TNP #25 - Keysight E36313A Triple Output Programmable DC Power Supply Teardown & Repair
Вставка
- Опубліковано 8 вер 2024
- In this episode Shahriar repairs a broken Keysight E36313A with a blown fuse and burnt smell coming from the unit. The teardown shows the dual-board design of the unit as well as the large toroidal transformer. The magic smoke has escaped from a dual Schottky diode power device with a Vf = 0.45V at If = 5A. The replacement eBay component is first verified to meet the specifications and installed in the power supply. The Keysight unit is then tested at a few output voltages and current to ensure proper functionality.
www.TheSignalPath.com
/ thesignalpath
www.Patreon.com/TheSignalPath
13 Watt standby for a modern piece of test gear with no fancy crystal oven to keep warm. That is not cool. It should have a mechanical power switch.
Makes me think there is something else wrong with it.
@@why78536 2 other comments measured similar consumptions on their units, so it's just a design oversight.
We are using "old" E36xx that are probably 20 years old by now and we had no issue with them whatsoever. From what I hear, the "new" E36xx just can't live up to that quality and reliability.
@@FooBar89 Yes, and it will still deliver that DC pulse if it loses AC power.
@@FooBar89 we tested several different supplies (as we decided not to buy the E36 anymore). It turns out, that most cheap supplies glitch during turn-on, the Rigol are loud (fan), R&S are too expensive for what they offer, but the Siglent supplies are actually really good. Stable no-bs supplies with LAN for a good price.
Nice when the problem reveals itself so clearly.
Shahriar, I measured my E36313A at 15 W when on standby, so that's at least two datapoints that this doesn't point to any other potential failure mode.
You mentioned doing a video on the shunt resistors and why they are shaped like that-
That would be interesting
Concur. Especially when combined with other non-trivial design issues with ultra-precision / high power resistors (tempco, material, moisture ingress, drift).
Excellent work once again! We have a few of these at work. Compared to the older E36XX series, it seems mostly like a face-lift with independently enabled channels and a graphical UI. The standby power consumption is quite horrible on these as well, the unit is actually warm when off and doing nothing. Always prefer mechanical mains power switches on gear like supplies.
Also, going to start videoing more repairs too. Just uploaded a repair of an MS2721B which I know you are also familiar with! 24 ripped pads on the digital board front panel connector.
Love the data logger on this model, nice to be able to see your V/I/P over time
Cool. I recently got one of these as well, with the rationale that its front panel / UI roughly matches Keysight's other offerings (DMM, signal generator, universal counter, etc.). One super nice thing about this PSU is that it is nearly silent. It is also really good at measuring low currents (great for IoT work).
Love the repair videos!
Yes, please do a video on those current shunt/sense resistors.
I bought one of these power supplies, and absolutely love it. Being able to auto-bond channels 2 and 3 in serial or parallel mode without jumpers is fantastic.
Lovely PSU, I like how it's built, good quality components and attention to detail. Thing of beauty, joy for ever.
Excellent work Shahriar !!! True professional procedure !!
Suggestion for our curiosity: On your next high energy equipment repair: Please add a FLIR shooting at full power (after servicing) so we can see which components have more thermal cycling and eventually more prone to failure.
Excellent video! I have one of these and indeed the standby power is terrible. But the supply is very good.
You can switch the model number in one of the setting menu which changes one of the 24v to a -24v supply
I lOVE repair videos! 😀
It is always interesting to watch videos about power electronics.👍
Great video. Wow so you can use the source meter as a "Curve Tracer". I wondered what replaced that old Tektronix oscilloscope type curve tracer of the past. Source meter are amazing but unfortunately outside of my price range...
What a great find, lucky!
Keysight quality is going down and I wonder who let these out for production. We bought some and tested the noise output using a spectrum analyzer. The noise performance of these is even worse than a $500 Rigol. Please may you do a noise comparison of power supplies when u have time :)
Excellent
I liked it enough after the firmware fix and the recall, that I bought a second one.
Great video and nice repair, as always!
When components blow up such way, it's easier to find by look and by nose.. 😂
change that 100nF RIFA cap !!!!
Nice video. I've been using one of these daily for the last 7 years without issue, however, the screen now is starting to go. Otherwise it's been a great PS and quiet, you can barley hear it when running.
They've only been around for 5 years.
@@jaro6985 you are correct, i checked my invoice, ws 2018 when i got one. makes me more worried as the screen is going on it in only 5 years.
Good short!
love it, ty Shahriar
I'm pausing the video at the start to post my guess on what failed: I bet it's the power supply!
Studying the analog boards, I was never able to pinpoint any DAC's, so I suspect that they're using the STM32 to generate PWM or possibly delta-sigma modulators to generate the V/I set points and limits.
Lovely repair! 😂
After the repair, run the supply to maximize the power dissipation through the dual-Schottky (high line?) and observe the device with a thermal camera. I think you will find that these twin Schottky rectifiers have a bad habit of a die thermal hot spot, along with the device high reverse leakage can at times let the smoke out. Maybe move up to a TO-247 power package for the dual Schottky it looks like it would just fit on the heatsink.
Edit: Using the SMU 2460 see what the device reverse leakage is at 100V reverse and see if this makes sense for the design, the pair together may runaway in reverse?
Those type transformers are really broadband - probably a voltage spike from the line transferred into that device. They need to increase the surge amp rating to double or triple of whatever it is now.
Muito interessante parabéns 👏👏
Is it still generally safe to use variacs for bringing up switch-mode supplied, microprocessor driven equipment? I heard rumours that you risk the switch controllers going into an undefined state and for example leaving the FETs/triacs on and overvolting the equipment. Any thoughts? Thanks in advance
That sounds like a rather poor design, even simple flybacks have undervoltage lock out. The grid itself is not exactly stable.
Even $5 bricks when ran on overloaded generator or through series lightbulb just do equivalent of power cycling on undervoltage, I don't think there's topology in commercially available product that would die from this.
Rifa X-cap in a new design?? 😳
That's a nice schottky.
bring it up carefully (on a variac?) on a programmable AC source.
I wonder if that 13 watt standby indicates another problem lurking. That seems really high for a modern piece of hardware, especially with Keysight's name on it.
Not really, just the no load magnetising and core losses of the transformer, as it is always powered on. Big high power toroid, will have some power use always. No need really for a CM choke as well, seeing as you have such a massive inductor in the primary side as well, though they could have replaced one with a NTC inrush limiter, and thus allowed the mains fuse to be a lower rating and not have it blow with power connection. However fuse did the job and blew when the diode pack shorted out.
I remember Dave Jones practically losing his mind over a scope using 6.5W in standby.
From my knowledge, Schottky diodes are quite sensitive to reverse voltage/current.
Maybe it doesn't have a good enough protection?
13W power consumption in standby. That is not good at all. A bench PSU doesn't even need standby in the first place. My one has a real mains switch and starts up instantly.
It certainly won't be ErP compliant. ErP is a european directive that says standby power consumption must be
@@simontay4851 I think that's only for consumer devices, so not all too relevant here.
@@barthettema7323 That's just as bad as Weller leaving out the mains fuse from some of their products on some legal technicallity. Seriously, nobody expects their modern, brand-name instruments to have this kind of standby consumption.
Where did you get that Bananafy converter for BK precision electronic load?
👍👍
Always looking for low forward voltage diodes.
Looked up the NTST40120CTG diode. Both Digikey and Mouser say it's obsolete.
Why would Keysight use it in a new design? Maybe is was still active at the products release.
This was designed over 5 years ago, so probably.
What was the original fault. I have had 2 of these and my second one has just failed, bang on switch on smoke but only ch1 has no output have you any service info, ps i am a tech btw, thanks
1:00 Interesting. Does the Green "ground" terminal imply that you can power the unit with up to a +/- 240V DC offset from ground? If so I don't think I've seen a unit that can do that. Usually that terminal is tied straight to ground ???
Yes it is at least 200vdc
I guess it means the channels can be connected up to that voltage(they are isolated) from ground, the limitation is probably the transformer with a big safety margin.
Most linear PSU can do the same, although it is not always specified in the manual.
Lightning strike?
7:22 Lower left corner of screen. A wisp of smoke escapes the shrink wrap on the white cable when the load is enabled.
There is no smoke. The shrink tube is just holding the cables together, there isn’t even a break in the cable.
@@Thesignalpath It's invisible on a small screen but on my low-end 24" monitor, it's visible at around 7:23. Between output 1 and output 2, there is a binding post for ground with a green ring. If you stare at that, the moment you say "five amps", a small wisp of smoke floats upwards and past binding post. I've become unsure of the origin but it could be the cable. Just to be clear, I'm open to the possibility that there is no smoke and I'm losing my mind. I just found this to be an amusing observation - nothing more.
Your test equipment tear down and repair content is of the highest caliber. Thank you for this channel and your hard work.
You're right, I see it too!
Can you drop a link to that “bananafy” thing you have attached to your load unit? I could really use one, but my search-fu is failing me.
A few options: www.tindie.com/search/?q=DC+electronic+load+banana
Damn you are lucky
There is no way i can find broken stuff like this in Portugal 😒
Getting it from eBayUS is not worth the ship and tax. 😔
why all these colors? I hope this Power Supply won't suddenly identify as a Waveform Generator
The colors represent channels, and they match the ones on my DSOX1204G scope: Ch1 = Yellow, Ch2 = Green, Ch3 = Blue