Author Topic: Repairing a Rofin laser power supply (Read 5440 times)

macona · « **on:** April 10, 2021, 06:29:59 am »

I picked up a Rofin CO2 laser condition unknown today. Its a 300W unit that has a head and a separate power supply. the head is connected to the power supply with a big 7/16 style RF cable and a DB25 for signals and control. Looking at the manual I saw that the power supply is composed of three individual 48v power supplies which are in series so thinking I could rewire it to run off single phase, just parallel the power supplies.

So I took off the back cover when the smell hit me in the face, a strong chlorine smell. And then I noticed that the insides where covered with a brown stain. NOT GOOD

IMG_1583 by Jerry Biehler, on Flickr

I disconnected the in and out leads and slid the power supply out. Top cover has more of the staining so ti was definite coming from there. I popped off the cover to find this:

IMG_1589 by Jerry Biehler, on Flickr

IMG_1587 by Jerry Biehler, on Flickr

One of the output inductors is totally fried. So two things I notice, one, it is not three individual power supply like the manual shows, it is one giant 48v power supply with rep output transformers with he primaries in series and the secondaries in parallel after rectification. All driven by a mosfet h-bridge. The mosfet drive board can be seen at the top center above with the fan which has seen better days. Second thing is that other than the smoke the damage is limited to the output inductor, the hall current sensor, and the fan.

But why only one inductor fry? The transformers are in series so if one dies they both will die so the load can't become uneven that way, if the RF power supply had an issue the whole unit should have faulted out since the current sensor is on the output. I pulled the buss bar off the output of the diodes and checked them and they are all fine and also the transformers which appear to be fine, I need to grab my 34401A with the kelvin clips to check them out for sure, it is too low to measure with my hand held. Maybe a bad connection on the inductor cause a runaway heating event and just killed it?

I think its repairable, The ferrite itself won't care about the heat, I can rewind it with suitable wire and make it match the existing one with my old LCR meter. The current sensor turns out to be a Chinese knockoff of a LEM125-P which is about $30. And the little fan is a generic 12v 40mm fan, I probably have one of those lying around someplace.

So first thing I am going to try and clean off some of the boards, the mosfet driver is the worst since it just sucked the smoke through. Get that back together and then check out the current sensor, easy enough to test, I need to get a new connector for the ribbon cable, it melted.

A couple more pics of the carnage, the first one is the inside of the top cover, you can see how the little fan circulated the smoke.

IMG_1597 by Jerry Biehler, on Flickr

Current sensor:

IMG_1594 by Jerry Biehler, on Flickr

Mosfet h-bridge board.

IMG_1599 by Jerry Biehler, on Flickr

This thing smells sooo bad.

macona · « **Reply #1 on:** April 11, 2021, 05:16:24 am »

I have been told that the problem might have been a failure in the toroidal core of the inductor, if it cracks it can cause localized heating like this. After unwinding I found a crack though the circumference of the ferrite. Maybe that's it?

I put the good one on the LCR meter and it measured in at 46uh at 1khz, I found a core the exact same size and looks like if I get one with a 60u permeability I should be able to match it with the same amount of turns. I need to see if I can get ahold of one from the manufacturer, the only place I can find with stock is in the UK. The windings are composed of 8 14ga enamel wires which works out to a little less than 6AWG in cross section. Should be easy enough.

Everything else is looking pretty good so far. I found that mineral spirits dissolve a lot of the deposited gunk, so I got some more and ordering some brushes to at least clean the PCBs. Replacement fan is ordered. I cleaned up the mosfet driver board and the LEM board. After prying the melted plastic off the old IDC ribbon connector I yanked off the two tag tant caps and checked them, they are supposed to be 10uf, one was good, the other measuring at 3uf. I hooked the sensor to a small +/-15v supply and the output to a ammeter and a lead through the hole from my bench supply, the knockoff LEM still works following the current set by the supply pretty accurately.

The cap board which has the 10 3.9k resistors on standoffs seems to be in decent shape, there are 10 33uf caps in parallel and they measure in at 354uf so that seems to be in tolerance. I am just going to replace all the resistors. No point in keeping them.

So a few more parts to order and cross my fingers this thing works, there is a guy on ebay selling three of these power supplies... for $3500 ea. Hah

macona · « **Reply #2 on:** April 15, 2021, 06:53:42 am »

Got a new core coming from Micrometals. I need to scrounge up some 6 AWG wire to wrap a new one. Replacement parts for the damaged components have started to come in, the new fan is installed. Cleaned up the mosfet driver board with mineral spirits, its about the only thing that cut the oily reside from the burning heat shrink. I tried cleaning another board but it didnt make much of a difference. I grabbed a meter for checking ESD stuff like mats and packaging and used that to check if the residue is conductive and its not, so since there has been no corrosion in the years this has been sitting I think I am pretty safe leaving it as is.

I found the manual and dug through it. They actually had a wiring schematic for the laser with a lot of wires marked. Found out that the had two different power supplies that they used, both rated 48v at 150A. The other one used three 50A supplies in parallel to get the 150A. With the info from the schematic I found that the little board that interfaces the laser umbilical to the power supply unit supports both supplies and managed to figure out the control cable to the power supply I have. Its pretty simple, there is a 12v aux out, ground, two wires that when shorted enable the power supply and one analog signal out. I am not sure what it is, its not documented, but if I had to guess it is probably the output current of the power supply. The other power supply does not have this connection so I can probably ignore it.

The good news is that if I can't repair this thing I can just use a off the shelf 48v power supply, well, a few of them. A friend has some 48v 73 amp supplies that should work just fine.

laserschematic by Jerry Biehler, on Flickr

TheMG · « **Reply #3 on:** April 15, 2021, 01:46:47 pm »

One thing to watch out for, look closely at the good inductor. Are the individual wire strands enamel coated? If yes, then you can not just replace with a regular stranded wire.

Quite often individual enamelled strands are used especially in high frequency high current applications to reduce skin effect.

ace1903 · « **Reply #4 on:** April 16, 2021, 07:35:28 am »

Could you please pictures of the RF part of the unit(or all the rest parts)?
I have some LDMOS transistors from GSM base stations and wondering how difficult is to make RF part for laser or some plasma experiments.

macona · « **Reply #5 on:** April 17, 2021, 08:06:35 am »

OK, thanks for letting me know about that, Ill pick up some 14 awg magnet wire.

I don't plan on opening the RF section unless I have to. But there is a guy that has done some work on the same model and has pics here: https://translate.google.com/translate?sl=auto&tl=en&u=http://www.altortech.es/wp/recargas-laser-refill/rofin/recarega-laser-co2-scx30-rofin/

macona · « **Reply #6 on:** April 18, 2021, 06:51:01 am »

Toasty components have been replaced. Resistors on the filter board and I did wash that board and the current sensor board with mineral spirits audit cleaned it up pretty good like. I also got some G10 and made another inductor mount to replace the burned one. I bought a roll of 14awg magnet wire and a bunch of heat shrink to make up a winding with 8 strands like the original. Hopefully that will all be here around Tuesday. The RF cable I had made showed up today too, 92" long from KMR-600 type cable with 7/16 ends. So hopefully this coming week I will be able to find out if this thing works. Ill probably just use a garden hose to keep the thing cool.

IMG_1637 by Jerry Biehler, on Flickr

macona · « **Reply #7 on:** May 01, 2021, 08:08:36 am »

Things are getting a little closer, the magnet wire and ferrites showed up Monday and I matched the length of the old wire and wound them. Only to find I was way too short. So I just wasted a bunch of magnet wire. Ordered some more and that showed up today. Today I made it much longer, soldered a lug on one end and then wrapped it. I decided to use kapton tape instead of heat shrink since my wire is slightly heavier which is why I think I came up short the last time, the extra diameter made it take more wire. I just wrapped it till I got to the inductance I was looking for, ended up at 47uh which matches the other within a very small margin of error. Clamped that to a new base made from some 1/8" thick FR-4 and put the thing together.

Powered it up. Nothing, eventually I see a led blinking and hear a ticking sound from the low voltage power supply section of the power supply. That's a pretty standard sign the power supply is overloaded. I almost gave up here, but I decided to disconnect other boards that feed off the power supply, pulled the mosfet driver connector and the current sensor connector. Powered it back up and no ticking and the main relay engaged. So I hooked the mosfet board back up. Fan comes on and green power supply good light comes on. Woo!

Hooked up the control cable to the interface and jumped the interlock and check the output. 60v and the new indictor is not getting hot. Im not surprised the voltage is high, the design has no bulk capacitance on the output.

Looking at the sensor board and I remember the issue, I had ordered the wrong kind or ribbon wire-to-board connector and did a make it work thing. Well. It didnt, it shorted the +/- rails right to ground. So I need to fix that. Too bad I had to take it all back apart to get to the sensor.

So the next thing is to get a big cap on the output and put a load on it. I have a 48v input 1kw inverter that I can at least put some load on it to see what happens and make sure the power supply pulls down to where it should be. The power supply is also water cooled so I have to deal with that.

IMG_1674 by Jerry Biehler, on Flickr

IMG_1686 by Jerry Biehler, on Flickr

IMG_1681 by Jerry Biehler, on Flickr

macona · « **Reply #8 on:** May 03, 2021, 04:55:23 am »

Got the IDC connectors today and installed it and put the thing back together.

Its alive. Output dropped to 47.9v once I connected the sense leads. It had looked like the sense was connected internally but it looks like it was set up just to limit open circuit voltage if the sense was not hooked up.

I hooked up the inverter and plugged a blower into it which draws over 1kw durning startup and the power supply never noticed it. https://youtu.be/Vp-OsIASEmw

I plugged the head umbilical to it and a couple LEDs came on so it looks good so far.

Next step is to roll it all outside and hook it up and plumb the water cooling, for now I will just use a garden hose to supply water. It has a flow switch in the head which needs to be satisfied before it will power up the RF section.

Untitled by Jerry Biehler, on Flickr

PA4TIM · « **Reply #9 on:** May 03, 2021, 08:35:06 am »

Do you have a link to the manual of this units ?
I repaired 2 of them. I am not afraid of toasted repairs (like holes in PCBs and rebuilding inner layers) but respect that you managed to get this thing in this poor shape back to work. https://schneiderelectronicsrepair.nl/?p=667 I have some pictures on my site. I have more pictures if you need some details

macona · « **Reply #10 on:** May 03, 2021, 09:26:40 pm »

Here is a link for the manual. Rofin did transfer to a new company but they will not support the older lasers. https://drive.google.com/file/d/1lG4_BgPkRlXXt4Co4yaVshGmox7OPLwZ/view?usp=drivesdk

Looks like yours is slightly different. Mine is the HPC-818-1 vs your 816. Wonder if the last number is the capacity in KW. Looks like your main transformer and output inductors are the main difference.

I’ll post pinouts for the db-15 later.

james_s · « **Reply #11 on:** May 03, 2021, 09:54:41 pm »

Wow, that thing really cooked, I'm surprised it didn't have any sort of protection circuitry to catch the failure and shut it down. RF excited CO2 lasers are crazy beasts, the power supply is pretty much just a big radio transmitter outputting an unmodulated carrier.

macona · « **Reply #12 on:** May 04, 2021, 04:43:38 am »

There is a LEM type current transformer on the output that would detect a short on that end and there is a CT on the primary side of the transformer but this failure was undetectable because it was before the output and when it fried it was probably just getting hot enough to burn off the heatshrink of the wire wrapped around the torrid, eventually it burned off the enamel from the magnet wire and shorted itself out.

Yep, big ole radio transmitter, 81MHz at about 7kw. Peak pulse power of the laser is about 700w if it was running on the 10.6um line and 300W QCW. This one runs on the lower gain 9.3um line and is supposed to be 225W QCW and probably 575w pulsed. I have a couple others, a little coherent GEM50 that uses a 100MHz 600w supply and a Coherent 150 that has it's RF power supply integral. It uses a 48v, 3kw power supply.

PA4TIM · « **Reply #13 on:** May 04, 2021, 08:43:13 am »

Thanks for the manual.
The PSU was, according the customer for a 6 kW RF, but efficiency of lasers seems to be very poor zo that is not the laserpower itself. This was a huge table/bed (2x4 meter or so) and it cuts/engraves things like paper and plastic. Metal cutting lasers must be real beasts. Pretty bizar if you think of it, cutting metal with light (I know the physics so I know why it is possible but it is still pretty cool)

I worked on a RF while it was connected to the laser. We had a old Rigol scope standing near it while the RF cabinet was open. That was pretty useless.
I do not know much about lasers. I just repaired some laser stuff for a customer who repairs and service these kind of beasts but not on board level. That is my work. I do not do field repairs but I made an exception because I find lasers very interesting. So it was very cool to see one operating. I had no clue the laser itself was not in the moving head. So I told him, somethings goes wrong when something in the the right front side of the laserbed started to glow and smoke. Turned out he used a brick to block the laser while we worked on it. So he showed it, the brick was glowing. Very impressive.

PA4TIM · « **Reply #14 on:** May 04, 2021, 08:49:42 am »

Do you have experience with the RFs, how critical is the matching of the mosfets ? And do you have experience adjusting them ? 6kW is nmot something I like to do by trial and error

macona · « **Reply #15 on:** May 04, 2021, 09:30:39 am »

I have worked on CO2 lasers up to 4400W, those are a whole different design, flowing gas with turbo pumps and DC excited. That machine would cut up to 1" of steel. The company I work for now builds laser micro machining systems that drill via in pc boards, process flex PCBs, and a few other things.

Im hoping I can cut some thin metal with this, I am sure I can cut thin steel with oxygen assist. Also do laser welding with it. I have a big pulsed yag laser welder that is almost 40 years old, if this works I might replace it with this laser.

I have no experience with the RF on these units. Im crossing my fingers it is in working condition. I should know in the next couple days.

There is this guy that did repair a RF section of one of these lasers, the site is down right now: http://www.altortech.es/wp/recargas-laser-refill/rofin/recarega-laser-co2-scx30-rofin/ There seems to be a partially complete copy over on archive.org https://web.archive.org/web/20210124150945/http://www.altortech.es/wp/recargas-laser-refill/rofin/recarega-laser-co2-scx30-rofin/

The construction is... unique, I guess. I have not seen any other 6kw RF power supplies so I can't really judge.

PA4TIM · « **Reply #16 on:** May 04, 2021, 12:02:52 pm »

I have picture of the insides of an RF on my site (link in post above). It is completely "dead-bug" build. The PCBs are in parallel and watercooled. On the other side is a similar unit. Together 6kW. On the backsitre of each unit is a combinere and directional coupler. The units on the picture had resistor problems due to some fluid and a rough transport to Italy and back. Some of the bondings where loose or only making pressure contact. That part of the repair was succesfull. The rest of the parts I checked as much as possible. A problem that still exist is that both units work but there is too much power difference. Maybe the gate bias is adjusted with the bad directional couplers, so the control electronics thought both gave enough power. I wanted to measure the current and gate voltages but as soon as that thing starts to transmit the scope became useless.

The guys in Italy say the mosfets are special versions due to correct phase matching, a guy in England says it does not matter. The factory uses mosfets with different Vgs specs in one unit so I think Vgs matching is not a big thing and is adjusted using the 10 turn pots. It regulates power by switching the RF output for a short or longer time period.

Are you sure about the 80 MHz ? I had no frequency counter there but according the owner is runs at 40 MHz. That sounded valid to me because the 40 MHz band is used for all kind of things.

macona · « **Reply #17 on:** May 05, 2021, 04:45:32 am »

Says 81MHz in the manual. Coherent uses 81 on other lasers they make too, so that's probably it. The power supply you are working on is for a different laser so maybe it runs at a different frequency.

Well, one step forward and two steps back. Hooked everything up today, ran water lines, hooked up the rf cable, made a little board setup that I can use to provide the PWM signals to the laser and satisfy the interlocks and... nothing. The light for the interlock went off and you could hear the big power supply turn on but nothing else. No other diagnostic light. I checked the fuse on the power supply that supply the low current 48v to the head and it was bad. Replaced it and it fried too. Opened the head up and did smell something a little funny.

IMG_1699 by Jerry Biehler, on Flickr

Well crap. Thats not good. Lol, I managed to find one of the pieces that was blown off and it was a LMLM2576HVT switching regulator. But which one? Tracing out the circuit it parallels in and kind of ORs in with the 12v aux coming from the main power supply, so pretty good chance it is 12v, I went back outside where I took it apart and got incredibly lucky and found the other piece and it is indeed 12v out.

But what caused it to fail? I pulled the IC out and checked the output trace to ground and a dead short. I pulled the output cap off and still a dead short, pull a little black SMT component off in parallel with it and the short it gone. This little bugger:

IMG_1702 by Jerry Biehler, on Flickr

After some digging I figured it is made by General Semi then eventually I found the code and it's a SMBJ13A‎ which is a TVS diode, 13v. So something happened, maybe when the main power supply went haywire it shorted this out and killed the regulator IC. Everything else looks good and the replacements are on order from DIgiKey. I did also find this laser has 5k hours on it, supposed to be good for 10k. So crossing my fingers again.

macona · « **Reply #18 on:** May 14, 2021, 03:45:46 am »

Got the parts today, digikey is taking forever to ship stuff right now. Put the part in and hooked it to my bench power supply at 48v and the leds on the board lit up and a relay clicked, so it looks like that fixed it. Hooked the laser all up again and got water flowing. Shutter opened and all the warning light went off except for a yellow "modulation over" which is supposed to light up if the modulation goes past 50% duty cycle. Odd, but yellow lights won't stop operation. I enabled the rf supply and applied a signal... and nothing. I start checking signals, its getting 5v on the enable line to the RF supply, differential signal to the rf supply as well. It should be working with just those signals.

So I pop open the power supply. Everything looks good. There is a little box where the DB15 for the control signals come in and pop that open. Look at those components and notice the 8 pin dip looks a little funny. I pulled the board out and yup, she's fried, and it's the voltage regulator. That and the little 4.7uh SMT inductor is open as well. So ordered those parts as well, as well as a couple extras of all the ICs in the box. My worry is that when the regulator went it took out things downstream. If it killed the microcontroller im SOL. Probably. All the outputs on the feedthough caps are labeled so it might be possible to reverse engineer and get it working. It appears to limit the duty cycle of the signal, there is a jumper that sets it to 40/50/60% and it monitors the vswr to shut things down if things become unmatched.

The culprit:

IMG_1728 by Jerry Biehler, on Flickr

Rest of the board:

IMG_1721 by Jerry Biehler, on Flickr

Top of the RF Supply

IMG_1725 by Jerry Biehler, on Flickr

One of the individual drivers

IMG_1727 by Jerry Biehler, on Flickr

macona · « **Reply #19 on:** May 25, 2021, 09:21:53 pm »

I got the parts the other day and installed the new vreg ic and powered the board up on the bench. Had the power supply to limit to 200ma @48v and immediately the new ic got hot. Tried to see what what shorting but the vreg died before I could see what was getting hot using a thermal camera. So I yanked it and put power directly to the 5v rail on the board and watched with the camera. One IC lit up like a Christmas tree. I had ordered spares of all the ics on the board, though one was obsolete so I had to order some from china though ebay. Guess which one died? Yup. That one. a DS8922AM which is a dual differential line transceiver and has no crosses. Tried contacting the seller to see when they were going to ship and still have not gotten a response so I ordered some more from a different seller with faster shipping. Though I just realized its the same seller so I may have screwed up.

I had pulled out what I thought was a microcontroller before powering up the board because that part would be unobtanium if it was fried. Turns out it is a FPGA. Pulled the fried IC off and soldered in a new Vreg and put the FPGA back and and powered it through 5v and then 48 and it seems happy. Now I just have to wait for the ICs to show up.

IMG_1738 by Jerry Biehler, on Flickr

macona · « **Reply #20 on:** June 02, 2021, 05:57:29 pm »

Got the ICs from HK last night and installed it on the board and put the RF supply back together. This morning I hooked everything up and it works. No smoke other than what I burned with it. I have not put a power meter on it, I need to get the water rigged up to cool the power head I have that can read this high. Now to put all the screws back in and button it up.

I honestly can't believe I managed to fix this thing.

https://youtu.be/PSc9BsQCB3E

atorrico · « **Reply #21 on:** November 29, 2023, 03:35:37 am »

Could you help me test a DC source for a Laser machine, I am looking for a manual or electrical diagrams


EEVblog Main Site	EEVblog on Youtube	EEVblog on Twitter	EEVblog on Facebook	EEVblog on Odysee

Author Topic: Repairing a Rofin laser power supply (Read 5440 times)

Share me