made the output short to GND, and then I saw a spark and after that I disconnected OPA549 output from GND immediately and disconnect it from DCDC EVM.

After the operation, when I reconnect the output to DCDC EVM, I cannot get the right square waveform (has a bias, and not very "square") if I probe the output of OPA549. And if I wait for few seconds, the waveform becomes a DC waveform.

However, if I don't connect the output to DCDC EVM, the square waveform of OPA549 output is very good as expected.

During all tests, DCDC EVM is with no load. Thus I think there should be no current will be pull out from OPA549. And after the "short" operation, DCDC EVM still works well if I test it with normal condition. To exclude the impact of DCDC EVM, I used a new same type DCDC EVM to test OPA549 board, the results still the same.

As above description, if it is possible that OPA549 would be destroyed from my "short" operation? 

Or does OPA549 have a PSpice model? If there is a PSpice model for it, I can use it to simulate from my side.

Thanks.

Bing

Well, Bing, the spark is just the prove of having an inductive kick-back 

You urgently need the protection diodes, as I mentioned earlier.

Another issue is the DC/DC modul. These moduls are designed to have a decoupling cap at the input for providing a defined source impedance. It's no good idea to operate them without this cap.

And why are you torturing the DC/DC modul with a 50Hz 40V square wave at the input? 

Kai

Hi Bing,

Adding to Kai's comments. Can you tell us what application the circuit is for? I have additional questions about the circuit.

1. If you are going to use OPA549 as voltage regulator, you do not need DC/DC module part. You can drive a load directly up to 8A continuously. You can use OPA549 power amplifier as regulator, though the power efficiency will be poor. 

2. With  OPA549 + DC/DC module, the power efficiency is determined by OPA549 (lowest one), which is approx. 15-30% depending on the power configuration, assume DC/DC module is 90-98% efficient. If OPA549 has power efficiency at 15% and DC/DC module has the power efficiency, say 90%, the overall power efficiency will be 0.15*0.9*100 = 13.5% (here I assumed 90% in DCDC module as an example. You really need to check the specification of the power module, which is likely a lot lower under certain operational states).  

3. The DC/DC module is designed to operate at high input switching frequency, PWM switching frequency is likely >50kHz, more likely in 100kHz-200kHz or even higher. The DC/DC converter is to converter PWM switching signal to DC regulated voltage. 

At the output stage of DC/DC converter, there is a LC filter, its resonance frequency or poles, 1/(2*pi*sqrt(LC)), needs to be 1/10th to 1/20th of PWM switching frequency. This is to filter out the PWM harmonic frequencies. Also, there is Q at the output of LC filter, which you need to deal with vs. DC/DC module's load.

With 50Hz PWM frequency, that switching PWM harmonic noises cannot be filtered out via the LC filter, which it rolls off at a rate of -40dB/decade.  

Please let us know the type of application you are working with. We can assist you to resolve possible issues. 

Best,

Raymond

Thanks Raymond, Kai

Sorry maybe it confuse you. OPA549 is used to help test the input line transient for the DC/DC converter. By adjusting the output of signal generator, I can get different square waveform voltage rise rate on the output of OPA549, then I can use this output to test the line transient feature of the DC/DC. To test the line transient feature repeatedly, I set the square frequency at 50Hz as I described.

Not sure whether this usage is OK for OPA549?

Thanks.

Bing

Hi Bing,

why so complicated? I would do this quite old school with a simple hand switch

Select R1 and C1 to set a defined rise time. (R1 must not be set too high, of course, since it must pass the DC/DC converter current.) Set the scope to single trigger.

Having C1 allows the DC/DC concerter to operate under standard condition. I already told, that the input cap is part of the DC/DC circuit and must not be omitted. You want to test the DC/DC converter like it's used later in the application? Then it will also have the cap at the input. Testing the DC/DC converter without the input cap will give wrong results anyway.

You might want to add a big resistor in parallel to C1 to allow C1 to be fully discharged before the next edge is arriving.

Kai

Thanks, Raymond and Kai. I will take it as references.

Thanks!

Bing