EEVblog #641 - Dumpster Cash Register Teardown

[EEVblog]

Teardown Tuesday.
Dave tears down a cash register found in the dumpster.
Does it have any useful parts to salvage?
German brand Quorion QMP 3282

Datasheets:
http://www.powertipusa.com/pdf/pg12864ak.pdf
http://www.altera.com/literature/ds/m3000a.pdf
http://www.paradigmtools.com/docs/R8820.PDF
http://www.st.com/web/catalog/sense_power/FM142/CL851/SC1790/SS1555/PF63141
http://www.st.com/web/catalog/sense_power/FM142/CL851/SC1794/SS1498/PF63236
http://www.atmel.com/images/doc0313.pdf
http://www.onsemi.com/pub_link/Collateral/MC34164-D.PDF

[winfreak]

Great teardown, I have always wondered how they look from the inside...


BTW: How would you go about on displaying anything on that display you salvaged?

[EEVblog]

BTW: How would you go about on displaying anything on that display you salvaged?

The LCD is easy of course. The fluoro one a bit harder. You'd capture the lines while it's working and decode the pinout and data required.

[winfreak]

BTW: How would you go about on displaying anything on that display you salvaged?

The LCD is easy of course. The fluoro one a bit harder. You'd capture the lines while it's working and decode the pinout and data required.

Sounds tricky but quite doable!
I'm looking forward to working with Arduino + such displays...

[dentaku]

It was even more interesting than I imagined.
I like those displays.

The big keypad could be used as a crude very low resolution push (as opposed to touch) sensor of sorts that knows where i's being pushed on.

[nomis]

Does it have any useful parts to salvage?

About the paper cutter: Note that there is a small rectangular notch in the center of the blade. By controlling how far the blade cuts down you can either cut the paper completely or leave a small connection between the receipts to avoid creating a mess of loose paper snippets.

I once had to deal with this kind of printers and it was quite fun to do not-really-complete-cuts in sub millimeter increments in the paper. Instant paper festoons! 

Bye,
        Simon

[Hole]

Hmmm.

This German thing has a 16 bit RISC cpu with all the bells and whistles it can have. And an 8 bit cpu with not so much bells. And all that for adding some numbers every other second. And then do some odd multiplications or divisions?

Sure a big canon to shoot that small bird.

[homebrew]

Hi Dave,

nice teardown! But to my perception it is not engineered that well, when it comes to safety. In no way I can think of this device should be conform to any safety standard...

1) Obviously some parts of the case are grounded (like the drawer) and others should be as I don't see ANY form of double insulation from the mains. There isn't even the otherwise typically installed insulation foil between the PSU and the chassis.

2) Mains input wiring directly goes to the PSU, including the ground wire. From there the only connection to the rest of the system is through the wimpy metal bushing around the plastic PCB standoff. It looks like being made of copper and if it is pressing against the tinning of the PCB only (Video 14:40). Thus the whole ground system is held together by a PCB mounting screw that grabs into plastic...

Baaahhhh ...

[SeanB]

What Dave did not notice is that the power supply only delivers 24VDC. the individual boards have built in voltage regulators to generate the supplies they need from that. The VFD probably uses a serial data stream that generates the characters, i got 2 from Kolbep that use a single 64 output VFD driver that drives the various columns and rows of the VFD. That probably is the job of the one ASIC on the main board. You also will find around 64 or 128k of battery backed SRAM as well to store both the written description of the various items ( tea, coffee, tuna sandwich, BLT, Smoothie etc) as well as the assignations for the various keys on the keyboard, along with keeping a tally roll and day totals for the register.

The printer being a separate board is to enable the same main board to be used with the thermal printer, a dot matrix printer or even to drive a standard external parallel printer, you will probably have a centronics interface on the pins, along with the 24V supply. You would have a small board with a set of buffers and a 5V regulator if you had the external printer. The reason for the extra grounds is because they did not consider that those print heads draw a peak pulse current of around 200mA per pixel element, thus you see the multiple tabs on the head that are at the side that are the ground connections, you can have 5A or more flowing through those pins, and on the top side you will see a big ( though very low profile) decoupling capacitor on the 24V rail to the print head. You probably have 512 elements on that head, and they are driven at around 1MHz to clock the serial data into them along with the PWM value per element ( to give a gray scale of around 16 values from nothing to full black) and this has to be done for each line. The 2 motor drivers are for the stepper that does the paper feed and the cutter, and the cutter driver also drives the register open solenoid. You print an ASCII bel character ( ctrl G) and the drawer opens, typically for a POS printer.

Older POS printers used a single heater element running on a spiral track on a rod, that is driven by a DC motor and an optical encoder to scan along the paper, with the one end of the track having a ratchet that moves the paper up one pixel for each full stroke of the head. Slower and the head driver is then on the board with only a 4 connection flex to the head, 2 power and a pair to measure the head temperature for feedback to control the density. On the integrated heads the sensor is on the flex and feeds all the drivers to keep the print density constant ( same final instantaneous temperature ) as the head assembly heats up in use.

[sergey]

Nice teardown indeed. Wouldn't really expect such a computation power in things like this. You can easily fly to the moon with those DSP

The good thing is even if you fail to decode the serial interface of the VFD you can always decode individual segments of VFD itself, so no worries at all and you definitely need to salvage parts like this!

[phenol]

The thermal head is made by Kyocera using thin film technology. The potted driver ics are basically shift registers with low side mosfet switches on every dot. The dot resistance is typically between 600 and 1500 ohms on those 24v heads. looks like a 200dpi head, each dot is 1/8mm big. The head can be clocked at speeds much higher than 1MHz, typically 5 up to 30 in faster models, and yes, it does draw a lot of current when a lot of dots are fired simultaneously. Today there are thermal printers that can operate at voltages as low as 2.7-3V.

the cutter can do partial or full cuts just by reversing the polarity on the dc motor inside.

[dexters_lab]

never seen so many ferrite rings in one product!

[magiccow]

never seen so many ferrite rings in one product!

yeah nice collection of ferrites...

[Dave Turner]

I think I spotted another bodge on the main board - a flying resistor and capacitor at about 12:18.

[synapsis]

I picked up two standalone IBM thermal receipt printers off eBay for dirt cheap. I used one to make an obnoxiously huge Welcome Home banner for a friend (200 feet long).

The cutter and printer would probably be good for making mailing labels or small instruction sheets for uCurrents.

[nitro2k01]

never seen so many ferrite rings in one product!

This one beats it by a mile. 39:41 into that video. Granted, it's a one-off product, but look at those ferrites!

[David_AVD]

The little release lever at the back of the cash drawer is common.  It's how you open the drawer in case of power failure.

[Rasz]

awww
that board to board connector screamed 'get a hotair and demonstrate how to desolder me !!!!111!seven"


I cant picture in my head how a thermal head like that works. I mean I know how it works, it just doesnt compute, you would need a big thermal mass/heatsink to dissipate energy quickly between pixels, otherwise you wont get sharp edges ... but this thing has no problem with sharp edges, so wtf ?

btw thermal paper is great when you need to locate hot spots but have no access to thermal printer, just plaster your board in thermal paper, turn it on and wait for black to appear near hottest chips/shorts.

[Paulusthewoodgnome]

Make a nice microprocessor controlled toilet paper dispenser with the paper cutting module

[miguelvp]

Make a nice microprocessor controlled toilet paper dispenser with the paper cutting module

I'm pretty sure there is no market for very narrow toilet paper.

[cengland0]

Hey Dave, love your videos.  Just wanted to point out that you missed another bodge near the backup battery:

[German_EE]

That VFD display is a thing of beauty, you could make a very nice frequency counter using that module.

[Niels]

Why do they (RDC) denote the 80C186 clone as an RISC/DSP processor?

RISC architecture have more done in parallel to reduce execution time, this costs extra gates/silicon. In order to reduce the silicon the number of instructions is reduced.
The 80186 burns a lot of clock cycles to do noting..

And an DSP processor is expected to be able to perform Multiply-Accumulate on a vectors, the 80186 can't do this.

The 80186 has been used a lot in embedded applications, even if it have some drawbacks.
It does not have JTAG debug interface, it does not even have debug registers (Introduced on 80386)

Debugging was at best performed with a debug monitor in ROM, connected with a serial interface.
The debug monitor had to overwrite the code with INT3 instructions.
This required the code to be located in RAM, and often there was not enough RAM to support this feature.

The reason for using the 80186 is its days, was the possibility to run DOS, and/or use DOS development tools and library’s.

Paradigm (The site you were linking to) makes tools to locate, and debug 80186 a target without DOS.

[phenol]

The company i work for makes all sorts of thermal printers, the lightest one being 27 grams or so.
We also have a big variety of printers with built-in driver boards the same size as that printer in the video. all you need to start printing is power and paper. it can work as a windows printer or independently printing ascii chars rx'd over serial or usb.

The print head itself does not necessarily need large thermal mass to dissipate heat between lines; the dots are designed to cool down sufficiently between print cycles. If they fail to do so, the image bleeds. Even then there are software tricks to mitigate bleeding, ie historic control.

As for the VFD, i recently made a freq counter with 8 7-segment ussr-made VFD tubes...all in all, filament voltage, grid voltage and whatnot, it draws a lot less current from the 5-v rail than 8 7-segment led displays

[nixfu]

I was wondering what he was going to do with that big thing that had been sitting up on top of the shelf, I noticed they had to move it when they were making the new bench the other day.  I am guessing the removal of that shelving unit is where the idea for this teardown came from.  Either tear it down or throw it out.

>Why do they (RDC) denote the 80C186 clone as an RISC/DSP processor?

I will tell you why.  RISC was a major buzzword in the 80s and they were simply trying to jump on the bandwagon.