Keithley 2306 Dual Battery Simulator Repair (maybe)

[Smokey]

I just scored a battery simulator from the ebays.  Advertised as "powers up, no display".
https://www.tek.com/en/products/keithley/dc-power-supplies/2300-series

https://download.tek.com/manual/2306-901-01%20(F%20-%20Apr%202008)(Ins).pdf

Dave did a teardown of the one channel model here: https://www.eevblog.com/forum/blog/eevblog-1005-keithley-2302-battery-simulator-teardown/

I don't think there is a repair manual available.

Got it on the bench and turned it on.  The fan comes on, and as described there is nothing on the screen.  The corner of the VFD looks like it arc'd inside.  That may be the only issue, but I'm still waiting for the GPIB adapter to ship so I decided to poke around inside and see what I can see.

(The lighting in the pictures I took sucked, so I'm using screen shots from Dave's teardown)



There are some test points for the output voltages of the transformer. 
+/-15V, +/-5V, +25V

The +/-15V and the +5V are regulated, and measure good. 

The -5V and +25V both appear to be unregulated and measure high:
+25V Testpoint measures +40V
-5V Testpoint measures -9V

The transformer primary is a regulated 12V chopped into a 60kHz square wave by those two devices on the left of the transformer.  There are +/-40V, +/-25, and +/-9V square wave secondaries.



The strange thing in this case is that there are two nearly identical channels (both populated on the 2306 model I have here) and BOTH of them measure the same high voltages for the +25V and -5V rails. 

Does this make any sense?  I don't see anything obviously wrong with the diodes or caps doing the unregulated filtering.  Since both channels are acting identically, I would be surprised if there was a transformer drive or secondary issue. 

If these rails are supposed to be +40V and -9V, why label the test points +25V and -5V?

[edpalmer42]

I can't find a service manual, but here's a 367 page Instruction Manual that includes parts lists, component placement drawings aaannnndddd no schematics.   

https://images-na.ssl-images-amazon.com/images/I/B1fzKhl8h3S.pdf

Also, lots of info & documents here for you to wade through.  There's bound to be some treasures.

http://electronicsandbooks.com/edt/manual/index.php?dir=Hardware/K/Keithley+www.keithley.com/

Ed

[Smokey]

Ya, I've done a bit of digging and I haven't heard anyone talk about those voltages.  I know there is another repair thread with a different problem here: https://www.eevblog.com/forum/repair/keithley-2306-dual-psu-battery-simulator-repair/

So there are a few people with these simulators.  I would have thought someone would have measured the +25V test point and said something about that voltage being +40V if it was supposed to be like that. 

[Smokey]

Sunday afternoon Keithley bump....

[picburner]

I am one of those who wrote in the other tread.
I still have the K2306 and should do some maintenance on it.
If you have patience, as soon as I can, I open it and take the measurements in the points indicated.

[gamalot]

The -5V and +25V both appear to be unregulated and measure high:
+25V Testpoint measures +40V
-5V Testpoint measures -9V

The strange thing in this case is that there are two nearly identical channels (both populated on the 2306 model I have here) and BOTH of them measure the same high voltages for the +25V and -5V rails. 

Does this make any sense?  I don't see anything obviously wrong with the diodes or caps doing the unregulated filtering.  Since both channels are acting identically, I would be surprised if there was a transformer drive or secondary issue. 

If these rails are supposed to be +40V and -9V, why label the test points +25V and -5V?

I just opened my 2306 and measured those voltages, they were pretty close to yours, measuring 36V at +25A/B and -9V at -5VA/B.

[gamalot]

I put the upper PCBA back and connected it to the lower one, the voltage of +25VA measured at this time is 28V.

---

In order to verify the guess of that the +25V rail voltage will decrease when the load is increased, I connected my 2306 to the electronic load. Sure enough, when the channel #1 source 8V/5A, the test point +25VA voltage drops to 25.2V.

[picburner]

Measurements made with respect to TP102:

TP100 =  +4,97V
TP101 = +14,90V
TP103 = -15,06V
TP104 =  -6,64V
TP111 = +28,20V

If you need other measurements just let me know.
Can you take a picture of your display?

If there is a problem on the main cpu side on the display you can read "INITIALIZING" a sign that the display controller is unable to communicate with the main cpu.
If, on the other hand, the display is completely dark the display pcb may be missing 5V, the display is broken, the display controller's xtal is broken or there are other problems on the display pcb.

[Smokey]

You guys are the best!  Thank you for checking that.  Gives me new hope it's just the display, which shows no activity at all. 

The remote display should be here soon.  I saw another thread somewhere where some dude replaced the vfd with an lcd module and minimal hacks.

[Smokey]

Thanks again for checking on those voltages.  Well this may be the easiest repair of all time.

I couldn't find a 2306-DISP remote display, but there are some 2304-DISP modules on ebay and I picked one up for $15USD. 

Just got here, plugged it in with a standard RJ45 ethernet cable, and it works!.  Simulator is alive.  Outputs looking good.  I'm not sure if I'm missing anything by using the 2304-DISP instead of the 2306-DISP but it looks like all the menus are available.

Still waiting for the GPIB adapter for full tests, but looking really promising. 

[Smokey]

Update..... Not out of the woods yet. 

The supply only successfully powers up intermittently.  Especially if it's been sitting for a while and is cold, it will just hang at the "initializing" part of the remote display routine (as mentioned in the other repair thread, the display has it's own processor and is responsible for that "initializing" text).  The remote display actually doesn't show the word "initializing", it just shows two flashing boxes in the middle, but I think it's the same thing.
Power cycling sometimes results in the supply powering up successfully, and once it does it seems to power up consistently until I let it sit again.  I confirmed the output is good when it's working and I can talk to it over GPIB.

There is also one more piece to the puzzle, which is while the unit is working, sometimes the front panel display will actually come alive for a few seconds and show "Initializing", then die again.  It never actually displays anything else even when I unplug the remote display while it's running, still just "Initializing" randomly flashing on the front. 

I had the thought that maybe something on the front panel display is shorting a supply rail or something, but having it completely unplugged has no apparent effect on the intermittent power up issue.

I can also confirm that the issue looks like it's in the bottom processor/supply board (not the top output board).  Even with the top board off, it still only powers up intermittently with the remote display showing the flashing boxes when it fails, and an error code when it does power up without the top board.

A few people replaced the SRAM (R1LP0408CSB-5SC) in the other thread, but they are presently out of stock unless I want to buy a minimum of like 20 pieces at $15-$20 each.  I'm still unsure why that would be causing an issue like this though.

[picburner]

The two vertical flashing boxes you noticed on the display are normal and are shown just before the final splash screen appears.
It should do this when everything is working normally.
I had to replace the SRAM because the VFD display was stuck on 'initializing', this is not necessarily your case because at least sometimes your instrument starts up.
.....Unless this is a symptom of an upcoming demise.

[Smokey]

When the unit fails to power up, the two vertical boxes on the remote screen never go away.  They just keep flashing until I power cycle. 

When it does power up, they are only there for a second or two before it kicks into operational mode.

[Smokey]

Ok.  New update. 
First a confession... The unit I'm working on is actually a 2308, which is the updated version of the 2306: https://forum.tek.com/viewtopic.php?t=94760#p153847

But... The bottom board that is giving me trouble is shared with the 2306, just with a different firmware build (it actually has the 2306 check mark on the silk screen for the board model).  I haven't seen anyone here with a 2308 and I didn't want to scare everyone off since the board that's having issues appears to be the same.

Update:
I hooked a logic analyzer up to the serial lines on the display connector and captured the messages on Pin4 and Pin5: [UART decoding at 9600 8n1].  Pin4 is coming from the main processor because it gets traffic even when the cable to the display is unplugged.

During successful power up without the top board installed:
Pin4 has the following messages:
\x0C\x01\x06\0  2308 ADDR  6  A01   00000 60HZ
\x05\xFF\x06\0 start response   model INVALID
\x05\xFF\x06\0 start response    rev INVALID 
\x05\xFF\x06\0 start response    key INVALID 
\05\0\06\0  0.000 V #1 OFFOVERFLOW   
\05\0\06\0  0.000 V #1 OFFOVERFLOW   
...[repeating]

Pin5 has the following messages:
2304 B02  \x80



The overflow message is because the top board isn't installed.  That doesn't happen when the top board is installed.

When the unit doesn't power up correctly I get these messages, then nothing...
Pin4 has the following messages:
\x07
\t\x17
\x07
\x0C\x01\x06\0\x20

Pin5 has the following messages:
2304 B02  \x80
2304 B02  \x80



With the remote display plugged in, I get the following message on Pin4:
\x0C\x01\x03  REMOTE PANEL        ENABLED


Does any of that set off any alarms for anyone?
Any more ideas on ways to troubleshoot this?  I haven't see any other diagnostic interfaces.

[Smokey]

I'm not sure if this is relevant, but it is interesting. 
I took the front panel off of the unit and powered the unreliable internal VFD display from a 5.00V bench supply.  When I power up the display this way, it pulls about 230mA and powers up every time.  Right into "INITIALIZING".  Nice and bright.  This is very different from the random intermittent flashing that happens when it's being powered while connected to the unit.

So I figured I would see what minimum voltage would turn on the display.  And... It starts to operate unreliably at about... 4.95V.  It won't turn on under 4.94. 

Well when I measure the +5VD rail at the display cable, it measures..... +4.95V.  So that appears to be the (an?) issue with the display. 

Is this a design flaw where they used some parts in the display board without sufficient minimum voltage margin?  Or has something aged to where it's unreliable now?  Is this related to the system not powering up somehow (like whatever components won't power up in the display are also refusing to consistently power up on the rest of the digital section?).

::Update 7-24:: See msg bellow: https://www.eevblog.com/forum/testgear/keithley-2306-dual-battery-simulator-repair-(maybe)/msg4319182/#msg4319182
It looks like the undervoltage monitor IC on the display board is not working correctly....

[Smokey]

Plot thickens....

The +5VD rail is powered by an LM2940CT:
https://www.ti.com/lit/ds/symlink/lm2940-n.pdf

The input to that regulator is +6.5VD, which measures about 6.40V at the input to the LM2940CT.
As I said in the last post, the +5VD rail is measuring about 4.95V.  Normally I would say this is pretty damn good.

Test = power the 5V rail from an external supply:
So far, when I power the whole system +5VD rail from an external supply, and confirm with measurement that the voltage of the +5VD rail at the display connector is the full 5.00V, the supply powers up consistently every time as confirmed by the external display.  When installed in the system, the internal VFD display only powers up every time consistently when I boost the +5VD rail to 5.13V.  With the whole system running, the +5VD rail pulls about 575mA.

It just seems insane that having the +5VD rail be only 50mV low would make the whole system unreliable.   

Any ideas?

::NOTE::  I should add that I have looked at the 5V rail on the scope and it is really solid.  No excessive ripple or noise or anything nasty.

[Smokey]

I'm thinking I can replace the LM2940CT with something that has a variable output and just bump it up to about 5.1V, but that seems like a band aid for some underlying issues.

This looks like it might be viable: https://www.digikey.com/en/products/detail/microchip-technology/MIC29302WT/771593

I can probably hack this in the same footprint without needing a carrier board since they are both through hole.  I can solder the new part into the IN,OUT,GND pads of the existing TO-220 footprint, and just air wire the adjust and enable pins.

[Smokey]

Looking at the LM2940CT 5V part datasheet, the recommended input voltage range is 6.25V to 26V.  This is surprisingly close to the 6.4V that is actually on the input to the regulator (even through the dropout plot does say it's rated at 0.5V dropout at 1A).

The output voltage range, depending on the current level, can be anywhere from 4.75V to 5.25V.  So the normal operating range extends way into the non-functional zone at the low end of the regulator output tolerance.

There has to be something else going on.

[AVGresponding]

I take it you have checked for dry caps on the output side of the regulator, and the display board, which could easily cause such symptoms? What is the ripple voltage?

[Smokey]

The LM2940CT has maybe 50mv of high frequency ripple at about 200kHz, which I would think is coming from the square wave drive into the transformer primaries for the output voltages.  I don't think it looks that bad at all really.  When I power the unit from the bench 5V supply, the character of the voltage rail doesn't change at all, it just increases the voltage a little. 

That's what so strange to me.  The voltage rail looks solid at 4.95V.  It's not like its dipping whenever there is a current demand.

I actually don't think there are any electrolytic caps on the 5V rail... but there are a ton of tantalums.
With that said, the in circuit resistance across the 5V rail is about 750ohms, but that's in circuit.  The total current on that rail is under 600mA, which is well within range for the regulator.  It's not even getting very hot.

While the display board and the system board are both showing what looks like essentially the same symptom (extreme sensitivity to under voltage on the 5V rail), the main board has the problem even when the display board is completely disconnected.

[mnementh]

Have you tried isolating the input to the LM2940CT and powering it with a slightly higher voltage; like 8-12V? What you're describing sounds to me like it may be right on the verge and dropping out.

mnem
 

[Smokey]

Have you tried isolating the input to the LM2940CT and powering it with a slightly higher voltage; like 8-12V? What you're describing sounds to me like it may be right on the verge and dropping out.

mnem
 

That's a good test.  I'll set that up. 

One thing I don't understand is that the LM2940CT datasheet lists this part as 0.5V dropout, and the dropout plots clearly show 0.5V at 1A as the dropout.....
But the electrical characteristics for the 5V version clearly show the input voltage as:
Input voltage : 5 mA ≤ IOUT ≤ 1 A: Min=6.25, Max=26
https://www.ti.com/lit/ds/symlink/lm2940-n.pdf (page5)

I don't necessarily just want to start replacing parts that aren't showing failure symptoms, but the LM2940CT is probably going to be the first to go.

[AVGresponding]

Have you tried isolating the input to the LM2940CT and powering it with a slightly higher voltage; like 8-12V? What you're describing sounds to me like it may be right on the verge and dropping out.

mnem
 

That's a good test.  I'll set that up. 

One thing I don't understand is that the LM2940CT datasheet lists this part as 0.5V dropout, and the dropout plots clearly show 0.5V at 1A as the dropout.....
But the electrical characteristics for the 5V version clearly show the input voltage as:
Input voltage : 5 mA ≤ IOUT ≤ 1 A: Min=6.25, Max=26
https://www.ti.com/lit/ds/symlink/lm2940-n.pdf (page5)

I don't necessarily just want to start replacing parts that aren't showing failure symptoms, but the LM2940CT is probably going to be the first to go.

Have you checked the input ripple voltage? If it's reading close on DC, it might be falling under the dropout voltage at the bottom of ripple voltage. Though the output you measured looks good to me; I'd be surprised for anything designed to work using a standard 5V linear regulator like this to stop working at 4.94V if it's not faulty in some way.

[Smokey]

I've been out all day.  I'll check the input soon.

But I agree 4.95v is usually pretty stellar for an ldo.

I had this idea that there may be some adc or comparator somewhere that is confirming rail voltages before enabling the system and a voltage divider resistor is out of spec causing the thing to never turn on even though the actual voltage is fine.
But there is no schematic....

[Smokey]

Considering that both the +25V and -9V test points are expected to have unloaded voltages much higher than what is listed on the silk screen....

...I wonder if the +6.4VD testpoint that feeds the 5V regulator is actually supposed to have much higher voltage than that as well. 
::Update:: ... its' not.

I know you guys probably already put your supplies back together, but if you could measure one more voltage, it would be interesting to know what a good unit measures at the +6.4VD testpoint on the board right next to the fan.
::Update:: No need for this.  It's just the output of a switching regulator is that is programmed for 6.4V.  It's doing what its supposed to.