Hybrid IC questions

[et...@whidbey.com]

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

[tabb...@gmail.com]

To your 2 questions: no chance.


NT

[John-Del]

The hybrid ICs that I have done postmortems on usually had the semiconductor 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 die 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 mitigaged) and solder SMD IGBTs to the substrate after carefully cutting away the blow dies, assuming there was room.

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

[Cursitor Doom]

On Tue, 18 Dec 2018 09:36:50 -0800, tabbypurr wrote:

> To your 2 questions: no chance.

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
protocols, whether for profit or not, is conditional upon a charge of
GBP10.00 per reproduction. Publication in this manner via non-Usenet
protocols constitutes acceptance of this condition.

[Pat]

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

[Jon Elson]

On Tue, 18 Dec 2018 09:16:36 -0800, etpm wrote:

> 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.

> 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?

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

[Jeff Liebermann]

On Tue, 18 Dec 2018 09:16:36 -0800, et...@whidbey.com wrote:

> 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.

Yep. A local machine shop had a Mori Seiki 5 axis mill with a Fanuc
controller. Lousy photo at top of:
<http://www.santacruzprecision.com/equipment.htm>
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.
<https://www.ve7ca.net/TstH86.htm#HK14>
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:
<https://www.ve7ca.net/TST/H86/hybrid.png>
Not the lack of detail, such as xsistor part numbers.

> 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.

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.

> 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?

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.

> I have looked for datasheets and pinouts on the hybrids but no luck
>yet. Maybe I need to look harder.

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:
<http://www.learnbydestroying.com/jeffl/pics/CNC-conversion/index.html>
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:
<https://www.cnczone.com/forums/fanuc/>

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





--
Jeff Liebermann je...@cruzio.com
150 Felker St #D http://www.LearnByDestroying.com
Santa Cruz CA 95060 http://802.11junk.com
Skype: JeffLiebermann AE6KS 831-336-2558

[et...@whidbey.com]

On Tue, 18 Dec 2018 12:52:31 -0800, Jeff Liebermann <je...@cruzio.com>
wrote:

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

[Jeff Liebermann]

On Tue, 18 Dec 2018 14:09:12 -0800, et...@whidbey.com wrote:

> 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.

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.

> And you are correct Jeff, I shoulda provided the control info. The
>control is a FANUC 3T. Tha part is a FANUC A-RV06.

Thanks. Looks like there are a few available from the parts cannibals
on eBay (from Poland, Canada, and New Jersey):
<https://www.ebay.com/sch/i.html?_nkw=FANUC+A-RV06+FA8113+2F3+INTEGRATED+CIRCUIT>
<https://www.ebay.com/itm/FANUC-HYBRID-A-RV05-/252764564343>
<https://www.ebay.com/itm/FANUC-CT1183E-CT1183E-RISCN1/312110064342>
<https://www.ebay.com/itm/A-RV05-CT1183E-HYBRID-FANUC-ID15323/172110952943>
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:
<http://www.cnc888.com/houmo.asp>
Lots of Fanuc boards and parts on the site.
<http://www.cnc888.com/pcb.asp>

> 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.

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:
<http://www.kaibader.de/exposing-a-chip-on-board/>
<https://www.youtube.com/watch?v=hQ5hXEK35WI>

This looks like the back of the hybrid:
<https://i.ebayimg.com/images/g/glcAAOSwgSpbs5Pj/s-l1600.jpg>
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.

[et...@whidbey.com]

On Tue, 18 Dec 2018 20:24:22 -0800, Jeff Liebermann <je...@cruzio.com>
wrote:

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

[John-Del]

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 able to spot something. If you need to do any soldering you MUST preheat the crap out of it as the ceramic substrate will steal heat like an Amazon porch pirate doing his Christmas shopping.

[Jeff Liebermann]

On Wed, 19 Dec 2018 08:28:08 -0800, et...@whidbey.com wrote:

>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.

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:
<http://cncmanual.com/download/1647/> 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.

> 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.

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.

[et...@whidbey.com]

On Wed, 19 Dec 2018 09:55:06 -0800, Jeff Liebermann <je...@cruzio.com>
wrote:

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

[Jon Elson]

On Tue, 18 Dec 2018 14:09:12 -0800, etpm wrote:

>
> 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.

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.

> The problem with changing the control is that I need enough time
> when the machine is not needed to make the swap.

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

[Jon Elson]

On Wed, 19 Dec 2018 08:28:08 -0800, etpm wrote:

>
> 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.

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

[et...@whidbey.com]

On Wed, 19 Dec 2018 16:36:06 -0600, Jon Elson <el...@pico-systems.com>
wrote:

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

[Jeff Liebermann]

On Wed, 19 Dec 2018 16:40:14 -0600, Jon Elson <el...@pico-systems.com>
wrote:

This might help with the parts counter function:

"Parts counter M code?"
<https://www.practicalmachinist.com/vb/cnc-machining/parts-counter-m-code-290803/>

[Jon Elson]

et...@whidbey.com wrote:

> On Wed, 19 Dec 2018 16:36:06 -0600, Jon Elson <el...@pico-systems.com>
> wrote:

> > 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.

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 http://linuxcnc.org/docs/html/gcode/o-code.html#ocode:looping 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.

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

Jon

[et...@whidbey.com]

On Sat, 22 Dec 2018 15:26:13 -0600, Jon Elson <el...@pico-systems.com>

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

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

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