Hybrid IC questions

I have a CNC machine with a FANUC control. The control uses what the web calls "hybrid ICs". It looks like these devices consist of some sort of IC and/or others ICs and pasive devices. The control is old enough that getting new boards is kind of a crap shoot. And I have replaced bad boards with good used boards only to have them fail sometime later. I want to have a stock of known good boards. So I need to test the boards but I don't really have the expertise. I'll need to figure it out as I go. One thing that might help is to find out what is under the epoxy or whatever is covering the devices on the hybrid ICs. These are parts from the 1980s. Can the coating be removed without damaging whatever is underneath? Is it likely the devices on the hybrid board will be identifiable once exposed? I have looked for datasheets and pinouts on the hybrids but no luck yet. Maybe I need to look harder. Thanks, Eric

Reply to
etpm
Loading thread data ...

To your 2 questions: no chance.

NT

Reply to
tabbypurr

On Tuesday, December 18, 2018 at 12:16:02 PM UTC-5, snipped-for-privacy@whidbey.com wrote :

The hybrid ICs that I have done postmortems on usually had the semiconducto r dies directly on the substrate, so that there would be no individual part numbers, polarity markings, or road mapping.

I used to repair plasma IPMs, but most of the time you could see the IGBT d ie blown right off the board. I used to put those on a heat block (ceramic substrate, **very** hard to wet solder on if the heat soak wasn't mitigage d) and solder SMD IGBTs to the substrate after carefully cutting away the b low dies, assuming there was room.

But those only had a viscous snot potting the circuit and was easily remove d. Others like Sanken STK hybrids did not use potting beneath the outer co ver.

Reply to
John-Del

Some ingenious folks manage it, though! I believe they use a strong acid, then when all the muck is washed away they reverse engineer the chip underneath. Fucked if *I* could!

--
This message may be freely reproduced without limit or charge only via  
the Usenet protocol. Reproduction in whole or part through other  
 Click to see the full signature
Reply to
Cursitor Doom

Hybrid might also refer to a digital IC on the same die as an analog IC. It doesn't matter, though. As others have said, it is unlikely that you could ID parts and replace just what went bad. It might be easiler (but still very difficult) to use one of the newer single chip micros or hobby boards to emulate the entire bad circuit board's function.

Pat

Reply to
Pat

Sorry to be crude, but it is a "fool's errand" to try to keep one of these old controls running. I know, I did the same, but moved to something maintainable as soon as it was available. You ought to look at LinuxCNC. It can almost certainly do more than your ancient Fanuc control, and it is entirely repairable. You can use the existing motors, Fanuc's were VERY good! Some of their encoders are proprietary, but my company (Pico Systems) makes converters for most of them. You can do a complete retrofit for less than a couple used boards will cost.

I've been using LinuxCNC since 1998, have had to upgrade old PCs several times as they got obsolete and cranky. And, I've upgraded the LinuxCNC software several times, as new, "gotta have" features were added.

Jon

Reply to
Jon Elson

Yep. A local machine shop had a Mori Seiki 5 axis mill with a Fanuc controller. Lousy photo at top of: I was the unofficial repairman. The owner of the shop sold it about 3 years ago and retired. He's still around, but all the Fanuc paperwork and spare boards went to the buyers.

I also used to design and make hybrid electronics in the late 1960's for Alpha Electronics, a manufacturer of 2-way radio accessories. The hybrids were used in CTCSS encoder/decoder boards and in early LED digital watches. Under the hybrid is essentially a custom "circuit board" build on an alumina ceramic substrate, with conductive traces and resistors silk screened and baked onto the substrate. All of them were custom made for some specific purpose, which makes them very difficult to source should the supply dry up. For example, the HP8640B RF signal generator has a hybrid RF power amp output stage that tends to self destruct. HP no longer makes or sells them, so creative repairs and replacements are epidemic. You're probably going to be faced with a similar problem. It might be possible to replace a hybrid with a work alike device on a PCB, but then you would need a schematic of the hybrid internals, which you probably don't have and is probably difficult to find. Here's one attempt to reverse engineer the HP8640B hybrid: Not the lack of detail, such as xsistor part numbers.

I did some of that many years ago. Similar problem, but no hybrids or exotic parts. I ended up building a test fixture with a spring pins at all key test points. The fixture simulated the operation of the machine, which was a controller for an old rotary blade lumber mill. You could do the same with the Fanuc controller boards, except I seem to recall that there were about 8 boards in the controller cage, which would turn this into a major project. I only had to make one fixture.

No and no. You could probably soften the coating, but if you try to pull on it, it will tear apart the wire bonds even if the transistors and IC's are coated with junction compound.

Have you considered removing all the Fanuc stuff and replacing it with a more modern CNC controller and motor control? I did that a few months ago on a Shizuoka vertical mill replace an aging Bandit controller: Total cost of the 3 servos, 3 motor drivers, controller, and a mess of incidentals was about $2,000. The choice was to try to keep the Bandit alive a little longer, or to rip it out and start over. Starting over turned out to be better, easier and cheaper.

However, if you want to continue to do battle with the Fanuc controller, you might check on the Fanuc forum on CNCzone:

Hint: If really you want help with something, it's usually helpful to provide the model number of the machine and controller.

--
Jeff Liebermann     jeffl@cruzio.com 
150 Felker St #D    http://www.LearnByDestroying.com 
 Click to see the full signature
Reply to
Jeff Liebermann

To Jon and Jeff, Yes, I have considered replacing the control. And I consider it more every day. I have done this before and it worked well. This machine has a turret, tailstock, and counter that would need some sort of PLC programming to interface with the control. I might need to learn ladder programming again. And you are correct Jeff, I shoulda provided the control info. The control is a FANUC 3T. Tha part is a FANUC A-RV06. The problem with changing the control is that I need enough time when the machine is not needed to make the swap. The bigger problem is that the original control has, until recently, been robust and completely trouble free, and I am nervous about ripping it out. I just need to do the research on new controls, how to interface the turret and other stuff, and then go for it. But now I have parts to run. Thanks eveyone, Cheers, Eric

Reply to
etpm

Ok. It's a lathe. Any particular maker and muddle number? Servo or stepper drive? What drives the spindle motor? Do you have a VFD phase converter running it which can be remotely programmed?

If you want specific help, names and numbers are needed. If you want general help, strategy, and sympathy, just leave out the specifics.

Thanks. Looks like there are a few available from the parts cannibals on eBay (from Poland, Canada, and New Jersey): There are probably others listed, but are difficult to find because of all the creative titles and part numbers.

This looks like a potential OEM source: Lots of Fanuc boards and parts on the site.

Actually, your biggest problem will be decoding the English translation of the manuals from Chinese.

I don't think you'll have any problems interfacing to a tool changer, tool turret, or power tailstock. The counter can be replaced by a shaft encoder. The lathe version of the cheap controllers that we used had provisions for all of these and more.

However, you're correct to question whether it's worth replacing a working and reliable machine simply because of the non-availability of one hybrid. Such conversions tend to be a one way affair, where it it impossible to put things back after the conversion begins.

I would take a known dead A-RV06 hybrid and do whatever it takes to reverse engineer the circuitry. I couldn't find anything with Google search. Initially, an X-ray photo would be a good start. Then, do a chemical attack on the conformal coating which I can't identify from the photos and can't remember from the real thing. If it's hard, it's probably epoxy. If soft, silicone epoxy or urethane. Try heat, acetone, nitric acid, or hot MEK to soften the epoxy.

This is for a COB, but offers a clue as to how it can be done:

This looks like the back of the hybrid: and might be useful to identify the large parts.

Once you have access to the wiring and some of the components, measure the passive parts. Look for laser trimmed resistors. Try to identify the active components. Eventually, build a schematic. Then, try to design something that works in the same manner, but which uses modern components. I once reverse engineered the hybrid RF front end for a Cushman CE-6(?) service monitor, which I ended up replacing with two MMIC chips.

Good luck.

--
Jeff Liebermann     jeffl@cruzio.com 
150 Felker St #D    http://www.LearnByDestroying.com 
 Click to see the full signature
Reply to
Jeff Liebermann

The reason I wondered about the hybrid is not to fix it but to find out what it does so that I can figure out how to test it. The same board keeps failing in the control and I don't know why. The counter is different than just a a part counter. It increments by an M code. So only if the program has this code does it count. Once it reaches a pre-set number it send a signal to the machine to stop once the program running ends. The control itself has a switch for continuous running that works in conjunction with the counter. For many jobs this feature is a huge plus. I load stock and the machine keeps making part after part. I can program the machine to make some many parts and then stop with another M code but if the program needs to be restarted for any reason, which is common when setting up, the count is also restarted. The counter only counts when the program actually finishes a part. So whatever new control I put on must have this feature. It is very convenient and saves lots of time. Eric

Reply to
etpm

On Wednesday, December 19, 2018 at 11:27:09 AM UTC-5, snipped-for-privacy@whidbey.com wro te:

If the hybrid is potted in epoxy then I don't see any way of getting lucky. Any attempt at depotting epoxy will likely add to it's problems.

If it's potted in a soft silicone snot or not potted at all, you might be a ble to spot something. If you need to do any soldering you MUST preheat th e crap out of it as the ceramic substrate will steal heat like an Amazon po rch pirate doing his Christmas shopping.

Reply to
John-Del

Third try: If you want specific information, models and numbers are needed. You get zero points for being as vague as possible.

Any particular board in the Fanuc 3T that is failing? Do you have the Fanuc 3T maintenance manual? With a schematic, you can probably guess what the A-RV06 hybrid does for a living by looking at the input/output pin labels. Ah, found the manual: 24.5MBytes Ugh. Not much in the way of schematics and no mention of A-RV06.

At one point, we were blowing up servo amplifier boards and modules. By chance, I put my clamp ammeter around the drive lines and found that the common mode current was not zero. I scope later showed some rather large spikes. What had happened was that the frame and case grounding had gone intermittent because the screws had rattled loose. A little oil and rust in the under the ground lugs, and fairly soon there was no connection. I couldn't see it with a continuity tester so I systematically cleaned and tightened every ground frame screw that I could find. It worked.

M54 for Fanuc, M30 for most others. A quick skim of the Chinese mill controller manual shows M30 is supported, but not M54. Also, I was thinking of the spindle turns counter in the lathe version of the Chinese controller that is useful for cutting threads. Sorry for the confusion.

--
Jeff Liebermann     jeffl@cruzio.com 
150 Felker St #D    http://www.LearnByDestroying.com 
 Click to see the full signature
Reply to
Jeff Liebermann

Boy Jeff, I am totally not trying to be vague. I was looking for just generic help but really should have provided as much info as possible. Especially since there are all sorts of hybrid ICs and they can even be on regular circuit board substrates, from what I have seen. So, to be specific and complete, which I should have been doing all along, here is the info: The board is the A20B-0008-0630/02A. It is also known as PC1 for this control, the 3T. The board handles the I/O from and to the machine. I do have all the generic 3T manuals from Fanuc as well as the manuals from Miyano. All the wiring manuals available. They do not show any circuit board specifics, just where wires and connectors go. The first time this board failed, a couple years ago, the machine had various alarms that said a certain switch was open. I forget exactly which switch. But it was for a door or cover. Anyway, my machine does not have this switch. Yet the board thought that it was open. Just to make sure I traced the pins on the board connector to the cable connector and added all the optional switches. Every switch would show its state in the diagnostic screen. and all would change a bit when on or off except for the switch that was supposedly causing the alarm. Replacing the board fixed the problem. Just recently I posted about a parameter changing and how changing it back made the machine work properly. I was being hasty. The fix lasted one part. The machine went back to ignoring programmed speed commands and the signal to the part counter went away. The changed parameter did not change again though. Fanuc thought that the PC1 board might again be the problem so I bought a used one and the machine works properly again. Fanuc thought the board might be the problem because it handles the I/O and because the control thought that it was executing the M25 code. The start light flashed and the machine didn't hang up as it would if the control didn't get a FIN signal after executing an M code. And even though the speed signal to the spindle drive comes off of the main control board, the speed control dial and all the switches all go throug the PC1 board. Thanks, Eric

Reply to
etpm

LinuxCNC contains a real-time version of Classic Ladder, a ladder logic implemented in software, that can be interconnected with other components in the motion and G-code interpreting sections. This is great for tool turrets and similar systems.

Yes, certainly the classic conundrum! But, once you do it, you will be greatly relieved to know that the system is 100 X more reliable, and you can fix anything that goes wrong with modern and affordable parts.

Jon

Reply to
Jon Elson

I'm pretty sure somebody did this about a year ago in LinuxCNC. It should be pretty trivial to implement this even in the G-code, with the entire program in a do while loop, and the counter as a numbered variable. You could also display the counter value on the virtual control panel.

Jon

Reply to
Jon Elson

I appreciate your posts Jon. I have been looking at LinuxCNC as has my son. In an effort to convince me to upgrade the control. You said: "I'm pretty sure somebody did this about a year ago in LinuxCNC. It should be pretty trivial to implement this even in the G-code, with the entire program in a do while loop, and the counter as a numbered variable. You could also display the counter value on the virtual control panel." I don't know what a do while loop is. I'm sure it's easy enough to find out though. As long as the programming doesn't change I don't really care. I mean, as long as all my old programs will work the same without needing any changes. Having the count displayed on the main display is fine but it would be best if the original counter operated the way it does now. That is, a number is entered via the thumbwheels on the counter and then when the displayed count reaches that number the counter sends a signal to the machine and it stops. Because the counter is not involved in the programming, except that it is incremented by an M code, the count can be changed while a part is running. Which I have done more than once. If I was able to read the ladder programming in the eproms that control the machine would I just be able to cut and paste it into the LinuxCNC software? Man, would that make life easy. If a breakout board can be made to act just like the existing boards then maybe the control will be pretty close to plug and play. Can it really be that easy? Eric

Reply to
etpm

This might help with the parts counter function:

"Parts counter M code?"

--
Jeff Liebermann     jeffl@cruzio.com 
150 Felker St #D    http://www.LearnByDestroying.com 
 Click to see the full signature
Reply to
Jeff Liebermann

LinuxCNC has added a number of features to the basic G-Code language, turning it into an actual programming language. So, you can write G-code that is "parameterized", so you can set a few variables and then the G-code loops and computes the actual coordinates based on those numbers given at the start.

In your specific case, you would enclose YOUR entire G-code program within a couple lines, like this: #1 = 500 (do 500 parts) O100 WHILE [#1 GT 0] < your program goes here > #1 = #1 - 1 O100 ENDWHILE M02

See

formatting link
for more info. But, the sceme here is you set user variable 1 to 500, then the WHILE loop executes your program 500 times, and decrements variable 1 at the end of each run of the program. When variable 1 decrements all the way to zero, the machine stops. And, to get tricky, LinuxCNC has a virtual control panel that can put indicators and buttons on the screen. I use it to show touch probe status and spindle speed, but you can put nearly anything on there, such as the value of that variable.

That may involve adding a couple IO points to the hal configuration, but could certainly be done.

Jon

Reply to
Jon Elson

I can already do the above with an M code. The machine counter works a little differently.

Our power went off during a windstorm last week and then holiday stuff came up but today I finally got back into the shop and had a look inside the control cabinet to get an idea of what would be involved, doing a swap. There are 25 connectors from the main control board going into the machine side. I want to use these connectors, all of them. So if Honda connectors are still available then it looks to me like the best way to go is to build a breakout board to interface the new control to the machine through the existing Honda connectors. Some connectors only use a few of the pins in the connector while others use most or all. So there might be as many as 30 connections being made in one connector. Many of the connectors are for connecting a bunch of relays to the control. These relays in turn often connect to bigger contactors. A lot of the connectors bring the inputs from switches to the control. 66 switches in total not counting the various limit and pressure switches. And many of the switches have multiple positions. With a new control a new keyboard could be used, and probably needs to be unless Linux CNC will emulate a Fanuc 3T keyboard. Anyway, it looks like a big project so I need to start getting everything together and ready for a plug and play swap. Plug and play is a big reason I want to use the original connectors. That way the new control can be plugged in without removing the old one. Testing can be done and if I hit any major snags then the old control can be plugged in and the lathe can make parts while I work out the bugs. Eric

Reply to
etpm

ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.