Author Topic: EEVblog #371 - Universal Programmer Teardown (Read 29749 times)

EEVblog · « **on:** October 16, 2012, 12:12:11 pm »

Dave.

MadManMarkAu · « **Reply #1 on:** October 16, 2012, 01:27:19 pm »

Hi Dave,

Did you manage to get the firmware reflashed onto the device? I'd like to know how that went. Did the firmware upgrade came as a .hex file or some other directly uploadable format, or was it possible to extract the directly uploadable format from the upgrade package?

Monkeh · « **Reply #2 on:** October 16, 2012, 03:14:00 pm »

Looks like the processor let the magic smoke out.

The soldering on that thing is ugly.

tom66 · « **Reply #3 on:** October 16, 2012, 05:34:30 pm »

18:50 those are BAS70 double-diodes.

I would guess it drives the IOs with whatever - 5V or so - and relies on the diodes and resistors to protect the FPGA.

I think the FETs are on the top board.

pickle9000 · « **Reply #4 on:** October 16, 2012, 07:05:18 pm »

One thing I like about the wellon programmers is that most can test and ID 74 series and 4000 series logic.

jpelczar · « **Reply #5 on:** October 16, 2012, 08:00:46 pm »

Maybe the FPGA bitstream is loaded from PC, when you run the programming software ?

EEVblog · « **Reply #6 on:** October 16, 2012, 10:01:24 pm »

Quote from: tom66 on October 16, 2012, 05:34:30 pm

18:50 those are BAS70 double-diodes.

Yes, BAV70's, I missed the silkscreen.
Even less flexibility than I mentioned then.

Dave.

free_electron · « **Reply #7 on:** October 16, 2012, 11:26:00 pm »

pretty poor circuitry ...

i'll rip my Dataman and Hi-Lo apart tonight and post some images.... that's a whole other can of worms. ( i bet this wingpangpong cannot even program a real PROM like an 82S182 or a triple voltage eprom like a 2508 ) as for the programming alogrithms , those may be off as well. Too bad i don't have my old Data I/O 3900 anymore. that was a beast when it came to internal electronics...

One thing these machines invariable claim is 50000 devices or 70000 devices , but by the time you strip out all the 27c128 of all the different manufacturers and all the 27c64 and 2402 and all other duplicates you end up with about 800 devices....

westfw · « **Reply #8 on:** October 17, 2012, 01:52:09 am »

What is the most simultaneous voltages that any programmable part has ever needed? I can't think of more than three - a normal supply for internal logic, a higher voltage for blowing fuses, and a perhaps-different higher voltage for activating programming mode...

SeanB · « **Reply #9 on:** October 17, 2012, 04:58:38 am »

2508 needs 5V, 0V and a -21V substrate bias. To program it you need a 21V program pulse, along with a 6V verify applied to Vcc to shift the level to make sure. They do retain data well though, 40 years and no failures over the full temperature range no problem. 4 supply voltages for a 40 year old device, if there was a NMOS version ( not sure) it would need an extra -9V supply and a 12v supply as well.

EEVblog · « **Reply #10 on:** October 17, 2012, 05:00:14 am »

Quote from: westfw on October 17, 2012, 01:52:09 am

What is the most simultaneous voltages that any programmable part has ever needed? I can't think of more than three - a normal supply for internal logic, a higher voltage for blowing fuses, and a perhaps-different higher voltage for activating programming mode...

That's why I drew two as an example on my diagram.
DAC-per-pin is complete overkill in most cases, you could have a very flexible programmer with just two or three DAC "busses" that could be switched to any pin.

Dave.

notsob · « **Reply #11 on:** October 17, 2012, 05:09:43 am »

Looks like a suspect product anyway, although you can still get them on fleabay, the wellon site does not appear to exist [www.weilei.com], have fun getting software updates for it.

Same for that other product from genius - software download site doesn't exist, unless they are supported elsewhere.

EEVblog · « **Reply #12 on:** October 17, 2012, 05:22:42 am »

Quote from: notsob on October 17, 2012, 05:09:43 am

Looks like a suspect product anyway, although you can still get them on fleabay, the wellon site does not appear to exist [www.weilei.com], have fun getting software updates for it.

The website works fine for me, and updates seem to be weekly. Last software update was yesterday.

Dave.

notsob · « **Reply #13 on:** October 17, 2012, 05:51:03 am »

interesting, I get page not found or server not available, even though weilei.com is a valid dns with internic

Found it http://www.weilei.com.cn/indere.htm

free_electron · « **Reply #14 on:** October 17, 2012, 11:52:38 pm »

as promised
- dataman48LV ( low voltage 48 pin machine)
- old-skool printerport interface.

1st picture : outside view
2nd picture : pin driverboard. a bunch of allegro microsystems and holtek chips , diodes all under control of a Xilinx FPGA
3rd picture : UDN2987A and Holtek SP720A and associated resistor and cap networks
4th picture : the zif socket itself is socketed. as these things do wear out they are easily swappable
5th picture : the brains of the outfit. an 8032 cpu with external ram and rom. combined with some big-ass i/o chips (UCN5833A : 32 channel , serial input drivers ) , a bunch of reed relays and transistors in TO126 body ( BD679 )

the 8051 ( 8032) runs a partially fixed / partially adaptive program. it has enough code on board so that you can 'upload ' the correct missing part to the 8032 and the fpga. that code is then jumped to. this allows for easy algorithm updates. the pc program holds a massive pool of binaries and loads the relevant code to the cpu depending on the chip that needs programming.

this machine has manufacturer certified programming algorithms and can do many devices that 'cheapo' programmers never can handle ( because they can't get the programming specs )

There is only a couple of 'real' programmer makers out there. Hi-Lo systems , Data I/O, BP microsystems , Dataman and a few others.
The others are approximations. Sure anyone can do the classic eprom / flash and serial proms and some pal / gal as the algorithms for these are public domain. But try doing a program of a CPLD like an altera 900 series.... or a uPSD device .. the algorithms and methods for those chips are subject to NDA and chip manufacturers only give them out to reputable programmer makers. Ain't going to find them in wing-pang-pong. Often the chip manufacturer has to work with the programmer maker to verify correct operation of the algorithm.

The waferscale PSD ( see mikes xray teardown ) were only ever avaialble on Hi-Lo and BP microsystems and DataI/O as those were the
only ones that they worked with to get the algorithm in there. The gold standard for a long time was the Data I/o 2900 and 3900 and later the Labsite / ChipLab programming system. Those beasts could really do any chip out there. They were very complex machines full of asics. don't bother buying them on ebay. they are worthless without the driver software which you can't get .. the software came on floppies that were paired to your serial number. You could not run on different hardware and yuo needed the correct hardware options for your 'build'. horrendously expensive, but the gold standard. any production facility had those machines. A used 3980 with software goes for several thousand of dollars. these machines have swappable top-plates that allow you to program esoteric packages such as PGA's , tssop , plcc and others without having to futz with adapters.

the socket for plcc's and tqfp is made with a kind of elastomer. very weird stuff.

They have 88 pin drivers each with their own DAC and can program devices with supplies as low as 1 volt. verification for correct programming is done at vccmin and vccmax. they are really production beasts.

free_electron · « **Reply #15 on:** October 18, 2012, 05:09:21 am »

here's the teardown of the Hilo machine.
1 : top view
2 : device module. these pop off . there is one for 48 pin and a qfp one as well ( i only have the 40 pin)
3: guts of the device module. some ttl logic 573 style
4: top of the device module : lots and lots of hybrid style pin driver modules.
5: closeup of the pin driver modules
6: the brains in the bottom of the machine
7: essentially a souped up 87c52 ( on board flash ) with an external 32k ram for code and 1 meg ram for data. just like the dataman they upload the programming algorithm as a chunk of 8051 code , upload the data to be programmed and this thing takes off and does all the timing stuff. they use an ftdi 245 for the usb interface.

EEVblog · « **Reply #16 on:** October 18, 2012, 05:19:37 am »

I think those oversized aluminium heatsink plates are begging for trouble...

Dave.

free_electron · « **Reply #17 on:** October 18, 2012, 05:48:10 pm »

trisonic · « **Reply #18 on:** October 27, 2012, 01:07:43 pm »

I want to buy an universal programmer called TL866CS, what do you think about it?
Is anyone use it?

EEVblog · « **Reply #19 on:** October 27, 2012, 01:23:42 pm »

Quote from: free_electron on October 18, 2012, 05:48:10 pm

?!?

The plates appear to be directly laying on the board, overlapping the ground plane, with only the solder mask separating them. Any burr on the heatsink could easily puncture the mask.

Dave.

alm · « **Reply #20 on:** October 27, 2012, 02:14:17 pm »

I would expect both the plates and the PCB area below it to be already connected to ground, although there does appear to be a separate area right below the TO-220 packages. PCB and case might already be connected via those screws.

free_electron · « **Reply #21 on:** October 27, 2012, 04:40:56 pm »

ah now i see what you mean. heatsink is ground. non-ground transistors have isoation material on them. the slug under the heatsink is probably floating.

anyway. they've been making this model programmer forever. hilo knows what they are doing.

coromonadalix · « **Reply #22 on:** July 25, 2015, 03:22:36 pm »

Dataman 48.. and labtool-48 .. and the chipmaster 6000 are the same units, rebranded, don't know wich one is the original ?

Their software have the same gui, different logos, and for the licensing i don't know...


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

Author Topic: EEVblog #371 - Universal Programmer Teardown (Read 29749 times)

alm

Share me