EEVblog Electronics Community Forum
EEVblog => EEVblog Specific => Topic started by: mariush on March 19, 2013, 11:02:29 pm
-
EEVblog #440 - Atten PPS3205T-3S Triple Output Power Supply Teardown (https://www.youtube.com/watch?v=Ju9Q05y_3V0#ws)
-
Sorry to hear about the camera trouble.
And thanks a lot for redoing the shoot, much appreciated. :-+
-
His voice at the beginning talking about metal support rods, especially the words "although it's a hope", have a tone of pure despair.
-
Dave, I hope you appreciate the irony in using your Atten hot air smd rework station to fix up some of the dodgy joints in the Atten PS :-//
Thanks for the perseverance given the camera issues.
Rob
-
The quality of the boards is just... horrible.
Yeah, give them credit for actually using connectors and gluing the cables but so ugly surface mount..
The 10000uF capacitors are probably fake, the quality of the text is just bad.
I saw some Aishi capacitors somewhere on the front panel which are low quality in general and the rest of the caps spread on boards are equally no-name - I doubt they would have paid about 6-10$ a piece (http://uk.farnell.com/united-chemi-con/ekmh630vsn103ma50t/capacitor-alum-elec-10000uf-63v/dp/1682011) for those 10000uF capacitors
Wouldn't it make more sense to turn the two boards 180 degrees and have the heatsinks at the sides of the case, with a lower speed/ more silent fans?
This way each heatsink and fan would have clear path all the way to the front of the case and the board for the 6v output would also be less cooked.
Not to mention the two boards could then be merged in the center and they'd save some money on the boards and some components like linear regulators that could be shared.
I guess it costs too much to stamp another hole in the case and pay for a second fan, or something like that. I don't see any other reason...
-
Hard to say what exactly happened with the files but i'm guessing the camera records onto a SDcard or similar.
90% of the time those kind of problems are caused by failings in the fat32 filesystem, it's really prone to damage when not unplugged/unpowered gracefully.
The real kicker is the damage doesn't always show up when it happens, it's not until later on (sometimes months later) when files begin corrupting.
-
Another interesting video Dave. It looks that the negative lead of the 100uF, 16V capacitor at the 31:32 point in the video is unsoldered (unless it is just a weird black reflection in their shiny solder joint).
Frank
ps Maybe you could dump the video clips from the camera to your PC (if it is near you) at regular intervals so that you wouldn't lose too much footage if the camera gets wacky. I'm always having problems with corrupted thumb drives.
-
Dave, not my original quote, but here it is, just the same:
"Technology" is simply stuff that doesn't quite work right, yet. There was a time when wheels were technology, but they've been debugged now.
ATTEN you MUPPETS! You've very nearly got a good product here, but you've screwed up. Fix all the stuff Dave's told you about, and come back and tell us when you have.
-
I would think that the biggest parameter that a fake supply cap would not be able to match to the original would be it's Equivalent Series Resistance (ESR). Dave, if you have the strength, can you measure one or two of those "ChemiCons" <snigger>
To measure a "true" heatsink temp with a non-contact IR device, simply stick a 25mm square piece of black insulating tape on the heatsink, and measure that. Works for me.
-
Interesting that the SIL pad for the 5V series pass transistor also has heatsink compound on it.
I was always under the impression that you shouldn't use compound in conjunction with a SIL pad. All it does is add thickness and impede the thermal transfer.
-
Interesting that the SIL pad for the 5V series pass transistor also has heatsink compound on it.
I was always under the impression that you shouldn't use compound in conjunction with a SIL pad. All it does is add thickness and impede the thermal transfer.
Is the SIL pad even worth using? Since the transistor is screwed to the heatsink, it's still got the screw to conduct through...? ???
-
There're likely bushings on the screw to prevent contact.
-
Yep. Those caps are fake big time.
-
Interesting that the SIL pad for the 5V series pass transistor also has heatsink compound on it.
I was always under the impression that you shouldn't use compound in conjunction with a SIL pad. All it does is add thickness and impede the thermal transfer.
Is the SIL pad even worth using? Since the transistor is screwed to the heatsink, it's still got the screw to conduct through...? ???
The SIL pad provides the insulation for the metal backing of the transistor. The hole in that particular package is insulated. Photo of 2 pin version attached.
-
I was REALLY hoping you would disassemble the bloody thing and remove that damn eyelash from the display. Seeing it irritates me. :scared:
-
To measure a "true" heatsink temp with a non-contact IR device, simply stick a 25mm square piece of black insulating tape on the heatsink, and measure that. Works for me.
I was thinking the same thing watching the video. Doesn't even have to be black tape - any colour electrician's tape works as well, or even paper-based masking tape. They are all black at the IR frequencies the temp meter works at. Kapton tape is probably better. Even a black waterproof marking pen does a good job.
-
Thanks for redoing the video so we could see the insides.
My end thought on this is that the engineers wanted to do a god job but after the whole thing went to shit. I will look elsewhere because the user interface is crazy, and the build quality had many issues. Too bad that the whole thing got let down by the idiots building and programming it after the hardware design was done.
-
Thanks for the Teardown. I think that Atten use two 2SD1047 in parallel for the 30 V 5A supplies because of the Safe Operating Area limitations. The ST Microelectronics data sheet Figure 2 shows 2 A at 50 V for DC. I guess that is with the junction temperature = 25 degrees C. I guess that with the transformer tap switching it should only ever be subjected to those conditions for a few ms so there is some margin for the Safe Operating Area getting smaller at higher junction temperatures.
-
Hmm, are those thermostats normally open or normally closed? (Yes, it ought to be normally open with that configuration, but that's why I'm asking...)
If you were trying to get a full load on CH2/3, shouldn't you have used the lowest setting on the CH3 rail to maximize the drop over the transistor? I see you're using the resistor for this though. So maybe even just put it into constant current mode byt shorting it out.
-
Hmm, are those thermostats normally open or normally closed? (Yes, it ought to be normally open with that configuration, but that's why I'm asking...)
I was wondering the same, usually these things are NC but that would not make sense in this configuration.
-
Searching for "KSD302" shows N.C. and N.O. types.
Alexander.
-
I understand the camera rant in this video, but it was a bit more than a little over the top. Also it seems more likely the memory card is to blame than the Canon, but I digress. Thank you for redoing all the video regardless - it's much appreciated by us fans of the show.
-
I understand the camera rant in this video, but it was a bit more than a little over the top.
Eh, who cares? It was funny, and felt good for anyone who's had something similar happen.
-
I was looking forward to you measuring the temperature of that electrolytic cap that's way too close to the heat sink.
TI Power Tip 50: Avoid aluminum electrolytic capacitor pitfalls
Power Tip 50: Avoid aluminum electrolytic capacitor pitfalls (https://www.youtube.com/watch?v=QgtLYSkczSg#ws)
-
Oh yea - regarding the cap that was "way too close". Electrically that may still be the case for adequate isolation (I don't know), but thermally speaking - it's at the opening to the (channeled) air flow for the heat sinks. If anything that should be one of the best cooled capacitors in the case regardless of proximity to the heatsink. The leading face of that heatsink should also be the lowest temperature point on it and emissivities of both shiny parts are not going to be great so thermal radiation isn't going to be terribly significant.
The massive cap near the side of the heat sink on the other hand may have a nice gap, but it will experience basically zero air flow for cooling. I would expect that cap to get much warmer than the one in the center of cooling air flow path.
-
I believe that another capacitor killer is heat conduction through the PCB trace from the rectifiers and power transistors directly to the capacitor terminals.
-
I always change the card out after around 100 erase cycles, never trust that the card will go anywhere close to rated life, especially with a FAT filesystem that can do multiple writes to the FAT per block of data written to the card. The cards are cheap enough, especially if you do as I do and go for the lower capacity cards, and keep a few on hand. I still have a half dozen 512M cards that I got for $1 each on sale, used for older cameras that only understand simpler formats. I just use 4G or 8G cards in my camera, keep a spare always in the bag.
-
I understand the camera rant in this video, but it was a bit more than a little over the top.
I sympathize with Dave on this one; the few times I lost an entire morning or afternoon's work due to equipment failure I had the urge to scream my frustration to the world. The difference is that he can do that (as he has a Youtube channel) and I can't (or wouldn't create a Youtube channel just for that... :)
As others mentioned, I would also suspect the card as well. Years and years of suffering with corrupt data and cheap brands of flash memory cards, nowadays I tend to stick only with SanDisk and Kingston models coming from reputable suppliers (too many fakes around, similarly to the elec cap on the supply). They seem to have better endurance.
-
They did a pretty good job on hardware design. Everything seems to be done properly. I actually think those engineers have much harder life than for example those in Agilent or Tek, because they are pretty much limited to industry standard parts. I guess if that was an Agilent you would see at least OP07 or some expensive, high-spec opamps from LT or AD or even some ASIC (nice to have an in-house ASIC fab, right? O0). Agilent/Tek doesn't care, potential customer is prepared to pay a shitload of dollars for the device. I bet it took those Atten guys quite a lot of time to design this, according to "on time, on budget, working... Pick two" rule.
I think that those opamp/dac/msp430 boards may have been actually hand soldered. In such design you typically find a broad range of different component values, and maybe the cheap-ass assembly house had only a very old pick and place equipment with very limited number of feeders. And it's much harder to overheat resistor or cap than opamp or microcontroller. So they probably assembled ICs on pick and place and then got some minimum wage (in chinese meaning) workers to hand solder the rest. That would explain difference in soldering quality between ICs and passives.
MSP430F2013 is a very nice chip. I have used it before for temperature readout and stabilisation. Internal reference is quite good too. On the other hand clocking setup is pain in the ass, as it has like 6 different clock signals for all sorts of power saving schemes.
-
you replace a SD card after 100 uses seanB?
damn man. Then again I am the type of person that washes ziplock bags. :-DD
I am using my cards which are like.... years old.
Also, why are they using ultra low power MCU in a linear supply? I find the thought amusing.
-
From what I have seen so far, on paper the ugly Rigol DP832 (http://www.tequipment.net/RigolDP832.html) at $409.00 has this thing beat by a mile even though only 3 amp output. The user interface is certainly in another world even if it is the sick pink/orange color.
-
you replace a SD card after 100 uses seanB?
damn man. Then again I am the type of person that washes ziplock bags. :-DD
I am using my cards which are like.... years old.
I have been burnt by cards that died suddenly. Only one that I have used hundreds of times is a Sony MS in a camera, not going to pay the price for a new one, even though it is an 8M memory stick. I once did an experimnent with a Kingston SD card on my EEEPC, used it as the drive running Ubuntu, including swap partition. Left it running overnight, and did nothing with it, just running in idle. After 2 days looked at screen to see a kernel panic message about being unable to access swap space. Card dead, bad blocks all throughout. New card with Puppy Linux then went in it's place. Now I can use the card slot again, all runs in RAM.
-
What do you think about the placement of the circuitry for Ch3?
To me it seemed almost like an afterthought, they first designed the PSU with two channels, then at last minute the usual manager/marketeer from hell show up and claimed that they must squeeze a third channel in, and so instead of re-design the whole thing they just found the few free cm3 where to fit the additional hardware.
-
From what I have seen so far, on paper the ugly Rigol DP832 (http://www.tequipment.net/RigolDP832.html) at $409.00 has this thing beat by a mile even though only 3 amp output. The user interface is certainly in another world even if it is the sick pink/orange color.
What the hell! This looks like one of those awful Sony stereos:
(http://www.tequipment.net/ProductImages/rigol/dp832/media/dp832_image_1.jpg)
(http://static.betazeta.com/www.fayerwayer.com/up/2011/07/genezi.jpg)
But hey, at least the user interface is actually usable!
-
What the hell! This looks like one of those awful Sony stereos:
http://www.tequipment.net/ProductImages/rigol/dp832/media/dp832_image_1.jpg (http://www.tequipment.net/ProductImages/rigol/dp832/media/dp832_image_1.jpg)
But hey, at least the user interface is actually usable!
WOW! It is a mixture of an old rotary telephone with the Space Shuttle cockpit!
-
Dave, I feel sorry for the lost files.
IMO the inner design of the PS matches with the product performance; on the other hand, the construction quality matches that of the firmware. A real pity.
Did you see the quality of the soldering of the big components? As horrible as the SMD ones.
Charlie.
-
It certainly looked like a fairly decent product. Too bad about the crappy UI. Dave, what other supplies did you consider before buying the ATTEN unit? The nice thing about the Internet is it often gives you the chance to find out about a product and discover any problems it might have (like a bad UI) before you buy it. If there were no reviews pointing out the problems with the unit before you got one, at least the EEVBlog teardown will warn others about the product.
-
It certainly looked like a fairly decent product. Too bad about the crappy UI. Dave, what other supplies did you consider before buying the ATTEN unit?
None really.
I hadn't seen another precision triple output supply for anywhere near that price in Oz. I wanted to see if it was too good to be true.
-
Hi -
Would anyone have a schematic or service information for either this unit (Atten PPS3205T-3S) or better - the CSI version (CSIPPS55S)? The channel 1 output on mine has died (possibly the relay) and I'd like to get some information before I dive in. CSI had no information and Atten (3 different offices) never replied.
Thanks in advance.
-
Did someone upload your video again with your permission?
Someone reuploads a whole bunch of your videos, plus other youtube creators and adds spam...
https://www.youtube.com/channel/UCCWPDYbCdcxxWda0VEFhaQg/videos (https://www.youtube.com/channel/UCCWPDYbCdcxxWda0VEFhaQg/videos)