Hi Sven.
You seem to have linked to various documents to do with various clocks of mine, so I better respond.
The circuit I came up with is a hybrid of the typical circuit diagram used to drive a 6 or 7 digit seven segment display and the circuit used to drive the 16 digit alphanumeric displays in pinball displays.
Those displays as used in pinball machines had a system involving -100v 0v and 90v. There were obsolete UDN chips running the high and low side. In practice a lot of these display boards failed with the chips working close to their maximum and people have resorted to workarounds to replace them. I looked at the design for my BK2K, Black Knight 2000.
1] You get the values for those resistors by looking at the circuit diagram and parts list for any pinball machine that uses them. It's just one resistor change to adjust the brightness balance for the segments. Otherwise I used the high and low side drivers as in the 6 and 7 digit Bally machines. As you say, there's a PDF of these diagrams on my Clock-It website at:
http://www.clock-it.net/neon/bally/6-Digit_Display_Driver_(AS-2518-21).pdf and
http://www.clock-it.net/neon/bally/7-Digit_Display_Driver_(AS-2518-58).pdf
I used 1/2w resistors.
No I didn't get the values from Babcock, nor do I see any values on the Babcock document you linked to. I got them from my BK2K pinball schematic and parts list - only the parts list has the values, so you have to find each segment on the schematic and match up the resistors. I'd have to look to see what values I used, I probably wrote them down somewhere. The 100k resistors on the digit drivers should be 1/2w too, although pinball manuals say 1/4w. They cook and burn out.
2] I don't use pin 1. My display works just fine, no it is not necessary and I don't think pinball machines use that pin either, from what I can see. Pin 90 does not exist on my Cherry displays, so N/C it certainly is.
3] The 100v zener will be fine. It's simply a pull mid. The techies here will tell you what that is all about. I'm not sure, I think it's about fast firing or reducing ghosting or something. I used a 110v zeners because I pulled one from a stack of dud display boards I had handy.
You need to sweep the display quite fast to get it flicker and ripple free, you are doing a 16 multiplex. I used a PIC microcontroller to do this with a counter outputting a 4 bit code to a 1 of 16 decoder with latch and inhibit. You could simply make a four bit counter and have it free-running the 1 of 16 decoder to sweep the display. You could still use the inhibit pin to blank the display when necessary. This would take the load off the microcontroller and free it up to do other things which you then don't have to interleave with the display routine. I will probably do this in future as it can be tricky with various scrolling routines taking different times and as a result varying the brightness - characters start to jump out at different brightnesses.
When adjusting the high voltage it will tell you when the voltage is correct. Too high and you will see some arcing and ghosting on unlit segments. Too little and you get flickering and uneven illumination. There is a sweet spot, you will know it.
My routine seems to work well for a rock-steady display using a PIC. I added a bcd to 7 segment chip in case I want to simply send numbers rather than segment patterns individually, but in the end never used it. I defined all the character shapes in a word array corresponding to the ascii code for that character - send the ascii number for that character and it switches those segments on.
John S
> You could simply make a four bit counter and have it free-running the 1 of 16 decoder to sweep the display. You could still use the inhibit pin to blank the display when necessary. This would take the load off the microcontroller and free it up to do other things which you then don't have to interleave with the display routine.
This isn't going to work, as you still have to synchronise the bit pattern for the segments at the same time as you select the appropriate digit, so the complexity is still there - all I've done is increase the chip count. Bright ideas late at night...
John S
> Hello Group!
>
> I received
> http://de.mouser.com/ProductDetail/Vishay-Dale/PD007D07001S51/?qs=sGAEpiMZZMvkC18yXH9iIurp2WUOVpc7D8%2fd1LKTOnY%3d
> some days ago, wondering how to drive it.
Sven, now I am confused...
You label the the thread Vishay/Dale 16 digit... and give a link to what you bought, a 7 digit display. Do you have the pinout for the display you have bought?
John S
> Now I'm in the process of doing the PCB layout and getting my BOM together.
Sven, do NOT go by my power supply layout. The two 1 ohm resistors should be replaced with a 0.05ohm. Also my layout is really bad. It works now, but I had to change the feedback path, ie move the pot. It would be better to move the FET so it sits at one end of the MAX1771, not at 90º as I have it. Generally, if you are going to get a double sided board made, go by Nick de Smith's layout on his website:
http://www.desmith.net/NMdS/Electronics/NixiePSU.html
The MAX1771 is very picky - I don't get much success with it. It is prone to spikes on that feedback loop and won't regulate.
John S
I'm sorry, I forgot to mention that Mouser got their data wrong (7 segments instead of 14+2). I even notified them.
Since Vishay hasn't a web page about it, I made one where I try to collect data. I'm so free as to publish the production documentation they were kind to send me, I hope they don't mind:
http://sven.killig.de/displays/Panaplex
As you can see, it's very similar to your's, but there's a discrepancy that makes me headaches: on page 49 in
http://mirror2.ipdb.org/files/311/Williams_1989_Black_Knight_2000_Operations_Manual.pdf
I can recognize 3 different resistor values for the segment areas, whereas in the PDF from Vishay/Dale there are nine...
X-)
It could work with your discrete continuous multiplexing added to one of the circuits in the app note
http://catalog.osram-os.com/catalogue/catalogue.do;jsessionid=?act=downloadFile&favOid=02000002000047e5000100b6
You're right. I'm sorry, I mixed it up. The correct URL would have been
http://de.mouser.com/Search/ProductDetail.aspx?R=PD016A07004S51virtualkey61300000virtualkey71-PD-16A070-4
I'm yet too unexperienced in EE to layout a DC/DC-converter, so I ordered the HVPS-H from
http://www.tayloredge.com/storefront/SmartNixie
I'm just wondering whether it will run from (dual) USB power, as with 5 V supplied it will only provide 5,5 W.
> As you can see, it's very similar to your's, but there's a discrepancy that makes me headaches: on page 49 in
> http://mirror2.ipdb.org/files/311/Williams_1989_Black_Knight_2000_Operations_Manual.pdf
> I can recognize 3 different resistor values for the segment areas, whereas in the PDF from Vishay/Dale there are nine...
You are correct, Williams Pinball use only three values as follows. You need to look at the lettering in your diagram for the segments.
As you would expect, smaller segments have less current and a higher resistor value.
Segments:
Dot, Comma, g, m = 18k
e, f, c and b = 10k
r, p, n, d, k, j, h, a = 8k2
Which makes your 16 segments in total.
Note that the right hand two numbers on the display do not have commas, no need for them with numbers under 1,000. You might need to know this if you are intending to scroll sentences with commas!
I believe Data East used the same displays around this time. Robocop is a pinball I had that I am sure had 16 digit alphas. I have the schematics for that somewhere but really, just go with the three Williams values, they are fine. If you want to get picky, just adjust them yourself afterwards if you see one segment that seems brighter or dimmer than the others. They look fine using the three values, as far as I am concerned on the board I made. The Black Knight looks OK too, I cannot see any difference in brightness among the segments.
The displays are pretty bomb-proof unless you bust that nipple off the back. Glue a piece of tubing over it at the very least. You don't want to be breaking these displays, they are costly.
On old displays it's quite easy to rip the metal tag connectors out. Even if you do pull one out, they will usually push back in. There is a bead of resin to hold the connectors into the glass, but this has often flaked off in places and can lift quite easily if you bend the metal tags. I've resurrected displays that had a dozen or so connectors ripped out and it was OK after pushing in some contacts from a broken donor display. Some go in one way round and some the other. Hot glue them in if they are loose. I've even heard of pinball people using a dremel to grind away some glass when a connector has broken off inside the glass. Conductive paint can be useful too.
John S
FWIW, there's actually a table in the datasheet that gives the surface
area of each segment, and I'm going to speculate that the current is
directly proportional to the area.
Bob Armstrong
Yes, but Williams lumped in the two short middle segments to be the same as the dot and comma, which don't really have the same surface area. I think any difference in appearance is so minimal they decided to ignore it.
John S
thanks!
Since this is my first PCB, it would be kind if you have a look at my schematic PDF
http://sven.killig.de/displays/Panaplex/Panaplex.pdf
or the Diptrace file
http://sven.killig.de/displays/Panaplex/Panaplex.dch
I plan to order the PCB at ITEAD, where they can only be ordered in batches of five. So if you're interested, I could send you one when it's finished.
Personally I could make use of one. I think I have another three good displays here, so yes please.
I can send you a bunch of 90v zeners if you want them. Used, but good.
Use 1/2w for the 100k resistors. Nicko told me I didn't really need to use 9k1 resistors, 10k will do just fine. 9K1 was on the Bally board design I copied.
ITEAD use the same factory as Seeedstudio (yes three eees) so it's pretty much down to any cost variation whom you choose.
I cannot read a DCH file, so I cannot comment on that. Can one of the EEs here look the schematic over? It looks OK to me, but I am not very clever.
You may need to experiment with a small off period to avoid partial ghosting on other segments. I do my all display routine calculations during the off period, fishing for the correct ASCII character and wanging the correct segment pattern out, so I don't actually have any additional delay as such. I find there is a sweet spot where you can get a crisp bright display by adjusting the voltage.
Good luck with sweeping those sixteen digits at a constant rate. If you get timing variations you will see the ripple.
John S
Then I'll try to free the PD0/SCL pin, so you can use the AT90USB1286's I²C controller.
The last obstacle is the power: As you can see on page 6 in
http://www.tayloredge.com/storefront/SmartNixie/DataSheets/Datasheet_1363-1364.pdf
the 1364 can only deliver 5,5 W or 27 mA when supplied with 5 V. Do you think this will be enough?
http://www.vishay.com/docs/37003/016a070.pdf
speaks of 4-9 W...
Nice coincidence: I just fetched your 1364 out of my mailbox :-)
> This display does not appear to need the high voltage directly (~170V)
> that a bare gas plasma display would require; the data sheet says this
> is generated this internally with its own DC/DC converter.
I might have been a bit confusing here: I just linked to the PDF because the only power consumption information I know of is in there. I have a bare display indeed.