Keithley 2306 Dual PSU / Battery simulator repair

[SparkyBruce]

Hello, I have just started a repair of a Keithley 2306 dual PSU/ battery simulator. I have not been able to find any schematics for the unit , but the service manual does have a BOM list and component location drawing.

The initial state was the unit would display Initialising on the screen after about five seconds but not anything else.

So far I have found that the rear fan has died and that the message display is coming from the micro on the front panel itself. The main micro on the lower digital PCB is running code from the Flash ROM (32PLCC) and I can see accesses to the main RAM , but nothing to either of the two I2C serial flash config chips.

There is no communications to the front panel board (looks like a two wire serial link), and only the main supplies (12V and 5V rails) are up.

If I remove the main Flash chip then the micro appears to just cycle through the address lines so it does look like it is running some code, just not talking to the front panel. I guess the code must be failing to configure something and it is just sat there in a tight loop waiting.

The dual PSU channels look like they use a front end 12V to +25V boost converter (isolated) per channel , then they do post regulation on the outputs of these on the analogue PCB.

I have not been able to find any pictures of the inside of the unit (2302/2303/2304/2306 or 2308) on the web so I will take some pictures and post them over the next couple of days.

There do appear to be a number of these units on the market at the moment due to the shutting down of a number of mobile phone development centres.

If any body has any suggestions as to my next step then please let me know.

Thanks

Bruce

[KJDS]

I've fixed two of those just by unplugging and plugging back together everything I could so unplug, though I assume that was the first thing you tried.

[SparkyBruce]

Hello,

Yes it has been to bits a few times mostly to clean out all the dust.

I was hoping that it monitored the fan for rotation , but that doesn't seam to be the case. I will check the thermistors tomorrow to see if any of them are tripping , but I would expect them to be flagged on the screen rather than a no-boot.

It could even be a failed flash upgrade , so I will try a new chip with the latest firmware that I have been able to download off the Keithley support site. It is still a shame that the schematics aren't available.

Thanks

Bruce

[picburner]

Hi,
 
this was the same fault of my 2306.
In my case the culprit was K6T4008 (U124) memory.
The instrument, at power on, performs a hardware self test.
If it finds something wrong it lock without giving any error message.
As you've already noticed "Initialized" is produced by the display circuit only.
The fault may be RAM, ROM (checksum error), or TMS9914.
If there is no complete activity on the address/data bus XTAL or CPU.
You can find the update firmware on Keithley forum.

Good luck
Giampiero

[SparkyBruce]

Thanks,

I have checked that initially the main micro reads from the Flash , and briefly does some SRAM accesses , then it looks to be reading from flash then accessing something else (not the SRAM or the GPIB chip from looking at the chip selects).

From a quick look at the actual chip selects for the micro I couldn't see any of them toggling so I need to check that I have the correct version of the datasheet.

So I will need to buzz out which chip selects go to which devices. I guess they have some simple byte wide I/O ports to control the rest of the system.

Thanks

Bruce

[picburner]

The pins 2 &12 of U107 are high or low?
If  high there isn't  clock for the power supplies, this means that the sequence of the diagnosis is not finished
for some reason, and the system is locked.
I attach a pdf with a partial schematic I started drawing (then when found the fault I left it incomplete...)
and the photo of section involving my fault (U124).

Regards
Giampiero

[SparkyBruce]

Thanks Giampiero,

The switchers are not running so I guess 2&12 of U102 are low.

I have buzzed out the following:-
CS0 is the FLASH CS which is what is expected
CS1 is the SRAM
CS2 is the GPIB controller
CS3 links to the serial flashes (U133)
CS4 goes to U138.9
CS5 goes to U138.1
CS6 is U107.2
CS7 is U201.2 & U101.2
CS8 is U138.5
CS9 is U138.1

The 50/60HZ mains detect connects into PE4/AS pin 106 of the micro.

I will scope out these signals to see what is going on. When it is sat there it is reading and writing to "something" not the SRAM or the GPIB so I need to see what it is talking to (it reads code out of flash all the time).

Thanks again for your help.

I am hoping after getting the micro happy all the rest of the complicated analogue "stuff" will be ok.

Bruce
 

[picburner]

Hello Bruce,

I was fortunate enough: the only fault was the SRAM.
I replaced the memory, performed the calibration and then the instrument is passed from photo 1 to photo 2.
I hope that you are so lucky.

Regards
Giampiero

[SparkyBruce]

Hi,

I have just checked for what happens if I lift the SRAM CE pin (and tie it high to it's inactive state).

The result is exactly NO change with what the signal lines are doing.

So it does look like the RAM is faulty (new one ordered).

So why does the Keithley kill it's SRAM?

Bruce

[SparkyBruce]

Well thanks to Giampiero my new sram chip arrived this morning.

After fitting and testing I can confirm that this has fixed my unit.

So everything else is fine.

In the end I had to replace the fan and the sram and now it is good to go so I just need to let it warm up and I can verify it's calibration this afternoon.

Thanks for your help

Bruce

[SparkyBruce]

Ah - I replied too soon.

Channel 1 started to get a negative offset after about 1 hr of use set to 3.000V
It went down to 2.780 volts on the display (and on my external DVM) , despite being set to 3.000V

Channel 2 is still happy.

So the monitoring side of things (voltage out and current) are still correct , there is just this difference between demanded voltage and output voltage.

I have quickly tried a calibration run at home but despite being better than before there is still an offset between demanded voltage and the output.

I am going to swap out the two elec caps (10uF 35V and 220uF 50V) tomorrow just in case it is one of them.

The channel 1 side of the heatsink does feel warmer than the other side so it could even be an issue with the power transistor driver.

Thanks

Bruce

[picburner]

Hi,

A few days ago, during a full load test on both channels, it has  suddenly restarted without giving error messages.
Perhaps it is the case of making a ESR control of all electrolytic capacitors.
In normal conditions, apart the previous case, my instrument has no problems of overheating in the analog power section and the voltage will not drop under load.

Giampiero

[SparkyBruce]

Sorry to hear your troubles.

I have been checking the zenners on my unit as it appears that channel 1 is running far warmer than the other channel , despite channel 1 mostly working.

The output voltage is a few hundred millivolts low from the demand voltage. So if I set 1V I get 700mV ish which does wander a bit , but the output current sits at about 0.

When I load it up it does drive the output and measure the current correctly so it is acting like the bias current for Ch1 is too high (hence why I am checking the zenners in the circuit).

I will be borrowing a friends thermal camera to see if that can shed any light on what is going on, but I guess I will have to map out the circuit.

The big fets are IRFP240 & IRFP940's.

Bruce

[SparkyBruce]

Ah fixed.

In the end after borrowing a friends thermal camera (thanks Steve) it was tracked down to one of the op-amps in the output voltage sense cct (an AD818). the chip was getting over 100C and after a false start with the main voltage control op-amp (which was also very hot) it proved to be the positive sense cct that was messing everything up.

After a quick re-cal all looks happy so I will run it over the weekend to see what it does.

Thanks

Bruce

[Prehistoricman]

Big thanks to those in this thread who found this solution and shared it with us all. I also had a Keithley 2306 display INITIALIZING forever and I ordered a replacement RAM on eBay that fixed the issue.

My unit also had a faulty fan. It works at 6V but higher voltages cause the fan to run slower. Perhaps one of the windings is open.

Thank you!

[picburner]

Interestingly, three defective SRAMs on three instruments.
Maybe a lot born bad?
If you can be interested in the attachment there is the B18 firmware.

[Smokey]

It's a shame the unit in the pictures here was used in a dust factory!