Alllllll - riiiiiiight..... worked a bit on the thing last night...
RecappingI received a bunch of caps from TME. They have really a great choice of electrolytic caps. Even when narrowing down the search to only Nichicon and 105°C rated caps, I still had the choice between several series of caps ! So what to choose ?! A normal person would just pick the cheapest, or roll a dice.. but of course an electronic guy has to make rational/informed decisions, so I had no choice but to download the data sheet for all these series I was offered. Mostly VX, VY, PM and PW series, then compare them and try to figure out what would be best.
I am hardly an expert so was not straightforward... but in the end I decided to go for the PW series where possible, else pick PM, but leave the VX and VY alone. VX and VY look like your cheap and short life high losses filter caps for linear supplies. The 'P' "series of series" so to speak, looks more like it's designed as low ESR for SMPS use, and longer life/endurance. If anything, PM/PW data sheets give you the actual ESR, down to the milli-ohm, for each and every voltage/capacitance combination possible... whereas the VX/VY series don't. They just give you a rough "losses" factor for the entire range of caps, no detail.
Then, when you open a datasheet (all 4 attached below) , they show you a little diagram that links the series to other series... so there you learn about even more series that you didn't even know existed ! More confusion ! Who knew there could be so many different models/types of electrolytic caps ?!
So to decide between PM or PW, the impression I got was that PW was the slightly improved version of PM, about to supersede it. Slightly lower ESR, longer life, smaller case... though if you look at the tables/hard facts, at least for the caps I was concerned with, the PW was exactly the same package/dimension, not smaller, and the ESR was so marginally smaller that it was not a decisive factor. Like 29mohms vs 30, IIRC.
Still, it was a bit better and meant to replace PM short term. Also, the PW was often much cheaper than the "old" PM.
Then for some of the caps, I swa a 'PS' series, which cost 10 times more than the PW... IIRC the PS is so outrageously expensive mostly because of its astonishingly high life expectancy, from memory, something like 20,000 hours when the PS and PW are in the 5000/8000 range, and the VX/VY series in the 1,000 range...
So, basically, not all electrolytic caps are the same...you get what you pay for and make an informed decision.
So , I ordered mostly PW, and PM only when the PW was either not available, or its minimum order quantity was like 100 or 500...
Sorry for the long paragraph but well... that's actually what one has to go through when ordering something as simple as caps... nothing is really simple, the more choice you are faced with, the more time you end up spending figuring things out...
Reading the manufacturer datasheets closely is a good way to learn about the subject, when like me you don't know all that much !
I would be interested to see a technical paper from Nichicon or other good brands, that explains the gory details of just what's involved in making a cap last longer than another, and what's involved in achieving a 105 or even 125°C rated cap, compared to a common 85°C cap.
Some caps I ordered 20 of them because they were so small, literally 5 cents a piece, so 20 was the minimum order quantity. Well, I have dozens of caps in my drawers now, can't hurt.
So I replaced all those caps. Went pretty smoothly. Some were a bit of a gamble because the track was on the component side, so you have no way to be sure that the solder flowed through the hole to reach the underside of the cap... all you can do is heat the lead/pad for a decent amount of time to make sure it's well heated, so the solder can flow well and hopefully get to the other side of the PCB.. and add more solder than usual, to make it fill the hole and have some left to make the hidden joint.
I also replaced the 2 caps on the pre-reg board, the ones I had already replaced the other day for test purposes, with crappy caps from my local electronic shop.
Only cap I did not replace, as discussed earlier, is the big 450V mains filter cap. However I did order a couple anyway... just in case I somehow decide to replace it. This I will have them handy. These aren't cheap t 5 Euros a pop, but... even I don't use them on this scope, I could still reuse them later on no problem, since pretty much all SMPS supplies have a 450V electrolytic cap to filter the mains... so these caps aren't wasted, no worries.
That's for electrolytic caps. Sadly TME has an extremely limited selection of dipped ceramic caps, either that or I am blind and did not find them. So I have not yet replaced the 3 little axial green ceramic caps that revolve around the PWM chip. Will have to order these from some other place, yet to be determined...
Anyway, it didn't bother me too much because the scope was working just fine even with one of these caps missing, and a doubt on the health on the other two. Plus, the pre-reg board is easy to access later anyway, not much involved, so the caps can be replaced any time.
MOSFETSeeing as the SMPS (main board I mean) was fully recapped, I wanted to put all the shield and pre-reg board in place, so as to have it in it's nominal configuration before testing the supply rails. This meant I had to remember... how to put the FET assembly back together again ! Tried to improvise, no good. Luckily I had a coupe pictures taken during disassembly, that clearly showed how it's put together, phew. Never be afraid of taking loads of pics when you take stuff apart... it will be your savior 6 months later when comes time to put it all back together !
You can never have too much pics... you can never know what picture will be the one that might save your bacon later down the line !
So one thing leading to another.. the legs of the old FET, broke as I had to bend them to put the FET back into its plastic holder/box. Didn't think that soldering the pins would be very elegant nor anything to be proud of, so I took that opportunity to replace the FET with the modern replacement I had bought. I didn't find the usual IRF something model, so had to improvise and just use a parametric search and pick what might be suitable... I stumbled on some weirdly named part I had never heard about, it's a
IPP60R280P7 made by Infineon... I thought these guys only made memory chips ?! Apparently not... you live and learn.
PDF (too large to be attached) :
https://www.infineon.com/dgdl/Infineon-IPP60R280P7-DS-v02_00-EN.pdf?fileId=5546d4625b10283a015b1f973023047bThis part as one would guess, is at least no paper much better than the old IRF820, which was 2.5Amps and 3.0 ohms RDSon. The Infineon part can sustain at least 8 amps, and RDSon is literally an order of magnitude lower, at 0.28ohm. Sounds almost too good to be true, but well, they had the guts to put that in their datasheet so I guess there must at least some truth to it ! Hell, even if it's only half as good as they say, it would still be much better than the old IRF820 !
Later I realized one pesky detail, ahem... there is one little "detail" that the old IRF820 DOES feature, that I didn't think of taking into account when I search for a replacement : the IRF has an INSULATED tab ! It's not connected to the drain. Oh noooo... need to search for another replacement then... and spend hours pulling datasheets, because the parametric search engine does not offer to search for tab insulation, bummer... would have been too easy eh ?!
so, I evaluated the risk I was taking by using that my non-insulated replacement...
For one, the FET is not fastened to a heat sink ! Far from that : it's covered by a pretty thick (1mm or so) ceramic "pad", itself covered by what must be (?) some kind of EMI shield, which takes the form of a piece of FR4/fiber glass... that's yet another layer of insulation. So at least it looked like the tab would not make the chassis live and kill me, a good start. Then I looked at the possibility of the tab coming accidentally in contact with whatever in the scope, a lose wire, anything...
but everything in the area is tightly held, and also the FET is not out in the wild far from that : it's completely isolated in its little brown plastic holder/prison... it's covered all around, no way a lose wire, or even a lose screw, could get inside the holder and come to contact the live tab... ).
So, I decided I would solder my replacement FET in place, and put the SMPS all back together again, all the shield and mounting hardware, all in back in one piece again.
Honestly I don't understand how Tek though it was OK to leave the FET swithching the juice all day long, with no heatsink and zero airflow, not even still air.. literally sealed in this plastic holder ?!
More incredible is that despite this, and the high RDSon of 3ohm, hence lots of losses/heat... this FET did not melt !
Or maybe it did and was already replaced once... or twice, or thrice... or frice... LOL ! what should we say after thrice...
Anyway, at leas the modern replacement with it's 10 times lower RDSon, can only help ! Maybe this one won't fry... we shall see.
TestingFinally, the moment of truth : will it work, will it blow, will I get electrocuted, or or or or.... ?!
Luckily none of the above, I am glad to report !
Scope still alive. I checked all the rails one by one. First the reference rail, -8.6V, because all the other rails are tied to this one. It was well within spec so I did not even consider touching the trimmer to try to make it better. That rail has a very tight tolerance +/- 40mV, and I was down by 15mV. So I left it at that.
All the other rails measure well within spec as well, as could be expected. As for ripple, it was getting late last night, so didn't feel like firing a scope to get the real picture, so instead, I went the cheap route : just checked with a DMM in VAC mode, to get a taste for it. It's got a very decent RMS bandwidth of 100kHz, and the SMPS switches well below that, at 40kHz for the pre-reg, and only 20kHz for the inverter. So the DMM would get a fair share of the spectrum, the reading would hold water, I thought. Obviously the manual states Peak to Peak values, not RMS, but well as a rough approximation, I thought well, let's say the RMS shall not be more than half the P-P value, say. Hardly scientific, but well, you get the idea.
So did that... and the DMM showed.... ZERO millitvolt ! Not one. A rock solid, steady zero mV ?! Too good to be true. Couldn't possibly trust that, could I... so I tried another DMM (same specs), and it agreed ! Eh ?! Hmmmm....
So in the end I did eventually fire up a scope, my 2232 as usual, to get a taste of what was really going on.
I was not disappointed, my DMM did not lie to me ! Scope has much higher BW of course, so I do get some ripple, but so tiny ! Max sensitivity of the vertical amplifier is 2mV pr division, and well, I get about one division P-P ! 2mV P-P ! Geez ! The manual requires less than 10mV P-P on all the low voltage supplies (5V and symmetrical 8.6V), and I get 2mV only !
30V rail allows for a bit more ripple of course, up to 50mV, and I measure a grand maximum of 25mV worse case, only 15mV when averaging the waveform to get rid of the random crap and get a more stable waveform. The 100V supply can do with up to 200mV ripple, and I measured a bit less than 100mV. So again well within spec !
Rail Tolerance Measured P-P ripple Measured
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
-8.6V +/- 40mV -8,585V < 10mV 2mV
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
+5V +/- 0.25V 5.03V < 10mV 2mV
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
+8.6V +/- 260mV 8.54V < 10mV 2mV
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
+30V +/- 1.5V 30.2V < 50mV 15 to 25mV
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
+100V +/- 5V 100.3V < 200mV < 100mV
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
Every body likes to see waveforms right, so here they are.
Now I left the best bit for dessert, as you do !
The SMPS is now.... completely.... SILENT ! I don't mean to say "more quiet", or " a lot more quiet", or "whisper quiet"... no, it is become absolutely silent, nothing to be heard what so ever ! I didn't quit expect that ! Was as shocked as I was surprised and pleased !
I left it running for a fair bit of time, as I was taking measurements with the DMM and scope on all the rails... but it remained quiet. Powered it off after I had finished testing all the rails. Some time later I powered it back up again, thinking I might have just gotten lucky the first time, and that the noise would reappear.. but it did not !
So.... my main project with this scope was to precisely use it as a test bed to investigate the SMPS whine, but now that said whine has suddenly vanished... well !!! I think I might keep it anyway, for the time being at least, to get some repair experience. So I still intend on giving a shot at fixing the remaining issues with this scope.
So... was it the FET or the new caps that cured the whine ? No idea ! But it sure motivates me to do the long overdue recap on the SMPS of my 2232 ! When I fixed its SMPS last summer, after it blew, I was happy enough that I managed to get it back up and running (was overwhelmed and clueless, so thanks to the guys who helped me back then !), so I didn't feel like messing any more with it... I needed a rest ! LOL
I think now is a good time to recap it :-)
Also, worth noting : I noticed that the 2232 SMPS was unusually quiet and much quicker to go to near silence, than it used to be. Took it say 2 minutes to quiet down, versus maybe 15 or 20 minutes usually ! Only thing I can think of, is that last night when I used it, it was really, really very hot here !
So this is consistent with Tautech's supposition that the whine might be related to a warm up/temperature phenomenon. Based on what happened with the 2215, I guess it's fair to say that it's either the caps and/or the FET that are involved. Given that I had already replaced the FET in the 2232, looks like the caps might be the culprit then ! Could still be the FET I guess, but I have ordered two of them. So if recapping the 2232 doesn't make it quiet, I might try replacing the (new but different model) FET, with the same I used in the 2215. So some more experiments to come !
Some pics and straight to bed