Author Topic: INSTEK spectrum analyzer GSP-827 -6 dBm down than the real value (Read 20404 times)

Adal · « **on:** March 02, 2015, 07:00:00 pm »

Hi everybody.
Recently I acquired an used INSTEK GSP-827 spectrum analyzer. During the initial test, I discover that the displayed level are -6 dBm lower.
For example the calibration level is:
-30 dBm @ 100 MHz

It´s appear
-36 dBm @ 100 Mhz.

The same level appears if I put external RF Level.

Fortunately, this machine include a input offset in order to compensate the "user" gain/loss. So I use this machine as is.
I hear your recommendations or if someone have the service manual for this thing, I can attempt a correction.
Best regards

G4JMO · « **Reply #1 on:** March 06, 2015, 11:02:01 am »

Hi,
I seem to have the same problem with my unit too!
Let me know if you find a solution or a source of service manual and I will do the same.
Best regards
Andrew

Adal · « **Reply #2 on:** March 06, 2015, 11:46:44 pm »

Hi Andrew.

I think is a software issue, not hardware. Because if I lower or increase the input RF level for example in 10 dBm, the spectrum follows the variation.
I write to Instek for a solution. Maybe you can write them too in order to "pressure" for a solution.

Best regards.

G4JMO · « **Reply #3 on:** March 08, 2015, 03:22:13 pm »

Hi, I've written to Instek too so hopefully at least one of us will hear back!
Will let you know if I do,
Best regards
Andrew

G4JMO · « **Reply #4 on:** March 10, 2015, 10:29:12 am »

I heard back from the manufacturer today - trying to point me to an authorised service centre and not responding to the question of whether the -6dB is a known issue:

Thank you for using GSP-827.

Could you please let us know the serial number of GSP-827?
We could trace and find the proper distributor to assist you soon.

As the GSP-827 need to be proceeded the complete calibration by ATE after repairing (include the components swapping),therefore, from this case, we would have the authorized distributor to assist you solve the problem

Regards

Andrew

ruffy91 · « **Reply #5 on:** March 10, 2015, 11:52:03 am »

-6dBm seems like a impedance mismatch. (-6dBm means exactly half the signal amplitude)

KJDS · « **Reply #6 on:** March 10, 2015, 12:10:08 pm »

Is the 6dB flat across the band?

Is it possible to turn the calibration off?

Fraser · « **Reply #7 on:** March 10, 2015, 02:20:01 pm »

The most damaged components in a spectrum analyser are the first mixer, pre-amplifier and, in extreme cases, the input attenuator. The input socket centre contact is also a potential victim of abuse.

Mixer death is usually accompanied by a larger signal loss than you are seeing. A MMIC gain stage at the input of the analyser would be a possibility as if can continue to operate but with no amplification. You then effectively lose its stage gain but all other functions such as the attenuator and mixer work normally.

One of the greatest errors that some spectrum analysers make is to connect the RF input to an open circuit type antenna. ESD and static present themselves to the sensitive front end electronics and cause al manner of damage. It is a good idea to always have a 10dB attached to the input of an analyser and only removing it when you are absolutely sure it is safe to do so. Many Spectrum Analysers need a button sequence to select 0dB input attenuation for just this reason. The manufacturer likes to keep a pad in circuit to protect the sensitive RF stages that follow.

As to a software fault, sorry, I don't buy that idea. It would initiate a recall from Instek and much talk in the industry.

My advice would be to carefully investigate the RF front end circuit.... they are usually pretty easy to understand and are often laid out in a linear fashion with interstage screening, stripline interconnects and 'weird' pcb filters

If you get lost, just post a picture and I may be able to assist in identification of possible culprits.

Aurora

Additional comment. The unit has front end protection..... normally a couple of fast diodes and a pad. Worth checking that this has not been destroyed.

On many spectrum analysers you can see the 'Zero' marker at 0MHz (its actually the the first L.O).
Is that at the correct amplitude ? Issues here can mean mixer damage.

Also, check your noise floor. Is it still within spec ?

AndyC_772 · « **Reply #8 on:** March 10, 2015, 08:01:53 pm »

Are you absolutely sure that the cal output isn't specified as an open circuit voltage, rather than the voltage into a terminated load?

Exactly 6dB across the range sounds too convenient to be a hardware fault.

Adal · « **Reply #9 on:** March 11, 2015, 07:24:23 pm »

Quote from: G4JMO on March 10, 2015, 10:29:12 am

I heard back from the manufacturer today - trying to point me to an authorised service centre and not responding to the question of whether the -6dB is a known issue:

Thank you for using GSP-827.

Could you please let us know the serial number of GSP-827?
We could trace and find the proper distributor to assist you soon.

As the GSP-827 need to be proceeded the complete calibration by ATE after repairing (include the components swapping),therefore, from this case, we would have the authorized distributor to assist you solve the problem

Regards

Andrew

Andrew.
I don´t receive yet any response from Instek. So I will attempt the repair. Remember, I acquire the GSP-827 used, so maybe the thing are damage.

Adal · « **Reply #10 on:** March 11, 2015, 07:28:26 pm »

Hi Aurora.

Ok, let me take out the covers and check for damage.

Somebody has a service manual?
Best regards.

Fraser · « **Reply #11 on:** March 11, 2015, 07:54:38 pm »

Sadly GW INSTEK service manuals are as rare as hens teeth. I have been in touch with the support team previously and they will not release service information to the public.

Aurora

KJDS · « **Reply #12 on:** March 11, 2015, 08:51:01 pm »

You should be able to get an idea of where the issue is from some testing.

If the -6dB is flat across the frequency range then the damage is after the first mixer.

If there is an auxillary IF output you can also check the level there.

Adal · « **Reply #13 on:** March 11, 2015, 10:54:55 pm »

Unfortunately the GSP-827 don't have a IF out. So I need teardown the thing and looking for damages in the input path. I take some pictures of my equipment and put here.
Best regards

Adal · « **Reply #14 on:** March 11, 2015, 10:56:18 pm »

Yes. The -6dBm error are flat across the frequency range

G0HZU · « **Reply #15 on:** March 11, 2015, 11:22:33 pm »

I had a quick skim of the features/modes in an online doc and this analyser can operate as 75R and also 50R.

For example, I think you are meant to calibrate this analyser in 75R mode with an external 75 to 50R attenuator inline? The manual calls it the 'ZZ adaptor'. This zz adaptor will have a loss of just under 6dB. If you calibrate in 75R mode and forget to fit the (ZZ adaptor) 5.9dB matching attenuator during the calibration then you could end up with an analyser that appears to be deaf by about 6dB after calibration.

G4JMO · « **Reply #16 on:** March 12, 2015, 08:27:50 pm »

Thanks for all the comments. It is some kind of hardware fault, not a 75 /50 mismatch or anything like that. It's behaving in a similar way with external input. I've also found out that the previous owner of my unit had to return it for repair on several occasions for this very fault. Hopefully when its skin is removed, it should be straightforward to find the problem !
Regards

KJDS · « **Reply #17 on:** March 12, 2015, 10:30:26 pm »

Quote from: Adal on March 11, 2015, 10:56:18 pm

Yes. The -6dBm error are flat across the frequency range

Then it is very unlikely that the problem is before the first mixer.

Adal · « **Reply #18 on:** March 13, 2015, 05:46:55 pm »

Thanks to all for your comments.

Andrew, I agree with your diagnostic. In fact yesterday when I turned my GSP-827, the internal 100 MHz -30 dBm calibration signal appears -34 dBm, only -4 dBm lower

And another fault, when I put an 100 MHz AM -30 dBm signal with 100 % modulation, I can´t resolve clearly the two side band, only the carrier.

I make a video in order to show us the test

My strategic for fix it will be
Check the receive path.
Check the IF train and the digitizer. I hear you suggestion.

videobruce · « **Reply #19 on:** March 14, 2015, 04:39:02 am »

Quote from: Aurora on March 11, 2015, 07:54:38 pm

Sadly GW INSTEK service manuals are as rare as hens teeth. I have been in touch with the support team previously and they will not release service information to the public
Aurora

The Chinese are tight lipped. I guess they are paranoid about the competition coping their design. Rigol is the same way. I tried more than once to get a real SM, but there isn't any (so I'm told). Both have something they call a "Service Manual", but all it is is a "verification' manual. The 827 is a older SA, there maybe a possibility doing a search to stumble across something outside of Instek.

Funny, with the real US companies, most had/have SM's available. Actually with both importers, you have to the equipment serviced in China.

videobruce · « **Reply #20 on:** March 14, 2015, 04:41:56 am »

Quote

We could trace and find the proper distributor to assist you soon.

Since when distributors repair equipment??

videobruce · « **Reply #21 on:** March 14, 2015, 04:51:50 am »

Quote

-6dBm means exactly half the signal amplitude

Isn't -3db half the level??

G4JMO · « **Reply #22 on:** March 15, 2015, 03:33:00 am »

Here's the RF PCA
I've not had time to do a thorough circuit analysis yet but here's a brief description (Refer to GSP-827 Front end photo...
RF input through capacitor into resistor pad R91,92, R93 (75R, 120R, 93R respectively). Then there's a cap to ground and what appears to be 2 x SOT23 back to back diodes marked 'EOF'...probably limiters.
The signal then feeds via C103,101 into U22 marked S169. This is a Skyworks Phemt GaAs SPDT DC-2.5GHz switch. Signal continues into a second S169 (U23), a third S169 (U24) and then into U21 marked S131 418. I think this is another Skywork's switch. Then onto another S169 (U25) into U26, another S131 418. The signal then gets +16dB gain from U18, an NBB400 MMIC. Looking at this components solder seems to suggest it has been the subject of some attention before!. I plan to isolate it, power it and feed a signal into it and check how much gain it is giving.
I'd be surprised if it was this component, being so far down the chain. I guess the first things to check are the protection diodes.
Regards
Andrew

Fraser · « **Reply #23 on:** March 15, 2015, 12:29:03 pm »

Do not be fooled by the position of the MMIC. I have seen many such amplifiers destroyed by excessive signal level. I have also seen partial failure as some are multi stage inside the package and suffer failure of one stage. The fact that this has received attention in the past increases my suspicions surrounding this component. It is worth checking how much current it is drawing and comparing the treading with the manufacturers recommendation. This can help identify failure and (when repaired) check that the MMIC is correctly biased in the design.

I managed to cook an MMIC in a test piece by passing broad band noise through it. It is not just the peak RF signal level that can be harmful, it is also the combined energy of a multitude of signals in a broadband signal source that causes high dissipation in the amplifier.

If this does turn out to be the failure point, I recommend some further testing after repair to see whether some modifications are in order to protect the MMIC.

Some interesting reading from NASA:

https://electrondevices.grc.nasa.gov/files/T-06-Dec-1996-JPL-96-25.pdf

NBB400 datasheet is here:

http://www.rfmd.com/store/downloads/dl/file/id/29224/nbb_400_data_sheet.pdf

http://www.farnell.com/datasheets/1499972.pdf

Aurora

grumpydoc · « **Reply #24 on:** March 15, 2015, 12:48:56 pm »

Quote

Isn't -3db half the level

-3dB is half the power, but if you reduce voltage by a factor of two then power is quartered which is a 6dB drop


EEVblog Main Site	EEVblog on Youtube	EEVblog on Twitter	EEVblog on Facebook	EEVblog on Odysee

Author Topic: INSTEK spectrum analyzer GSP-827 -6 dBm down than the real value (Read 20404 times)

Share me