Start Up Fails When HDD Connected

I agree. I can see that if the SM cap is leaking a tiny bit of current then the FET might not have enough gate volts to turn on fully, and then higher peak demand on the FET by a busy HD could cause a problem.

I don't have the right tools to hand to guarantee success and just tagging it "seems" to work. But the second machine I fixed has died again while on soak test so I thought I'd do as you recommended, and since it is only a spare what have I got to lose?

First pic is with SM cap removed, phew, and next pic is with another new 4.7uF leaded cap.

But... I should have checked all the supplies to the HD first because the +5V is missing, while the +12V is switching OK. Both are OK leaving the PSU card, so I need to find the +5V control chip. I remember it has been mentioned in this thread. Off for a look.
It looks like this one needs a new 5V chip. Input supply is 5.81V, but the feedback control pin is only 0.77V when it would be 1.25V if regulating correctly.

It looks tricky to remove without also removing the plastic HDD power plug so close. I need to buy more gas for my pyropen.

I may be some time fixing this one. :(
 
That's interesting, because it sounds more like the fault I was looking into in the first place, and which got overtaken by the capacitor fix.
 
That's interesting, because it sounds more like the fault I was looking into in the first place, and which got overtaken by the capacitor fix.

I fear I may not find enough time or enthusiasm to fix T2 no.2 as a total dismanling is needed to remove the HDD power plug, not a quick fix.

However this evening's exercise was not entirely futile. Taking off the 4.7uF capacitor was very much easier than I expected.

I fiddled about for a while on T2 no.2 masking off the surroundings from the SM C, then I had a lightbulb moment and realised how very thin the capacitor's internal wires would be compared with its SM tags soldered to the PCB pads.
So I gently, gently rotated the can back and forth on axis, no great force and it gradually eased, travelled further each twist, and eventually snapped off. The wires inside it are a mere 0.2mm according to my calipers.

Then it was easy to clean off the remaining SM tags from the PCB pads with couple of quick touches from a trusty old Weller TCP.

I've re-done T2 no.1 properly now. A fifteen minute job including unplugging and re-plugging and shaping the leads on the new C. OK so far.
 
So I gently, gently rotated the can back and forth on axis, no great force and it gradually eased, travelled further each twist, and eventually snapped off.
Nice work. Always worth considering the best way to remove a dud component. I used a similar technique recently after the Tektronix PAL vectorscope I was refurbishing made an interesting noise and started emitting copious quantities of the magic smoke - a filter cap across the mains input had died. I didn't fancy the work involved in removing the large PCB so waggled the cap until its leads fractured, then soldered a new (Class X2 type) to the plated-through holes (which I knew I'd find beneath). The board cleaned-up quite nicely.
 

Attachments

  • 11.JPG
    11.JPG
    334 KB · Views: 21
  • 8.JPG
    8.JPG
    324.8 KB · Views: 21
  • 2.JPG
    2.JPG
    320.2 KB · Views: 22
I used a similar technique recently after the Tektronix PAL vectorscope I was refurbishing
No diff phase there then, although the burst phase is a bit off. Is that Set90 or Set180 that needs a tweak... I forget now.
started emitting copious quantities of the magic smoke - a filter cap across the mains input had died...
Mother's food mixer/beater thingy suffered the exact same problem. She turned it off at the wall rather than on the device (she almost always did), then phoned me asking what to do and whether to try turning it on again. I said NO!
then soldered a new (Class X2 type)
Having removed the defective C, I was amazed to find the exact same part in the spares drawer. Fitted it and been working ever since, which is sadly more than can be said about her cooking days.
 
although the burst phase is a bit off
Yes, well spotted! The bars were coming from a Snell & Wilcox TBC11P set to output a test signal (also does multiburst) - could be its coder needs a tweak, rather than a problem with the vectorscope (which I haven't tweaked at all).
 
So cutting through the cap above the base with long-nosed snips would work?
Maybe. But I think a lot depends on how thick the wires are and how messy cutting through the cap might be.
Here's s pic of two twisted off SM caps and their bases (less tabs) next to a couple of junk-box mini-caps I had tagged on previously.
You can see how thin the wires are, and a couple still with a tiny bit of solder from the tabs.
 

Attachments

  • 20230308_131855_c.jpg
    20230308_131855_c.jpg
    134.8 KB · Views: 23
t looks like this one needs a new 5V chip. Input supply is 5.81V, but the feedback control pin is only 0.77V when it would be 1.25V if regulating correctly.
I had another look at T2 no.2 after studying the LDO reg data sheet again, and the feedback pin 5 is doing the right thing because the control pin 1 is low. I'm not going to be able to find where this signal comes from without a circuit.

Even without HDD this signal never appears. The T2 never gets as far as lighting front panel or showing boot loader on HDMI output. So there is an upstream problem in system controls.

Alas my diagnostic skills are down to about 1% of what they were thirty years ago.

The sad thing is that it did sort of work when I was given it, and I've not done anything silly with it!

My mains socket includes a power meter and this gives quite a good indication of how far through the boot sequence a box travels. This T2 only reaches about 16W and stalls there.

My "working" T2 no.1 goes 16W (boot screen) 21W, 48W (briefly), and settles at 22W SD or 23W HD.
However while this one will always record on schedule from deep sleep (0.5W) with video output off (16W) it has an intermittent problem with the display output being misplaced vertically and offset horizontally differently above and below the "blanking" being shown on the raster! eg here
Just sometimes it will stick at 20W when told to shutdown.
 

Attachments

  • 20230309_144629_c.jpg
    20230309_144629_c.jpg
    160.5 KB · Views: 16
Maybe. But I think a lot depends on how thick the wires are and how messy cutting through the cap might be.
Here's s pic of two twisted off SM caps and their bases (less tabs) next to a couple of junk-box mini-caps I had tagged on previously.
You can see how thin the wires are, and a couple still with a tiny bit of solder from the tabs.
I used the cutting the top off the cap method because i feared damaging the board. My cutters are very small & sharp. But i just found a scrap humax board (DTR-T1010) with similar caps & used a gentle twist backwards & forwards & got good results on each cap. I didnt force anything & just continued till they came off.
 
I have been 'playing' with WiFi dongles, and have observed the HDR-FOX is able to put up an error message on-screen if (I believe) the USB port is drawing too much current. That means the multiple power management circuits are monitored by the SoC and it can react to faults, suggesting a detected fault on the HDD supply is triggering a deliberate reboot.
 
My solution was to snip the top off of the cap in several goes with a very small pair of wire cutters & cut the side of the can, clearing out the insides & snipping the leads before pulling the remains off of the board.
I did much the same. Cut the top off, then gently pulled the rest using a pair of pliers and it came away from the board with very little force needed. So no damage and no mess, just leaving the soldered connections broken off at board level not causing any problems. I didn't fancy applying heat to remove them because of the proximity of those absolutely miniscule resistors.
It would be interesting to discover the outcome if someone who has a unit with this fault could just remove the 4.7uF capacitor and see what happens.
I didn't bother replacing it as I didn't see what it would achieve.
Yes, my faulty one doesn't totally match either yours or Newcoppiceman's.
It seems like it's the same problem though as it has been completely stable for over a week now. Previously it either wouldn't switch on at all (if warm) or would crash within 2 minutes, making it completely unusable. I guess the leak though the C was enough to kill the 12V disk power mostly all the time.
So either delaying HDD +12V switching off is vital or it's been copied without thinking from earlier designs, which happens.
It seems like it is not vital. I struggled to think how it would be.
 
Only a failed mathematician with a soldering iron here, but isn't it a soft-start thing to protect the power MOSFET?

Let's say that the SoC is not calling for the HDD, so in the circuit posted above the 12V input is on but the NPN transistor is off. The MOSFET voltages would be G, S, D = 12V, 12V, 0V.

When the SoC turns on the HDD, the NPN transistor will be on. The MOSFET voltages would be roughly G, S, D = 2V, 12V, 12V.

Getting from the first state to the second means charging the relatively large output capacitor without burning out (or over-stressing - device current should be < 4A cont/20A pulsed, according to spec (PDF)) the MOSFET's power channel. Since a resistor is not wanted in the power line, the mystery cap ensures that the MOSFET turns on gradually, first charging the o/p cap and then supplying the HDD. The time constant is1/RC ~ 200s, but the MOSFET should be on for any G voltage < 9V. The parallel 10kΩ resistor is potentially over-sized but 10V between S and G roughly halves the S-D resistance (to ~ 1Ω, maybe 0.5W with a 2TB drive), as well as, but unimportantly, the bias circuit power consumption, to ~ 12mW).

Shutdown/power-off could happen with the SoC turning off the NPN transistor, or with the power being cut. In the first case, the mystery cap gradually discharges. turning off the MOSFET; the o/p cap will discharge though the HDD If the power is cut, the sizes of all the many storage capacitors and of the cap at the base of the NPN transistor will become significant. At a minimum, the mystery cap could limit any discharge current from the o/p cap(s) through the MOSFET's reverse diode (must be < 1.3A) if the lost 12V input starts to show a low-resistance path to ground.
 
Last edited:
I've had the same problem as others with the cycling of the
POST sequence about fifteen times before returning to its standby state and the problem evaporating if the HDD poser plug was disconnected.

Thanks to @Newcoppiceman's hard endeavours to trace the problem I've now replace the capacitor as detailed by others & it all appears to be working a normal. I de-soldered the inner side of the original cap & was able to pry it away from the board with a cocktail stink. I saw no evidence of glue although it appears I left the iron near the plastic base a bit too long & partially melted the base. I the slid the iron down the other side of the cap to ensure I didn't accidentality touch the other components & with a quick dab I was able to flick the capacitor out of the way. I then retouched the solder pads on the PCB with some fresh solder, tinned the bent legs of the new 4.7uf/35 capacitor & fitted it in place as shown in other examples.

Thanks to all who helped find a solution.
.
 
I believe this might solve my issue but I’m not someone who repairs tech usually any suggestions where I could go to get it fixed etc? Or if anyone here can help fix it maybe?
 
I believe this might solve my issue but I’m not someone who repairs tech usually any suggestions where I could go to get it fixed etc? Or if anyone here can help fix it maybe?
If you've never wielded a soldering iron yourself, your best bet is probably a community repair shop or similar.
 
Back
Top