Teardown: Rigol DG2041A (Arb. Function Generator)

[olsenn]

Here are some photos of the insides of a Rigol DG2041A arbitrary waveform generator:

[saturation]

Nice! thanks for making this available.  I wish the photos were clearer though to make more detailed inspections.  If you see this thread, its photos like this that clearly say can have problems, but similar close ups can show good workmanship if its there.

https://www.eevblog.com/forum/index.php?topic=4757.msg62864#new

[nukie]

It looks like olsenn units are used and quite dusty inside, need some cleaning work before he can show us some high rez guts.

[olsenn]

Actually I purchased the unit brand new and it was drop shipped directly from Rigol. I think the fact that it is so dusty and sections covered in epoxy just shows how poor Rigol's manufacturing is for these devices.

[nukie]

Epoxy that secures parts is okay, the fact that epoxy cost money and dust is free.

Is it just the internal which is dusty or externally as well?

[olsenn]

No, just the inside is dusty. The device works perfectly (save for the software), it's just that the epoxy strings across the unit like a spider web in some places; plus the dust tells me it was probably left open in the manufacturing plant for a while before they encased it. All the solder joints look fine, so I guess I shouldn't complain

[olsenn]

Damn! I take back what I said about the solder joints being fine. After taking another look at the photos I took I notice the device is riddled with problems; jumper wires are soldered across surface mount components, caps are melted and poorly aligned; the workmanship on this was horrible.

This is unacceptable! I paid over $1000 for this device and ordered it brand new from tequipment (shipped directly from Rigol). I would be inclined to say that this must be uncommon, but the photos of the DMM show the same shit quality. If anyone here is thinking about purchasing a product from Rigol, especially if it isn't the well respected DS1052E, DON'T!!!!!!! Let the boycott begin!

[Bored@Work]

I paid over $1000 for this device and ordered it brand new from tequipment

I think you should discuss that with them. One argument that people make here is that one should exclusively buy from reputable sources like Tequipment. They clame because in case of problems you have someone to talk to. As opposite to buying directly from someone unknown in China. I think it is time to test the theory. Send Tequipment the images and ask them "WTF?".

On another note, Is there something like a PCB revision number on the PCB? And does the serial number carry a revision number? It would be good to know which PCB and hardware version Rigol fucks up like that right from the factory.

Rigol likes to use rather long revision numbers. The first two are the hardware revision, the second two are the logic revision, the next three are the software revision. E.g. 00.02.00.06.00.02.05 means hardware version 00.02, logic version 00.06 and software version 00.02.05. Although individual components might have shorter numbers.

[firewalker]

Contact Rigol asap.

Alexander.

[saturation]

As BAW says, ask Tequipment, if it looks shoddy to you, make an excuse to return it.  Not sure opening it voids the warranty, so keep it in mind.  Competing Agilent, Fluke or Tek arbs cost 2-3x more, the Rigol offers value if it performs as well as these brands.

I paid over $1000 for this device and ordered it brand new from tequipment

I think you should discuss that with them. One argument that people make here is that one should exclusively buy from reputable sources like Tequipment. They clame because in case of problems you have someone to talk to. As opposite to buying directly from someone unknown in China. I think it is time to test the theory. Send Tequipment the images and ask them "WTF?".

[olsenn]

I sent an email off to Tequipment, but they just said that their return policy is seven days after receiving the product. Rigol does not respond to their emails, and officially the warranty would be void by me opening the case.

However, despite my low confidence that this machine will last for a very long time, it is currently working perfectly, so I will just have to hope that it will survive long enough to make some good use of it. Since I just leave it on my bench anyways, it shouldn't be impacted too much from vibrations et cetera.

My theory is that Rigol ships their crap products to those who order from their educational program (10% off the regular price); I am by no means certain of this, but since the DM3061 multimeter shown on this site originally went to a school, and this was the only piece of equipment I bought from their edu program and coincidentally the only piece that is f@%&*# up, I'm thinking it is at least a possibility

[Hypernova]

Rigol does not respond to their emails

Wait wait wait, are you saying that Rigol don't respond to their own distributors either?

[amspire]

I can understand that you would like to see the highest quality of construction, but is there anything that is actually a problem?

Misaligned capacitors is probably not going to be a reliability problem.

"Melted Capacitors"?  Are you sure? In the photo, it looked like a cap may have been hand soldered and the flux was still on the board making the capacitor look messy. It is very hard to melt ceramic capacitors.

Jumper wires - it happens. 

Some companies do a fix when they see a problem.

Other companies don't do a fix - they send the devices out with the known problems and leave the fix for the next generation of boards.

I would prefer that they at least fix the known problems myself, even if it means the odd wire or resistor soldered onto the board. They may have been fixing a major bug, or perhaps the fix was just to lower noise or distortion by a small amount. Whatever the fix was for, I would bet the device runs better with the fix then without it.

With low production run devices - like expensive test instruments, you can see fixes on the boards particularly for first revision devices.

We are used to seeing high production run devices. By the time we usually get our hands on them, the initial problems have already been fixed. With low production run products, you cannot afford to throw out a batch of expensive boards just because an odd capacitor has to be replaced, and a link or two has to be added.

Richard.

[Bored@Work]

I sent an email off to Tequipment, but they just said that their return policy is seven days after receiving the product.

So much for the "you need to buy from a local distributor, because you can get help in case of problems" argument.

Rigol does not respond to their emails,

Muhahaha. Rigol being the same heap of shit as ever. The interesting question is why does Tequipment do business with a company that doesn't answer their mails?

and officially the warranty would be void by me opening the case.

That is not such a clear-cut thing, depending on jurisdiction and probably other circumstances. I vaguely remember cases here where companies who tried to deny warranties because computer cases were opened, got dragged into court and lost. But you need a lawyer and all that to get a definitive answer. And that is not fun.

My theory is that Rigol ships their crap products to those who order from their educational program

My theory (watch the fanboys cry out loud in a second ....), so my theory is Rigol has a non-discriminatory policy selling crap to everyone and talking to no one, whether in education or not.

[grenert]

Doesn't Tequipment advertise on this forum?  Boy, you'd think they wouldn't want the bad publicity of leaving this guy high and dry with a lemon product...  It raises the question of what exactly are you getting extra by buying from them versus rolling the dice with someone cheaper or ebay?

[saturation]

Tequipment is within their contract rights written on their policies when you buy from them, that is part of a sales agreement between the buyer and seller.  After you buy something, no vendor or seller is obligated to serve as a warrantor of the products they sell; Amazon for example because its online, extends this to ~ 1mo;  some credit cards like Amexco give you extended multimonth return-no-question ask warranties on purchases. 

In the US, you are protected by consumer protection laws that are have different details in every state.

Before you proceed, you will have to read the warranty carefully to see if what you want applies to you and that you are likely right and they are wrong.

Send 1 certified snail mail letter with return receipt, addressed to a specific named individual at the company such as the President or CEO of XXXX, America or USA, to insure he 'gets the message'.   Imply intent to seek redress by legal means to show your purpose.  Normally, such letters are written in lawyer letterhead to add more force to the letter.  Use 10 pt courier font and use as few words as possible.  Check the web for sample 'demand letters'.  If your state has guidelines on writing demand letters, follow them for your particular state.

http://www.mass.gov/?pageID=ocaterminal&L=3&L0=Home&L1=Consumer&L2=Consumer+Legal+Resources&sid=Eoca&b=terminalcontent&f=30_day_demand_letter&csid=Eoca

Give them an ultimatum to reply by 10 days via email, or whatever is customary in your state; typically its 2 weeks, or 5 business days x 2 = 10 days, implied business days.  Gently threaten them with also involving the BBB and the state Attorney General.

If you get no reply by the deadline, formally complain to the Better Business Bureau if they are a member, and the Attorney General of the state where the company is incorporated.

http://www.ohioattorneygeneral.gov/Consumers

At the same time, gather the paperwork to file a small claim lawsuit or via mediation.  Different states, different rules.  Here's one.

http://courts.phila.gov/pdf/brochures/small-claims-court.pdf

In general, the threat of a lawsuit resolves the issue 99% of the time with companies of some repute, because a large company lawyer's time will cost more that the value of a small claims suit and most can't afford bad publicity, even if the company is in the right.  BBB and complaints or suits go on their record, even if you haven't been able to collect, and this affects their chances from getting contracts with gov't or large companies, who will research their records for 'good faith dealings' with consumers and businesses.

Wait until you get a subpena to send them to show your intent incase your real intent is just a bluff.  The cost to file a small claim suit varies by state, but its typically ~ $40-100; its generally a DIY lawsuit, no lawyer needed and its fairly easy to do, once you get used to it.  Add to it the stamps and your time, determine if the effort is cost effective for you; for $1000 item, spending $100 to get $900 back may be worth it.

[alm]

In general, the threat of a lawsuit resolves the issue 99% of the time with companies of some repute, because a large company lawyer's time will cost more that the value of a small claims suit and most can't afford bad publicity, even if the company is in the right.

Gotta love the US.

Is it actually that clear cut that this is a warranty issue? Does it fail to perform to spec? Sure, it's ugly, it is a sign of shoddy quality and lack of attention to detail. There have to be reasons why Rigol is much cheaper than Agilent, I guess QC is one of them. Customer service is another, apparently.

[Bored@Work]

Is it actually that clear cut that this is a warranty issue? Does it fail to perform to spec?

Such questions make lawyers rich. I don't know US laws, and I am not a lawyer, but from some issues with suppliers I know that lawyers here like to debate implied quality, and what defines it. E.g. state of the art of workmanship.

[amspire]

I still don't get it.

I completely understand that you want something that looks like it should be in the Smithsonian as a reference for perfect construction.

But for me, before I got myself really worked up and upset, I would still ask myself "Is there a real problem?". I am not sure you have shown us anything yet to prove there is a real problem.

The unit works.  It has had a manually worked fix, which is much, much better then having a pristine unit without the fix. It looks like someone didn't clean the flux off the board after the repair, but that can be corrected in 20 seconds.

If you send it back, then what? Buy an Agilent at double the price? If you got a replacement, what if it was perfectly constructed, but didn't have the fix? 

If I didn't like the way the wire and the capacitor was soldered, I would probably replace the wire, and resolder the capacitor, clean the flux off and then I would feel 100% satisfied. The skewed capacitors - so what?

Sure it might one day fail, but that can happen anyway.

Richard.

[saturation]

Its the game here unless someone changes it.  Legal is just another tool in business.  Apple, HP, Google etc., are suing or being sued for one reason or another [ patent disputes to disgruntled employees], its the same with most companies in varying degrees.  Once you get used to it, you won't fear it, unless you are about to lose your shirt  because you opted not to settle, which most cases go to.

In general, the threat of a lawsuit resolves the issue 99% of the time with companies of some repute, because a large company lawyer's time will cost more that the value of a small claims suit and most can't afford bad publicity, even if the company is in the right.

Gotta love the US.

[0xdeadbeef]

While this thread kinda focuses on the build quality, I wonder how good the signal generator actually works.
I'm especially interested in the arbitrary part and rectangular signals and how's the general signal quality.
 
I.e. what is the magnitude of the jitter (DC/period) on a rectangular signal in the few kHz range?
Also I'd like to know if the device is able to load arbitrary signals via the front USB slot from simple text files (like ARB files on the Agilent generators). I figure the UltraWave software is able to import/export to Txt and CSV files, but I would assume I have to convert them to the binary RDF format to load then directly from an USB stick. Is that correct?

 If so: is something known about that binary format? Though some details would have to be clarified, it seems to store the values as 16bit unsigned values (little endian). Which means it stores neither the voltage level nor the frequency, while it does so in the CSV and text files. There are also some erratic values in the RDF files. Like when you create a straight PWM and the values should be only either 0x3FFF (5V) or 0 (-5V), there are single values of 0xbfff in the "high" values and single 0x80 values in the "low" range. If you change the "Quantify" setting to 32, the low values all become 0x80 and the high values become all 0xbfff. Which is a little weird, but yeah well...

Besides, it's interesting to see that the Ultrawave software from Rigol.com crashes when trying to install and that from Rigolusa.com is dated from 2006 and refuses to load the CSV files it saved. Is there a working newer version available?

[EDIT]
Ok, I just realized that the DC is limited to 20-80% for <8MHz (even worse above that), so this makes it useless anyway...
It really seems that anything below the Agilent 33521A is somewhat crappy.