How to confirm a dead TMS9118NL?

By barbeque

Supporter (8)

barbeque's picture

19-03-2020, 00:30

I received a Casio PV-7 from the Japanese auctions recently and it has a few flaws. The big one initially was that power would only work if I turned the plug in the jack just right - might need a new jack. After that, though, I still wasn't getting video.

Voltage at the 78005AP 5V regulator seems to be a stable 5.00-5.02V according to my meter, though the input voltage is a little higher than the wall-wart claims (~12V instead of 10V; I am using a 100VAC step-down transformer to power it) so I will probably replace that power supply in the future.

I took a LP-560 logic probe to the CPU pins. The CPU clock is present, MEMREQ is active, and RESET works as I'd expect (low for awhile, then becomes high). Address pins seem to be changing state. I believe the CPU works and the BIOS ROM as well.

At the TMS9918NL, XTAL2 and XTAL1 are always showing as low on my logic probe. They also show low at the legs of the 10.7MHz crystal that drives the TMS VDP. RAS is always low, and CAS is always high, on both TMS4416 DRAM chips for the video memory. This leads me to suspect that maybe the TMS9918 has died and is pulling the clock down.

Although it's anecdotal, I've also seen a post on a Japanese site (linked from the PV-7 page) where a hobbyist fingered the VDP as the source of a failure:


The system did not start even if the power was turned on when this PV-7 was obtained, but it was restored after replacing the VDP.

So my question is, are there any other tests I can do to confirm the TMS9918NL is really dead? I have some spare sockets and TMS9918ANLs from Aliexpress, so a replacement is not difficult, but I'd rather not have to desolder that big copper RF ribbon if I don't have to.


Login or register to post comments

By barbeque

Supporter (8)

barbeque's picture

19-03-2020, 03:16

Oops, typo: it's a TMS9118NL. I can't read upside down.

By hamlet

Scribe (3188)

hamlet's picture

19-03-2020, 06:54

I changed that typo in the headline.

By jltursan

Prophet (2303)

jltursan's picture

19-03-2020, 11:53

You can check the TMS9118 alone getting the composite video signal from its pin 36. Use a 330Ohm (or 470Ohm) pull-down resistor between pin 36 and GND and get the video signal from the pin. This way and mounting a female RCA you can easily check if VDP itself is generating a video signal.

If you don't get video doesn't means the VDP is faulty, usually VRAM causes black screens more frequently. Given this IC uses two 16Kx4 ICs, it's easier to check. If you have good-known replacement VRAM ICs, you can try to piggyback originals and check for changes. Of course, if some of the VRAM ICs are getting really hot and you can barely keep your finger over its surface, think that they could be faulty and it's worth the replacement.

By Wlcracks

Champion (428)

Wlcracks's picture

19-03-2020, 15:11

I repaired some msx1 computers.
Some had really big shorts and problems, but never i my cases the vdp.
Always (v)ram/rom or control-logic. No clock can be broken crystal, they are sensitive for mechanical shocks.

By Alexey

Guardian (2852)

Alexey's picture

19-03-2020, 16:22

I already encountered 2 broken VDPs. One v9938 was from Sony HB-700P that didn't generate INT and the other v9118 from Casio PV-16 that was completely dead. And the last one was dead since the 90s because the machine was relatively unused. Maybe some of those VDPs were just faulty.

By barbeque

Supporter (8)

barbeque's picture

24-04-2020, 18:34

Follow-up on my original post: I got some Chinese TMS9118s in the mail (they're surprisingly expensive compared to 9918s.)

I removed the Casio's original VDP, socketed it, and then installed one of these Aliexpress TMSes. The PV-7 works now, but I want to know for sure if the original TMS was bad or if it was something else I inadvertently fixed while removing it (e.g. the CPU clock bodge wire run over the sharp through-hole pins - although I didn't find any holes in the wire's insulation.)

I'll build a test circuit for it some time. Thank you everyone for the help!

By sdsnatcher73

Paragon (1094)

sdsnatcher73's picture

24-04-2020, 18:53

Can’t you just pop it in the socket, you now know everything works with the new VDP. So if it doesn’t with the old one in the socket it seems enough proof it is dead...

And on a side note, it seems you might have been able to use a TMS9918NL they are mostly compatible (might still be interesting to pop a 9918 in and see if it runs fine)

By barbeque

Supporter (8)

barbeque's picture

26-04-2020, 00:12

That's a good point - I just did that and it does appear that the old 9118 was dead. It's joined a bunch of Micron RAM in the "bad parts bin." I'll have to try it with a 9918 later!

I don't know what killed the chip, but it doesn't really matter as long as the computer is working. Frontrunner guesses are heat and ESD (through the composite jack?)