EEVblog Electronics Community Forum
Products => Test Equipment => Topic started by: inaxeon on December 22, 2021, 12:52:29 pm
-
Hello Community
Only a month after having purchased it, I've made the unfortunate mishap of damaging the input of my new Siglent 7.5 GHz Spectrum Analyser. Basically once I did this the warranty is toast. I was quoted EUR 3200 to fix it (they just replace the whole RF deck). Ridiculous. They helpfully linked me to this video:
https://www.youtube.com/watch?v=ToVJTKCyIU8 (https://www.youtube.com/watch?v=ToVJTKCyIU8)
And told me I could try repair it myself. Thanks. Not.
So instead I'm going to pull it open and see if I can sort it out. There's no pictures of the inside of one of these I could find so this should be interesting.
A first look reveals that it is a completely different instrument inside to the 3GHz models. Looks like it's setup for VNA too but I can't see any official upgrade for that yet?
After undoing 1000 plus screws I'm "In like Flynn". The main RF deck is a completely different design to the 3 GHz models.
I'm goint to start off by replacing the first, and maybe second RF switch. There ain't much else in there and see how that goes. It's an HMC1118.
-
As they say, "What do you have to lose?" Good luck.
Years ago, I forgot the dc block and damaged a spectrum analyzer. Some mistakes are soooo painful!
-
I blew it with an RF current probe. No DC involved here ;-)
-
And told me I could try repair it myself. Thanks. Not.
So instead I'm going to pull it open and see if I can sort it out.
Here is now some logic what I can not understand.
You did not want do it using Siglent help but after then you want do it just with your own?
I am guite sure if Siglent told you can also try repair it yourself, naturally after then they may also give some helpful instructions for it, perhaps perhaps even more...
Least I have been in this situation (exept that reason for failure was unknown but propably ESD to input connector) with other SA model and I get all needed instructions for diagnose and also for repair, they sent even needed components (rf switch and some pin diodes).
-
I did ask for technical details. I got this:
However I´m not allowed to provide much information about the device, therefore I added the link to the video from Shahriar.
So basically EUR 3500 or chance it.
-
It's an HMC1118.
Well, hope that fixes it for ya. If only you could fit an IC socket for next time ::)
-
I wonder how they determined that fault is not covered by warranty?
-
If J20 is the input then it seems the input has a DC block incorporated. It is strange that the input goes straight into a chip without any ESD protection. A better solution would be to have a good old relay operated attenuator before any semiconductors. My Advantest spectrum analyser defaults to having a 10dB attenuator enabled in order to have some input protection. It can be disabled but this is only necessary for measurements that needs the extra sensitivity.
-
I wonder how they determined that fault is not covered by warranty?
It could be due to the fact that I blew it by being a dumb arse... It's not like they don't say it a 1000 times when EMC testing to protect your spectrum analyser. The connector I had powering the DUT was loose, causing a power cycle during testing which popped it.
-
If J20 is the input then it seems the input has a DC block incorporated. It is strange that the input goes straight into a chip without any ESD protection. A better solution would be to have a good old relay operated attenuator before any semiconductors. My Advantest spectrum analyser defaults to having a 10dB attenuator enabled in order to have some input protection. It can be disabled but this is only necessary for measurements that needs the extra sensitivity.
J20 is the input from the FP. I don't think it's -that- strange. They likely didn't want to pay the performance price of protecting it.
-
An external attenuator (10dB) or RF limiter is a good idea to use together with what is -essentially- an open input spectrum analyser.
What you can try is to remove the first HMC1118 and feed an RF signal into the next stage (which seems to be an attenuator circuit) to see if only the first stage is broken.
-
What you can try is to remove the first HMC1118 and feed an RF signal into the next stage (which seems to be an attenuator circuit) to see if only the first stage is broken.
I was thinking that, and was very tempted to -but- for the cost of only £12 I'd prefer to chance it. If I start experimenting it increases the number of times that area is reworked, and the chances of damaging the PCB.
-
hum could the eevblog member or signal path member do a checkup on it ??
They are good in trouble shooting signals path etc ... send them an Email ??
-
hum could the eevblog member or signal path member do a checkup on it ??
They are good in trouble shooting signals path etc ... send them an Email ??
Hopefully I shouldn't need to. I already did a bit of testing on it. Superficially the digital attenuator (HMC424) and preamp appear to be still working OK. What I see is a fixed 40dB of attenuation on the input, so it is quite likely that it's one of the switches. Have ordered a couple more to try out.
But I could be wrong. Will find out in due course.
-
I did ask for technical details. I got this:
However I´m not allowed to provide much information about the device, therefore I added the link to the video from Shahriar.
So basically EUR 3500 or chance it.
Let me see if we can get some better guidance for you. Might take a day or so.
-
I wonder how they determined that fault is not covered by warranty?
It could be due to the fact that I blew it by being a dumb arse... It's not like they don't say it a 1000 times when EMC testing to protect your spectrum analyser. The connector I had powering the DUT was loose, causing a power cycle during testing which popped it.
Did you have the Preamp ON ?
-
Did you have the Preamp ON ?
It was at some point but I think I switched it out just before the incident. I it does still seem to be providing a consistent 30dB gain of what left I can get from the input - which is attenuated by 40dB or so. I -think- it's OK.
-
Did you have the Preamp ON ?
It was at some point but I think I switched it out just before the incident. I it does still seem to be providing a consistent 30dB gain of what left I can get from the input - which is attenuated by 40dB or so. I -think- it's OK.
Typically we would leave attenuation settings to be auto managed and have Preamp ON for EMI work when using a noncontact RF probe.
Where you in EMI mode when the incident happened ?
-
Did you have the Preamp ON ?
It was at some point but I think I switched it out just before the incident. I it does still seem to be providing a consistent 30dB gain of what left I can get from the input - which is attenuated by 40dB or so. I -think- it's OK.
Typically we would leave attenuation settings to be auto managed and have Preamp ON for EMI work when using a noncontact RF probe.
Where you in EMI mode when the incident happened ?
Not in EMI mode at the time. Manual attenuation 10dB.
-
Did you have the Preamp ON ?
It was at some point but I think I switched it out just before the incident. I it does still seem to be providing a consistent 30dB gain of what left I can get from the input - which is attenuated by 40dB or so. I -think- it's OK.
Typically we would leave attenuation settings to be auto managed and have Preamp ON for EMI work when using a noncontact RF probe.
Where you in EMI mode when the incident happened ?
Not in EMI mode at the time. Manual attenuation 10dB.
Hmmm, did the RF input overload warning sound ?
From the datasheet:
Maximum average power: 30 dBm, 3 minutes, fc ≥10 MHz, att > 20 dBm, preamp off
Maximum damage level: 33 dBm, fc ≥ 10 MHz, att > 20 dBm, preamp off
-
Hmmm, did the RF input overload warning sound ?
From the datasheet:
Maximum average power: 30 dBm, 3 minutes, fc ≥10 MHz, att > 20 dBm, preamp off
Maximum damage level: 33 dBm, fc ≥ 10 MHz, att > 20 dBm, preamp off
No buzzer. Wouldn't expect it on a transient, would you?
-
Hmmm, did the RF input overload warning sound ?
From the datasheet:
Maximum average power: 30 dBm, 3 minutes, fc ≥10 MHz, att > 20 dBm, preamp off
Maximum damage level: 33 dBm, fc ≥ 10 MHz, att > 20 dBm, preamp off
No buzzer. Wouldn't expect it on a transient, would you?
No, however the more info you provide the better a diagnosis plus more importantly this all serves as help for those that might have a similar issue.
Dealing with issues like this on a public forum provide a long lasting record that can help others many years down the track.
On of the most knowledgeable on SSA's is rf-loop whom replied earlier with some pertinent points. Like he experienced, let's see what Siglent HQ advise.
-
If you look at the HMC1118 datasheet (https://www.analog.com/media/en/technical-documentation/data-sheets/HMC1118.pdf), it quickly gets obvious that Siglent's overload specs tell only half the truth. They appear to be valid for frequencies of 10MHz and up, but below, there's a steep drop of the damage threshold of these switches as it seems. At a few tens of kilohertz, this limit can be as low as 10dBm. And a wonky power connector for sure may produce signals in this lower frequency range. An additional LF/DC block may be a good insurance to prevent similar "oopsies" from happening in future.
Anyway, good luck with the repair and thanks for sharing the information on the 7GHz SSA!
-
Hello Community
Only a month after having purchased it, I've made the unfortunate mishap of damaging the input of my new Siglent 7.5 GHz Spectrum Analyser. Basically once I did this the warranty is toast. I was quoted EUR 3200 to fix it (they just replace the whole RF deck). Ridiculous. They helpfully linked me to this video:
https://www.youtube.com/watch?v=ToVJTKCyIU8 (https://www.youtube.com/watch?v=ToVJTKCyIU8)
And told me I could try repair it myself. Thanks. Not.
So instead I'm going to pull it open and see if I can sort it out. There's no pictures of the inside of one of these I could find so this should be interesting.
A first look reveals that it is a completely different instrument inside to the 3GHz models. Looks like it's setup for VNA too but I can't see any official upgrade for that yet?
After undoing 1000 plus screws I'm "In like Flynn". The main RF deck is a completely different design to the 3 GHz models.
I'm goint to start off by replacing the first, and maybe second RF switch. There ain't much else in there and see how that goes. It's an HMC1118.
So you’re complaining that the device *you* damaged, and are reticent to have them repair for such a high cost, the same one for which they’ve kindly shown you TSP channel video for, to help you repair it and save you money? Nice. They didn’t HAVE to share that video with you.
-
Any chance you checked the impedance of the front end with a VNA before disassembly?
Replacing the first HMC1118 will hopefully fix it.
If you're not super comfortable with hot air rework I recommend timing how long it takes to remove the original IC so you can use the same heat/time when installing the new one. Obviously you want to use as little heat as possible but it will probably take quite a bit.
-
So you’re complaining that the device *you* damaged, and are reticent to have them repair for such a high cost, the same one for which they’ve kindly shown you TSP channel video for, to help you repair it and save you money? Nice. They didn’t HAVE to share that video with you.
The video they linked to is not about repair, or the instrument I have, barely even close. Additionally, a request for even a scrap of information like the part numbers for the switches was not honoured.
I'm happy to put it out there that I'm unimpressed with Siglent in this instance. I will however say a big thanks to the comments from other EEVBlog members who have been very helpful.
-
Any chance you checked the impedance of the front end with a VNA before disassembly?
Replacing the first HMC1118 will hopefully fix it.
If you're not super comfortable with hot air rework I recommend timing how long it takes to remove the original IC so you can use the same heat/time when installing the new one. Obviously you want to use as little heat as possible but it will probably take quite a bit.
I don't have a VNA unfortunately. Fairly comfortable with hot air rework, shouldn't be an issue.
-
...
So you’re complaining that the device *you* damaged, and are reticent to have them repair for such a high cost, the same one for which they’ve kindly shown you TSP channel video for, to help you repair it and save you money? Nice. They didn’t HAVE to share that video with you.
:-// You may not have noticed that the video is courtesy of "The Signal Path" aka Shariar, who's a member on EEVblog as well. It's freely available on YT and Siglent hasn't any involvement in it except having supplied the VNA hardware to Shariar for testing.
Moreover, the instrument in the mentioned video is the 3.2GHz variety whereas the OP's broken unit is a 7.5GHz SSA which (internally) is a completely different animal.
Of course, despite all this it's a kind move by Siglent to provide a link to Shariar's video even though it wouldn't help much regarding the repair of the OP's broken instrument.
Regarding the repair itself: Great care must be taken when doing hot air rework on these hybrid PCBs in order to not lift any pads or traces. This is exacerbated since lead free solder is used and the HMC1118 has an exposed pad that's soldered to the bulk of copper (GND) that's probably also VIA-stitched to underlying ground layers. I wouldn't even attempt this procedure without a decent (IR) heater from the other side of the PCB. Only hot air from the top most likely won't do the job.
For installation of the replacement part(s), I'ld definitely use leaded or even special low-temp solder in order to reduce the risk of collateral damage...
Edit: Just noticed that the OP commented on this topic while I was typing...
-
Regarding the repair itself: Great care must be taken when doing hot air rework on these hybrid PCBs in order to not lift any pads or traces. This is exacerbated since lead free solder is used and the HMC1118 has an exposed pad that's soldered to the bulk of copper (GND) that's probably also VIA-stitched to underlying ground layers. I wouldn't even attempt this procedure without a decent (IR) heater from the other side of the PCB. Only hot air from the top most likely won't do the job.
For installation of the replacement part(s), I'ld definitely use leaded or even special low-temp solder in order to reduce the risk of collateral damage...
Edit: Just noticed that the OP commented on this topic while I was typing...
IR heater? Wow man that is full on. I recently re-worked a QFN 24 on a non RF PCB which had heavy stitching to the internal GND layer. It was a piece of cake. You really think I need an IR heater? I usually just pass the hot air from side to side and heat it gradually.
-
I'm happy to put it out there that I'm unimpressed with Siglent in this instance.
As this is a very near new SA and top of this range, Siglent's position is such that only new replacement parts can can be sure of delivering guaranteed performance. Having their service shop repair it with new parts would also ensure the warranty remained intact rather than have a owner repair it.
I'm on my 3rd one having upgraded some steps from their first releases to the SVA1032X I have today.
The other 2 were sold to customers with some remaining warranty or warranty we covered ourselves.
-
Regarding the repair itself: Great care must be taken when doing hot air rework on these hybrid PCBs in order to not lift any pads or traces. This is exacerbated since lead free solder is used and the HMC1118 has an exposed pad that's soldered to the bulk of copper (GND) that's probably also VIA-stitched to underlying ground layers. I wouldn't even attempt this procedure without a decent (IR) heater from the other side of the PCB. Only hot air from the top most likely won't do the job.
For installation of the replacement part(s), I'ld definitely use leaded or even special low-temp solder in order to reduce the risk of collateral damage...
Edit: Just noticed that the OP commented on this topic while I was typing...
IR heater? Wow man that is full on. I recently re-worked a QFN 24 on a non RF PCB which had heavy stitching to the internal GND layer. It was a piece of cake. You really think I need an IR heater? I usually just pass the hot air from side to side and heat it gradually.
I don't think it is that bad. It is only a small QFN chip. If you apply heat on the other side of the board to pre-heat it with hot air and then desolder the chip it is very likely a piece of cake indeed. My approach would be:
0) set the hot-air to 330 deg C
1) Preheat the board from the opposite side (a 5 x 5cm area) using hot air
2) Desolder the chip by applying heat to a 4x4 cm area; not just the chip itself
3) Apply flux
4) Put the new chip on the board and align it
5) Heat again to solder the chip to the board (align by nudging it if necessary)
6) Apply extra tin to pads that don't have enough tin
Note 1: there is no step to remove tin; this is a process which can easely lift pads so don't do it
Note 2: I don't point the hot air so much at the chip itself but primarily onto the board. In the end the board needs most of the heat energy
-
I'm happy to put it out there that I'm unimpressed with Siglent in this instance.
As this is a very near new SA and top of this range, Siglent's position is such that only new replacement parts can can be sure of delivering guaranteed performance. Having their service shop repair it with new parts would also ensure the warranty remained intact rather than have a owner repair it.
I'm on my 3rd one having upgraded some steps from their first releases to the SVA1032X I have today.
The other 2 were sold to customers with some remaining warranty or warranty we covered ourselves.
I was told by Siglent's European support that there is no board level repair facility in this area. If there was I would have opted for it.
The options presented to me were:
A) Pony up €3200
B) Fix it yourself.
Impressive would be "We'll take care of it" or "Fix it for €300" for example.
-
Well FedEx have just been and gone with my Christmas bundle of Hittite. It's a success! I've fixed the 40dB attenuation and have it back to how it was.
A few notes, and lessons:
1) As was suggested to me, replacing a QFN on an RF PCB like this is a lot harder than a thinner digital PCB. I didn't end up heating from the underside but it still took a bit more heat than usual to do the job.
2) There are some foams around the SMA connectors which won't survive a heat gun. Cover them first.
3) Always protect your spectrum analyser with a transient supressor.
4) Don't expect any favours from Siglent if you end up in this situation
5) You can learn more about the input characteristic of the SSA 3075X Plus from the HMC1118 datasheet than you can from Siglent's, because, it's parked right at the input with almost no protection ;-)
Merry Xmas all and thanks for your help
-
Well done! :clap: :clap:
Now, convert it to a SVA.
-
Well done! :clap: :clap:
Now, convert it to a SVA.
Certainly no point in paying for software anymore now that my warranty is toast. Perhaps I'll leave reading the 1000 forum pages about this till next year ;-)
-
Well FedEx have just been and gone with my Christmas bundle of Hittite. It's a success! I've fixed the 40dB attenuation and have it back to how it was.
Great! Nice job.
-
4) Don't expect any favours from Siglent if you end up in this situation
Congrats on fix!
I just want to comment that you seem to expect that Siglent should provide service that is nonexistent with ANY other manufacturer out there...
Nobody does component level repairs anymore, for many years. They might replace board and send broken one for rework to central facility.
Anyways good to hear good news.
Marry Christmas to all!!
-
I just want to comment that you seem to expect that Siglent should provide service that is nonexistent with ANY other manufacturer out there...
I just looked back over all of my posts. I couldn't see any where I'd said I expect Siglent to provide such a service? I certainly would have been impressed if they did.
Have a look at this video. It's by a guy, who blew the input of an SSA, exactly the same way that I did, doing the exact same thing. Skip to the 1:30 mark.
https://www.youtube.com/watch?v=2iAmuLCg2wc (https://www.youtube.com/watch?v=2iAmuLCg2wc)
He apparently has been offered a free repair. If this is true, I certainly would like to know what qualifies him for that, but not myself? I was not asking for a free repair, just a bit less than a €3200 bill to sort out such a minor mishap.
Although he is Chinese, him and his business are based in the UK, just as myself. We ought to be talking to the customer same customer service team too. Why such different treatment? As I said I am assuming that assertion is true.
-
You just answered your own question.
-
IMHO if you contacted them and just said that it failed, you probably would be offered a free repair. But I guess you contacted them and said you blew up the input.
-
I just want to comment that you seem to expect that Siglent should provide service that is nonexistent with ANY other manufacturer out there...
I just looked back over all of my posts. I couldn't see any where I'd said I expect Siglent to provide such a service? I certainly would have been impressed if they did.
Have a look at this video. It's by a guy, who blew the input of an SSA, exactly the same way that I did, doing the exact same thing. Skip to the 1:30 mark.
https://www.youtube.com/watch?v=2iAmuLCg2wc (https://www.youtube.com/watch?v=2iAmuLCg2wc)
He apparently has been offered a free repair. If this is true, I certainly would like to know what qualifies him for that, but not myself? I was not asking for a free repair, just a bit less than a €3200 bill to sort out such a minor mishap.
Although he is Chinese, him and his business are based in the UK, just as myself. We ought to be talking to the customer same customer service team too. Why such different treatment? As I said I am assuming that assertion is true.
Thank you for clarification, I missed that.
I guess there is a difference between entry level inexpensive instrument and one that cost 5 times as much.
Also, we don't really know how that person reported problem to point of sale. Maybe it was accepted as a warranty claim, not as user damaged equipment...
In your place, I would not expect Siglent to repair for free. If they did, I would be happy, but certainly they wouldn't owe it to me to do so.
Also, don't take for granted what people with Youtube channels say, or think that the way how manufacturers treat them is same as everybody else. They sometimes not only get offered free repairs, but all kinds of free equipment in the first place in exchange for reviews or simply web presence...
But all that is a moot point now. Most important thing is that you, with your competence and knowledge, were able to repair expensive instrument. That was nice work.
I would, if I may, would like to invite everybody to a discussion as to how they protect their SA inputs from damage in various potentially dangerous situations. That would be one very useful topic for us all.
-
I would, if I may, would like to invite everybody to a discussion as to how they protect their SA inputs from damage in various potentially dangerous situations. That would be one very useful topic for us all.
For what I was doing, appropriate protection is something like this: https://www.tekbox.com/product/tekbox-tbfl1-transient-limiter/ (https://www.tekbox.com/product/tekbox-tbfl1-transient-limiter/)
It'd limit you to 600MHz but that's plenty for EMC work. Outside of this there probably isn't such a need for protection as you aren't likely to see unexpected overloads.
-
I have several overload limiters here depending on the application. See the image below. The one circled in yellow just contains two pairs of back to back 1N4148 diodes. This has very low insertion loss and very low VSWR. I think the the loss is about 0.1dB at 1GHz and the VSWR is typically 1.05:1.
It only provides effective limiting to about 500MHz though. It also only provides brief protection but should survive 5W of RF for a few seconds. It looks a bit ugly but the RF performance is very good.
The limiter to the left also has an RF fuse and a DC block fitted. This has slightly more loss but it should blow the RF fuse once the limiter diodes conduct. This limiter may well protect at >10W RF power but I'm not sure how quickly the fuse will blow. It should hopefully blow before the limiter diodes fail. I did test it numerous times and the fuse blew in a split second as long as the RF power was above 3W.
The other limiters are pretty much the same except one of them is suitable for AF through to about 20MHz. This has a DC block and a fuse.
For frequencies from VHF through many GHz I also have some PIN diode limiters but these will probably only protect up to about 3W.
-
As I wrote before: if the SA doesn't have a resistive attenuator as the first stage, a good start would be to fit an external 10dB attenuator so whatever protection is in the semiconductor has a fighting chance to cope with a surge coming in.
-
As I wrote before: if the SA doesn't have a resistive attenuator as the first stage, a good start would be to fit an external 10dB attenuator so whatever protection is in the semiconductor has a fighting chance to cope with a surge coming in.
These do have auto default attenuation settings that when turned OFF expose the unit to user errors.
There are times when you need auto attenuation OFF however they are relatively rare when instead you can leave the instrument to mostly protect itself.
When I got my first, a SSA3032X worries about damaging the RF input were very high so I got the Utility kit that has a selection of adapters, cables and an attenuator which after gaining some confidence I rarely used again.
https://siglentna.com/product/ukitssa3x-utility-kit/
-
As I wrote before: if the SA doesn't have a resistive attenuator as the first stage, a good start would be to fit an external 10dB attenuator so whatever protection is in the semiconductor has a fighting chance to cope with a surge coming in.
These do have auto default attenuation settings that when turned OFF expose the unit to user errors.
Did you even look at the circuit board? From the input the signal goes straight into a chip without any protection whatsoever. This is a big no-no in any design because it is extremely sensitive to damage.
-
I dug out my homebrew (precision) protection limiter (the one circled in yellow below) and measured it for insertion loss and return loss. This limiter was very carefully designed to allow me to use it ahead of a thermocouple power meter or a spectrum analyser and to not have to worry about adding correction offsets for the majority of measurements. The ultra low insertion loss and ultra low VSWR mean that it doesn't really contribute much to measurement uncertainty.
You can see the scale for insertion loss is 0.1dB and the return loss is better than 40dB out to 1GHz. The limiter circuit is really simple but the key to the RF performance is in the choice of decent Suhner bulkhead connectors and careful RF layout in the small metal enclosure. You can see that I just soldered the two bulkhead connectors the ideal distance apart and the tight metal enclosure helps to preserve the optimal return loss and insertion loss.
The limiter only provides robust protection up to about 500MHz but it does limit to some degree above this.
-
See below for the other limiter that is just to the left of the previous one. This one has a DC block included and also an RF fuse. This means the insertion loss is a bit higher and the RF fuse limits the bandwidth. However, it still has very good performance up to about 200MHz.
See the plots below. This is the limiter I use if I play with any amateur radio stuff from LF through to 145MHz. The RF fuse may be able to protect against really high power 'accidents' but I've only tested it to about 5W. It's arguably a better limiter than the precision limiter because it has the RF fuse included. This added protection is a good tradeoff against the 0.25dB insertion loss up at 145MHz.
-
Did you even look at the circuit board? From the input the signal goes straight into a chip without any protection whatsoever. This is a big no-no in any design because it is extremely sensitive to damage.
You are not wrong about that however we have to expect this for an instrument at this price point.
-
Did you even look at the circuit board? From the input the signal goes straight into a chip without any protection whatsoever. This is a big no-no in any design because it is extremely sensitive to damage.
You are not wrong about that however we have to expect this for an instrument at this price point.
What do you mean? Most of SAs out there would probably be damaged by same event, especially wide bandwidth expensive ones. SA front ends are not robust at all, and are made for best RF specs not survivability. There are very few that are bit more robust than others, but those are not norm.