LNA spf5189z issue

[radiolistener]

I just got LNA module from aliexpress, it is based on spf5189z chip.

It worked great, but not for long. It is died in several hours. Now it has short circuit on the output pin. And I don't understand the reason for that.

I powered it from lab power supply with 5.01 V with 0.2 A current limit settings.
It was connected to antenna through decoupling capacitor and over-voltage protection diodes (1N4148).
Output was connected to RTLSDRv3.
All equipment (PSU and PC) is grounded, include ground contact of the antenna.

I connected antenna output to BM867 in CREST mode, just to make sure that there is no high voltage pulses. It worked for a several hours and largest catched pulse is just 0.479 V.

See picture of PCB in attachment. Also I added schematic that I reversed from PCB.
Before use I replaced R1 (jumper 0 ohm) with ferrite bead (BLM18EG601 - 600Ω @ 100 MHz / 0.35Ω @ DC).

It worked well for several tests and it's current consumption was 0.07 A.
But after next power on, I noticed that current is jumping up to 200 mA (limited by power supply).
And that's it, after that it doesn't works

I also found this article related to the subject: http://pa0o-jaap.blogspot.com/2018/04/spf5189z-lna.html
But it's not clear what is the reason for burnout and how to avoid it?

Any idea why it happens?

[radiolistener]

I have idea that it is burned out just because it may start self-oscillating, when input is open and it leads to high current consumption. Since there is missing current limiting resistor, it consumed too high current and is burned out. It seems that my power supply limit 200 mA is too high for spf5189z.

Does this theory possible? How to protect it from such conditions?
I'm thinking about add 30-50 Ohm resistor for power line. Would it work?

[biastee]

I would not place oscillation at the top of a list of suspects because:
1. if an amp oscillates, the output is ~P1dB; i.e. below the abs max rating.
2. MMIC amps have a long history of being pressed into service as oscillators without problem.

Another example of MMIC amp used as oscillators
Did u switch off the power supply and then switch it on again? Turn on transients are known to be very damaging.
see pg 3 of this

[radiolistener]

Did u switch off the power supply and then switch it on again? Turn on transients are known to be very damaging.
see pg 3 of this

yes, it happens when I switch off and switch on several times in a row.
It was not power switch, I just touched wire with alligator clip connected to lab power supply.
And I used long wires (about 1.5 meters) to provide power for the module.

How to fix it? If I add 68 uF capacitor on amplifier PCB (after ferrite bead), does it helps to solve it?

What is the safe schematic for power supply filter on LNA module? (for spf5189z).

Originally I got it with just two capacitors and 0-ohm jumpers instead of resistors.
Before damage, I replaced R1 with ferrite bead (BLM18EG601 - 600Ω @ 100 MHz / 0.35Ω @ DC).



This LNA module has very nice aluminium shield. But it seems that it needs for some modification on power line.

What modification can you advise me to improve protection for power on/power off conditions?

[biastee]

1. Ya, adding the uF-range cap should help.
2. Twist together the 1.5m long wires to reduce the parasitic inductance

[Alexei.Polkhanov]

Any change you powered it up in reverse by accident? If it takes 200mA then this current is going somewhere so something must be hot. I would just replace SPF5189 by a new one - they are $6 from Digikey and places alike.

Did you measure it with 50ohm terminator(s) screwed on to the input and output?

L1 is meant to block RF from coming back but also should block some of the transient currents on turn on.

[radiolistener]

Any change you powered it up in reverse by accident?

no, the ground wire was always connected during series of measurements. I don't think that random reverse polarity was possible in such configuration.

But I connected power supply with wire touch, so there is possible series of short pulses on power line at connection time.

If it takes 200mA then this current is going somewhere so something must be hot.

No, there is nothing hot, may be because the chip has good thermal contact with aluminium shield. The voltage dropped down to < 1 V due to OC stabilization.

I tested it with multimeter, it shows 17.02 Ohm on the spf5189 input pin and 6.29 Ohm output pin. When it worked well it was about 1 MOhm on both.

So, it seems that short circuit happens at the output pin of spf5189.

Did you measure it with 50ohm terminator(s) screwed on to the input and output?

The input connector was opened when it happens. The output connector was connected to RTLSDRv3. But I'm not sure that it happens at power on. Before that input was connected to antenna.

Previously I applied power when input and output was open, power consumption was 0.07 A and it worked well after that. I tried it several times and after that it worked well.

L1 is meant to block RF from coming back but also should block some of the transient currents on turn on.

I suspect that L1 may lead to overvoltage when power on condition is not clean and has series of short pulses. Or may be static electricity applied on power pin of the module. Is it possible?

Here is my gain measurements for different power supply voltages - 0V, 3.21V and 5.02V which I performed before incident (input level is -40 dBm):

[tombi]

I had one of these from Banggood but it wasn’t in a nice block but just had a little can over the chip.

It died pretty fast too and I bought some cheap replacement MMICs from China.

My theory was that I was accidentally shorting the power alligator clip to the can and it was earthing some other way (and thus reverse biasing it). Soldering leads to the PCB seemed to prevent further deaths of amplifiers but to be honest I stopped playing with it shortly after.

Tom


Sent from my iPhone using Tapatalk

[Bud]

Try powering the amp with 9v via a 50 Ohm current limiting resistor. You can use two 100 Ohm resistors soldered instead of R2 and R3 in parallel. Use 0805 or better 1206 resistors if they fit. This way your amp will see 50 Ohm load at low frequencies, as oppose to being RF  short circuited. Pads for  R2 and  R3 were placed there  for that type of current limiting  in the first place.

[radiolistener]

Try powering the amp with 9v via a 50 Ohm current limiting resistor.

spf5189z has the following Absolute Maximum Rating according to datasheet:
Max Device Current: 120 mA
Max Device Voltage: 5.5 V

So, I don't think that powering it from 9V is a good idea 

Currently I don't have replacement, waiting for a shipment...

[Bud]

Per the device datasheet, the current through the device is 90mA @5V. With a 50 Ohm series resistor the voltage drop across the resistor will be 0.09*50=4.5V. With a 9V power supply the voltage across the device will be 9-4.5=4.5v

You measured 70mA current @5V , which is by the way already outside of the  datasheet spec, which may indicate you got a fake. Anyways, lets say you have 70mA, then the voltage drop across the 50 OHM series resistor will be 0.07*50=3.5V. With a 9V power supply the voltage across the device will be 9-3.5=5.5V which is within the device spec per your note above.

[Bud]

If you have a concern on overvoltaging the device, do as i said but power the circuit from a variable power supply and monitor the DC voltage across the amplifier. Increase the power supply voltage until you reach 5V across the amp, and you will see the supply voltage will be around 9V.

[radiolistener]

I think there is no guarantee that spf5189 resistance will be constant at power on transition. So, at 9V and 50 Ohm limiting resistor it may apply up to 9/50 = 180 mA current. It exceeds absolute maximum rating, which is max 120 mA.

But idea to use current limiting resistor is good. Since it works with almost the same gain from 3.2 V, a small resistor should not make big difference, but may protect MMIC from over-current.

[biastee]

As long as you avoid connecting the power supply with a wire touch, the problem should not recur. Making the same connection via the power supply's ON/OFF switch is safe because the voltage will ramp up gradually on a good lab supply. However, if you really want make the LNA bulletproof, then u will need to install a transient-voltage-suppression (TVS) diode (transil) close to the device.

[radiolistener]

Can I replace TVS diode with some small 5 V linear regulator installed within LNA shield?

For example - L78L05, it has 100 mA limit, probably it also may be used as current limiter, is it correct?

[biastee]

Yes

[radiolistener]

Some update. I got these spf5189z from aliexpress: https://www.aliexpress.com/item/33026418359.html

They come in tape reel package from cheap and dirty plastic which has noticeable quality difference from original tape reel package components. Also it has weak marking which is different from original one.

I replaced old burned spf5189 with a new one. But it turns that this is fake 

It works, but it's power consumption is 145 mA and gain below 100 MHz is 10 dB lower than original spf5189. It heating like hell, so the aluminium shell is pretty hot while powered. Also it has different frequency response than original one, fake spf5189z has a gap with gain 0 dB at 13.5 MHz (see picture). Original one don't have such gap on a frequency response.

I left it powered on for a night (for test) and at morning I found that it also burned out. I'm not sure if it was burned due to high current or some high signal from antenna (antenna input has DC block capacitor and voltage limiter with two 1N4148). But anyway this fake much worse than original one and has high current consumption which is out of specification.

I also found that fake spf5189 has about 3 kOhm on the input pin (I tested it with Brymen BM867 before soldering). As I remember, original one has about 1 MOhm (I tested it while it was ok). But I'm not sure, since original is burned out, so I cannot measure it now.

When burned out, fake spf5189z has about 2 Ohms on the input pin, the same as original.

If someone have original spf5189z, it will be interesting to know it's input resistance (measured for DC).

[radiolistener]

I want to modify original circuit for bias tee.

What is the best way to modify this circuit to provide power through bias tee?
I cannot get a new PCB, so I need to reuse this one:

[biastee]

That depends on whether you want to supply it with 5V as originally, or you want to use a higher voltage such as 12V. In the 1st situation, just replace C4 with a 0R jumper. In the second situation, solder a chip resistor on top of C4 to provide the required voltage drop.

[radiolistener]

biastee, thanks, looks pretty easy

But... Is it safe for MMIC to power it directly from long coax line?

[biastee]

i m not aware of any problem 

[desolatordan]

But I connected power supply with wire touch, so there is possible series of short pulses on power line at connection time.

If you are powering up the power supply, THEN connecting power to the amplifier, you are likely causing an over-voltage condition that's killing it.  The input inductance (including any length of coax) will resonate with input capacitance.

[radiolistener]

I suspect that the root of cause why it was burned out, is not a power. Probably it's just Chinese quality or strong signal from antenna. I tried to power on/off another module a lot of times and it still works ok.

I'm using two inverse-parallel 1N4148 diodes for over-voltage protection on the input of LNA module. But may be these 1N4148 are not fast enough.


Regarding to the bias-tee power mode, I still not sure if direct power from coax line is safe. I didn't tested it yet, I bought bias-tee module, but still waiting for it...

[OH2LIY]

Hi, I recently buyed these: https://www.ebay.com/itm/254599063723?ViewItem=&item=254599063723 .

From the beginning, I was sceptical what I get. Easiest thing is measure voltage/current. Datasheet says ~90mA@5V. These crappy chinese mmic what I got, takes over 150mA@3V (limited by power supply current protection, 3V measured from mmic output).
Real thing is measure noise figure/gain, and it`s way over what it should be.

I can say directly that these are Crappy Fake Relabelled  MMICs, and ebay refunded instantly after opening counterfeit case.

[radiolistener]

I can say directly that these are Crappy Fake Relabelled  MMICs, and ebay refunded instantly after opening counterfeit case.

My one is the same 150 mA and the same marking. I think this is remarked other type of MMIC.
May be TQP3M9009 or something like that.

[HighPrecision]

Hi radiolistener,
see this document from MiniCircuits:
https://www.minicircuits.com/app/AN60-034.pdf

This maybe help on your problem of burning SPF5189Z, in paragraph 4 there are some suggestions to mitigate this issue.

MiniCircuits PGA-103+ is declared as a direct replacement of SPF5189Z in datasheet.

Beware that these devices are NOT UNconditionally stable below 100MHz, maybe a good idea to place a filter for FM broadcasters and other signals in <100MHz band. This issue is described here:
https://www.minicircuits.com/app/AN60-064.pdf

[Max.Bz]

Also got a large number of fake LNA SPF5189Z (visually easy to distinguish by the shape of GND pin in the original it is shorter and skewed) in terms of parameters, this is far from LNA, but nevertheless they do not fail , since they had to use them in an amount of several thousand pieces, gain was 13-15dB, @~1GHz

[virtualparticles]

You normally cannot hang an LNA directly on an antenna. Various noisy bits in the air will eventually conspire to overload the input of the device. Put some sort of bandpass filter in front of it. I speak from 15 years experience designing Tower Top Amplifiers for Land Mobile Radio. I definitely know how to burn out LNA's. 

Best,

Brian

[cdev]

What do you think about built in ESD protection?

Some MMICs  ( like Mini Circuits PSA4-5043+) build in a moderate level of ESD protection.
But it seems that even with the manufacturers basic ESD protection, they often get taken out by ESD from sources like lightning.

An ungrounded conductive path (like a quarter wave monopole) going straight into the LNA, even if you have a DC blocking capacitor on the input (which can be awfully small) seems like its asking for trouble..

Even so, unless one has both lower and higher out of band frequencies attenuated in many tower settings especially, unless one if far away from other RF sources strong out of band signals seem likely to "steal dynamic range" from the desired application.

So a band pass filter and DC block as well as a DC path to ground seem like common sense, maybe almost mandatory?

[radiolistener]

What do you think about built in ESD protection?

I'm using two antiparallel diodes and 47 nF capacitor for DC block on the input

[A.Z.]

What do you think about built in ESD protection?

I'm using two antiparallel diodes and 47 nF capacitor for DC block on the input

two may suffice, but may also cause signal clipping

http://www.kk5jy.net/rf-clipper/

[cdev]

Mini circuits PSA4-5043+ has quite decent ESD protection (ive never lost one from ESD) and its just a three terminal device so its as simple as it gets for an LNA. Basically, you can make an LNA out of a little chunk of 50 ohm trace in just a couple of minutes. They cost around $2.75 each.

They are probably better than these Chinese bootleg spf5189 LNAs. I have both so if you want me to compare the two I'd be happy to.

[cdev]

WA5VJB also makes a small LNA PCB well suited for a mini circuits mmic or bias tee and sells them for very little. Maybe $1. Its an easy, half hour LNA with generic pads which would support all kinds of parts..

[genugeo85]

In order to limit the current to this mmic it is necessary to introduce a negative voltage of approximately -0.6--1V on the circuit input (similar to the supply of GaAs and HEMT transistors). The current will be reduced to 50-90mA.

[Gerhard_dk4xp]

No, it is not. Read the data sheet, it says clearly "single supply".
The chip is end of life, Mouser has no stock, nothing new to arrive.

Mini Circuits PGA103 works nicely, even less noise.

[radiolistener]

I also hear that adding bias offset on the input can help to improve performance for this fake/counterfeit Chinese chip. But didn't tried it

[genugeo85]

What the Chinese say in the datasheet is one thing, the reality is another ...
I also burned a few pieces until I injected a voltage of -0.8V at the input. At that time I measured a consumption of about 60mA.
I speak from my own experience ...

[genugeo85]

Some photos with the behavior of my modified amplifier ...

[cdev]

Some LNA  MMICs have built in ESD protection. I have used it in quite a few PSA4-5043+ devices. (just that device, no others)

At least ten were successful and the failures were not caused by ESD. On one I accidentally pulled a leg off.. (oversoldering)  . Oh, and I lost one in the messy area beneath my desk. Because they are very very small.

But none have had to sit up on a tower directly connected to a monopole..A high static risk situation. I bet a high value resistor to bleed off ESD to ground or any antenna with a DC path to ground would protect it.

Datasheet for  PSA4-5043+ attached.

Mini Circuits claims
*Class 1B HBM ESD rating (500V)

It seems like it would make a lot of sense to buy 20 and use it to retrofit each of your various cases with a simple, new LNA.

[m3vuv]

I have just got one of these lna moduals from aliexpress for my si 47xx homebrew radio,ive a couple of questions,1 are these ok for hf ,and 2 will these run from an 18650 cell?,are they ok for 2 meters as well?,ok 3 questions lol. TIA.

[cdev]

I have used two metods to stop oscillation.

1, inserting lossy elements in parallel with inductors in the power chain, for example, adding low value resistors to reduce the Q of the bias tee inductors,

2.) using flexible RF loss antistatic foam, you can attach it to copper tape (on its back) and then put a thin piece of cellophane wrap right on top of the circuit and press the antistatic form over that. Apply some pressure right on the device. This will introduce loss.

3.) make sure that both the input and output to the LNA go through a short length of decent 50 ohm transmission line, ideally where the shielding is well grounded. Ferrites around these two short lengths of transmission line may help too, all my gain stages are festooned with as many ferrites as I can reasonably fit on them. This may look funny but my experience is that it pays off in signal integrity.

4) Make sure the grounding on the MMC is top quality. It should be immediately under the MMIC, for example, place your vias right at its legs. Make sure the impedance of the vias is low. Grounding is the most important thing with newer 50 ohm LNA devices. They are super easy to use if you give them what they want.

[m3vuv]

I have used two metods to stop oscillation.

1, inserting lossy elements in parallel with inductors in the power chain, for example, adding low value resistors to reduce the Q of the bias tee inductors,

2.) using flexible RF loss antistatic foam, you can attach it to copper tape (on its back) and then put a thin piece of cellophane wrap right on top of the circuit and press the antistatic form over that. Apply some pressure right on the device. This will introduce loss.

3.) make sure that both the input and output to the LNA go through a short length of decent 50 ohm transmission line, ideally where the shielding is well grounded. Ferrites around these two short lengths of transmission line may help too, all my gain stages are festooned with as many ferrites as I can reasonably fit on them. This may look funny but my experience is that it pays off in signal integrity.

4) Make sure the grounding on the MMC is top quality. It should be immediately under the MMIC, for example, place your vias right at its legs. Make sure the impedance of the vias is low. Grounding is the most important thing with newer 50 ohm LNA devices. They are super easy to use if you give them what they want.

well looks like my question is unanswered!

[cdev]

The simple example of the Chnese copycat LNA that I have is still working, but my guess is that the noise figure is higher than the original. (which is specced at less than 1 db, [pretty good) Maybe 4 db or more is what it seems like to be just judged on how it performs with local sigansl here in the NYC suburbs. I can receie lots of signals down the Jersey SHore and even signals from NASA at Wallops Island.. (they lauch lots of satellites from there. So I find the copycat LNA  to be crap, it doesnt really make sense to use it say with an RTLSR which already has around a 4 dbm noise figure. The really weak signals from south of here and along the Long Island Sound, do come in better with the minicircuits LNA.

So whats the point?  Transmit, I dont know. I only receive. An 18650 fully charged might give you a day or two of enhanced reception. As stated, its not such a great LNA. Also, as radiolistener has pointed out they tend to die from ESD. So arent so great.

[cdev]

Not so great for below 50 MHz. They dont specify because the OEM specs end at 50 MHz. So they dont have to.

Lots of people find they die from ESD fairly young. My LNA4all has, in contast lasted me now more than eight years, getting used a lot in all that time..

But I have never used it outdoors.. Its never had to deal with ESD from an outdoor monopole.. Just lots of planar disk antennas, discones. etc.

The easy indoor life.

[Costas07]

Hi,
Bought one of these Chinese LNA supposedly based on SPF5189. It consumes ~ 160mA from 1 Li-Ion cell 18650 size (4.2V fully charged).
Based on the specs this is too much. Also the gain is lower and barely gets to 15dB contrary to datasheet. To me it seems these are not original MMICs.

[spider_provider]

Hey all,

I think there's two variations available on eBay etc: on the shield, the SPF5189z label is either in black text, or black outlined text.

From my experiments, the ones with outlined text want 120-160mA, and the one with solid black text wants 60mA. (Voltage ~5V, tested 3x outline and 1x solid variants)

So I'm guessing the solid black text ones are real. Only going off current, they both work the same to me, I don't have any spectrotron professor farnsworth stuff like you guys

Edit: Adding side-by-side. That's a pair of antiparallel 1N1418's under the heatshrink, and no I am not sorry about my soldering job.

[biastee]

When a device oscillates, its current will typically increase above normal. So, it is worthwhile to check for oscillation whenever the current is abnormal.

But none have had to sit up on a tower directly connected to a monopole..A high static risk situation. I bet a high value resistor to bleed off ESD to ground or any antenna with a DC path to ground would protect it.

Why suspect static only and NOT consider indirect lightning strikes? Another way the designer can protect the LNA is to choose a high pass L-network as the input match. This network has a shunt L that can bleed static & transients.