PCB Damage

secutest_jig2

Modified secutest cabinet for “easy” measuring

 

secutest_jig1

The white probe is de i-prober

Tĥis is the pcb from a G&M Secutest. It was not working like it should. The problem turned out to be some vaporized traces inside the 4 layer pcb, a LM358 with ventholes, a dead 555 from the boost converter, a dead TL082, a burned inductor, a blown fuse and 4 bad relais.

The the pcb above the LCD pcb turned out bad too. A short between the negative rail and ground. This was harder to find because it was not visible. I found it using my fluxgate based i-prober currentprobe. Repairing was easy, I milled a way a part of the top gnd layer, then the burned out part to remove the short and then filled all up. The new trace is now for safety reasons above the pcb.

Here some more info about the PSU PCB repair. First step was “opening” the pcb so I could inspect the inner-layer and remove the burned residu (that is conductive and can cause leakage current. Not something you want in this type of instrument.)

Here you see a part of the problem.  This was not the worst one but I could not made that visible on a picture because there was a groundplane that blocked the view.  You see the black “fog” around the trace. This is inside the pcb photographed with a lamp behind the pcb. The traces go to the pins of a relay. 4 traces were burned/vaporized on several places, bizar !

secutestPCB2

Close up

I milled and scraped “channels” so I could remove the burned stuff and layout new traces. I used 0,2mm wire for this. Look at the size of those sot23 transistors to see the scale of things. All this is done under a microscope. I use excavators and mini scrapers that I made myself. It is not difficult but rather time consuming and you need a steady hand and some mechanical experience. I have done a lot of mechanical things for many years, small scale like building scale models (cars, bikes, ships, trains and a small steam engine) but also restored (and complete rebuild) real bikes and cars.

Normally to much work for an old Secutest, but I just love doing this sort of things so now and then as extra practice.

secutestPCB1

first step

I had to remove some of the top traces.

secutestPCB3

An other close up of the channels with the new “traces”

 

secutestPCB5

overview of the area after it is finnished.

It looks a bit messy because I covered the new toptraces with epoxy.

secutestPCB6

Ready to solder the removed components.

And after testing all connections the parts are tested and placed.

secutestPCB7

The repaired pcb is under the orange relais

Teardown Isotech Venus dryblock

A peltier powered dryblock from Isotech that stopped working. There was no current flowing to the TECs (peltier elements). That can be caused by a dead controller, dead PSU or dead TECs. They should have a very low resistance so that is easy to test first. I measured a few megohm. The once used here should measure an AC resistance around 2,2 ohm.

I made pictures from the tear down for those who are curious about the construction of these instruments.

The underside

The underside

The first step was easy, removing the cabinet. This one has a motor with a rotor that holds a magnet. It is the thing in the Telflon ring on the picture above. It should be glued but had come loose.

Sideview

Sideview

After removing the motor and the rest of the cabinet we see the dryblock in its full glory.  Check the 4 screws on top of the cabinet, the whole block is only mounted by those tiny things

Cutting away the foam  is a dirty job but not difficult. Just be careful not to cut the wires.

Heatsinks removed

Heatsinks removed

Here most of the foam is removed. The pressure-plates are pressed together during mounting and then “locked” by the M3 threaded rods screwed in the brass studs. They are not used to set or change the pressure, only to keep the pressure constant.

Going deeper

Going deeper

The foam was burned where it touched the anodized dryblock core.

the first peltiers visible

the first peltiers visible

After removing the threaded rod the plates come loose and the TECs are visible. The core looks like it is milled with a flycutter, the alu plates too. The TEC’s are lapped and sealed versions. The sealing is nothing more as captan tape and kit.

Burned foam

Burned foam

Breaking away the last foam. The TECs are stacked in pairs of 2. There are 2 strings of 4 TECs. one string is made so there are 3 in series. “One” of the 3 is made of 2 parallel. A construction that is often used in dryblocks. All TECs are the same. Marlow DT-12-6 in the sealed and lapped version.

All the TECs removed. Three are toast

All the TECs removed. Three are toast

As said, 3 are toast. Nothing to loose and curious is a good combination so time for some experiments on the dead ones.

A TEC is made from a lot of small blocks semiconductor n and p material. They are all mounted in series (electrically, thermally they are parallel). This is done by soldering them on small metal strips. This is all mounted between two ceramic plates.  When a voltage is applied to the free ends of the semiconductors there is a flow of DC current across the junction of the semiconductors causing a temperature difference. The side with the cooling plate absorbs heat which is then moved to the other side of the device where the heat sink is.

So what was wrong with them ? I have seen and measured a few other dryblocks that where here for repair. Datasheets from the TEC state often the AC resistance. If a dryblock does not meet his specs anymore and the controller etc is OK, there is a good chance the resistance from the TEC is increased. I think they first come a bit loose, increasing the contact resistance and finally end open. You can measure the resistance per row and find the one that is wrong. In this case the problems where all on the outer rows nearby a corner.  As an experiment I managed to resolder them, that turned out to be not so difficult and the result is a resistance within the datasheet specs. I will only use the repaired ones for some experiments and tests to see how they hold.

 

E&H massflowmeter PCB repair

E&H flowmeter

E&H flowmeter

This one had a few fried resistors due to a short caused by damaged isolation from a wire.

Hot !!

Hot !!

I first removed the resistors. The picture above is after cleaning with IPA. The pads are loose but not damaged. The black stuff is burned epoxy. You can see the structure from the, now epoxy-less, FR4 glasfiber. You can pinch a needle through it as if  it were cardboard.

Bad things come in 2

Bad things come in 2

This is the other side.I milled away the damaged part, glued in a new piece. I then drilled holes and placed some vias. Then milled the new pads and traces and soldered new resistors in place.

The new traces

The new traces

the other side

the other side

The new resistors

The new resistors

Not very beautiful, but this sort of things are hard to photograph and always look worse then the reality.  I could have painted it and use some filler but that is a waist of time. It must be good and that it is.

 

PCB transplantation of a Fluke DMM

 

 

Before

Before

Fluke uses a resistor divider to tell the processor in what position the switch is. It is made from a sort of carbon film. At the picture above you can see it. This meter has also a problem with a partial dead IC so a nice victim to practice my PCB transplanting skills.

In this meter the black stuff is worn beyond cleaning, believe me I tried all the tricks from cleaning to pencil. I made a circle from very thin pcb, copied the layout from the black stuff to that. Then milled a “pocket” in the meters pcb  to hold  the new pcb. I made the diameter a few mm biger so there was room for resistors. Fluke changed this systemj later to one that used resistors. Glued it in place, soldered some wires to replace traces that had to go. The result was successful. It still does not measure but it now can switch into the functions.

After

After

Megger BT-51

Testing after repair

Testing after repair

This Megger is one that is made to measure milli and micro ohms. It uses a 2A current for this. The meter part uses a ICL7107, a multimeter IC that is around for many years. It is f.i. used in the Fluke 8020. It was dead but not a problem because they still make it.

When I can not find service documentation I must make my own schematic. For something like this it is easy. I make 2 pictures, fool around with them in the Gimp and this is the result:

the PCB

the PCB

The real thing

The real thing

Electrtechn laboratorium UH28M high voltage tester

UH28M_1

After repair

This tester is made unchanged for many many years. It can deliver 5kV and do 100mA. This unit was as dead as an instrument can be. The biggest problem was the power supply transformer that was dead. it has several windings including 2 that are over 2500 turns. Both of them where burned open. That is a bit to much to do by hand so I first made a winder. You can see it in action while winding this transformer.

The transformer

The transformer

The result

The result

Testing

Testing

Measuring voltage and current as a final test. Besides this it had more problems but now it is ready for another 20 years.

Valhalla 2701C DC calibrator

The front panel

The front panel

Repairing calibrators is always fun. This one had no output. It has the 120mA option but that was dead too. The fuse was blown, a logic-opto and a p-channel jfet were dead. The IC caps had a very high leakage current.  The jfet was obsolete but I found a usable substitute. The only problem was the sot23 package. That was easy solved because it fitted on the through hole pads.

The reference is a LM299 made by LTC. The 120mA option is located after the pushbutton section.

The 120mA option

The 120mA option and LM299

 

The inside

The inside

Bottom view with some modes

Bottom view with some modes

Fluke 123 scope meter

Na reparatie

Na reparatie (en wat oppoetsen)

A Fluke scopemeter made in 2010. It stopped charching and turning on resulted in a few seconds before it switched off again. A new battery did not solved the problem and that is why it arrived here.

According the owner it always had some problems. The psu is nothing more as a transformer , diodes and some capacitors. The voltage is 20V but drops to 15V under load.

Be careful when replacing this psu. Fluke has the nasty habit to use the center-pin for the negative rail. Besides that the connector body is a special size. The easy way is to reuse the cable and buy a 15V powerbrick that is screwed together.

The meter is like most Flukes build to survive a lot of abuse. The problem is you need to open it like a book. That is not bad as long as you do not need to look at the other side. And the connection between the two parts is not very strong so it is easy to tear of a connector or conductor. To tackel this problem I made a simple  frame to support the meter.Under the PCB and display the frame is open so it is easy to turn around and work on both sides

Laadstroom meting

“frame”

Een detail:

 

De bevestiging

 

The faults were a bit unusual. It did not charge because all three current sense resistors were open. Strange because they looked normal. After solving this it worked as long as you used the psu. But not on it’s battery unless it was fully charged. So there must be an other problem.

And then the manual does not help anymore and you are on your own. Also because the manual has some faults (or is for an other version). A resistor and two testpoints are  wrong. But this is easy solved with the scematic, chapter 3 and a calculator.

The cause of the problems are the smd 150uF/6,3V polymer electrolitics. For the ESR fanboys, an ESR meter was here of no use. The datasheet specs for ESR are for 120Hz (like for most caps) besides that they measured still within specs for all the parameters. There is a better way to look for cap related problems, using a scope to measure ripple. And that ripple was huge on the 3V3 rail. It looked more as AC with a small DC offset. A DMM with not enough bandwidth and crest factor would measure this as 3,1-3,2V.  I replaced the caps for a version that was capable of handling more current.  And this is not some magical thing, it is stated in the datasheet.

I do not know if this just was a bad series of caps or just a bad choice from the designers who are after all just humans like you and me. After that I checked all testpoints to see if there are other problem or almost problems. Then cleaning the pcbs and closing the cabinet. The final step is testing all functions and monitoring the charging over 20 hours.

Some PCB details for the fans:

De bovenkant

top

De onderkant

bottom

Hameg HM-8118 LCR meter

This LCR meter was dead. Hameg was always very helpful in providing manuals.To bad this is not  the case for the newer gear.  This is a very nice LCR meter with a lot of useful functions.

De voeding

The power supply

This is the power supply. The quality from the soldering is not very good, and it was dead. Strange enough the rest of the meter was build very nice.

After the repair was done it was time to test it

Testen met een ESI-1010 standaard

Testing, using an  ESI-1010 standard

Testen met een GR 100mH luchtspoel

Here a GR 100mH inductor standard

 

Behlman BL-1350

Heel dikke trafo's

Heel dikke trafo’s

Dit is een Behlman BL1350 1,35kVA AC powersupply.

Een erg zwaar apparaat dankzij 2  enorme trafo’s. Deze was gemodificeerd, waarschijnlijk voor een vaste meet opstelling. Het gevolg was dat hij niets meer deed. Dit was gedaan met een extra PCB van experimenteer printplaat en twee ceramische steuntjes met een stuk of 6 shuntweerstanden. Daarnaast was de Volt potmeter afgekoppeld. Afkoppelen was niet lastig maar ze hadden ook andere dingen gedaan waardoor het goed zoeken was.

Qua reparatie stelde het niet veel voor maar het was een erg imposant apparaat.Output is 0 tot bijna 300VAC.

Heel dikke MOSFET

Heel dikke MOSFET

Niet alleen de trafo’s waren groot, de mosfet kan er ook wat van.