Project Yaigol: Fixing Rigol scope design problems.

[Bud]

BTW you could probably get the 2 pdfs in one post by posting one and then the other by way of a "Modify".

Sorry I am not sure I understand.

[tautech]

BTW you could probably get the 2 pdfs in one post by posting one and then the other by way of a "Modify".

Sorry I am not sure I understand.

Sorry Bud if I wasn't clear, I meant having both the text and pictorial content of both pdf's in the body of reply #2  as well as being able to download it.
You'd probably have to compress all the pics to stay within forum limits but that wouldn't matter if the full size ones were available in the pdf's for download.

If you want, I'll try and then you can copy/paste into reply #2 after which I'd delete any unnecessary post/s.

[Alex Eisenhut]

Does anyone else think there should be a decoupling cap close to the power pin of the oscillator? Looks to me like there's a LC filter and a ferrite on the input of the regulator, as well as the usual big/small cap combo, but nothing on the load side. I see a little guy floating up there but not right next to the IC.
Unless the cap is on the bottom side, there seems to be vias for that there.

Granted, the thing is probably not drawing more than 10mA, but wouldn't it be best to knock out high frequency energy on the load side, rather than only on the input side of the regulator?

It might change nothing at all, but for a 1 cent part, I thought it was SOP to sprinkle power decoupling caps liberally. I mean you can always depopulate it after some empirical tweaking, no?

[Bud]

I meant having both the text and pictorial content of both pdf's in the body of reply #2  as well as being able to download it.

But you cant just post pictures in the body of a reply, you have to host them elsewhere and point picture URLs to them from your reply, haven't you?

If you want, I'll try and then you can copy/paste into reply #2 after which I'd delete any unnecessary post/s.

Sure that would be ideal, as I am not sure when I can get to creating and formatting in-reply copy of it. I can provide a MS Word source file if that helps.

[Bud]

Does anyone else think there should be a decoupling cap close to the power pin of the oscillator?

I tried one, it did not make a difference. The proper fix is in Part 3 which I just posted.

Granted, the thing is probably not drawing more than 10mA

Overall current through the LDO is 38mA, that includes the oscillator and the PLL.

[Bud]

I just found my old notes which says the home made inductors for the PLL should use 9mm long #30 wire (not 7mm as in the article). It also says I measured one made from 10mm length with the VNA and it had 3.25nH inductance at 1GHz. Seems I then reduced the length to 9mm to drop inductance a little.

[ve7xen]

I meant having both the text and pictorial content of both pdf's in the body of reply #2  as well as being able to download it.

But you cant just post pictures in the body of a reply, you have to host them elsewhere and point picture URLs to them from your reply, haven't you?

You can attach the pictures to the post, then click 'preview' to get the URL of the attached image (this might actually appear before you click 'preview' when you're uploading attachments), then use the `[img]` tag to post them inline. A bit of a pain, but possible.

Great work in this thread!!

[Bud]

Oh men, too much work attaching pictures...

[tautech]

Bud with MarkL's assistance bought the poorly implemented Yaigol PLL clock to our attention nearly 18 months ago (for those that were taking notice), there is much about it in the previously linked thread and since then has documented the ordeal in this thread for the benefit of all. 

That he has taken so long to present his findings in detail is confusing to me and I wait with great interest for any gems not yet revealed. 

Carry on Bud, the floor is yours. 

BTW you could probably get the 2 pdfs in one post by posting one and then the other by way of a "Modify".
Not sure, but it'd be worth a try rather than have to download the 1.5 mb which will a PITA for some.

The PLL power supply in DS2072A was oscillating.

 
 
Unbelievable

What else is there? 

[Emi]

Mmm...quite shocking this last issue!
if all psu in there "ocillate"  and the "always on"  1khz sig gen too ( why not let it switchable with a pushbutton ? ) it will be a big trouble 
this boy ,and his infinite harmonics, is very close to ch inputs !
what about pwm bcklight then? 
Bud, save us...

[senso]

Very interesting read.
Good job.
 

[German_EE]

Threads like this are why I find EEVBlog so valuable, there is some damn fine RF detective work going on here.

[nctnico]

Does anyone else think there should be a decoupling cap close to the power pin of the oscillator?

I tried one, it did not make a difference. The proper fix is in Part 3 which I just posted.

Using a tantalum is one option, the other is to put a small resistor (I always use 0.47 Ohm) in series with a 10uf MLCC capacitor to stabilise an LM1117 (or equivalent) LDO.

[wraper]

Does anyone else think there should be a decoupling cap close to the power pin of the oscillator?

I tried one, it did not make a difference. The proper fix is in Part 3 which I just posted.

Using a tantalum is one option, the other is to put a small resistor (I always use 0.47 Ohm) in series with a 10uf MLCC capacitor to stabilise an LM1117 (or equivalent) LDO.

Third option is to replace the LDO with MLCC tolerant type.
http://www.diodes.com/_files/datasheets/ZLDO1117.pdf

[Monkeh]

I haven't opened my DS1054Z, but it's easily confirmed from the teardown photos that the exact same LDO configuration is present and oscillating like a cheap motel bed.

[madires]

Thanks for the thorough investigation!  It's sad that they have made such stupid mistakes by not reading datasheets. At least we can fix the issues.

[bitseeker]

Bud,

Thanks again for all the work you've put into this. I also like your brand adjustment photo editing. Nice touch!

[CaptCrash]

Thanks for presenting this, I really enjoyed your image manipulation.  Well done.

[Bud]

Using a tantalum is one option, the other is to put a small resistor (I always use 0.47 Ohm) in series with a 10uf MLCC capacitor to stabilise an LM1117 (or equivalent) LDO.

Third option is to replace the LDO with MLCC tolerant type.

Yes can do that too but inserting a resistor is an extra work, may need either cut the pcb track or mount it the capacitor in a tombstone manner. Replacing the LDO is extra work too: Yaigol use 1117 adjustable type of regulator, so there is a voltage setting resistor divider with it. Replacing the regulator will require removing the divider or replacing with a new one depending if your new regulator is of a fixed or adjustable type. In both cases i'd think it is not necessary, just do a simple capacitor swap and you will be fine.

[Bud]

Thanks for presenting this, I really enjoyed your image manipulation.  Well done.

Thank you All for your interest and feedback. Yes wanted to make it a bit funny for easier consumption
 

[Monkeh]

Replacing the LDO is extra work too: Yaigol use 1117 adjustable type of regulator, so there is a voltage setting resistor divider with it. Replacing the regulator will require removing the divider or replacing with a new one depending if your new regulator is of a fixed or adjustable type. In both cases i'd think it is not necessary, just do a simple capacitor swap and you will be fine.

The datasheet he linked is for a drop-in replacement for a 1117. Also, pin 1 (ADJ/GND) is clearly firmly connected to ground in both DS1000Z and DS2000 scopes (it's a fixed 3.3V reg).

[Bud]

Ah yes, sorry, i confused it with some other circuit

[Bud]

After i posted Part 3 i found my old notes saying i actually measured the Yaigol capacitor with the VNA and it had 6 mOhm of ESR in its dip. A very far cry from the LDO specified limit of 33 mOhm.

[bitseeker]

Just finished reading Part 3. So, is it fair to say that making the corrections in Part 2 brings a 2000 scope's PLL into a similar condition as the firmware-fixed 1000Z and will lock? In other words, will both models be OK at that point? Then, fixing the LDO oscillation provides additional cleanup applicable to both models, but the values for the 1000Z are TBD.

The final spectrum comparison at the end of Part 3 is awesome. 

[mimmus78]

Thanks buddy ... this post is better than any Netflix series.