EG&G princeton 5210 lock in amplifier repair (almost repaired, frq offset)

[coppercone2]

Ok So I had a working 5210 lock in amplifier that came in off ebay, but when I opened it up I saw a gigantic capacitor leak. Like 4 square inches of PCB were covered with some kind of shit that flowed out of it, and it would not come off with a q-tip and alcohol, so I had to wash the whole thing in simple green after total disassembly. There were several chips that seem corroded but I did not replace anything but the capacitor because it was working before cleaning. The stuff leaked out on the chassis too. I am not sure I guess its also possible something spilled on the capacitor since there is a vent hole above it, but anyway it looks corrosive on the IC pins

I undid every single header, ribbon cable, pin, socketed chip and cleaned it carefully with alcohol after 24 hours of drying, and I replaced the capacitor (470uF) with a identical value. The header connectors near the capacitor were the most corroded.

When I try to power it back on, the digital meters read full scale and the front panel is unresponsive. If I connect a power circuit to where the backup battery used to be (it was dead, but I cut it off prior to cleaning), its just as unresponsive but a bunch of front panel lights light up, its built like a sony stereo which I fucking hate working on, so I suspect it might get binned

Does anyone have a manual for this thing? I have no idea what I did, maybe I should not have touched the socketed chips.

I found a random 5V rail on the logic and it seems OK.

How do I troubleshoot this? I have a logic analyzer and DIP clips, I thought a start might be to look at the processor to find a clock signal or something like that? Anyone have experience fixing princeton lock in amps? The displays sometimes miss a segment and sometimes the analog display is pinned to a random low value (1 out of scale of 10, but usually at 0). Does anyone have any ideas on which chips to start on? Particularly verifying if the EEPROM is intact because I know if that goes off I am fucked. I thought maybe a bond wire fractured when I was removing them, but I used a real IC extractor on it.

I am mega pissed off because if I just ignored the giant white pool and did not spend 3 days cleaning it, I would have had a working unit. fuck my ocd, but I can't help it since it looked like a horse splooged on the pcb


should I replace the corroded chips? try to figure out what the processor is doing? I figure it must have only been surface damage and I damaged something with ESD or mechanically (i did wear a wrist strap and used a real electronics dip puller tho and set the chips on a esd mat and used a wooden chair).

the leader chip looks like a mc68008 and the usual associates. when I unplug the big gang switch and the floating shield can from the PCB, it has no change, so I assume its something with the big square board on the bottom.

[coppercone2]

might be related to moisture, after 20 min in the oven I got 2 keys to work. maybe i underestimated the PCB, the guidelines are long, it able to tell and remember the GPIB and Serial port settings, maybe its not suffering from amnesia

doing an aggressive 2 hour bake cycle at 170F because thats the lowest the oven will go. the first time I opened it after 30 min it certainly seemed to let out some humidity. Maybe there is water trapped some where.

[Kleinstein]

A larger amount of water can be a problem, but especially the digital part should not be effected if the board is otherwise clean. Humidity can be a problem if there is still quite some residue from a leaking capacitor or batterie that should be cleaned. For cleaning the residue of leaking part alcohol is not very effective - plenty of water is usually the better way. Alcohol is only the last step, helping to remove the water residues.

Humidity, but also alcohol absorbed in the PCB can cause indirect effects from swelling and baord stress, effecting contacts and maybe poor solder joints.

Given the age of the instrument I would look for drop shaped tantalum caps - they are prone to failure, sometimes catastrophic if at a powerful supply.

I have no manual, but some 25 years ago I had a look at it, and it even included schematics, at least for some parts, so one could fix a flaw with the DAC ouputs, but this is nothing to worry right now.
First the CPU part should be working before worrying about the analog readings.
EPROMs are usual not sensitive to mechanical stress, if they come in the classic ceramic case. So no worries on the bond wires. It is more that if used at high temperature they may start loosing there contend.

[coppercone2]

Ok I got it working kinda, the dial indicator melted but this is whats going on (fucking stupid oven, oh well). I had it setup in a weird angle in the oven but when I left it like decided to slide over and touch the side wall of the oven. Maybe I will find a big mirrored analog gauge on a flip down bracket from some simpson or something there one day if  Iget it working. like the targeting system in the x-wing

When I turn it on, I need to press the Green function key then the other key then the green function key again to change anything. When it was working normally I could use the button to select voltage range, filter setting, so forth. Now I need to

1) Press green key - green light turns on
2) Press key that does whatever I want, including the frequency setting keys that are normally velocity controlled - nothing happens
3) press green key, whatever I did in step 2 occurs now


So its basically unusable. I checked all the keys for continuity (I thought one of them was shorting, but the buttons all make a multimeter peep). The local/remote switch disables the front panel but it does not make the LED next to it go off. Interestingly the 'red' shift key (like the green shift key) works, so it can do an 'auto' measurement. Setting the oscillator, offset, amplitude, etc.. is a fucking game now. Guess at how long to hold the button, feel like im playing some stupid boxing game on a NES. It seems that every key works, you just need to play games with the green key (labeled control)

I took all the keys off and tried to squirt deoxit on the little plastic they put on to act as a key, so my bad, at least they gave you push buttons, but anyway I don't think its a mechanical issue, I tested continuity on every key to make sure its not shorted out, don't think I really got water in that place. I wish you can desolder the buttons but it looks like careful reflow on plastic only 

Makes me think of some kind of interrupt or something. That the green key is making something check some kind of register whereas it should normally check that automatically. But I can't just push the green key after pushing the button, I need to 'prime' it by pushing the green key and then push the button then push the green key to send it through. wtf is this behavior. like the green key is substituting some refreshing clock? something with shift registers? I also did try shorting it manually with some probes to see if maybe the button is misbehaving with like bounce but the ones that work, work, and the ones that don't, have no difference.

You think its still something with caps causing this weird ass issue? any idea what the green shift key is associated with? its something with some kinda scan I think, there has got to be a missing periodic waveform I think. . it seems like an odd way to build a keypad, you would think everything there would cause a interrupt, why is it enabled and 'read' with the green key now? I feel like if I put a high frequency square wave as the input to the green key and disconnected the LED light it would essentially be fixed lol, but I hope there is a better way to fix it then to rig a 555 timer in there

[Kleinstein]

The describtion on the green botton is weired. I remember that there was a green key, but not what it was good for. It is quite a complicated Instuments with quite some functions, so working keys would be nice.
There are quite a lot of things that can go wrong with the computer part and it is hard to tell from the symptoms what is wrong.

I would not expect the keys to cause an interrupt. A keymatrix is usually scanned to a regular time and this signal should be measurable at the keys. There may be something like a missing pull up, that could cause floating pins and this way effects, like the need to "Prime" with the green key.

[coppercone2]

so like some how when you push the green key you are getting current into something by accident to charge gates?

the green key is like a shift button on a computer, some stuff is dual use so you press the green key and then press it and its supposed to do the alternative function. I.e. there is a setting button for display #2 that its green key function is the instrument lights.

So if you do the greenkey + pushbutton + greenkey method

you get
1) setting 1
2) setting 2
3) setting 3
4) setting + no lights (turns off lights, not sure if its 3 or 4)
5) setting 4 + turns lights back on
6) setting 5
7) setting 1

so when you get to selection #4 it fights between turning the lights off and getting to setting #4

normally when you push the button its supposed to change settings, and when you greenkey the button its supposed to trigger the light on/off. but they are merged now. so you think the master key is not causing the CPU to do anything with the program, its just keeping the circuits running by some kind of unwanted effect on current paths? I am not quite able to imagine the circuit you are talking about where the greenkey some how helps pull up other things in a matrix, but I don't work with this kind of stuff usually.

Do you think there will be a scan signal (pulse signal) going to the keypad on the ribbon cable that I should look for?

[Kleinstein]

With some 20 bottons going to the computer, I would expect them to use a kind of key matrix, though one knever knows and there are alternative ways.
The number of lines on the ribon cable could give a hint.

The PCB for the front should also give a hint on how the keys are connected.
Nomally there should be a scan signal, though there is also a way to first check for any key and only than check which key by a scan. So I would expect a periodic scan signal, but not sure.

Fighting with 2 functions may be something like a short between lines.

[coppercone2]

Ok I tried to probe around

It seems every signal in the 40 pin connector is either logic high or logic low outside of what looks to be some spikes that are hard to trigger on (logic high line spiking low every 600uS), maybe this is it? I can't freeze on it so I am not sure if its variable duty cycle or what, I can just measure it with the redicule while its flashing in an out of existence.

When I hold down the green button I see like a really high duty cycle square wave show up on some pins, around 99% duty cycle, probing the top if the IEC connector, wheras normally these pins are logic low

don't really know how to proceed, areas near the clock are buzzing with activity

[coppercone2]

Bought new  socketed stuff
P8254-2 (hope its drop in for P8254.. not sure what the -2 and -5 do). does this have to be programmed?
 MC68008P8 
TC5565P-15
MC68681P


and I will replace (think I have these in parts boxes, will find out today), they have signs of corrosion
cd74hct00

sn74hc03n

lm339n

The system is complicated so I will spend a couple of dollars to see if there are drop in replacements that work to possibly fix the problem, because when I see corroded parts in a broken system and have to trouble shoot it, its driving me fucking crazy, hopefully those intel parts will be fun to play with at least, because for a while I was thinking about doing a discrete system with a zilog or other discrete CPU, so I can tolerate the likely spare parts


when I do that and it still does not work I think I have a doomsday stockpile of tantalums that can pretty much do the whole unit, and i reckon that this is it

then when its still not working I can take out the logic analyzer and operate it with some semblence of sanity, since I have a floppy drive for my VNA and alot of DIP clips I can upload big logic communications plots with 128 inputs (pay off the old logic analyzer), if anything having no corrosion will facilitate the use of dip clips for trouble free telemetry

[coppercone2]

Hmm I did the 74 series logic gate and there is a significant amount of corrosion underneath the chip despite the extended bath. I wonder if a clock signal was being pulled to ground some where, but unfortunatly I installed it backwards and will rework it tomorrow, the unit failed to power on but no smoke etc, hopefully its not damaged because it was my last one

I think I should replace everything that was bathed in the white goo

I feel like I can probobly replace a good deal of these chips for small sums of money

[coppercone2]

reworked the chip because it was bothering me, looks like its not broken and the device is in the same state it was, but there is another nasty chip under neath and a few near by, I will order them on digikey

If I spliced in a logic analyzer to all 40 pins and posted a logic diagram, would that help someone trouble shoot my issue?

[ebastler]

This thread in another forum supposedly has a user manual (and some other "high level information" in the following post). I am not a registered user there, hence could not download the attached files, but it might be worth registering?

https://www.electronicspoint.com/forums/threads/eg-g-5210-lock-in-amplifier.291898/#post-1790845

[coppercone2]

I have both of those documents but they don't get into the discrete TTL based logic in these things

search for manual (its on a japanese site) and for technical notes (its on their website or one of their parent companies)

they just give you a overhead of the circuit but this is pretty specific user interface bug that probobly comes down to a single bad gate inside of something

before I started I did the 'verify your equipment is good using the internal generator routing into the input" thing and it worked fine.

this is more a thread on how to trouble shoot discrete gate/processor control boards then the lock in circuits, which I think should be fine. It's something really stupid because I can use the instrument if I wrangle with the green shift key. I have a feeling its either gonna be blind luck, a circuit trace or a logic analyzer investigation 

I just got a male-female 40 pin cable so I can splice in the logic analyzer to the ribbon cable  because right now its ridiculous to trouble shoot the way its built. The most similar thing I have seen is like non archaic but old wavetek units. its probobly built like that

[coppercone2]

https://www.ameteksi.com/-/media/ameteksi/download_links/documentations/supportcenter/signalrecovery/instruction_manuals/18978-a-mnl-a.pdf?la=en

This is a manual for like a earlier version but it looks like the same stuff is going on the front for switches

maybe someone can notice something, they were kind enough to give a schematic

page 119 shows what they use for their switch matrix, maybe that will be the most helpful, but there is no shift key

I am trying to think of what the shift key must do, it looks strait forward in that schematic.

119 goes to BD2 on page 117, which goes into a 74LS245, which goes into a MC68488. I think I want to put a tap on that equivalent circuit if I can find it on the other pcb.


Am I correct in thinking that the 68488 is locked up and not accepting the inputs from bd2345 unless the shift key is pushed first for some reason and that I should try to trace out that signal chain? So see what those traces are doing once I find them when the button is not pushed, after its pushed, and after the buttons are pressed after its been pushed and what happens when its been pushed again, but the 5410 should have the  MC68681P  in place of the 68488, which is a interface adapter, reciever transmitter sounds like the same thing as a interface adapter with different lingo

from the looks of the schematics, the types of parts they use, and so on, i am guessing the company migrates about as fast a lichen between designs, they usually try to drop in stuff that does the same thing in their mind, thats what I always see, the same paths just with different materials

BTW does anyone know anything about the P8254-2 part? The PCB just says P8254, the seller I buy from says 2/5, but the marking is 2. What the heck is this ?

[capt bullshot]

These are speed grades:
nothing: 8MHz
-5: 5MHz
-2: 10MHz

https://www.scs.stanford.edu/10wi-cs140/pintos/specs/8254.pdf

[coppercone2]

ok, I read the datasheet. I see what is going on now.

Too many little details. I got the correct part. I wanna keep things moving along without getting stuck with another disassembled instrument

This is a non eeprom chip right? i assume its programmed on power up every time, I don't see any mention of memory in the datasheet

This chip or one of the other peripherals might be a good way to see what the system is doing since I can see how its getting programmed on LA.

I also thought to setup a EEPROM dump. I have a data generator and a logic analyzer so I think I can write the program to a floppy. Or maybe I think the pickit does it so I can read it with that and save it.

[Kleinstein]

The 8254 has no internal EPROM or similar. So it is just programmed by the main CPU for use and a simple replacement with a new chip is possible.
The 68488 IO chip should control the GPIB interface and the MC68681 should be be for RS232. Both may also be used for some additional smaller tasks - maybe the keys.

I would no worry so much about the corrosion products shortening a clock - these are usually relatively low impedance. The more tricky part may be 74LS... chips, that us an open input to be logic H. This was sometimes used in the old days to safe a resistor.

[coppercone2]

One more observation,

1) the velocity control still works, once I prime it, I can hold the button down for a while, then trigger it, and it will offset whatever thing I want on channel 1 proportional to the key hold duration. Longer = faster, so acceleration works.

Channel 1 shows the correct thing, I can adjust reference level, oscillator, phase, so forth. I can display the input but since I melted the panel meter I don't know if they track 

 
2) Channel 2 display shows off scale +1.99, the only way I can get it to display anything else is to use shift + gpib or shift + rs232 to get like the port information printed.

I also verified that undoing every single connection, apart from the keypad ribbon cable, has no effect on the unit. I can disconnect all the shielded floating cans, etc, without a noticeable control system change.

3) Channel 2 has the up/down buttons to adjust offset. So in channel 1, if I select a function, then use the velocity keys, I can adjust frequency. I think on channel 2 if I have 'offset' high lighted, I should be able to change the offset in the same way. But, it is pinned on +1.99, no matter which setting I have it in, other then the 'print GPIB or rs232 channel' setting. So this means that the fault is related to here, since there is partial lack of function.

The manual also calls the shift key a 'dual key' function


And possibly evidence of damage, the light bulb that got hit by a stray which the circuit detective missed:
the local remote switch does not cause a light change when it is activated. It has a LED on it that is totally dead, but the key works. This might relate to the god damn thought you had about some kind of bias impedance being incorrect. I have a feeling this might be a very good place to start a diagnostic. The LED is going to be the highest power consuming device there, and it shows sign of a bad driver, unless the LED is broken. I will measure this guy. Maybe there is a low voltage/current limit that is preventing the LED from turning on

[coppercone2]

I traced the IC to a HC377 flip flop on the front panel, and a resistor network. The resistor network seems intact. The flip flop is going right to the ribbon connectors.


I need to figure out this flip flop, is it possible for a working flip flop to light neither of the LEDs up? I assume yes, since there is an option to disable all lights on the front panel, with the shift key, and it works. This function works. Since those 2 lights refuse to light up, it can either be a signal not reaching the flip flop, or a broken flip flop that is ignoring the signal, or not carrying it out.

But I am not getting this: how do you disable the lights with a flip flop? It should be stuck on either of them, right? So if its an octal flip flop, this must mean that its just being used as a buffer, not as a hard switch, right, so 1 flip flop circuit per LED rather then 1 flip flop sub circuit per 2 leds, which would leave one LED on unconditionally, but it would not be able to disable front panel lights. ?

I am being confused because the previous model uses a latch, but this uses a flip flop. why, they are similar. Maybe its less overhead since it uses a pulse. That would explain why the shift key does something, it must be causing the flip flops (there are multiple) to flip.

So being that they are flip flops, and the shift key is causing them to flip for some reason, a defective flip flop on the output should not effect the rest of the flip flops, unless its a issue related to input impedance and fanout? Most likely a circuit controlling the whole shebang is the issue, not the end flip flop, so the particular flip flop I found is just closer signal wise to a damaged PCB region (imagining a area that has a damage gradient across it with multiple signal paths), and maybe its being triggered my coincidental inrush spikes, from the shift key for whatever reason that happens to cause electromagnetic radiation, maybe the fact that there are green and red LEDs that are triggered that have a different voltage threshold then the yellow leds or possibly a higher current (they do seem alot brighter then the yellows) are emping it and making it shift? (controller is dead, carnosaur is in the control room smashing its tail across a panel every time someone is trying to key the radio? (carnosaur 2 analogy))

I will put soic clips on the flip flop to see what its doing

[coppercone2]

Any idea what the operating temp on those IC's is? Some of the bigger chips run hot, the eeprom is cool, the hottest chips are the two interface adapters, one of them is 126F, the other 115F. Well within normal semiconductor operating parameters but it stands out super hard on the PCB

The hottest chip is the TSM9901. It says its an interface adapter that provides interrupts to the CPU. I think that is the main contender for whats controlling the input to those flip flops. I wonder if it has a reason for being so hot?

[coppercone2]

looks like its fixed, god damn chip logic interface bent its leg over and it looked like it was fully inserted after I resocketed and cleaned the goo. fully responsive. oh well useless chips inbound from ebay , the unusually hot running temp made me investigate it. now its responsive but I need to put this back together

Usually the legs slide off to the side, but this one curled up under neath the chip and made it look from a sideways inspection that it went into the socket unless you get really close, I only noticed after I turned it upside down, its the one right near the back of the chassis with a wire over it.

at least I have replacements I guess, and its a step above to make a system without just reaching for a PIC one of these years

$46 down the drain but oh well. Might be worth while to replace that chip anyway because its been running a leg short for quite a few troubleshooting minutes, perhaps it will benefit from replacements... at least I got rid of most of the corrosion too, could have been a time bomb

[Kleinstein]

Some of the old inteface chips run quite hot (old N-MOS versions - not all are avialable as a lower power CMOS version). 74F...  need quite some power too and thus run quite hot.

[coppercone2]

Yeah I usually see high temp near high speed clocks, I know when I saw the high speed 74 series logic run at really high speed, it was running pretty hot, but I also noticed sometimes when I replace old 74 series chips, they run cooler, I think it has something to do with aging or maybe manufacturer (I keep track of HC, LS, etc, don't mix)

Do you have modern version of the same HC/LS/etc chip that uses different transistors with the same parameters? So you can have say a NMOS HC and a CMOS HC? I saw a big ~30F drop in temperature once in a spectrum analyzer repair when if I recall I replaced a working chip with the same exact chip name type (I thought it was broke, and it cooled down, but all signals were the same). I think it was a multiplier? maybe. I thought it was just thermal aging but maybe they improved the process some how. I also see it alot when I replace old discrete transistors (sometimes just replace 5-6 transistors near a broken one just to be safe), they often run significantly cooler.

[coppercone2]

So the only thing I need is unfortunately a replacement for a 2900 0105 panel meter, which unfortunately warped from the heat due to the oven mishap.

[Kleinstein]

One can not directly mix 74LS or similar (BJT based) with normal CMOS based (e.g. 74HC). The output  H level of 74 LS and simlilar is only marginal for the CMOS inputs. So one would need 74HCT or similar to work with the smaller input signal (or a signal coming from a 3.3 V domain).

The current demand of the logic series is quite different. 74LS , 74S, 74F and similar are BJT based and also use quite some power if the frequency is low, with the faster ones using more power. The modern CMOS versions like 74 HC, AC, AHC or similar use very little power at low frequency, but can be similar or in some cases even higher power at there maximum speed.

NMOS was used for some of the IO chips, AFAIK there is no standard NMOS logic series, only BJT based or CMOS. It is the P8254 or MC68681 where there may be a lower power alternative. I have not checked if the origianals are NMOS or already CMOS. The other point may be changing 74F... to ACT or similar. With different levels at the ouputs they can still behave different and the timing is also slightly different. I would not really bother.