Author Topic: FeelTech/FeelElec FY8300S Quick Tear Down  (Read 4044 times)

0 Members and 1 Guest are viewing this topic.

Offline nctnico

  • Super Contributor
  • ***
  • Posts: 20117
  • Country: nl
    • NCT Developments
Re: FeelTech FY8300S Quick Tear Down
« Reply #25 on: January 18, 2020, 01:02:47 am »
Those square waves look OK. What happens if you use pulse generation mode (if available)?
There are small lies, big lies and then there is what is on the screen of your oscilloscope.
 

Offline CDaniel

  • Frequent Contributor
  • **
  • Posts: 349
  • Country: ro
Re: FeelTech FY8300S Quick Tear Down
« Reply #26 on: January 18, 2020, 03:37:38 am »
Square wave looks ok , but turn slowly the frequency to see if some "odd" values have jitter
If is jitter free inded it is abig step forward
 

Offline rf-loop

  • Super Contributor
  • ***
  • Posts: 3354
  • Country: cn
  • Born with DLL21 in hand
Re: FeelTech FY8300S Quick Tear Down
« Reply #27 on: January 18, 2020, 04:21:15 am »
Dear customer, we are very apologize for the trouble and thanks for your suggestion. We could accept your suggestion and replace the screws. If there is any problem with the product, please feel free to contact us.
Email: service@feelelec.com


FeelElec

It is bit confusing. There reads FeelTech FY8300S

You are FeelElec and talk like you are from this company who make these.

FeelTech in China in Henan Province do not know this FY8300S at all in their own web sides  ( http://en.feeltech.net/  ) (sidenote, fun that company do not use even https ). There can not find at all find this 8300S model. I wonder how many copycats there (here) in China is for these... taobao etc is full of these or lookslikes fhat print front panel labels as Feeltech or Feelelec  because we live in a culture of imitation.
ETA: But downloads section there can find FY8300S manual and other materials.

So who/what is FeelElec and what is FeelTech who is original (or are they same) manufacturer for example FY6900 and then their webside in 6900 front panel read FeelElec. What is this "guess who..." game for. 

How about your products CCC. You do not care  C E  or UL or TUV or what ever but looks like you  do not care even CCC.  Who we hang after first bad happen with these electrical shits even if low voltage electronics parts are not so bad in price. 

China Compulsory Certification (CCC). The CCC certificate was introduced in 2002 and applies to imported goods as well as to Chinese products. Why this information is missing everywhere where FeelTech or FeelElec who ever these are are giving informations about their products. Is it so that you think about these things...so what...who cares. Later you may find that also in China you can not anymore do how ever just because sky is high and caesar is far away. 

Is it so that that your products do not meet even domestic standards what is today nearly like mandatory in China afaik and you need do sales so that no one is responsible, just like P2P makets. How you round over China Customs (yes I know it is still today quite easy but later more and more difficult after (noname) is fully integrated inside.
Some nice similarities with European Union C E...
"Failure to comply with CCC regulations will lead to goods being detained at the border or returned to the sender. Likewise, improper use of CCC licenses or printing a CCC-mark on products without certification will also result in fines and penalties." 

And

"The two most important elements of China Compulsory Certification (CCC) certification are product testing (products that are to be certified are sent to test laboratories in China) and factory audits (inspection of the product manufacturers by Chinese auditors). Once a certification is granted, it is valid for several years but must be maintained through annual follow-up audits."

Now, please show these products CCC and documents about audit.  And so that all your products are traceable to it.  I think it is simple because of course you have these.


also it is good take seriously one previous message: //Posted by: rhb, on: January 16, 2020, 11:39:50 am//
« Last Edit: January 18, 2020, 10:22:46 am by rf-loop »
If practice and theory is not equal it tells that used application of theory is wrong or the theory itself is wrong.
-
Harmony OS
 

Offline rf-loop

  • Super Contributor
  • ***
  • Posts: 3354
  • Country: cn
  • Born with DLL21 in hand
Re: FeelTech FY8300S Quick Tear Down
« Reply #28 on: January 18, 2020, 10:51:58 am »
My mistake.

Here's a 9.32566MHz square wave. Triggering on the leading edge with persistence set to 10 seconds.



I'd be getting a thick trailing edge if there was jitter wouldn't i?

This do not look bad at all.

Normal way to measure jitter is measure pulse to pulse, (period to period)  jitter. It can do with scope what have low trigger jitter. Trig to rising edge but watch next rising edge. Also good to check example jitter inside period, example trig to rising edge and watch falling edge.
In data sheets jitter is typically pulse to pulse or period to period jitter. Mostly these are today told using jitter rms. But true circuit is many times critical for peak to peak jitter.  Of course manufacturers like write rms jitter because this number is more nice looking. It is very near zero rms jitter if generator generate million pulse 50ns period pulses with zero jitter and then one period is 70ns. Nice but our some test setup may fail totally even when our generator have very low, nearly zero p-p time jitter rms number. p-p jitter is very important in many cases. Long persistence with good trigger setup tell it after wanted time period. I have long time cricised companies, independent of brand class about these informations. System math produced time jitter need tell peak to peak specially because it is not at all random distribution and system designer naturally know it exactly and then can tell rms for sales advertisements if like. 

« Last Edit: January 18, 2020, 10:53:50 am by rf-loop »
If practice and theory is not equal it tells that used application of theory is wrong or the theory itself is wrong.
-
Harmony OS
 

Offline nctnico

  • Super Contributor
  • ***
  • Posts: 20117
  • Country: nl
    • NCT Developments
Re: FeelTech FY8300S Quick Tear Down
« Reply #29 on: January 18, 2020, 11:20:13 am »
Well... I don't think you have to expect stellar jitter performance from such a low cost generator so I really wouldn't bother measuring it extremely precise. The square waves are likely made using a sine wave to square wave converter (comparator) which may have issues at really low frequencies because the rise/fall times of the sine wave becomes to slow.
There are small lies, big lies and then there is what is on the screen of your oscilloscope.
 

Offline CDaniel

  • Frequent Contributor
  • **
  • Posts: 349
  • Country: ro
Re: FeelTech FY8300S Quick Tear Down
« Reply #30 on: January 18, 2020, 12:14:49 pm »
Normally the square wave is generated digital from FPGA , but an older FY generator has jitter like in your oscillogramme with trigger ... for allmost all frequencies except for some that divide exactly with the clock
« Last Edit: January 18, 2020, 12:19:02 pm by CDaniel »
 

Offline TurboTom

  • Frequent Contributor
  • **
  • Posts: 907
  • Country: de
Re: FeelTech FY8300S Quick Tear Down
« Reply #31 on: January 18, 2020, 01:31:28 pm »
On today's "Magic Pulse", "SiFi II", or whatever the manufacturer may call their version of edge accuracy improvement technology, the slopes don't consist of a single "step" of the DAC output (i.e. outputting a series of -1V samples and the next series of samples directly outputting +1V -- this would permit much faster slopes), but the edges are "stitched up" of a certain amount of samples, ususally 2..3 and the manufacturer makes use of the reconstruction filter to smooth that out to a "more or less" steep, homogenous edge. To compensate for the mis-match of the sample frequency with the signal frequency, the arrangement of these "slope samples" is calculated (or looked up) new for each edge.

As a result of this approach, you will always find some slope steepness wobble with these generators (which may be more or less pronounced on a certain instrument, I emphasized on it quite a little on my "rants" on the SDG6000X which permits outputting faster slopes in certain modes than what the reconstruction filter is designed for), yet with the zero crossing (or centerline crossing) of the signal being virtually jitter-freeand thus resulting in the very good phase jitter figures.

This approach may also lead to some ringing on the "plateaus" of the square (quite visible with the Rigol DG800 / DG900 series) and as it appears, with the AWG that's the subject here as well.

Did anybody test if this three-channel AWG permits outputting full length 8kSa arbitrary signals simultaneously on all outputs? To tell from the FPGA specs, this shouldn't be possible (only enough internal ram to store slightly more than 2* 8kSa, and I don't find any external sample memory attached to the FPGA - unless it's located on the other side of the PCB).
 

Offline grizewald

  • Frequent Contributor
  • **
  • Posts: 515
  • Country: se
Re: FeelTech FY8300S Quick Tear Down
« Reply #32 on: January 19, 2020, 08:01:29 pm »
That sounds like exactly what is being done on this generator. In the shot where I'm triggering off the sync output, I think you can see each edge of the square wave on channel 1 being adjusted slightly on each pulse.

The sync output is unadjusted, and the original 4ns jitter still appears to be present there.

I'll try installing the software and uploading a waveform of my own to check all three channels at the same time. Are there any features you think I should include in the waveform?
  Lord of Sealand
 

Offline edigi

  • Regular Contributor
  • *
  • Posts: 164
  • Country: hu
Re: FeelTech/FeelElec FY8300S Quick Tear Down
« Reply #33 on: January 21, 2020, 01:33:27 pm »
Thank you for the detailed tear down.
As a result, I've pulled the trigger on one of the FY8300S (financially it's a bad idea to read reviews  ;) ). It's on its way now.

Trading the jitter for interpolated edges (varying rise/fall time) is an acceptable compromise for me. I could not justify even the cheapest Siglent/Rigol generator so far (still significantly more expensive although much better DAC and probably logic as well) as I suspect it will see too little use with me, but it's good to have one sometimes.

It's a pity that they don't include a decent PSU even in this variant, but luckily there is plenty of space for a replacement (that I'm not very keen to do though).
It's really good that finally they use a standard 10 MHz clock (so that it can be easily replaced with a better one), although allowing for an external reference is really a missed opportunity.
For a little cost quite much value could have been gained...
 
The following users thanked this post: grizewald

Offline grizewald

  • Frequent Contributor
  • **
  • Posts: 515
  • Country: se
Re: FeelTech/FeelElec FY8300S Quick Tear Down
« Reply #34 on: January 21, 2020, 01:50:09 pm »
Don't forget to give it a good shake when it arrives. You don't want a screw shorting out the supply or anything else first time you power it up!

I also noticed that the rear panel is just a little bit too thick and this makes it very difficult to connect a BNC connector to any of the rear BNC plugs. I'm not sure how I'll address that problem yet.
  Lord of Sealand
 

Offline edigi

  • Regular Contributor
  • *
  • Posts: 164
  • Country: hu
Re: FeelTech/FeelElec FY8300S Quick Tear Down
« Reply #35 on: January 21, 2020, 02:16:55 pm »
It's pretty clear based on your review that one has to start with disassembly (any hint there is welcome so that plastic is not broken) and go through your fix list. Even if nothing is obviously rattling a screw can be stuck already where it should not or it can fall out later on as well unexpectedly. There is no point taking risks here.
Thank you for the fix list.
 

Offline grizewald

  • Frequent Contributor
  • **
  • Posts: 515
  • Country: se
Re: FeelTech/FeelElec FY8300S Quick Tear Down
« Reply #36 on: January 21, 2020, 02:41:34 pm »
It's pretty clear based on your review that one has to start with disassembly (any hint there is welcome so that plastic is not broken) and go through your fix list. Even if nothing is obviously rattling a screw can be stuck already where it should not or it can fall out later on as well unexpectedly. There is no point taking risks here.
Thank you for the fix list.

It's a pig of a thing to dismantle. The ends of the case snap on at the centre points both at the top and bottom and the sides. The plastic is fairly soft, so there's no risk of snapping anything, but you need four hands to release all four catches at the same time. In the photo below, I've circled the catches on the rear end panel.



Obviously, each catch has a partner on the opposite side, so all you need is two larger prying tools for the long edges and two much thinner ones for the side catches.
  Lord of Sealand
 
The following users thanked this post: edigi

Offline grizewald

  • Frequent Contributor
  • **
  • Posts: 515
  • Country: se
Re: FeelTech/FeelElec FY8300S Quick Tear Down
« Reply #37 on: January 21, 2020, 03:18:58 pm »
Oh, and don't forget the screw which holds the upper and lower halves of the case together. The head of the screw is hidden under the label on the bottom of the case.
  Lord of Sealand
 

Offline edigi

  • Regular Contributor
  • *
  • Posts: 164
  • Country: hu
Re: FeelTech/FeelElec FY8300S Quick Tear Down
« Reply #38 on: January 27, 2020, 09:03:04 pm »
Due to super fast shipping I've already received the device and have naturally disassembled.
There is a neat trick for this and you need only 4 small coins (the smallest that you can find, in case of the front side only 2 can be used due to the handle). Gently push the middle of the case inwards and push the coins in (see picture). After this, push neighbor coins away from the device (or top/bottom coin + flat type of screwdriver in case of front). Only 1 minute is needed to completely disassembe the case (naturally without grizewald exploring where the catches are it would have been very hard).

Nothing has fallen out but the fan is not rotating either... It has supply power so probably it's faulty (DoA). What I've noticed that the mains cable is very loose, almost falls out so a good fixing would require replacing the whole mains socket (if I knew where to buy one exactly the same size)...
I can understand the safety concern of the forum members: the PE (Protective Earth) wire is probably failing to reach the minimum cross-section requirement even for enclosure (although I could not find what it is in EU) it's so thin.
As I could not see how to make a final solution I've just used hot glue gun to stress relief the wires and prevent the fan or the screws falling out.

Unfortunately the rear BNC connectors are way too short (as noticed already in some posts) and it's not easy to fix it since the case probably supposed to fix them in place. Without making the plastic thinner however there is a huge risk that either the BNC socket breaks or the cable plugged in breaks. So I had no other choice but to use my router (originally planned for wood...) as I don't have such large drill that would have been needed for this. Initially looked quite ugly, but when assembled it's barely noticeable.

Next thing was the quick check of the functionality. I like how the SW developers have made an efficient use of the small screen space. It's really good (just like the documentation). Most signals don't have jitter (with a quick check I have the impression that log signals are an exception). This is very good at this price point.
Frequency accuracy is however quite bad, even though it's still within specification (20ppm).  After switching on it's roughly accurate, but warm-up means roughly 100Hz drift at 10 MHz. What's worse is that changing signal level or signal shape or anything that changes the heat generated creates a drift in the Hz range. For a signal generator this is too bad.
One thing that would require SW fixing is the printout of the measured frequency.  It can correctly measure GPSDO 1 Hz output duty cycle, however it prints only 1.0 Hz for frequency, where it could print 8 digits as well (like 1.0000001Hz) as the resolution is clearly there.
I plan to replace the TCXO (?) with an OCXO (means separate PCB) if its startup current (0.55A @5V) fits into the budget of the 5V rail. The current TCXO (?) is 3.3V with square output. Unfortunately 2 of its pins (which have 3.3V) are close to a plastic socket so probably neither desolder station nor a hot air station is an option (Argh!.!.!). The pin that is close the closest chip is the output.

I've checked the modulation with PMR device using harmonics (thus square as carrier) and it works nicely if radio is put close to the generator.

If the developers are reading this some improvement ideas:
- Safety first, use decent PE wire and cable isolation
- Rear BNC socket is too short even with the supplied cable, it must be fixed
- Better stability reference and possibility for external reference is a must
- If the data is there use it for proper frequency resolution

All in all value for the price is OK for me but it could be improved so it needs no fixing by the customer. It clearly cannot compete with the more expensive devices nor is it probably made for that target.
« Last Edit: January 28, 2020, 08:41:13 am by edigi »
 
The following users thanked this post: grizewald, duckduck

Offline grizewald

  • Frequent Contributor
  • **
  • Posts: 515
  • Country: se
Re: FeelTech/FeelElec FY8300S Quick Tear Down
« Reply #39 on: January 27, 2020, 11:42:53 pm »
Nice trick with the coins, I'll give that a try next time I need to take mine apart.  :-+

With mine, I put the cut wire back where it came from as I had the correct crimp pins for the connector on hand. I just crimped some 24AWG wire in a new pin and soldered the other end of the wire to the original cut ground wire. Then I took a proper sized and insulated cable and connected that (with a 10K resistor inline) to the PE pin on the socket with a proper spade lug and soldered the other end to the two ground pins on the underside of the power supply. With a 10K resistor, I measured about 30mV AC between the outside of the BNC sockets and earth.

Like you, I noticed that the internal 10MHz timebase isn't anything to write home about, but then again, neither is the one in my HP 5334A - in fact, it's probably worse!

If you want to wire it up so that you can feed it with a proper 10MHz reference from a GPSDO, just cut the legs on the clock chip. Then there's no risk of melting sockets and you're hardly going to use or need the internal reference again if you can connect a proper reference signal.
  Lord of Sealand
 

Offline edigi

  • Regular Contributor
  • *
  • Posts: 164
  • Country: hu
Re: FeelTech/FeelElec FY8300S Quick Tear Down
« Reply #40 on: January 28, 2020, 07:52:56 am »
If you want to wire it up so that you can feed it with a proper 10MHz reference from a GPSDO, just cut the legs on the clock chip. Then there's no risk of melting sockets and you're hardly going to use or need the internal reference again if you can connect a proper reference signal.

I'd not like to rely on GSPDO always and I'd rather have a stable than an absolutely accurate reference.
My aim for accuracy is somewhere in the 0.1 Hz range and the refurbished OCXO that I plan to use for this has voltage control with 10 Hz/V tuning sensitivity (because of this it's not good for any serious use) so if adjusted with a small trimmer it's probably good for a year or so. I have still not decided that if it worth the trouble of the external reference input as it would require some kind of signal detection and switchover.

With better stability references available nowadays the tendency is that even in case of more expensive devices the external reference input is the first victim of cost-cutting (a very unfortunate tendency btw.).
« Last Edit: January 28, 2020, 08:43:23 am by edigi »
 

Offline grizewald

  • Frequent Contributor
  • **
  • Posts: 515
  • Country: se
Re: FeelTech/FeelElec FY8300S Quick Tear Down
« Reply #41 on: January 28, 2020, 09:22:37 am »
I think I'll probably just use a switch to select between external and internal references and replace the internal reference with an OCXO. It will probably need an add-on power supply for the OCXO as I don't think the built in one has enough reserve to drive the OCXO's heater.
  Lord of Sealand
 

Offline nctnico

  • Super Contributor
  • ***
  • Posts: 20117
  • Country: nl
    • NCT Developments
Re: FeelTech/FeelElec FY8300S Quick Tear Down
« Reply #42 on: January 28, 2020, 10:25:18 am »
Just best sure to add some ESD protection to the external input. A 100 OHm series resistor and a TVS diode for example.
There are small lies, big lies and then there is what is on the screen of your oscilloscope.
 
The following users thanked this post: grizewald

Offline edigi

  • Regular Contributor
  • *
  • Posts: 164
  • Country: hu
Re: FeelTech/FeelElec FY8300S Quick Tear Down
« Reply #43 on: January 28, 2020, 07:56:47 pm »
It will probably need an add-on power supply for the OCXO as I don't think the built in one has enough reserve to drive the OCXO's heater.

It's hard to tell without having detailed technical specification of the generator.
I've checked my OCXO and it heats up much faster than I've expected. It uses 0.5A for around 45 sec and then it drops quite fast to 0.2A before 1 min. is reached.  That 0.2A is probably less than what is used by the fan (that I cannot measure since it's faulty).
If the fan is sacrificed (it's probably anyhow not very efficient even if it works, especially for cooling the opamps since they are in the front side of the generator) and use some ventilation holes instead it has to survive the higher load only for 1 min. or less.
I've calculated with an extra 50mA for stabilizing the voltage for the tuning input but that can probably pushed down if really needed.
Unfortunately I don't have much time for this before the weekend, but I don't see it totally hopeless.

If I have to replace the PSU it needs a dedicated PCB due to the screw hole constraints and that means weeks at earliest...(not that I would be in a hurry)
 

Offline edigi

  • Regular Contributor
  • *
  • Posts: 164
  • Country: hu
Re: FeelTech/FeelElec FY8300S Quick Tear Down
« Reply #44 on: January 29, 2020, 06:36:50 pm »
I just couldn't resist making a quick check of why the fan in my device is not working so I've desoldered it (not without surprise: I saw already earlier that one of the wires is soldered to a pin of a regulator, but I did not expect the other wire to be soldered to a small capacitor as it was hidden by the wire).

Checking with lab PSU the fan started to work from 6V up. I cannot really blame the vendor of that fan that it does not start less than half of its rated voltage...
Unfortunately I've also grossly overestimated the amount of current that can be gained by not using the fan. It probably does not use more than 30mA from 5V.
 

Offline grizewald

  • Frequent Contributor
  • **
  • Posts: 515
  • Country: se
Re: FeelTech/FeelElec FY8300S Quick Tear Down
« Reply #45 on: January 30, 2020, 08:44:07 am »
Interesting! The fan in mine is a Sunon and it starts up just fine at 5V.
  Lord of Sealand
 

Offline edigi

  • Regular Contributor
  • *
  • Posts: 164
  • Country: hu
Re: FeelTech/FeelElec FY8300S Quick Tear Down
« Reply #46 on: January 30, 2020, 02:14:23 pm »
This is what I've received. Maybe they've run out of their supply or found a different, even cheaper source.
If I start it by hand it can keep spinning, but it cannot start by its own from 5V (I've checked this also yesterday).
Also when running from 6V the airflow it generates converges quite strongly to nothing.

Btw. I've also noticed that it's soldered to a different place in your picture. This seems to me a rather ad-hoc cosmetical than a real solution, so I'll probably vote for ventilation holes even if I'm forced to use a different PSU because of the OCXO.
 

Offline ronbot

  • Newbie
  • Posts: 3
  • Country: us
Re: FeelTech/FeelElec FY8300S Quick Tear Down
« Reply #47 on: August 29, 2020, 02:34:26 am »
I got mine a few days ago, some of the mentioned issues have been factory resolved.
The rear BNC's now have a stepped hole (recessed on the inside of the back cover), so that the connector protrudes properly now... no fitment problems with BNC attachment.

Here's a shot I just took for those wondering about jitter... (please excuse my phones camera... it has noise issues)
Gen set for 10.00MHz output:
Main trigger timebase: 100ns/div
Delayed trigger timebase: 5ns/div
Delayed runs 1080ns after main trigger
Showing Envelope of 500 samples

 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf