EEVBlog #497 - Siglent SDG5000 Function Generator Teardown

[skycarl]

Plasma or grind,,, that shouldn't be there period.

[hobbs]

Re the cutouts in the ground planes--there are a lot of misconceptions about that.  The designer of thiis instrument seems to have fallen into the most elementary one: making  the cutout the same size as the pad.

The parallel plate capacitance formula breaks down really badly when the dielectric thickness is comparable to the pad size.  The fringing capacitance dominates so much in that case that the true capacitance can be several times the parallel-plate value. 

In order to really reduce the capacitance from a pad to the ground plane, you typically have to extend the cutout one or two dielectric thicknesses in both directions.

I do a lot of high-Z front ends, so I run into this problem all the time.

Cheers

Phil Hobbs

[dentaku]

Has Dave ever reviewed a function generator other than way back in episode #5?

[Bored@Work]

The middle line in E sign have to be shorter than the top/bottom lines -> and that's the case for the logo on Siglent
The distance between C and E have to be that long as the length of middle line in E sign -> and that's almost the case here

Ohhhhhh, common guys, are you serious? Legal, illegal, legal, illegal, legal .... this is EE forum not of some stupid Bar association I have whole bag of those labels around somewhere for all countries - they were included as "extra" with project box I ordered online from China. You can print any labels as much as you wish but your your brand reputation is not printable and that what matters.

You stop making fun of it if customs holds and then rejects your import and you are asked if you (a) want to pay for the destruction  of it or (b) want to pay for the return to the sender.

[rf-loop]

Re the cutouts in the ground planes--there are a lot of misconceptions about that.  The designer of thiis instrument seems to have fallen into the most elementary one: making  the cutout the same size as the pad.

You can now show how this affect this signal generator output signal quality?
It do not work? Do you thisnk they have tried minimize capacitances as low as possible? No, they have not and next I ask why they should. They have designed output circuit and in practice it works without problems.   I have also true measured data from it. So, what exactly is now wrong there? 

I'm interest about what problems you find in output signal now (related this PCB layout) and what changes exactly need do and what is improvement in signal quality after your designed changes.
 

[Rubi]

Hi

The metals is anodised which acts as an insulator. To get the shielding electrically connected they filed the coating away.
Of course then it loses its protection and starts to rust. A good workaround would be the use of aluminium.

Cheers Rubi
http://rubines.blogspot.co.at/

I thought that ferrous metals are difficult to anodize, and when they are the most common process is a black ferric oxide finish - commonly seen with screws and hardware, but not common at all for steel sheet.

Honestly I have no idea, maybe it is just a protective coating.
But imho there must be some insulation layer, because otherwise the file marks would make no sense.

Would be interesting if grandmaster David Jones could tell us if the no rusting portions of the metal frame have an isolation layer.

Cheers
Rubi

[Orange]

But there is no reason to.

Yes, there is.
It is totally unprofessional, and not found in any other professional brand test gear that I am aware of. If they cut corners there, what else are they willing to cut corners on?, or simply not care about?
It does not instill a not confidence in the QA and manufacturing processes.
Can you imagine if Agilent, Tektronix, or Fluke offered up rusty exposed metal chassis on their latest gear? Their arse would be hauled over the coals.
If Siglent want to play with the big boys and be taken seriously like for example Rigol are now, this is not the way to do it. Small things matter, engineers are naturally pedantic.

I agree Dave,
Those gold AQ stickers made me suspicious.
Anyhow, can someone confirm that the used TxDAC is genuine Analog Devices. Its missing the AD logo on it. They might be cutting corners here as well.

[EEVblog]

Anyhow, can someone confirm that the used TxDAC is genuine Analog Devices. Its missing the AD logo on it. They might be cutting corners here as well.

It would have to be. I doubt you'd be able to get a fake that worked like the real thing.

[tinhead]

You can print any labels as much as you wish but your your brand reputation is not printable and that what matters.

but Siglent's brand reputation is good enought for LeCroy, that matters Alexei.

[tinhead]

Anyhow, can someone confirm that the used TxDAC is genuine Analog Devices.

yes, i can confirm, they are genuine.

[flolic]

I thought that ferrous metals are difficult to anodize, and when they are the most common process is a black ferric oxide finish - commonly seen with screws and hardware, but not common at all for steel sheet.

Only aluminium and magnesium can be anodized, ferrous sheet metals are either hot dip galvanized or electrogalvanized.


 

[hobbs]

Re the cutouts in the ground planes--there are a lot of misconceptions about that.  The designer of thiis instrument seems to have fallen into the most elementary one: making  the cutout the same size as the pad.

You can now show how this affect this signal generator output signal quality?
It do not work? Do you thisnk they have tried minimize capacitances as low as possible? No, they have not and next I ask why they should. They have designed output circuit and in practice it works without problems.   I have also true measured data from it. So, what exactly is now wrong there? 

I'm interest about what problems you find in output signal now (related this PCB layout) and what changes exactly need do and what is improvement in signal quality after your designed changes.

Wow, guy, take a pill--I wasn't the one that slapped yo' mama.   

 I haven't tried one of these things out, myself, but watching the review, I thought it was an interesting observation.  It's amazing how little actual thought people put into things like decoupling strategy and minimizing capacitance. 

If you don't believe me, try doing some measurements of how large a hole you need in the ground plane to reduce the capacitance of a small pad by as much as half.  (Hint: it's a lot bigger than the pad.)

Cheers

Phil Hobbs

[robrenz]

Only aluminium and magnesium can be anodized,

Plus stainless steel, titanium and zinc.  meaning they will form a useful oxide layer with the correct procedure.

[ddavidebor]

Yep, only metal with a coerent oxyde can be anodized.

[Short Circuit]

Re the cutouts in the ground planes--there are a lot of misconceptions about that.  The designer of thiis instrument seems to have fallen into the most elementary one: making  the cutout the same size as the pad.

You can now show how this affect this signal generator output signal quality?

He's not saying that it affects output quality. Actually, he tells those cutouts do next to nothing at all.

[PA0PBZ]

He's not saying that it affects output quality. Actually, he tells those cutouts do next to nothing at all.

If the cutouts should do something but they don't it could affect output quality 

[RupertGo]

That rust won't just be fugly, it's got the capability of introducing a large number of non-linear junctions into the device shielding. Which, if there's any amount of RF energy floating about (in an 80 MHz arbitrary waveform function generator with GHz clocks, that isn't much of an if at all), could create some interesting side-effects with a very high fun factor in the tracking down thereof.

At least it proves that Siglent and its suppliers are honest and don't cherry-pick or hand-tune units to go out to reviewers. Every cloud has a silver - or oxidised - lining.

[free_electron]

Re the cutouts in the ground planes--there are a lot of misconceptions about that.  The designer of thiis instrument seems to have fallen into the most elementary one: making  the cutout the same size as the pad.

The parallel plate capacitance formula breaks down really badly when the dielectric thickness is comparable to the pad size.  The fringing capacitance dominates so much in that case that the true capacitance can be several times the parallel-plate value. 

In order to really reduce the capacitance from a pad to the ground plane, you typically have to extend the cutout one or two dielectric thicknesses in both directions.

I do a lot of high-Z front ends, so I run into this problem all the time.

Cheers

Phil Hobbs

rule of 3

[free_electron]

You can print any labels as much as you wish but your your brand reputation is not printable and that what matters.

but Siglent's brand reputation is good enought for LeCroy, that matters Alexei.

pot calling kettle ?

[komet]

Did I miss it or did Dave forget to say "ta-da" on this occasion?

[tinhead]

pot calling kettle ?

yeah, probably. The point is however what means reputation and who deserve it.

[tinhead]

How can you confirm it?

confirm what?

[WattSekunde]

yeah, probably. The point is however what means reputation and who deserve it.

Reputation arises from the user community and discussions like this. Mostly, reputation decide on the amount of the sales.

[LDM]

With the full external part of the chassis mounted it doesn't seem like mechanical stability would be a problem.

But still, rust is never good...


LDM

[maiakaat]

1.Chassis: The solution seems easy enough, they investigate the supply chain, and switch to chassis more typical of the quality expected, add $20 to cover the costs, and re-release as a revised version of the current brand, issuing an apology and agreeing to replace the chassis for anybody who has problems within 3 years specifically with the chassis who bought from approved dealers.

2. C E, I feel this has been answered and ignored, the C E certificate is available for this device on this thread, and China should be investigated by the EU for creating a virtually identical sticker which may have been deliberate

As an additional note, all countries should do something to stop unsafe electronics coming into our respective countries, to increase local sales, as clearly the current customs processes are not adequately funded, Ebay and Amazon pay no tax, and neither do any of the sellers, and this is undermining quality of goods, salaries, national and economic balance and well-being, not to mention safety where goods are dangerous (ie cheap chargers and things)