Fixed Attenuator Repair / Mods

[joeqsmith]

Just another "working" eBay attenuator.    Note the crater.    Thought I would have a look.    I assume these have BeO but was able to just warm the encapsulation with an iron set to fairly low temperature.   The series section was good but the shunt had failed open basically right in the center of it. 

The edges look like a total hack job.  Not at all what I was expecting. 

If you do ever have the chance to look at one, just keep in mind that BeO would have been fairly common.  Don't go in there with dremel and start grinding away at it.

I plan to rebuilt it using some better parts with different value that will make it useful around the lab. 

[Tony_G]

Any ideas on what caused the crater? Overpower event maybe?

TonyG

[SilverSolder]

Technically, the attenuator does work -  It just attenuates rather a lot more than expected! 

[HB9EVI]

made similar experience:

so called '100W' 40dB (3GHz) Attenuator got a crater from ~50W @ 432MHz, no idea why that happened; still waits for repair; indeed it still works - either as simple dummy or as Att with much, much higher attenuation

a 10W 30dB Att I overloaded I got fixed with bigger SMD resistors

[radiolistener]

I also have some experience with Chinese attenuators.
30W 40 dB burned out from 15 W power.
10W 40 dB burned out from 5 W power.

Both burned out during test, just out of the box.

Now I'm not sure if it worth to buy 100W version


On the other hand, several years ago, I bought Chinese 10W dummy load:
https://www.aliexpress.com/item/32771021842.html

It has 49.87 Ohm, good VSWR and pretty flat response. Sometimes I'm using it to terminate 15W power for 10-15 sec. It still alive and it has very nice good quality SMA connector with air insulator, it is very different from cheap "golden plated" Chinese SMA.

[George Edmonds]

But you are not using Chinese Watts, they are just the same as their Amp/Hour capacities of batteries.

George G6HIG

[joeqsmith]

Any ideas on what caused the crater? Overpower event maybe?

TonyG

The crater was due to the excessive heat/pressure when the shunt was damaged.  It was open and I would think there was some arcing going on.  I assume the PO over powered it.     

Technically, the attenuator does work -  It just attenuates rather a lot more than expected! 

It certainly does "eBay work".   Beause the shunt failed open, it attenuates much less than expected. The response is not very flat with that custom made stub hanging off it.   

Trying to find a source for flange mount resistors without a year lead time is the challenge. 

I've bought parts from this company in the past.  They have a few in stock but nothing I can use.
https://www.surplussales.com/RF/RFAtten-6.html

[Grandchuck]

But you are not using Chinese Watts, they are just the same as their Amp/Hour capacities of batteries.

George G6HIG

Good one George 

[joeqsmith]

One of the connectors was damage and required replacement.  Rather than just replace the part, I decided to remove the boards, sand the heatsink and paint it.  I used some of the high temp paints I use with the bike engine parts. 

The new element is rated for a much higher frequency than the attenuator originally was.  I'll make a label for it, once I have an idea how well it performs. 

So, I basically bought a decent heatsink with spreader, a couple of Teflon boards and a connector.

[joeqsmith]

Some time ago I built this 50 ohm load.  Basically, it's a 250W flange mount 50 ohm resistor followed with a 4.95K in series with the SMA (40dB).  A bit of magnet wire to compensate it.   It was made to test a current probe but I use it with the SA and VNAs to look at small amplifiers.  The problem is the BW is limited to around 100MHz (limited by the tap) and the power it can dissipate is fairly low.   

Because the JFW's original network's shunt burned open, the plan is to modify it to provide about a Watt.  A second attenuator is added to get down to the working range of the VNA.   The new network provides about 50% derate. 

[joeqsmith]

The new parts arrived.   Used DC to check the heatsink.   I may end up mounting a couple of fans to it. 

It's fairly flat out to a GHz where the JFW was originally spec'ed.   The new part has a +/-1dB flatness spec or to 3GHz.  It's not a perfect mechanical fit and the launch isn't quite right.  Very usable now.

[SilverSolder]

Impressive!  - is there any contribution from the test gear in those charts, or is it really possible to measure RF this precisely?

[joeqsmith]

Impressive!  - is there any contribution from the test gear in those charts, or is it really possible to measure RF this precisely?

Hum...  Well, that's a very good question.  Yes, the equipment will have some contribution.  This includes the standards, connectors, cables, VNA.   Some error would be from the setup.   Contrary to the belief that hand tightening connectors is good enough to 6HGz, I don't subscribe to that.  However, the torque wrench would introduce some error.   Then there is the untrained, unexperienced  human (me).   

If you feel that the data is impressive, you should have a look at a post from member rf-messkopf.  Follow their link to see a metrology grade 20dB attenuator.     
https://www.eevblog.com/forum/projects/20db-rf-attenuator-seeking-feedback-to-improve/msg2964070/#msg2964070

I don't have anything like that in my home hobby lab to try and verify the performance of my setup.  I'm using home made  SO standards and a cheap Mini-Circuits load.  If you read through that thread,  you will find I was getting some help from rf-messkopf.       

Yes, there's lot's of room for errors and the higher the frequency, the more critical everything gets.       

[SilverSolder]

LOL now that's what I call a flat response, on that 20dB attenuator! 

Even at 400Hz, I can detect sub-pF capacitance effects when playing around with bridge circuits, for example.

I can barely imagine how controlled the "electrons' environment" has to be to contain the reaction at GHz frequencies... 

[joeqsmith]

Within 0.03dBm to 4GHz, you know they weren't handing those out for free.  For low frequency work (<GHz), I have a set of home made standards to get me in the ballpark.  I wouldn't mind having a few high end standards to use above a GHz.  I tried to make a T-check but it was a total waste of time.  Similar to trying to make a load.   

[Mechatrommer]

Contrary to the belief that hand tightening connectors is good enough to 6HGz, I don't subscribe to that.  However, the torque wrench would introduce some error.

really? do you have to bring this notion to this thread? afaik there is only one "guy" in this forum swear by it. you are measuring S21 btw

[SilverSolder]

Within 0.03dBm to 4GHz, you know they weren't handing those out for free.  For low frequency work (<GHz), I have a set of home made standards to get me in the ballpark.  I wouldn't mind having a few high end standards to use above a GHz.  I tried to make a T-check but it was a total waste of time.  Similar to trying to make a load.

The physical dimensions and tolerances just get impossibly small at some point.  Wonder how the high end devices are actually made...   photolithography?

[SilverSolder]

Even at 400Hz, I can detect sub-pF capacitance effects when playing around with bridge circuits, for example.

or the lack of instrument's resolution or sensitivy is one possibility. unspecified Through kit is another, recompensating reality...

I can get to sub-PPM resolution and sensitivity, but at that point things are so sensitive that you almost can't breathe in the same room without affecting something.  So just like with high frequencies, the mechanical and electrical properties become a challenge to make stable over a range of temperature, humidity, etc. - the physical implementation of the ideal laws becomes a real issue!

[joeqsmith]

Contrary to the belief that hand tightening connectors is good enough to 6HGz, I don't subscribe to that.  However, the torque wrench would introduce some error.

really? do you have to bring this notion to this thread? afaik there is only one "guy" in this forum swear by it. you are measuring S21 btw

It's actually very relevant to SilverSolder's question on sources of error.   It also seems to be a very common opinion as I have had several people claim this after I posted the first video on the Nano.  You are not alone but you are the most recent.   

You get five stars for knowing and making clear that was S21.   

[joeqsmith]

Within 0.03dBm to 4GHz, you know they weren't handing those out for free.  For low frequency work (<GHz), I have a set of home made standards to get me in the ballpark.  I wouldn't mind having a few high end standards to use above a GHz.  I tried to make a T-check but it was a total waste of time.  Similar to trying to make a load.

The physical dimensions and tolerances just get impossibly small at some point.  Wonder how the high end devices are actually made...   photolithography?

Especially looking at the 3.5 standards I borrowed.  The person that loaned them to me told me about when they first started their career in RF.   Connector care was a mandatory lecture from the old German guy who managed their lab, before you were allowed to touch any equipment. 

I'm guessing a few of these would be sold on eBay as working, with an asking price of half of new. 
https://www.microwaves101.com/encyclopedias/how-to-not-trash-a-calibration-kit

The unknown thru is commonly used. 
http://ena.support.keysight.com/e5071c/manuals/webhelp/eng/measurement/calibration/advanced_calibrations/unknow_thru_calibiration.htm   

[Yansi]

Trying to find a source for flange mount resistors without a year lead time is the challenge. 

I've bought parts from this company in the past.  They have a few in stock but nothing I can use.
https://www.surplussales.com/RF/RFAtten-6.html

Shit ton of these on Aliexpress. Mostly, both new and used. Did not test any of them yet, but have a 10W 20dB attenuator on order.
Aliexpress lead time is just about a month.

[Bud]

This proves the validity of the 11th testament:

"Thou shalt downscale all Chinese specifications by a factor of 2"

[Yansi]

I am kind of surprised, that there would be any problems with the chinese power attenuators. The only thing I would direct concerns at, would be the flatness.

Even 10W is a lot of heat and when improperly heatsinked it is possible to burn out even much higher specified part.

If it burns out even after a very brief moment load, than oh well .. .

[joeqsmith]

I am kind of surprised, that there would be any problems with the chinese power attenuators. The only thing I would direct concerns at, would be the flatness.

Even 10W is a lot of heat and when improperly heatsinked it is possible to burn out even much higher specified part.

If it burns out even after a very brief moment load, than oh well .. .

I don't have any experience with them, which is why I asked about your testing when yours arrive.   It's very possible that the parts mentioned above were not properly heatsunk.  They don't provide any details.   

In my case, if the attenuator fails in such a way that more power is output than expected, it could damage the device it's attached to.  Similar to how that original attenuator I show had failed.   This particular attenuator will be followed with a second one.   

[Yansi]

Those chinese ready-made attenuators (have a bunch of N-connector type ones) are definitely quite small for the rated power, would guess the temperature rise would be extreme.

I also use them mainly as a protection of equipment when probing unknown stuff and am aware of the bad heatsinking. To be honest, I have never disassembled any of those I have, so I am not even sure, what is lurking inside.  2512 SMD resistors are certainly not good enough for a "10W" attenuator, regardless of heatsinking.

I can test them, but only with DC voltage, I do not have any suitable higher powered RF source.