Philips PM3070 100MHz Scope Repair No.2

[dazz1]

Hi
Thanks to Tautech I have another couple of scopes to resurrect. 
One is another PM3070.  Older than the No.1 and a little more wear and tear but from the same organisation.

I am now comfortably familiar with this model scope and it only took a few minutes to remove the power supply board.

[dazz1]

One of the first things I noticed was a small daughter board.  This is an option that outputs the:

• Main Time Base gate,
• Delayed Time Base gate and
• Main Time Base sweep

to BNC connectors on the rear.   My working No.1 PM3070 doesn't have that option so I am going to move it across.  I don't expect I will ever use these outputs, but not hard to transplant them.

[dazz1]

So on an initial inspection, I might have found a couple of issues.
What do you think?

[dazz1]

All of the Rifa caps were cracked and on their way to failure.

There was also another part that didn't look like it was all there.

[dazz1]

And the fan was missing.  It looks like it was never fitted.

[dazz1]

So now I need to figure out why the NTC blew out. 
It had to be feeding into a sustained short circuit or similar. 
The attachment shows a simplified schematic of the power converter.  The NTC resistor supplies T6001. 
Occam's razor says the starting hypothesis is that V6014 or V6018 has failed to a short circuit causing excessive current in the NTC. 

Fixing this is going to take a while because I need to buy parts in batches to avoid paying shipping costs.  So I expect to run projects in parallel until I have a long enough list of parts to make a purchase.

[dazz1]

OK so I checked the MOSFET and nearby NPN transistor today (V6014, V6018).  Both measure OK. 
I also checked all the diodes, and they check out OK. 

Something has drawn enough current to burn out the NTC anti-surge resistor R6001.  Now I am looking at this and wondering if the two DC Caps C6007, C6008 are faulty.  There is no obvious outward sign of heating or leaking.  I have noticed that the blue plastic sheaf has shrunk.  This could be old age or it could be heat. 

It is possible that the NTC resistor simply died alone but I don't have a spare.  If there are other faults, I want to find them before ordering parts.

[macboy]

It's rare for electrolytic caps to fall short but possible. See a thread I posted a few weeks ago, I had a power supply with a dead-shorted 1500uF 200 V Philips brand electrolytic. It took out a triac and a fuse. This supply has a PTC (Not NTC) for inrush protection, bypassed by the triac after a short time.
https://www.eevblog.com/forum/repair/repair-of-tti-tsx1820a-power-supply-(spoiler-shorted-electrolytic)/

[dazz1]

So another step taken in diagnosing the burnt out NTC thermister.

I set 10V to the 320VDC with a 0.3A limit.  When the current was limited, the voltage was limited to 5V.    That is 16ohm. 
With 320VDC, the current would be 20A.  So maybe that is what roasted the NTC.   This could be a red hearing because of the non-linear semi-conductor devices.

All of the components in pink hi-lighter are lifted to take them out of circuit. I wanted to isolate the power switching circuit from the DC smoothing caps. I split the circuit to see if the smoothing caps were a problem.

This scope hasn't been powered up for a very very long time so I started by reforming the capacitors. 
I started at 10V with a 0.1A current limit, and gradually ramped up the voltage until I maxed the power supply, at about 55V.  Enough to get the electro-chemistry right. 
The current was effectively zero.  So the caps are OK. 

I checked the really badly discoloured leaking cap on the secondary side that I had removed.  This had failed to an open circuit, so could not cause the NTC to burn out.

I also need to check the transformer. If this is faulty, the scope is a brick.  There is no way I could fix or replace a faulty transformer.

[dazz1]

It's rare for electrolytic caps to fall short but possible. See a thread I posted a few weeks ago, I had a power supply with a dead-shorted 1500uF 200 V Philips brand electrolytic. It took out a triac and a fuse.

Interesting.  I needed to eliminate cap failure as a cause.  Right now, the caps are sitting at 60VDC and practically zero current. They are OK.

This supply has a PTC (Not NTC) for inrush protection, bypassed by the triac after a short time.

Usually circuits have an NTC to anti-surge.  When the thermister is cold (max resistance) and the power turned on, the cap charging current is limited to something reasonable.
As the NTC warms up (reduced resistance) the power loss is less.

https://www.eevblog.com/forum/repair/repair-of-tti-tsx1820a-power-supply-(spoiler-shorted-electrolytic)/

[dazz1]

https://www.eevblog.com/forum/repair/repair-of-tti-tsx1820a-power-supply-(spoiler-shorted-electrolytic)/

I don't like putting a light bulb in series with the power supply for switch mode power supplies.   The main problem is that the feedback loop will try and supply full voltage and current output.    If the input voltage is half, the input current will be 2x.  That could fry parts that weren't fried before.

 Good SMPS will have undervoltage protection to stop the power supply running until the input voltage is above a threshold.  That could make it look like the power supply is faulty.

For my power supply, I have taken out enough parts to prevent the power supply switching.  At present, there is no current flowing in the switching circuit.

[tautech]

https://www.eevblog.com/forum/repair/repair-of-tti-tsx1820a-power-supply-(spoiler-shorted-electrolytic)/

I don't like putting a light bulb in series with the power supply for switch mode power supplies.   The main problem is that the feedback loop will try and supply full voltage and current output.    If the input voltage is half, the input current will be 2x.  That could fry parts that weren't fried before.

 Good SMPS will have undervoltage protection to stop the power supply running until the input voltage is above a threshold.  That could make it look like the power supply is faulty.

For my power supply, I have taken out enough parts to prevent the power supply switching.  At present, there is no current flowing in the switching circuit.

OTH a good sized (wattage) bulb will allow a SMPS to start but sure we wouldn't use a 40 or 60W like we might for a linear supply.
However a buddy in London said to never use a DBT with a PM3217 as he has popped their SMPS using one.
There are times IMO when you need use a higher wattage DBT bulb than you might think appropriate.

[dazz1]

Hi
After running power-off static checks of the components most likely to fail, including lifting leads, I have not found any fault in the power supply.
I have refitted all of the lifted leads except for the inductor that supplies the HT transformer primary drive circuit.  This disables the 2kV and 16kV parts of the power supply board.  MUCH safer to work on.

At present, I am working on the secondary side of the power supply.  I am reforming the electrolytic capacitors and testing the secondary side for faults.  With no mains power applied on the primary side, I am connecting my bench power supply to each output.  The bench supply is set to limit current to 1mA.   The rectifier diodes (if not faulty) stop the transformer windings shorting the bench supply.   

This procedure reforms the capacitors and also checks each individual power supply output in a way that will avoid the risk of magic smoke escaping.

The next step is to apply mains power with the dummy load.
Switch mode power supplies generally don't like running with no load.  The load provides control loop stability and stops output voltages climbing or/and oscillating.

I need to simulate a NTC device because I don't have one.   The specs say the nominal load is 50W.  At 220VAC, that is 310VDC ~0.16A. 

If I temporarily fit a 56ohm wire wound resistor, the voltage drop will be 9VDC.    That is well inside the specs for the normal operating AC supply (100VAC-240VAC).
The peak current to charge the Caps will then be 15.4A.  The BYV95 diodes in the rectifier are rated to peak surge current 35A so that is safe.

The 56R resistor will then be dissipating 9W at normal load. 

Of course I won't be connecting the power supply to the scope initially.  I will use the dummy load I made in accordance with the manual.  This dissipates  49W so that is OK.

So the temporary replacement of the blown NTC thermister with a 56R 10W resistor will allow safe power-on testing of the complete supply when connected to the dummy load.   I have 2x 22R 5W resistors in my junk box. 

The only part not tested will be the disabled HT section.   

So the difference between a resistor and a light bulb is the unknowns.  Going through the exercise of checking that any particular light bulb wouldn't cause stress to the power supply would take longer and be more expensive than just using a resistor.   Some basic maths tells me that using a resistor will keep everything operating well with spec limits.

[dazz1]

So having failed to find any defects on the power supply, I resoldered the lifted parts, except the HT supply, and connected the resistors making a stand-in for the NTC surge protector.
I connected up the dummy load.

With great trepidation, I plugged in the scope to the mains.  Nothing happened. 
I checked the power supply outputs and it was obvious that something had happened.
The power supply was working. Huh Rah!.

It wasn't all roses and cream.  The -6.4VDC supply is only -6.0VDC.  Far enough out of spec to justify further investigation.
That is the next step.
 

[dazz1]

Hi
It was a day of one step forward, 5 steps back.

After checking the DC outputs of the power supply, there was suddenly nothing to check. 
The fuse had blown in an almost nuclear sense.  Not quite the disintegration of atoms, but certainly the separation of atoms from other atoms.
This fuse flowing event was instantaneous.    Perhaps >5Amps.  That is a lot of power.

There was no smoke, no pop, no clue. 

I replaced the fuse.  It was old and could have just retired.
The 2nd fuse copied the first.  Hmmm.

So I am back to running static checks.
The 2x 22R resistors used for a temp substitute of the NTC also blew open circuit, so the fault has to be downstream of them.
The transformer windings are good.
In circuit checks of the two large transistors indicate they are OK as well.
The inductors between the two main DC caps are OK.

At least I now know that the NTC thermister failed because of a fault, not because it was faulty.  Big difference.

The fault could be a loss of closed circuit control.  A prime suspect would then the opto-isolator.  If this is faulty, it may cause the power rails to spike, but the secondary side has overvoltage protection.  I can't tell if this has been activated.
So some more investigation work to do.

[dazz1]

I am working on the hypothesis that the primary side of the switching converter is going open loop unstable.  When (if) this happens, the transistors won't switch off, shorting out the 300VDC and blowing the fuses.
This is most likely due some form of heat or voltage breakdown.

No point in replacing the fuse, it will just blow again without providing anymore info.  Also, there is a lot of risk pushing that much current and power through the circuit.  Something is bound to melt in a bad way.

Ideally, I would need to connect to a 300VDC current limited power supply.  I don't have one of those.
I do have an RD6006 which can comfortably put out 50VDC.   That would reduce the voltage by a scale of 1:6
That voltage is out of spec and too low to run the power supply.  I need to fool the power supply into thinking it is running "normally".  I will do that by temporarily increasing the current sensor resistors by 6x  (5R6).   Doing that will also provide some short circuit current limiting through the transformer added to the reduced supply voltage.

I also need to scale the load.  If I tried to run the full 50W load, on 50VDC, it would definitely stress the power supply.  Around 6.5W would be ideal.  Most switch mode supplies are not designed to deliver volts without current.  They need a load to keep stable and stay in a normal mode.

When I look at the circuit, it needs the +17V and -12V supplies.    Normal load for these supplies totals about 12W. 

So I will adjust the dummy load to provide about 6W total on the +17V and -12V secondary supplies.

So the plan is that this will allow me to run the supply until it fails.   With current limit engaged to something safe, I will then be able to probe around and find the problem.

[tautech]

When I look at the circuit, it needs the +17V and -12V supplies.    Normal load for these supplies totals about 12W. 

Can you supply these rails with your bench supply ?
Surely that would be a good check to see if there isn't a secondary side issue.

[dazz1]

Hi
Yes that is a good test.  I did that earlier this week.  I did not find any problems with the secondary side. 

At present, my most likely suspect is the primary thyrister.  If this doesn't fire, then the power transistors will keep going until the transformer core saturates.  At that point, the primary current is effectively unlimited and unconstrained.  Then the fuse blows.

Adjusting the circuit to operate at 50VDC should result in exactly the same timing diagram, but with lower amplitudes.   I am just hoping the fault doesn't need a full 300VDC to trigger.

[dazz1]

Hi
I have reconfigured the power supply so it thinks it is running normally.
I swapped the 0R5 current sensor resistor for a 3x parallel 10R = 3R33.  If 1A flows, the circuit thinks about 6.6A is there.
I disconnected some of the loads from the secondary to scale back the total Watts seen by the primary.

It works.

I have just done some basic checks so far.  I can vary the supply voltage from 40VDC to 50VDC and the secondary outputs are constant.
The 10V reference is working and stable.

Tomorrow I will start looking at wave forms.

I had a thought today that the blown fuses coincided with me using the  mains switch on the power supply board.  I will investigate to see if that is faulty.

If the fault appears again, I can now keep the current-limited power on while I trace the fault.  No fuses to blow. 

[dazz1]

Hi
I checked the switch today. It was a dead short, but in a polarised diode sort of way.
I (rechecked) the bridge rectifiers again and two of them were now shorted.

When I started to remove the diodes, one was split in two.  Now things make sense.

I think one diode has half failed to short.  This has sent AC through the NTC thermister turning into a fuse and burning it out.
Maybe the over current has then cracked it, but not enough to be obvious. 
Then I have come along and it has run for a few minutes before reverting to old habits and shorting out.    This has blown the fuse, but also the neighbouring diode that would have taken a shorted current.  Maybe the slight flexure of the board when I pushed the power switch was enough to reveal the fault.

So when I temporarily rig the power supply to run on 50VDC, it just works.  I still can't be sure that the diode fault was because the diode was faulty, or because it was made faulty by a down circuit fault.  The simplest, and most likely explanation (Occam's Razor) is that the diode failed alone.

So the plan is to soak test the power supply on 50Volts just to shake out any more latent faults.
I will replace all 4 of the rectifier diodes with an equivalent.  Any garden variety diodes should be OK for such a light duty power supply.  Only 50W. 

I will still need to order parts (X2 caps, NTC thermister)  so I still need to wait until I have a batch of parts to order.

[dazz1]

Hi
I used my brand new, really old, recently repaired No.1 Phillips PM3070 to record some reference wave forms on 60V supply.  I will be able to compare these to the wave forms when the power supply is restored to its normal self.

It would be fair to say that the wave forms in the manual are impressionist in their representation of reality.  That could be a by-product of the low supply voltage, but I don't think so.

So the waveform timing and general shape are just as they should be.  The power supply thinks it is running on 300VDC (not 50VDC) and thinks the load is a full 50W.
Varying the supply voltage shows the output volts are correctly regulated.

So it appears the DC side of the power supply is working as it should.

[dazz1]

Hi
Having finished repairing and modding the PM3070 No.1 scope, I reverted to fixing the fault on the PM3070 No.2 scope.
I replaced the cracked and the blown rectifiers plus the leaking capacitor. I happened to have exactly the right rectifiers in my parts stock.
I left the wire wound resistors temporarily fitted to act as current surge suppressors.
The power supply was restored to normal voltages, current sensing etc.

First step was to power up the power supply driving the dummy load.
I did that and checked all the voltages.  They were all within tolerance.    All the magic smoke stayed inside.

With the power supply disconnected, I connected each of the different voltage loads of the scope to 5V, current limited to 0.5A.  This scope hasn't been powered for an unknown number of years.  The electrolytic capacitors needed to be reformed.  Reforming the capacitors with a current limited voltage source reduces the risk of damage due to electrode to electrode conduction.   5V is just enough to check for major faults and to start the electro-chemical reform process.  I powered each voltage for about 15minutes.

The photos are a little blurred because of the length of exposure in low light, but they are proof of life.

One thing that is very noticeable is the feeble graticule and LCD back lighting.    The graticule barely shows even in low light with full illumination.  The LCD back light is also inadequate. 

At present, I am running the scope with power-on to flush out any latent faults.  I still need to fit replacements for the Rifa caps and I need a replacement NTC thermister.   

I now have three operational oscilloscopes, and one more to check over. 

[tautech]

After visiting Dazz today as I was in his area we had a good chat about this 3070 and his other one with the new LED backlight that looked amazing. 
I have a # of old PC SMPS with thermistors intact so will work with Dazz to find some that might be suitable and pop them in the mail.

[C-47]

dazz1, I'm a bit late seeing your post, but in your post #19 I had exactly the some problem some time ago on a 3065 with a split diode and it took a while to find it, I posted some where on the forum about it. Good job in saving it from the bin. Alan.

[R.J.]

The daughter board is an extra option that serves the Y-out at the back of the scope.