Author Topic: Keithley 2400 SourceMeter review and teardown (Read 53208 times)

TiN · « **Reply #25 on:** March 01, 2016, 02:45:44 pm »

Revive, revive...

Time to recalibrate my SMU. To do so in correct way, worth to spend some python-spaghetti-coding for calibration app for 2400 + 3458A.
So far both instruments talking well, ironing bugs..

Once done, it will be posted as usual

Not sure if anyone need this

ManateeMafia · « **Reply #26 on:** March 01, 2016, 02:58:58 pm »

Hmmm, sounds like a good idea.

TiN · « **Reply #27 on:** March 01, 2016, 04:51:35 pm »

Okay, first beta sorta works.
Verification readings are not correct, but calibration is ok.

Usage is simple.

On execution it will check for GPIB:3 = HP3458A ID and GPIB:24: KI 2400 ID.
If ID matches , it will save current calibration constants of 2400 into txt file.
Then user required to connect 2400's HI/LO to DMM HI/LO.
DMM will be ACAL ALL and then DCV calibration will execute.
2400 will beep when change to current input at DMM will be needed.

Python app for Raspberry Pi + 3458A + KI2400

Calibration log TXT data

z01z · « **Reply #28 on:** April 19, 2016, 09:20:33 am »

@TiN

After seeing your teardown, I am wondering about the infamous caps. Isn't it advisable to replace the electrolytic caps here, too?

If it is not too much trouble, would you be so kind as to capture the output of your 2400 with a scope? If possible both with light (20V, 100kOhm) and heavy load (20V, 20Ohm), or whatever load resistor you have handy. I'd like to compare your results with the device I have access to, to see whether it works well or not.

Thank you in advance!

plesa · « **Reply #29 on:** April 23, 2016, 08:28:20 am »

Just bought Keithley 2410 for $300 for parts. It seems to have issues in source part.
Thanks TiN for calibration script, it will be handy after repair.
I expect some damaged transistor in source section and zener diodes. Display is quite dim, another 7001 will be harvested:)
Unit is made in 2008 and hac C27 firmware.
Pictures will follow soon.

TiN · « **Reply #30 on:** April 23, 2016, 08:33:57 am »

You filthy man

I was watching that one. Well, sorta. Hopefully it's just a broken FET. Full teardown is expected, counting on ya

plesa · « **Reply #31 on:** April 23, 2016, 09:55:35 am »

I cannot resist to buy it for parts and try repair it. Faulty 2600 series is still too expensive and 2400 is fanless, which is big advantage.
It must be some kind of addiction:)

TiN · « **Reply #32 on:** April 23, 2016, 10:12:56 am »

Well you can break yours working ones and then u will have parts

Careful with 2410, it have 1.2kV on top end rails.

plesa · « **Reply #33 on:** April 23, 2016, 10:52:01 am »

Because you know how you damaged it it is less fun:) With buying meter with unknown state it is more troubleshooting from scratch. Thats more fun I suppose.
Yep, I will use the experience from K6517A repair. I expect they are using same Sanyo 2SK1412 transistors and one or more will be dead and this will also damage components close to them.

BTW another K6517A electrometer is waiting for repair, but It is more challenging, it did not fall down from rack like first one ( broken VFD), but it has blown input stage by overvoltage spike.
There is input protection (few series resistors and LS313 ( dual NPN ) and input stage is made by LMC6001. I found one faulty 2N3904 transistor but nothing else yet.

HighVoltage · « **Reply #34 on:** April 23, 2016, 11:56:44 am »

Hello TiN,

I have the 2420-C 3A version of this SourceMeter
Will your calibration script maybe work on this one as well?

Really nice work you do.
Thanks.

TiN · « **Reply #35 on:** April 23, 2016, 12:06:05 pm »

Lol, you guys were waiting all this time to hop in?

Here is the idea for you all. How about all of us make a comparison of 24xx series, in terms of output stages and perf? Making python app to support all 24xx's should be easy enough. I can do that, also supporting different DMMs (perhaps K2001/2002, 34470,3458, etc.)

Could be good article on SMU design practicality out of this.

HighVoltage · « **Reply #36 on:** April 23, 2016, 12:31:27 pm »

Please include the 2420-C in combination with the Keysight 34470A

TiN · « **Reply #37 on:** April 23, 2016, 12:53:04 pm »

Well, I'll include, but all the testing and tweaking will be your job to do, as I have none of HP meters except 58.

etrash2000 · « **Reply #38 on:** May 14, 2016, 11:28:38 am »

Quote from: plesa on April 23, 2016, 08:28:20 am

Just bought Keithley 2410 for $300 for parts. It seems to have issues in source part.
Thanks TiN for calibration script, it will be handy after repair.
I expect some damaged transistor in source section and zener diodes. Display is quite dim, another 7001 will be harvested:)
Unit is made in 2008 and hac C27 firmware.
Pictures will follow soon.

Did you ever got to work?

If not, would you like to sell me the EEPROM. I have a functioning but bricked 2400. I dont have the equipment to re-flash the EEPROMs so I would like to buy the EEPROMs.

If anyone is interessted to sell these 2400-803 and 2400-804 EEPROM then please contact me at etrash2000@yahoo.se

plesa · « **Reply #39 on:** May 14, 2016, 07:46:42 pm »

Quote from: etrash2000 on May 14, 2016, 11:28:38 am

Quote from: plesa on April 23, 2016, 08:28:20 am
Just bought Keithley 2410 for $300 for parts. It seems to have issues in source part.
Thanks TiN for calibration script, it will be handy after repair.
I expect some damaged transistor in source section and zener diodes. Display is quite dim, another 7001 will be harvested:)
Unit is made in 2008 and hac C27 firmware.
Pictures will follow soon.

Did you ever got to work?

If not, would you like to sell me the EEPROM. I have a functioning but bricked 2400. I dont have the equipment to re-flash the EEPROMs so I would like to buy the EEPROMs.

If anyone is interessted to sell these 2400-803 and 2400-804 EEPROM then please contact me at etrash2000@yahoo.se

Not yet, but in progress. I can made image or program chip with C32 (C33 has few second delay on startup).

TiN · « **Reply #40 on:** May 15, 2016, 05:30:37 am »

Give man a fish, he will be fed for one day.
Give man a net, he will be fed forever

etrash2000, I'd suggest to get USB programmer with adapters, such as TL866 and some 2MBit Flash chips (e.g. Digikey's SST39SF020A.). This will make you almost never worry again about programming ICs.
I used TL866 myself on most of my repairs, it covered it's cost manyfold already.

etrash2000 · « **Reply #41 on:** May 15, 2016, 06:48:21 pm »

Quote from: TiN on May 15, 2016, 05:30:37 am

Give man a fish, he will be fed for one day.
Give man a net, he will be fed forever

etrash2000, I'd suggest to get USB programmer with adapters, such as TL866 and some 2MBit Flash chips (e.g. Digikey's SST39SF020A.). This will make you almost never worry again about programming ICs.
I used TL866 myself on most of my repairs, it covered it's cost manyfold already.

Thanks TIN, I think I will buy the TL886A and the memories as they were dirt cheap.
Do any of you have the memory map for the 2400, for example, were are the device serial number and calibration values stored?
Is there anything I should worry about or be careful with?

plesa · « **Reply #42 on:** May 15, 2016, 07:28:32 pm »

Yep, buy TL866A with ICSP port. The calibration should be stored in U17 (24LC16B) including serial number and calibration dates.
When you have programmer connect ICSP and make backup.

etrash2000 · « **Reply #43 on:** May 17, 2016, 08:13:32 pm »

The TL886A is now ordered so I should have the parts in 1-3 month from china. $:=\$

I also checked the original memory in the 2400 just in case anyone else is interested in this subject, it was a AM28F020-120JC (with date code 9811).
TIN/Plesa, do you know if I can use a 8M/4Mb memory instead of the 2Mb, would it benefit the equipment in any way or am I forced to use the same size as the old memory.

Regards

z01z · « **Reply #44 on:** May 18, 2016, 06:54:15 am »

Quote from: etrash2000 on May 17, 2016, 08:13:32 pm

do you know if I can use a 8M/4Mb memory instead of the 2Mb, would it benefit the equipment in any way or am I forced to use the same size as the old memory.

No benefits, only hassle.
It can work if you only want to manually program the memory. You'll have to check how the extra address pin(s) are connected though. If you'd simply put the same data to the extra memory areas, only the unconnected pin might cause a problem.
If you're planning to use Keithley's flash programmer program however, the chances are it won't like it. For that matter, the same might be true for any random replacement. I remember seeing a list of different types (probably those are supported by the program), either in a txt or in the exe itself. It's better to choose one that is supported.

Edit: In fact there's a flshdata.txt file, the following are supported by the flash wizard. TiN's suggestion is included, so that should be fine.

Code: [Select]

' Number of Flash Parts Supported
47
' AMD29F040 (+5V, 4M)
&H1A4&,&H1A4&
' AMD29F002NB (+5V, 2M, Bottom Boot)
&H134&,&H134&
' AMD29F002NT (+5V, 2M, Top Boot)
&H1B0&,&H1B0&
' IS28F020 (+5V&+12V, 2M, Single Sector)
&HD5BD&,&HD5BD&
' AMD28F020A (+5V&+12V, 2M, Single Sector, Embedded Programming)
&H129&,&H129&
' AMD28F020 (+5V&+12V, 2M, Single Sector)
&H12A&,&H12A&
' M28F201 (+5V&+12V, 2M, Single Sector)
&H20F4&,&H2011&
' M28F201A (+5V&+12V, 2M, Single Sector, Embedded Programming)
&H20F5&,&H2012&
' SST28SF040 (+5V, 4M, 256byte page erase, byte write)
&HBF04&,&HBF04&
' SST29EE020 (+5V, 1M, 128byte pages)
&HBF10&,&HBF10&
' SST39SF040 (???????????????????? it's a new chip)
&HBFB7&,&H1F0B&
' SST39SF020A
&HBFB6&,&H1F0B&
'MX29F040C (+5, 4M)
&HC2A4&,&H1F0B&
' W29C020 (WinBond, +5V, 2M, 128byte pages)
&HDA45&,&HDA45&
' TMS28F020 (+5V&+12V, 2M, Single Sector)
&H89BD&,&H89BD&
' ATMEL29C020 (+5V, 2M, 256byte pages)
&H1FDA&,&H1FDA&
' ATMEL49F020 (+5V, 2M, Bottom Boot)
&H1F0B&,&H1F0B&
' ATMEL49F002T (+5V, 2M, Top Boot)
&H1F08&,&H1F08&
' ATMEL49F040 (+5V, 4M, 8 sectors)
&H1F13&,&H1F13&
' CAT28F020 (+5V&+12V, 2M, Single Sector)
&H31BD&,&H31BD&
' W49F002B (+5V, 2M, Bottom Boot) -> AMD29F002NB
&HDA25&,&H134&    
' W49F002N (+5V, 2M, Top Boot) -> AMD29F002NT
&HDA0B&,&H1B0&
' W29C040 (+5V, 4M, 256byte pages) -> ATMEL29C020
&HDA46&,&H1FDA&
' BM29F040 (Bright Micro Elec/Winbond, +5V, 4M, 8 sectors) -> AMD29F040
&HAD40&,&H1A4&
' M29F040 (ST, +5V, 4M, 8 sectors) -> AMD29F040
&H20E2&,&H1A4&
' M29F0002NT (+5V, 2M, Top Boot) -> AMD29F002NT
&H20B0&,&H1B0&
' SST27SF020-70/90-3C-NH (+5V&+12V, 2M, Single Sector) -> TMS28F020
&HBFA6&,&H89BD&
' SST29EE020A (+5V, 1M, 128byte pages) -> SST29EE020
&HBF24&,&HBF10&
' V29C51002B-55J/90J (Mosel, +5V, 2M, Bottom Boot) -> AMD29F002NB
&H40A2&,&H134&
' V29C51002T (+5V, 2M, Top Boot) -> AMD29F002NT
&H4002&,&H1B0&
' V29C51004B (+5V, 2M, Bottom Boot) -> AMD29F040
&H40A3&,&H1A4&
' V29C51004T (+5V, 2M, Top Boot) -> AMD29F040
&H4003&,&H1A4&
' V29LC51002-90J (Mosel, +5V, 2M, 512byte sectors) -> AMD29F002NB
&H4082&,&H134&
' AS29F040-55LC/70LC/70LI/90LC/90LI (Alliance Semi, +5V, 4M) -> AMD29F040
&H52A4&,&H1A4&
' HY29F040AC-55/55I/70/70I/90/90I/12/12I (Hyundai, +5V, 4M) -> AMD29F040
&HADA4&,&H1A4&
' TE28F800C3B (Intel, +3.3V, 8Mb, x16, TSOP, 8-8kB + 15-64kB sectors)
&H008988C1&,&H008988C1&
' TE28F160C3B (Intel, +3.3V, 16Mb, x16, TSOP, 8-8kB + 31-64kB sectors)
&H008988C3&,&H008988C3&
' TE28F320C3B (Intel, +3.3V, 32Mb, x16, TSOP, 8-8kB + 63-64kB sectors)
&H008988C5&,&H008988C5&
' Numonyx 28W320FCB -> Intel 28f320C3B
&H002088BB&, &H008988C5&
' TE28F640C3B (Intel, +3.3V, 64Mb, x16, TSOP, 8-8kB + 127-64kB sectors)
&H008988CD&,&H008988CD&
' AMD29F400BB (+5V, 4M, Bottom Boot, x16, TSOP)
&H000122AB&,&H000122AB&
' AMD29F400BT (+5V, 4M, Top Boot, x16, TSOP)
&H00012223&,&H00012223&
' HY29F400BB (+5V, 4M, Bottom Boot, x16, TSOP)
&H00AD22AB&,&H00AD22AB&
' HY29F400BT (+5V, 4M, Top Boot, x16, TSOP)
&H00AD2223&,&H00AD2223&
' ST29F400BB (+5V, 4M, Bottom Boot, x16, TSOP)
&H002000D6&,&H002000D6&
' ST29F400BT (+5V, 4M, Top Boot, x16, TSOP)
&H002000D5&,&H002000D5&
' MX29F400CB -> AMD29F400BB
&H00C222AB&,&H000122AB&

etrash2000 · « **Reply #45 on:** May 18, 2016, 09:45:47 pm »

Thanks for the list. I will have look on that.

A question, according to the data-sheets, SST39SF020A is a 5V only device for reading/writing while the original AM28F020 requires 5V for reading and 12V for writing. How can the Keithley 2400 write to the memory since the Vpp is not connected on the SST39SF020A?

z01z · « **Reply #46 on:** May 19, 2016, 08:43:00 am »

Well, what surely works is using the same flash as the original. Any other type is a gamble.
Any same sized flash most probably works if programmed externally, I'd still check pin layout and waveforms.
Whether in circuit programming works or not will only turn out if tried, it is only a theory that if the type is supported by the flash wizard, it will be programmed.
What you've notice isn't the only difference, the programming word sequence also differs. It is possible to read manufacturer and device ID from these chips. Based on this info the correct algorithm can be executed, if implemented by the programming bootloader. But again, this is a gamble.

etrash2000 · « **Reply #47 on:** June 09, 2016, 09:19:43 pm »

A short update for the interested ones.

I manages to save my bricked Keithley 2400 device by using the recommended TL866A programmer with a 4 MBit memory module SST39SF040-70-4C-NHE.
I did the programming directly on the TL886. I did not dare to use ICSP since I did not have clear instruction and the pinout of the Keithley.
The programming step was straight forward and worked on the first try. I used the even+odd C31 firmware supplied by TiN on xDevs.com

The calibration seems to be OK but I will probably do a new calibration of the unit this summer by using the script TiN has written for HP 3458A.
The next step will be to do a noise analysis of the unit as z01z suggested. Pictures will be up within a few weeks (too much work right now).

So, a big thanks to all and specially TiN for the files and instructions.

Kind regards

Colnago · « **Reply #48 on:** November 23, 2016, 11:52:57 am »

2400LV stopped functioning while measuring.
Digital part seems to function to the extent that I could even upgrade the firmware to C33 but still no readings, only '----'.
When switching on output it shows +23,977V when measuring voltage, 0,0003A for current and overflow for resistance. Hardly any change when holding leads together (voltage signs changes, current slightly decreases).
All buttons are functioning, reset to factory defaults, communicates (*IDN?, *TST? etc) through GPIB but no readings.
Keithley have a €1800 flat rate which is outside the budget so any suggestion is more than welcome!

TiN · « **Reply #49 on:** November 23, 2016, 12:08:38 pm »

Standard, take it apart and check pass transistors on power stage. You can check gate resistance to source/drain first.
If all are ok, then you can measure live voltages, to see if all power rails are ok.
Be careful, there are +220/-220V voltages present, pay attention to correct ground referencing during measurement.


EEVblog Main Site	EEVblog on Youtube	EEVblog on Twitter	EEVblog on Facebook	EEVblog on Odysee

Author Topic: Keithley 2400 SourceMeter review and teardown (Read 53208 times)

Share me