Transformer blows input fuse with unconnected secondaries

[Cicada]

Hi
Please note. The original attachment was incorrect. I have added the correct block diagram.


I know this is a long post but I really need help on this. I tried to be thorough in my description.


I need some help with a faulty circuit. See the attached schematic block diagram.

With the original setup (all secondaries connected ) the 250mA rated fuse F101 blows every time that I power the unit up with S101. The LEDs on the front display come on momentarily. The display circuits are powered by the 5v display & 5v logic rails.

I then proceeded to disconnect the secondaries one by one to see when the fuse stops blowing. I started with 1400V and then 140V. During these two disconnection and testing procedures, the fuse blew just as it did with all the secondaries connected.

I then disconnected the secondary at the (5V display & 5V logic) section. After switching on I could hear the transformer hum for the first time. So the fuse did not blow. I left the unit switched on for hours before returning to it. It was still humming away.
I then proceeded with fault finding and this is where I lost the testing plot a bit. I assumed that it must be the circuit connected to this (5V display & 5V logic) secondary that caused the problems. So during my connecting and disconnecting the secondary here while removing and testing components I never checked that the unit can be switched on multiple times when the secondary is disconnected.  It turned out that the fuse blows even with this secondary disconnected. Later it turned out you just need to cycle the power multiple times for the fuse to blow.

So after the confusion of this last test, I decided to disconnect all secondaries. I.e. the last one to be removed was the (+/- 15V and 24V).
Now with all primaries disconnected and a new fuse, the unit switched on. I left it on for a while and switched it off for 3 seconds. Then switched it on again and the fuse blew. I repeated this test with multiple 250mA fuses. The only variation was that sometimes I had to cycle the power more than once to blow the fuse. But not in one case did a fuse blow on the initial switch-on.
I also tried a 500mA fuse as well. This one blew on the first attempt.
The 250mA fuses are from old stock. Brand = Fusit.
The 500mA fuse is new. Brand = OneHungLow
Their all fast-blow fuses.

I then checked the transformer with a Agilent U1733C
Primary: L=4.4 Henry R = 27 Ohm  As measured on the live and neutral input pins of the line filter. These seem like decent values.
Sec 1400V: L = 4.1 Henry R=2.18k Ohm
Sec 140V: L = 3.9 Henry R= 81 Ohm
Sec 24v&15V: L = 336 milliHenry R = 3.7 Ohm
Sec 5V: L = 27milliHenry R = 0.4 Ohm
All inductances measured at 100Hz.

The Sec 5V: R = 0.4 Ohm seems a bit low?

I also measured the insulation between all primary to chassis ground and all secondaries to chassis ground. There are no shorts. I don't think I need to measure this with a megger since 250mA current to chassis would have triggered my earth leakage breaker.

I have not tested the line filter. Part number = 3EF1 by Systron donner.
It looks like this part. https://www.ebay.com/itm/133555373217
I assume it is this part. https://www.mouser.co.za/ProductDetail/TE-Connectivity-Corcom/3EF1?qs=sfrgH6goU2Jyc6W8Z1nXxw%3D%3D
I assume the filter should be fine since the fuse sits after the line filter. So any short between live and neutral in the filter will not be seen by the fuse.
Also, I have measured the primary resistance and inductance (as shown above) through the live and neutral pins of the line filter and the values seemed fine. No shorts. I have also measured the capacitance on the live and neutral pins = 660nF @ 100Hz.

What could be the problem here?
Does anyone have ideas on where I should look next?

The complete user manual for the instrument can be found here.
http://ftb.ko4bb.com/getsimple/index.php?id=download&file=06_Misc_Test_Equipment/Systron-Donner/Systron-Donner_M107_precision_DC_Voltage_source_Service_Manual.pdf

Thanks
Cicada

[Cicada]

Here is the updated diagram.

[Cicada]

I forgot to ask in my original post. Could it be that the cause is higher inrush current due to some degradation of the transformer.
If so what type of degradation is it?

[Benta]

It's a common problem. See this thread:
https://www.eevblog.com/forum/projects/soft-starting-large-transformers-(lt1-kva)-and-power-supplies/

[david77]

Have you tried a slow blow fuse? Fast acting fuses are usually not suitable for big transformers. Depending where on the sine wave you switch on your device, the inrush current may be to high and thus blow your fuse.

[srb1954]

I then checked the transformer with a Agilent U1733C
Primary: L=4.4 Henry R = 27 Ohm  As measured on the live and neutral input pins of the line filter. These seem like decent values.
Sec 1400V: L = 4.1 Henry R=2.18k Ohm
Sec 140V: L = 3.9 Henry R= 81 Ohm
Sec 24v&15V: L = 336 milliHenry R = 3.7 Ohm
Sec 5V: L = 27milliHenry R = 0.4 Ohm
All inductances measured at 100Hz.

The Sec 5V: R = 0.4 Ohm seems a bit low?

These inductance values for each winding don't seem consistent with the voltage ratios between windings. The ratio of inductances between any pair of windings should be the square of the unloaded AC voltage ratios between those windings. In particular, the 1400V winding seems low inductance compared to the 140V secondary inductance.

Part of the problem may be that you are measuring with the line filter still connected and the internal capacitor connected across line and neutral may be confusing the LCR meter. To get accurate inductance readings you need to fully disconnect each winding from any other circuitry. Simply opening the power switch will isolate the line filter from the transformer.

The resistance of the 5V sec is probably OK since this winding is probably rated at a much higher current than the other windings and will be wound with thicker wire.

[trobbins]

The fuse is identified as 3-AGSB, or 3AG slow blow, so you need to use a slow blow, as reference to the manual may have saved you a lot of effort.

If there is still a problem, then perhaps divide and conquer.

Disconnect all secondaries and power just the transformer primary windings via a variac and note the primary current as applied voltage is raised up to nominal mains voltage level.  Do the same with just the mains input emi filter being energised.  Confirm that the no load rms current increases with voltage and at nominal mains is acceptable for both loads.  If ok then that removes some doubt that those parts are somehow not ok.

[bdunham7]

I've been down this road a few times with people that just wouldn't spring for the right fuse.  They are a bit pricey and not as readily available, but you simply need the specified MDL-1/4 fuse and life will be good.  I promise.

https://www.mouser.com/ProductDetail/Bussmann-Eaton/MDL-1-4-R?qs=PdaBPmf3Bcj4cjsAdjsxMw%3D%3D

Yeah, almost $7.  But substitutes like ceramic or other seemingly heavy-duty or non-standard fuses won't work either.  Unfortunately you aren't local or I'd give you a few--I have a box of 100 roaming around here somewhere.  Your issue is not that uncommon.

[Cicada]

The fuse is identified as 3-AGSB, or 3AG slow blow, so you need to use a slow blow, as reference to the manual may have saved you a lot of effort.

If there is still a problem, then perhaps divide and conquer.

Disconnect all secondaries and power just the transformer primary windings via a variac and note the primary current as applied voltage is raised up to nominal mains voltage level.  Do the same with just the mains input emi filter being energised.  Confirm that the no load rms current increases with voltage and at nominal mains is acceptable for both loads.  If ok then that removes some doubt that those parts are somehow not ok.

Hi TRobbins
Yes, I see on page 2-1 they specify:
a (500mA 3AG) slow-blow for 115Vac operation and a (250mA 3AG) slow-blow for 115Vac operation.

I looked at the schematic on page 7-11. At the fuse it says:   .5A - 115V      .25A - 230V
So I used a 250mA fuse since I work with 230Vac
My mistake was to assume that I should use a fast blow fuse.

So yes I should have looked a bit wider. I will in the future.

I also looked at the BOM on page 6-9 which says 1/2 A
Was not very helpful.

The problem is that a 250mA slow blow does not work.
I have blown x5 250mA slow blow fuses. Some at the first power on some at the third or later. But they all do blow eventually.

With a 500mA slow blow there is no problem.

So something must have changed/degraded? over the years to cause a bigger inrush current requiring a fuse bigger than the specified 250mA slow-blow. Am I right?
If so, what could the mechanism be? I can't think of anything that can change in a transformer that would cause a bigger inrush current as it ages.

I have attached a picture of the voltage over a 1-ohm resistor in the neutral line. See "Inrush spike" attached. It shows the peak current = 6.4 Amp. This is with all the transformer secondaries connected and without the output card A6 installed. There is still a problem there.

The steady-state voltage ripple over the 1-ohm resistor is also shown in "Steady state" attached. I have measured the RMS voltage over the resister as 95mV. So the RMS steady-state current should be close to 95mA @ 230Vac

Cicada

[jmelson]

Every time you turn a transformer off, it leaves a remanent magnetic field in the transformer's iron core.  Depending on the exact part of the mains cycle you turn it off, the amount of remanent field varies.  When you turn it back on, at a random part of the mains cycle, flux bulds up in the iron and sometimes increases the magnetization into saturation.  Your scope trace clearly shows this, as the saturation event occurs after the peak of the mains cycle.
Jon

[Cicada]

I've been down this road a few times with people that just wouldn't spring for the right fuse.  They are a bit pricey and not as readily available, but you simply need the specified MDL-1/4 fuse and life will be good.  I promise.

https://www.mouser.com/ProductDetail/Bussmann-Eaton/MDL-1-4-R?qs=PdaBPmf3Bcj4cjsAdjsxMw%3D%3D

Yeah, almost $7.  But substitutes like ceramic or other seemingly heavy-duty or non-standard fuses won't work either.  Unfortunately you aren't local or I'd give you a few--I have a box of 100 roaming around here somewhere.  Your issue is not that uncommon.

Hi bdunham7

How is this fuse good for my specific case?
Is it because of the "time-delay" feature of the fuse?
How does this "time-delay" feature differ from slow-blow?  Do you have a good source of information explaining the difference?

I have posted a screenshot earlier of the inrush current on the AC input measured over a 1 ohm resistor. Maybe you can have a look at that.


Cicada

[edpalmer42]

Since nobody has asked the dumb questions, I will.  Did this unit ever work or are you repairing a unit that was dead when you got it?  Is the line switch set for 230V?  If you measure the primary resistance at the line filter or the power plug, does it go from your measured value of 27 ohms down to about 7 ohms when you switch from 230V to 115V?  That's the end of the dumb questions.  Hopefully the others are a bit smarter.   

Is your 95 ma value for just the transformer by itself i.e. with all secondaries disconnected?  If so, that means the transformer by itself is dissipating about 21 watts.  The entire unit is specified as drawing 35 watts.  That could be okay, but it seems odd.  This is a relatively small transformer and small transformers aren't as efficient as larger ones.

The only thing I can think of that hasn't been mentioned is the possibility of shorted turns in one of the windings.  The 1400V winding will have many turns of small gauge wire.  You might have a short there that wouldn't have a noticeable effect on the output voltage or draw enough power to heat the transformer but still could increase the starting surge.  I don't know how you'd look for something like that.

Ed

[bdunham7]

How is this fuse good for my specific case?
Is it because of the "time-delay" feature of the fuse?
How does this "time-delay" feature differ from slow-blow?  Do you have a good source of information explaining the difference?

The MDL fuse is what they use originally and is specifically designed to tolerate inrush current.  I don't know how it differs from the slow-blow fuse that you are using.  I would think that any time-delay or slow-blow (AFAIK that is just two different terms for more or less the same thing) would work, but perhaps some are different.  What exact fuse are you using?

6.4 amps does seem like quite a bit of inrush and edpalmer42 might be on to something with the idea of a shorted turn or section in the HV winding.  However, if you look at the MDL datasheet you'll see that the time-current curve shows 40-50ms @ 7A and an I²T of 0.447.  Your peak is 6.4A @ <2ms and an I²T of maybe 0.05 (just a SWAG), so according to that the MDL would be pretty unlikely to blow at those levels unless perhaps you cycled the switch very rapidly.  Also note the the MDL-1/4R has a cold resistance of about 3 ohms. 

You might just leave in the 500mA slow-blow fuse you have for now--that is unlikely to cause a house fire.  Also, are you having this issue with any plausible normal use case?  Or are you rapidly cycling the power?  A 250mA slow-blow fuse will warm up a bit at 95mA and would be much more likely to blow on a very quick power interruption than say a 5-second off-on cycle.

[Cicada]

Since nobody has asked the dumb questions, I will.  Did this unit ever work or are you repairing a unit that was dead when you got it?  Is the line switch set for 230V?  If you measure the primary resistance at the line filter or the power plug, does it go from your measured value of 27 ohms down to about 7 ohms when you switch from 230V to 115V?  That's the end of the dumb questions.  Hopefully the others are a bit smarter.   

Is your 95 ma value for just the transformer by itself i.e. with all secondaries disconnected?  If so, that means the transformer by itself is dissipating about 21 watts.  The entire unit is specified as drawing 35 watts.  That could be okay, but it seems odd.  This is a relatively small transformer and small transformers aren't as efficient as larger ones.

The only thing I can think of that hasn't been mentioned is the possibility of shorted turns in one of the windings.  The 1400V winding will have many turns of small gauge wire.  You might have a short there that wouldn't have a noticeable effect on the output voltage or draw enough power to heat the transformer but still could increase the starting surge.  I don't know how you'd look for something like that.

Ed

First off. I have to make it clear. The unit is now powering up with a 500mA slow blow fuse. The spec requires a 250mA slow blow for 230Vac. That is what caught me off guard.
All cards are installed except the output card A6 i.e. amplifier card. On A6 card the 1400V output arm burned out. See attached picture A6.jpg.
I now know, through debugging, that this A6 card failure is most probably caused by the fact that relay K2 (range change relay) on page 7-23 is not switching. This could have caused the damage on the card due to incorrect feedback settings.

Now to your questions:
Did this unit ever work or are you repairing a unit that was dead when you got it?
It has been in use for the past 30 years I guess. It had some recent failures that were fixed. I am not sure exactly what failed. Guy left the company. Heard from someone it was the range change relays.

Is the line switch set for 230V?
Yes.

If you measure the primary resistance at the line filter or the power plug, does it go from your measured value of 27 ohms down to about 7 ohms when you switch from 230V to 115V?
Yes

Is your 95 ma value for just the transformer by itself i.e. with all secondaries disconnected?
No. This is the current draw for the whole system with all the secondaries connected, excluding card A6. Because this card is damaged and unplugged.

If so, that means the transformer by itself is dissipating about 21 watts.  The entire unit is specified as drawing 35 watts.  That could be okay, but it seems odd.  This is a relatively small transformer and small transformers aren't as efficient as larger ones.
This is the complete power consumption of the unit minus the A6 card.

[edpalmer42]

Okay, those answers make me feel better.  21 watts with an almost complete circuit is much more reasonable.  When you hit the power switch you've got to set up the magnetic field in the transformer and then charge the bulk capacitors in the supplies.  I see 350 uf on the +140V supply and 1500 uf on each of the +24V and -20V supplies.  I assume you've checked those capacitors.

One more dumb question.  Have you checked your line voltage? 230V +- 10% shouldn't be a hard spec to meet, but if you're running at the top end of that range you'd have higher surges.

You've got a circuit that's almost working.  Without knowing how much power each supply should need it's hard to track down any one source.  It's possible that everything is just a little tired and leaky and when totalled up and added to the transformer surge, it's just too much for the specified fuse.  Bdunham7's suggestion to run with the 500 ma fuse is probably the only practical alternative.

Ed

[bdunham7]

I forgot the MDL datasheet link:

https://www.mouser.com/datasheet/2/87/eaton-mdl-time-delay-glass-tube-fuses-data-sheet-1608815.pdf

[srb1954]

Is your 95 ma value for just the transformer by itself i.e. with all secondaries disconnected?  If so, that means the transformer by itself is dissipating about 21 watts.  The entire unit is specified as drawing 35 watts.  That could be okay, but it seems odd.  This is a relatively small transformer and small transformers aren't as efficient as larger ones.

The actual dissipation in the transformer will be lower than 21W due to the power factor. Real power, P = V x A x PF

That is a seriously spiky current waveform, which will have high harmonic content giving rise to a poor power factor, maybe less than 0.5.

[trobbins]

Let's wait for the nominal mains voltage, but worth noting that if that was high then initial inrush from mains voltage only could be higher, as well as the energy throughput for capacitor charging. If you disconnect the loads, and do an inrush current capture then you may not see the very large current spike, which I suspect is cap loading and not related to saturation but maybe too early to tell.

Also fuses wern't very standardised back then, so just saying it is 250mA slow blow is quite ill-defined, as it is best to buy known parts from a manufacturer that states the standard the fuse complies with, and keep each such fuse quarantined and not add them to the bin of assorted 250mA fuses.

[Cicada]

Let's wait for the nominal mains voltage

235Vrms

If you disconnect the loads, and do an inrush current capture then you may not see the very large current spike

The current spike with all the secondaries disconnected is 3 Amps. It is a little less than half of the 6.2A with all secondaries connected and only card A6 not present. For pictures of the current spike refer to an image that I posted in an earlier post. "Inrush spike.jpg"

[Cicada]

I can give an update now. All the cards are installed. I use a 500mA slow blow. It works consistently.
A 250mA slow blow (as specified) only works for a couple of switch-on operations.
I forgot to mention in the previous post that the transformer has quite a loud hum on it.

It can be seen in an earlier photo that the 1400V string on the A6 output amplifier card is damaged. The string of resistors from R20 to R25 are all burned. The tracks on the board are still intact.

I have measured the transistors on card A6 to see which transistors are damaged. I have attached the result of my measurements conducted with a fluke 177 using the diode test feature of the DMM. The measurements for transistors shown in red indicate that the voltage measured is not what is expected. So the incorrect measurements could be due to the unknown influence of other components in the circuit or due to damage to the component.
I only deemed transistors to be good if there was no ambiguity to interpret the measurements. It can be seen that Q1 and Q5 > Q10 might be damaged. They are all in the 1400V string. So I have disconnected the 1400 V input to the card. I have plugged the card in and the unit powered up successfully.

Then I investigated further and found that relay K2 and K3 on the analog motherboard A5 do not switch. One relay has a broken input coil and the other relay is stuck in the off position, its coil is ok. Darlington driver U10 on A5 is working fine.
K3 is the polarity relay and it is stuck in its + polarity position so I could use this relay as is in the system.

K2 is the upper bit of the two range selection relays.  In the table below 0 = relay not enabled, 1 = relay enabled

K2       K1     Range
0         0       1V
0         1       10V
1         0       100V
1         1       1000V

So I could only change between 1V default and 10V range since only K1 switches.

In the 1V range, the output is from the 1V circuit on the motherboard A5. This circuit is to the top left of relay K1. The values set on the input correspond to the readings on my Fluke 177. So all is fine here.

In the 10V and higher ranges, the output is through the (damaged?) card A6. I have a very big offset that changes very little as the output is changed from 1 to 9V in steps of a 1V. Again I have measured it with my Fluke 177. Below is typical results.
It must be noted that these measurements were done with a unit that was not warmed up for 2 hours.
Set   Measured
1V    1.292V
2V    2.294V
3V    3.295V
4V    4.295V
5V    5.295V
6V    6.294V
7V    7.29V
8V    8.29V
9V    9.29V



So what remains:
1. I need to calibrate the unit using the specified calibration procedure which I have not read yet. I do have access to a 7 1/2 digit Keysight DMM. So I will do this as soon as possible. I am a bit apprehensive to start calibration too soon. I am a bit unsure how the unit's accuracy can be out by 297mV. That is a very big offset. It can't be natural drift over time. I am not sure if it could be due to partial damage to some of the transistors on the output amplifier card A6. I.e. to transistors Q3, Q4, Q11 and Q12 which I deem to be working fine based on my measurements. These transistors are relevant to the output on the 10V scale.

2. The user only needs the +10V scale to be functional. So strictly speaking I need not replace K2(range) and K3(polarity) but I am going to do it anyway.  https://www.mouser.co.za/ProductDetail/TE-Connectivity/R10-E1Y6-V430?qs=8wHch9UpSvZ05zrT%252B3HUpQ%3D%3D

3. I am wondering what internal failure could have caused the failure (burn out of resistor string) on card A6. My only suspicion now is that a failed relay K2 could cause some problems with the feedback circuit which in turn can cause too high a voltage output to the amplifier card. Which in turn could cause higher than expected output voltages. But I struggle to see this.
Then again the failure could also have been from some external causes.