Keithley 2000 Blown Fuse

[Hendrik]

Hello everyone,

my Keithley 2000 quit its job today with a blown fuse. As I am still a novice in electrical engineering I would like to ask for some advice that might help me diagnose the problem. For the past few days I have been reading low DC voltages with the DMM and logging them over RS232, nothing too wild.

I had an identical spare fuse at hand and swapped it with the damaged one. After that I was able to turn on the device for about 3 seconds before it blew the next fuse. In those 3 seconds I saw the front display properly light up and starting to initialize (I also think that the buzzer made the startup sound).
Would this suggest that the +5V and +37V rail are working properly since they are required for the vfd to work? (would this also imply that the power transformer is working properly (eg. no shorted turns)?)

On visual inspection with my untrained eye I couldn't see any obviously damaged components, burn marks on the pcb or any other flaws.
Neither did I smell anything burned.
From how fast the fuse blows I would think that the problem has to be a direct short to ground somewhere?
I would appreciate any suggestion on where to start looking and how to proceed next.

(I am referring to this repair manual: https://download.tek.com/manual/2000-902-01(B-Mar1997)(Repair).pdf )

Kind regards,
Hendrik

[Kleinstein]

The blown fuse shows excessive supply current. This can be from different parts of the circuit. The display part seems to work, though may still consume too much current.
A first would be to check if the main voltage setting is correct - so not set to 120 V, but that is unlikely.

One could try out the transformer on it's own, so no board connected. A shorted internal winding is possible, though not very likely. This would often cause excessive current and heating, but output could still be high enough to make the rest of the meter work. The transformer alone should not consume much more than some 3 W. So a fault would be rather obvious with a plug in power meter.

A part that tends to fail short are the rectifiers. So it is worth checking the rectifiers, especially for the input part. This can be done while still on the PCB with the diode test.
Another point is to careful look for leakage at the large electrolytic capacitors.

Chances are the 5 V and 37 V for the output and display side work. One could still try out the 5 V part by powering that part from from a lab supply with some 8-9 V instead of the transformer winding.
One could try out the output side part without the input side part (J1016) connected. This could narror down the fault by testing that half, that is likely still OK.

Similar it should be OK to try out the 5 V part of the input analog part (pins 1 and 2 of J1016) with some 10 V DC instead of the transformer.
The +-20 V to +-15 V regulator part is a bit more tricky, but may work one side at a time with the other shorted.

[Hendrik]

Thank you Kleinstein for your comprehensive answer.

The blown fuse shows excessive supply current. This can be from different parts of the circuit. The display part seems to work, though may still consume too much current.
A first would be to check if the main voltage setting is correct - so not set to 120 V, but that is unlikely.

Yes it is and was on 220V. The fuse was also rated for the 220V setting.

One could try out the transformer on it's own, so no board connected. A shorted internal winding is possible, though not very likely. This would often cause excessive current and heating, but output could still be high enough to make the rest of the meter work. The transformer alone should not consume much more than some 3 W. So a fault would be rather obvious with a plug in power meter.

Without a working fuse and without messing with the wiring I don't see how I would attach the transformer to power. I am about to order lots of spare fuses from mouser, so I could test that once they arrive.

A part that tends to fail short are the rectifiers. So it is worth checking the rectifiers, especially for the input part. This can be done while still on the PCB with the diode test.
Another point is to careful look for leakage at the large electrolytic capacitors.

I have never "really" used diode mode, so I am not sure if I tested this correctly. CR102 & CR104 read ~1.04V in Diode mode with red lead to negative and black lead to positive. CR103 reads 0.59V. Shouldn't at the very least CR102 and CR103 be the same since they are the same component? Besides the new fuses I will also order new electrolytic capacitors just to be on the safe side.
(Datasheet for the diode (I think): https://www.mouser.de/datasheet/2/427/dfm-1768184.pdf)

Chances are the 5 V and 37 V for the output and display side work. One could still try out the 5 V part by powering that part from from a lab supply with some 8-9 V instead of the transformer winding.
One could try out the output side part without the input side part (J1016) connected. This could narror down the fault by testing that half, that is likely still OK.

Similar it should be OK to try out the 5 V part of the input analog part (pins 1 and 2 of J1016) with some 10 V DC instead of the transformer.
The +-20 V to +-15 V regulator part is a bit more tricky, but may work one side at a time with the other shorted.

Will try that next. But I am a bit nervous not to fry anything.

[Kleinstein]

Looking for a bad diode is looking for an essential short (e.g. both directions < 100 mV in diode teste, usually < 10 mV). Smaller differences with > 200 mV can be from other parts connected.
Getting > 700 mV usually means the direction for the diode to block and than only other parts, like the regulators or capacitors.

The input side is pretty isolated and should have no issue at all with only the 5 V.

The +5 V for the input side may have a slight isssue with the DG211 and DG408 switches, that can get a 5 V side control signal and may have diodes to the +15 V supply. I don't think the logic chips are strong enough to cause any read damage though. one could consider linking the +5 V to the +15 V for that test, so the diodes would not power the +15V. Worst case the DG211 and DG408 are not that expensive and availabilty got a bit better.
Only the +-20 V and not 5 V should not cause an issue from chips with both supplies.

[iMo]

Doublecheck all those electrolytic and tantalum capacitors, could be shorted.
Is your fuse properly rated?

[Per Hansson]

Without a working fuse and without messing with the wiring I don't see how I would attach the transformer to power. I am about to order lots of spare fuses from mouser, so I could test that once they arrive.

I have a suggestion: don't use fuses but instead replace the fuse with say a 60W incandescent light bulb.
If the light bulb lights up bright it means you (still) have a short.

A part that tends to fail short are the rectifiers. So it is worth checking the rectifiers, especially for the input part. This can be done while still on the PCB with the diode test.
Another point is to careful look for leakage at the large electrolytic capacitors.

I have never "really" used diode mode, so I am not sure if I tested this correctly. CR102 & CR104 read ~1.04V in Diode mode with red lead to negative and black lead to positive. CR103 reads 0.59V. Shouldn't at the very least CR102 and CR103 be the same since they are the same component? Besides the new fuses I will also order new electrolytic capacitors just to be on the safe side.
(Datasheet for the diode (I think): https://www.mouser.de/datasheet/2/427/dfm-1768184.pdf)

The datasheet is correct, but really any datasheet for a diode bridge will be fine, they are all the same internal construction...
So please look at the datasheet again and identify the four diodes.
Realize when you measure between some legs you will have two diodes in series, thus a 1.4v drop.
And if you measure across one diode only you will see 0.7v drop.
Now the issue is when there are parts in the circuit (and there always are) it will skew the results.
So if you are unsure you must desolder the bridge rectifier and test it off board...

[floobydust]

Something will be getting hot yet I see no burnt parts and the multimeter seems to work for the few seconds.
I would rule out a power transformer with shorted turn. The primary fuse I believe is 5x20mm slow-blow 125mA (240V) Keithley p/n FU-91.
Unplug the secondary-motherboard connector and run the multimeter, the transformer primary connected as usual and energized, see if the fuse holds for a minute with it just powering the disconnected transformer.
You could use a light bulb tester to save on fuses.

[chekhov]

One could try out the transformer on it's own, so no board connected....

That's really a good step to start with ...
Here is an example of what good transformer is going to show when not connected to the board (with all wires connected to the input selector (for your case mains should go to White-Gray for 220V, or White-Red for 220V, btw if it's 230V in your mains, I suggest using 240V selector).

Code: [Select]

                  No load    Load,  Volts AC
4-pin connector
brown-brown        ~8.8      ~8.0
yellow-yellow     ~42.3     ~39.6

5-pin connector
green-green        ~9.8      ~9.2
gray-orange       ~21.1     ~19.7
gray-orange       ~21.1     ~19.6

Digital part (from brown-brown) takes around 0.6A under normal conditions, haven't measured analog one, but that should not be more.
Internally, transformer uses very thin (0.125mm) wire for 220V part, and it is on the very bottom, so not blowing your transformer is very important, when you get the idea from where you have the short, I suggest using laboratory power supply with current limiting to substitute it on the winding in question.

[Hendrik]

Doublecheck all those electrolytic and tantalum capacitors, could be shorted.
Is your fuse properly rated?

Now I have a stupid question: How would one do that correctly in circuit? Measuring resistance of the 1000uF capacitors for example starts at 0 Ohm then rising up to few MOhm over time (I am using a very cheap multimeter here). Would a shorted capacitor just sit at 0 Ohm?
Both fuses were 125mA rated slow blow fuses. I took the second one directly from a working Keithley 2700.

I have a suggestion: don't use fuses but instead replace the fuse with say a 60W incandescent light bulb.
If the light bulb lights up bright it means you (still) have a short.

I don't think that I have any incandescent light bulbs since we pretty much have changed everything to led.
Would these work?: https://www.amazon.de/-/en/Halogen-White-Kimjo-680LM-Dimmable/dp/B07VH9RC93?th=1
How would one connect these to the fusebox?

The datasheet is correct, but really any datasheet for a diode bridge will be fine, they are all the same internal construction...
So please look at the datasheet again and identify the four diodes.
Realize when you measure between some legs you will have two diodes in series, thus a 1.4v drop.
And if you measure across one diode only you will see 0.7v drop.
Now the issue is when there are parts in the circuit (and there always are) it will skew the results.
So if you are unsure you must desolder the bridge rectifier and test it off board...

Thank you, it makes a lot more sense with the schematic at the top of the datasheet.

Something will be getting hot yet I see no burnt parts and the multimeter seems to work for the few seconds.
I would rule out a power transformer with shorted turn. The primary fuse I believe is 5x20mm slow-blow 125mA (240V) Keithley p/n FU-91.
Unplug the secondary-motherboard connector and run the multimeter, the transformer primary connected as usual and energized, see if the fuse holds for a minute with it just powering the disconnected transformer.
You could use a light bulb tester to save on fuses.

Thank you for your suggestion, and yes you are correct about the fuse.
I already ordered some new fuses and will try that as soon as they arrive.

One could try out the transformer on it's own, so no board connected....

That's really a good step to start with ...
Here is an example of what good transformer is going to show when not connected to the board (with all wires connected to the input selector (for your case mains should go to White-Gray for 220V, or White-Red for 220V, btw if it's 230V in your mains, I suggest using 240V selector).

Code: [Select]

                  No load    Load,  Volts AC
4-pin connector
brown-brown        ~8.8      ~8.0
yellow-yellow     ~42.3     ~39.6

5-pin connector
green-green        ~9.8      ~9.2
gray-orange       ~21.1     ~19.7
gray-orange       ~21.1     ~19.6

Digital part (from brown-brown) takes around 0.6A under normal conditions, haven't measured analog one, but that should not be more.
Internally, transformer uses very thin (0.125mm) wire for 220V part, and it is on the very bottom, so not blowing your transformer is very important, when you get the idea from where you have the short, I suggest using laboratory power supply with current limiting to substitute it on the winding in question.

Thank you for the detailed information. I also switched the selector to 240V as my main read 231V.
Now a stupid question from my side again: I haven't worked with AC yet, so all my equipment is DC, like DC Load, DC lab power supply. Is that even of use when working with a transformer, since we aren't connecting it to the pcb where the rectifier would be, where I then could connect my DC load.
 

[Kleinstein]

The halogen bulbs should work. The cold resistance is lower and this can actually be positive compared to a normal bulb.

The bulb would be wired in series and I would wire it as a seprate box to have between mains (or idally an isolation transformer for some tasks) and the DUT. So the ouput would be more like an outlet / extension cord. For such tests I have 2 outlets wired in series powered from a single cable. One gets the DUT and the other gets a lamp of choice suitable for the load.
So the bulb part would be in series to the whole instruments and not just replacing the fuse. It only would kind of take over the function of the fuse. It would still be good to have a fuse in the circuit, as a failing bulb can have an at least temporary near short (arc). One may get way with a slightly larger on for the test if one takes care to replace it with the correct one later.

A shortet  capacitor would stay at near zero ohms. When tested in the circuit one would still not know where on the rails the short would be (e.g. could still be the regulator or rectifier). One would at least know which supply part is effected, e.g. the +20 , -20, + 15,-15, 5 (input side) or unlikely 5 V output side.

One can use a lab supply to power parts of the circuit and this makes absolute sense. Ideally one would need more than 1 lab supply, e.g. to get +-20 V and +10 V for the whole input part.
A rectifier also works with DC in. Usually there is enough reserve to be OK with this and not have the diodes alternating for slightly less local heating.
Alternatively one could provide the power to the large filter capacitors directly and skip the rectifiers.

[chekhov]

Transformer is just transformer ... verify its voltages w/o connecting it to the board in order to see whether it is still healthy.
And you may do that with any other fuse - if xformer itself is dead, fuse won't help ...

AC from transformer is only used in front panel VFD for heater, really haven't tried to use DC power supply for this type of front paneld, and in order to detect mains AC crosses zero volts, so in order to verify things it's ok to just connect DC instead of transformer for main board (using the same voltage as measured in AC mode), you always can disconnect front panel.
Just, if you only have 1-3 seconds until fuse blows, that's not right investigation path to continue using transformer.

Also you can connect one plug at a time - connect only digital 4-wire plug, verify voltages, then for 5-wire plug, do the same. This way xformer will see less stress. But do it very quickly - connect meter to measured part first, turn on, see readings, turn off.

Anyway, do whatever you can to detect whether the short is Cool part is that there coululd be no 'short' at all, but some component that starts drawing current starting from some voltage, so in diode mode may look ok, but not in real scenario.

If capacitor is ok, voltage will raise, that's a good sign.

[Andreas]

Have you already checked the mains line input filter?

[Kleinstein]

The Mains input filter is indeed a good point to check - one of the very first points on the list of suspects.

[Hendrik]

Chances are the 5 V and 37 V for the output and display side work. One could still try out the 5 V part by powering that part from from a lab supply with some 8-9 V instead of the transformer winding.
One could try out the output side part without the input side part (J1016) connected. This could narror down the fault by testing that half, that is likely still OK.

Similar it should be OK to try out the 5 V part of the input analog part (pins 1 and 2 of J1016) with some 10 V DC instead of the transformer.
The +-20 V to +-15 V regulator part is a bit more tricky, but may work one side at a time with the other shorted.

As suggested I connected 9-11 DCV to pin 1/2 to supply the 5V rail. The current draw stays constant at ~200mAmps. I would assume this suggests that there is no fault on the 5V rail?
I have an Agilent E3644 which can't supply negative voltage and a maximum of 20V. To test the +-15V Rail I would supply 20V to Pin 3/4 and short VIN of U119 to Ground for example (or how would that work?).

Have you already checked the mains line input filter?

I am afraid to ask, but where would that component be located? I guess very close to the AC plug?

[Kleinstein]

The mains input filter is between the mains input and transformer. I don't know where it is with the K2000.  It can be part of the IEC socket together with a fuse and thus before the switch. Often it is behind the mains switch and ideally behind the fuse.


200 mA for the 5 V part sound reasonable and would not blow the fuse.

With only a single supply one could try the +20 and -20 V one at a time. Ideally one would short the other regulator on both the input and the output side, so that the -15 V would not get positive if there is load or the short from the +15 to the -15 V.

[Hendrik]

The mains input filter is between the mains input and transformer. I don't know where it is with the K2000.  It can be part of the IEC socket together with a fuse and thus before the switch. Often it is behind the mains switch and ideally behind the fuse.


200 mA for the 5 V part sound reasonable and would not blow the fuse.

With only a single supply one could try the +20 and -20 V one at a time. Ideally one would short the other regulator on both the input and the output side, so that the -15 V would not get positive if there is load or the short from the +15 to the -15 V.

Some more photos of the case, cabels and fusebox. I don't see how I would check anything here.
I will check the +-15V lane tomorrow.


Edit: Upon further reading of the repair manual (p19/109):

AC power is applied to the AC power module receptacle (J1009). Power is routed through the
line fuse and line voltage selection switch of the power module to the power transformer. The
power transformer has a total of four secondary windings for the various supplies.

Is there even a mains input filter 

[floobydust]

I don't see a mains filter. The middle module is the on-off switch, then the fuse and line voltage selector.
These bench multimeter power transformers typically have a (grounded) electrostatic shield pri-sec, or split-bobbin winding so mains EMI does not get into the instrument.

Powering this up using bench power supplies in place of the many rails - I would not do. The power-up/down sequence is critical and I think easy to do damage by accident.

OP, if you have a signal generator and scope, you can inject a sine wave of a few volts into the transformer and see what you get out. We know the approximate turns ratio.
Otherwise, I'm not sure how comfortable you are with loose mains wiring but I'd just use jumpers to connect that light bulb across either the on/off switch (with it off) or the fuse to put it in series with the transformer primary, for the quick test. If it lights up there is a problem I would say with the transformer.

[macboy]

My favorite power supply, an Agilent 3620A, started randomly blowing fuses. Sometimes immediately at power on, sometimes after a few hours. Eventually I replaced X safety rated capacitor (Rifa branded) across the line, and that solved it. These capacitors, and MOVs, have a hard life absorbing transients. It is never a bad idea to replace them with new parts.

[Hendrik]

I don't see a mains filter. The middle module is the on-off switch, then the fuse and line voltage selector.
These bench multimeter power transformers typically have a (grounded) electrostatic shield pri-sec, or split-bobbin winding so mains EMI does not get into the instrument.

Powering this up using bench power supplies in place of the many rails - I would not do. The power-up/down sequence is critical and I think easy to do damage by accident.

OP, if you have a signal generator and scope, you can inject a sine wave of a few volts into the transformer and see what you get out. We know the approximate turns ratio.
Otherwise, I'm not sure how comfortable you are with loose mains wiring but I'd just use jumpers to connect that light bulb across either the on/off switch (with it off) or the fuse to put it in series with the transformer primary, for the quick test. If it lights up there is a problem I would say with the transformer.

Thanks for the confirmation. Yes I do have a signal generator and oscilloscope. I will test this tomorrow!
I try to avoid mains wiring since I am just getting into ee and have mainly tipped my toes into low voltage dc.

My favorite power supply, an Agilent 3620A, started randomly blowing fuses. Sometimes immediately at power on, sometimes after a few hours. Eventually I replaced X safety rated capacitor (Rifa branded) across the line, and that solved it. These capacitors, and MOVs, have a hard life absorbing transients. It is never a bad idea to replace them with new parts.

I ordered new caps alongside the fuses today, better replace them all !

[Kleinstein]

With a classical transformer power supply there is no absolute need to have classic input filtering with an X capacitors. There are common mode chokes on the secondary side for the output related part.
From the pictures it looks like there is not filtering in front of the transformer, if at all a relatively small capacitor in the case with the fuse, but I doubt that.

Power supply sequencing could be an issue, but not too bad. The main point that would not be good is if the input part  +5 V are higher than the +15 V rail.

The input and output side are well isolated and should have absolutely no problem when powered on there own.

A possible fault is a sharted MLCC, this can be tested already woth a rather low voltage (like 0.3 V) and thus before power supply sequencing becomes an issue.

[Hendrik]

I don't see a mains filter. The middle module is the on-off switch, then the fuse and line voltage selector.
These bench multimeter power transformers typically have a (grounded) electrostatic shield pri-sec, or split-bobbin winding so mains EMI does not get into the instrument.

Powering this up using bench power supplies in place of the many rails - I would not do. The power-up/down sequence is critical and I think easy to do damage by accident.


OP, if you have a signal generator and scope, you can inject a sine wave of a few volts into the transformer and see what you get out. We know the approximate turns ratio.
Otherwise, I'm not sure how comfortable you are with loose mains wiring but I'd just use jumpers to connect that light bulb across either the on/off switch (with it off) or the fuse to put it in series with the transformer primary, for the quick test. If it lights up there is a problem I would say with the transformer.

I have tried to test the transformer with my signal generator and scope. Here are the results:

Input: White-Grey 10V Pk-Pk (146,2 Ohm)
            

Measure1	Measure2	V Pk-Pk	@220V Vrms
Brown	Brown	0,43	6,71
Yellow	Yellow	2,06	32,04
Grey	Orange1	0,92	14,31
Grey	Orange2	0,92	14,31
Green	Green	0,46	7,15

I tried to connect my function generator to White-Red as it had the highest resistance with 172,8 Ohm (most windings? - With the input selector to 240V this would be the selected winding length I guess?), but I wasn't able to attach the probes to that cable. So I injected the sin wave over White-Grey which had the second highest resistance with 146,2 Ohm (which would be 220V input selector?).


All results stayed stable on the mentioned value. I haven't worked with Transformers yet, but the results are fairly close to the needed voltage at the rail. Would this suggest that the power transformer is working properly?

[Kleinstein]

To no supprise the turns ration still looks OK. One can howver not yet exclude a shorted turn, especially not one with a non ohmic contact (e.g. isolation damage leading to sparcs).

[floobydust]

I have tried to test the transformer with my signal generator and scope. Here are the results:

Input: White-Grey 10V Pk-Pk (146,2 Ohm)
            

Measure1	Measure2	V Pk-Pk	@220V Vrms
Brown	Brown	0,43	6,71
Yellow	Yellow	2,06	32,04
Grey	Orange1	0,92	14,31
Grey	Orange2	0,92	14,31
Green	Green	0,46	7,15

I tried to connect my function generator to White-Red as it had the highest resistance with 172,8 Ohm (most windings? - With the input selector to 240V this would be the selected winding length I guess?), but I wasn't able to attach the probes to that cable. So I injected the sin wave over White-Grey which had the second highest resistance with 146,2 Ohm (which would be 220V input selector?).


All results stayed stable on the mentioned value. I haven't worked with Transformers yet, but the results are fairly close to the needed voltage at the rail. Would this suggest that the power transformer is working properly?

Those numbers are good but only for input on one primary winding, so 110VAC in. The other primary winding needs to be tried. You could leave 10Vpp input on White-Grey and see about the same 10Vpp coming out on White-Red. The primary windings' ohmmeter readings should match, but you are getting a mismatch with 146.2Ω verses 172.8Ω which makes me wonder what is going on.
It's hard to know where a shorted turn is or across how many turns, or between which windings. You could have a pri1 to pri2 short as well and this test would not catch that.
I'm not great with the magnetic's theory on shorted turns, I use a Ring Tester because you still get transformer action despite a short. So I am saying the transformer is still suspect IMHO.
When the new fuses come in, I would still try run the transformer with the secondary disconnected.

If the transformer seems ok, do you have access to a thermal imaging cam? That can spot the hot parts on boards within seconds and save much work and grief.

A pic of the TR-299B transformer? https://www.eevblog.com/forum/buysellwanted/wtb-keithley-tr-299b-transformer-for-model-2000-multimeter/
Same part number as in Model 2700. Note the weird construction with a dual-bobbin, one winding outside the other.

The chinese knock-offs on eBay do not sound promising, or look like they are made properly, this $95+$15 shipping review hurts:
"Replacement transformer for Keithley 2000 DMM. Counterfeit sticker, this item was made in China somewhere, not by Northlake Engineering. Build quality/QC is also pretty dubious. Much thinner wires than original and the insulation on 2 of the secondary wires was knicked/bare. Took 7 days to ship, and 10 days to arrive. Not terrible, but for $35 they could do a lot better. On the positive side, it does work and is obtainable. Would not recommend this seller if there is a viable alternative."

[chekhov]

Transformer's primary winding consists of 4 sections - White 100V Blue 20V Violet 100V Gray 20V Red  in series.
For 100 and 120V "White 100V Blue 20V Violet" section used 0.175-0.18mm wire, for 220-240V additional sections "Violet 100V Gray 20V Red" - 0.125mm wire or so, so their resistances won't match (number of turns the same - 612 + 122 turns) - no issue with that.

[Hendrik]

If the transformer seems ok, do you have access to a thermal imaging cam? That can spot the hot parts on boards within seconds and save much work and grief.

New fuses should arrive tomorrow. Will test the transformer alone then and report back. Sadly I don't have a thermal imaging camera at hand but I thought about buying one a couple of times now. Depending on the results tomorrow I might order one.