The Power supply of Bias Tee damages the LNA

[mhsj]

Hi, I'm a beginner to RF. Now I'm meeting a pratical problem which I think it needs RF knowledge.
That I have a weak signal, and change quickly, so I use a Bias Tee to capture it and give it -9V. Then I send it to a LNA(+60dB). When I choose different power supply to Bias Tee, the LNA may break down.
I test ten times, and I lost 9 LNAs.
I find that when I use RIGOL DC Power Supply or Battery, the LNA goes well. Then I use other power a linear power adapter, The LNA Damaged, there is no Amplified signals.(It used to work fine, but it take away my 9LNAs)
Then I use Chassis power which gives other subsystems, the Spectrum of signal is too broad(noise).
My LNA is Two-stage amplification.
What's the problem with my LNA? I Unpacking LNA, it use BGA2869, my signal is 6MHz.
Just I think that the power supply of previous Bias Tee can not influent the latter LNA, I tried use a SMA DC Block, but it's useless.

 

[BigBoss]

There may be many reasons that blow your IC but I guess there was a Transient Voltage that has been killing Input/Output. ESD can be another reason. Output Level might be too high etc.
Don't use 2 LNA in Cascade to obtain this extreme Power Gain. 60 dB Power Gain can damage the IC when a captured anything/any signal comes to them from somewhere.
These can be Transient Power Supply spikes, ESD Disturbance , Reflections, anything.

 

[vfone]

According to the datasheet the maximum input power (before damage) of BGA2869 is +10dBm.
At 32dB gain, if you cascade two BGA2869, the maximum input level of the first stage should be less than -22dBm, otherwise the second LNA will be damaged.

BGA2869 require DC blocking capacitors at the input and at the output, and it has the supply voltage pin separated by the RF input and RF output pins. In this situation, I don't understand why you use a bias-tee.
Bias-tees are used on the RF ports when one of them (or both) require DC feed. But BGA2869 use separate pin for DC feed.

Post the schematic you use, and may get more information.

 

[volker@muehlhaus]

I use a Bias Tee to capture it and give it -9V.

(...)

I Unpacking LNA, it use BGA2869

This amp requires positive supply up to 7V, so I'm surprised that you feed -9V? Why that negative supply?
What does your amp data sheet specify regarding voltage, polarity and DC coupling?

 

[dick_freebird]

Many, but not all, small RF parts want blocking caps to
prevent DC bias from jacking the signal or damaging the
part (or whoever's attached). A supply apiece certainly
increases the opportunity for Bad Things.

Since RF usually doesn't care for or want DC on the signal
path, maybe step back and DC-block each node.

Now, bias-T blocks often are asymmetric and the DC
may be blocked from one signal port and not the
other. In fact that's kinda the whole point of having one.

So, be sure you understand the internal connections
of the particular piece in hand, and what such a piece
should be expected to impose if (say) you swapped
the goesinta and the goesouta. Pop goes the weasel?

 

[mhsj]

There may be many reasons that blow your IC but I guess there was a Transient Voltage that has been killing Input/Output. ESD can be another reason. Output Level might be too high etc.
Don't use 2 LNA in Cascade to obtain this extreme Power Gain. 60 dB Power Gain can damage the IC when a captured anything/any signal comes to them from somewhere.
These can be Transient Power Supply spikes, ESD Disturbance , Reflections, anything.

I may be wrong, I unpacked the LNA and found a similar chip to the one pictured

https://twitter.com/x/status/1407266420929581058

. So I think it's the same.
I only used 1 LNA and the internal structure is two-stage amplification.
I tested 1 LNA 10 times, lost 9 LNA.
I also checked the power supply noise with an oscilloscope and found nothing. It is a very small ripple noise.

 

[mhsj]

(...)


This amp requires positive supply up to 7V, so I'm surprised that you feed -9V? Why that negative supply?
What does your amp data sheet specify regarding voltage, polarity and DC coupling?

My weak signal is curruent from a PD. Bias Tee can change it to a voltage signal. Then I could use LNA.
When I choose a different negative bias, it gives a respond to 6MHz signal.

 

[mhsj]

Many, but not all, small RF parts want blocking caps to
prevent DC bias from jacking the signal or damaging the
part (or whoever's attached). A supply apiece certainly
increases the opportunity for Bad Things.

Since RF usually doesn't care for or want DC on the signal
path, maybe step back and DC-block each node.

Now, bias-T blocks often are asymmetric and the DC
may be blocked from one signal port and not the
other. In fact that's kinda the whole point of having one.

So, be sure you understand the internal connections
of the particular piece in hand, and what such a piece
should be expected to impose if (say) you swapped
the goesinta and the goesouta. Pop goes the weasel?

I'm not really understand the internal of LNA structure, but I'm sure the external connection is correct(except DC Block), so I use it at the input and output of LNA. It goes well with normally.

 

[mhsj]

What can I do for these structure, may be a diodes in Vcc?

 

[mhsj]

Perhaps I was vague, the LNA use 12V, Bias Tee use -9V. And I use oscilloscope to check 12V or -9V, when I poweing up, I find anything about noise voltage or pulse. After these testing, I lost 9 LNA. And all power supply has ±10mV of ripple noise.
Do I need additional diodes in my LNA for Vcc?

 

[BigBoss]

But it's difficult to catch a short spike or something on VCC line. Why do you use -9V and Bias Tee ?

 

[mhsj]

But it's difficult to catch a short spike or something on VCC line. Why do you use -9V and Bias Tee ?

I need a PD to detect a weak light signal, it varies so fast, and this PD needs a offset voltage to respond. Anyway, this -9V is a offset for PD. When this weak signal pass bias tee, it becomes a voltage signal. Then I can amplifier it.

 

[FvM]

Can it be that either connecting the -9V power supply or PD operation causes a negative input transient that damages the LNA? What's the input coupling capacitor value?

Your circuit may require input overvoltage clamping.

 

[dick_freebird]

I know of RF CMOS/SOI mixers that had a 10V HBM ESD failure threshold. The RF people don't like baggage on the input.

So hot-plugging to 9V through a similar value blocking cap (or bare) might well punch out a gate if it's a low voltage CMOS platform under the product.

 

[mhsj]

Can it be that either connecting the -9V power supply or PD operation causes a negative input transient that damages the LNA? What's the input coupling capacitor value?

Your circuit may require input overvoltage clamping.

I'm not sure about input coupling capacitor, I bought it from internet, but a add a aditional diodes at RF in, it works, but I still find any pulse from RF in with oscilloscope

--- Updated Jun 8, 2022 ---

I know of RF CMOS/SOI mixers that had a 10V HBM ESD failure threshold. The RF people don't like baggage on the input.

So hot-plugging to 9V through a similar value blocking cap (or bare) might well punch out a gate if it's a low voltage CMOS platform under the product.

Before I cause short circuit with -9V, hot-plugging is not a problem. And I found that no damage was done with my system, even I replace a new -9V(3 times), the problem out. Maybe a shor circuit change the power supply environment?