TI99/4A starts to distort sound and picture after 10 minutes

[jschaap26]

11 hours ago, HOME AUTOMATION said:

Dragonstomper! I didn't notice, I'd been promoted.

Ooops. I'm new to the forum and it shows. Apologies.

[jschaap26]

40 minutes ago, wierd_w said:

Speaking of which, it looks like the default choice in the late 70s was the white junk. Blech.  Put a nice chipset heatsink on there with some silver compound. Will probably get the overheating issue sorted.  Another thing to check would be reseating the IC. It could be working harder than it should be, due to poor signal fidelity, and generating heat.

Do you have suggestion(s) on where I could find a good heatsink for the 99? And I am guessing that this would mean somehow removing the existing "heatsync" (rectangular metal slab) attached to the RF shield?

[jschaap26]

33 minutes ago, --- Ω --- said:

Yes, but the best way to eliminate the overheating problem (IMHO) is...

(ducking & running)

I've heard about these. Are they still for sale today?

[Omega-TI]

1 hour ago, jschaap26 said:

I've heard about these. Are they still for sale today?

 

Not to my knowledge, but it's successor will be coming out in the future and Matt is always watching, so who knows what the final product will have under it's hood.

[Tursi]

3 hours ago, jschaap26 said:

Do you have suggestion(s) on where I could find a good heatsink for the 99? And I am guessing that this would mean somehow removing the existing "heatsync" (rectangular metal slab) attached to the RF shield?

Why not keep using the one on the RF shield?

 

The QI machines used a clip-on heatsink which to me seemed like crap, but was clearly adequate.  Something like these: 

https://vetco.net/products/40-pin-dip-clip-on-heat-sink-nte448g?gclid=EAIaIQobChMI9vDpms2u5gIV0xZ9Ch2BjQh8EAQYAiABEgIGrfD_BwE

(disclaimer: never used this company)

 

 

[JB]

1 hour ago, Tursi said:

The QI machines used a clip-on heatsink which to me seemed like crap, but was clearly adequate.

Yeah, the 9918 doesn't need a LOT of heatsink. It does need SOME, though. Otherwise it slowly cooks itself to death.

[wierd_w]

10 hours ago, jschaap26 said:

Do you have suggestion(s) on where I could find a good heatsink for the 99? And I am guessing that this would mean somehow removing the existing "heatsync" (rectangular metal slab) attached to the RF shield?

 

I would need proper dimensions for the VDP, but this is where I would be headed...

 

https://www.jameco.com/z/205-00036-001-Heatsink-Dip-14-16-Pin-ICs-1-18-X0-64-x0-65-Black_2150934.html? CID=GOOG&gclid=CjwKCAiAxMLvBRBNEiwAKhr-nC5fJPDyRn0hRso8aXI8Ia6bmYHYdPCbgem4NfKxawia82xDcAx_XBoCt1AQAvD_BwE

 

or

 

https://www.mouser.com/Thermal-Management/Heat-Sinks/_/N-5gg0?P=1yztn6z

 

Again, need to know the size, and I don't have that much time to doddle with it at the moment-- but quality sinks for DIPs are out there.

Edited December 12, 2019 by wierd_w

[+Lee Stewart]

Considering that the TI-supplied heatsink is comparatively thin and touches both the chip and the metal shield, I should think a hole would need to be cut in the shield to accommodate the larger heatsinks.

 

...lee

[wierd_w]

If it can be found in an appropriate package, thermal pipe sounds ideal then.

 

 

Say, one of these radiators, with one of these individual heatpipes.

 

You could then route the heat away from the VDP and still stay in the allowance. One of those mickey-mouse clip on sinks would be sufficient to hold the pipe down on the vdp.

Edited December 12, 2019 by wierd_w

[ChildOfCv]

You'd get a heat sink for a PDIP40 package, of course.  Console5 sells one, but I defer to the experts about how good it is.

 

https://console5.com/store/dip40-heatsink-glue-on-thermal-epoxy-on-style-dip-40.html

 

Mouser's entry doesn't have a picture, but here you go:  https://www.mouser.com/ProductDetail/Aavid/508700B00000G?qs=%2Fha2pyFaduj7EOP6GqZbpgmvVdNiANdareWH1lroxyMDOrpAKb8qVQ%3D%3D

[JB]

That's the beauty of this use. It doesn't need to be a very good heatsink. The chip only needs a LITTLE extra help cooling. It isn't exactly dissipating two hundred watts of power here. If it were a larger package, it wouldn't need a heatsink at all.

[wierd_w]

I am not convinced of the practice.  More than once I have seen the sheilding used in this capacity, and seen devices overheat due to "pizza box" designs.  While the TI has lots of empty space inside, and reasonable ventilation in the housing, if the RF shield becomes uniformly hot, it will radiate equally both inside the enclosure, and outside.  If the rate of dissipation is outpaced by the rate of accumulation inside, it becomes an oven.

 

Given the variability of period electronics, this gives me pause.

[JB]

I can't imagine a situation where the VDP is dumping more heat into the RF shield than it can radiate away. Really, the RF shield is overkill for cooling just the VDP. (see: tiny clip-on sinks for the late systems)

 

That most 4a consoles haven't cooked themselves to death over the decades is in and of itself proof of the viability of RF shield heatsinking.

Devices that didn't properly heatsink the 9918 were unreliable, and failed rapidly once put into service(I know of at least one such product, an X10 controller called "Homeminder")

[jschaap26]

8 hours ago, Lee Stewart said:

Considering that the TI-supplied heatsink is comparatively thin and touches both the chip and the metal shield, I should think a hole would need to be cut in the shield to accommodate the larger heatsinks.

 

...lee

Yes, that was my thought too.

[wierd_w]

The statement was in general. not in the specific.

 

In the specific, I have seen several tablets try this trick.  They cook pretty good.

 

Other things that might cause issue are bad capacitors overheating inside the sheilded area, failing voltage regulators getting hotter than they should, etc...  Overheating like that is a sign of serious bad juju, but that isn't the same thing as "not happening".   Most of those older devices have large enough structures in their silicon/run slowly enough that they don't really get that hot.  However, multiple unhappy devices in an enclosed area together could conceivably cause net overheating.  I've seen it in other devices. 

[jschaap26]

Last night I was able to get a digital multimeter test the power supply. After spending a lot of time on the Internet trying to find the pinout and voltages and coming across a lot of conflicting information I saw that the voltages were written on the wall wart (see attached pictures). Pins 1 & 2 were 19.77 VAC and 2 & 4 were 8.93 VAC. I am not sure if this is pretty much what I should expect or significantly out of spec. Any wisdom on that point would be much appreciated.

 

Also, I did run the machine (with a fan on the VDP) for a good while and the 4624N did not get hot. It stayed room temperature the whole time (@HOME AUTOMATION).

[HOME AUTOMATION]

Sounds good! Probably is just a somewhat pesky 9918.

 

I once bought a 4a w/o the wall wart. I found a very similar one. However, I exchanged the high and low voltages... all seemed fine at first. I left for while. When I returned... smoke, negative function, not happy! 

 

The five volt regulator failed(closed), allowing about 18vdc onto the five volt rail. That sound chip burned it's package away, destroyed the socket and burned the board badly enough that, after cleaning up there was a hole burnt through!

 

The 4a still worked... sans sound.

 

I think the sound I.C. is the only hybrid circuit on the 5 volts. It contains an analog amplifier.

[ChildOfCv]

6 hours ago, JB said:

I can't imagine a situation where the VDP is dumping more heat into the RF shield than it can radiate away. Really, the RF shield is overkill for cooling just the VDP. (see: tiny clip-on sinks for the late systems)

Yeah, the RF ought to have plenty of heat dissipation capability, certainly more than air does.

 

The uncertainty though, is that it needs to be secured to the 9918 chip in order to work as advertised.  I can easily see situations where the shield may be on the computer, but not pressing against the 9918 and therefore not providing the heat wicking that it needs to do.  Unless the shielding has some kind of reliable spring force on the chip specifically, I'd be wary of it.

[Tursi]

16 hours ago, wierd_w said:

You could then route the heat away from the VDP and still stay in the allowance. One of those mickey-mouse clip on sinks would be sufficient to hold the pipe down on the vdp.

Now you have me envisioning a liquid cooled TI... we could run it by the clock chip and the power supply at the same time.

[Tursi]

11 hours ago, wierd_w said:

I am not convinced of the practice.  More than once I have seen the sheilding used in this capacity, and seen devices overheat due to "pizza box" designs.  While the TI has lots of empty space inside, and reasonable ventilation in the housing, if the RF shield becomes uniformly hot, it will radiate equally both inside the enclosure, and outside.  If the rate of dissipation is outpaced by the rate of accumulation inside, it becomes an oven.

 

Given the variability of period electronics, this gives me pause.

Many of us have run 4A consoles 24/7 for extended periods, and the RF shield doesn't get warm from the VDP (and sometimes clock) using it. Only the power supply under the coffee warmer ever really gets notably warm on a normal system. Those I did end up using a case fan on my BBS system - I just mounted it over the vents above the cartridge port to suck air out. Even the coffee warmer stayed cool with just that.

 

 

[Tursi]

3 hours ago, ChildOfCv said:

Yeah, the RF ought to have plenty of heat dissipation capability, certainly more than air does.

 

The uncertainty though, is that it needs to be secured to the 9918 chip in order to work as advertised.  I can easily see situations where the shield may be on the computer, but not pressing against the 9918 and therefore not providing the heat wicking that it needs to do.  Unless the shielding has some kind of reliable spring force on the chip specifically, I'd be wary of it.

Hah, had this too! A friend loaned me a machine for my BBS once to run while I used my machine for development work (it had the 16-bit RAM mod). His machine cooked itself in 24 hours of runtime. When we popped it over, one of the wires was running between the VDP and the heatsink - it had never been properly connected as far as we could tell (because the heat sink grease was otherwise unmolested ). I believe he replaced the VDP to prove that was it, but I wasn't there for the repair part.

 

[AwkwardPotato]

Try applying new thermal paste on the stock RF shield/heatsink and see if the problem goes away. If the system still acts up afterwards, there's a good chance that the 9918A is failing (chips aren't always completely non-functional when they're damaged).

 

If a VDP can't stay stable even with refreshed thermal paste on the stock heatsink, it's a better idea to replace it than cut up the shield/install heat pipes/custom heatsink, since the problem will likely only continue to get worse.

[jschaap26]

On 12/15/2019 at 4:04 PM, AwkwardPotato said:

Try applying new thermal paste on the stock RF shield/heatsink and see if the problem goes away. If the system still acts up afterwards, there's a good chance that the 9918A is failing (chips aren't always completely non-functional when they're damaged).

 

If a VDP can't stay stable even with refreshed thermal paste on the stock heatsink, it's a better idea to replace it than cut up the shield/install heat pipes/custom heatsink, since the problem will likely only continue to get worse.

Thanks for your thoughts. Yeah, I have applied modern thermal paste and reattached the shield/heatsync but, unfortunately, after 15 minutes it still starts to act up unless I keep a running fan over the vent holes. That being said, where is the best place to find 9918A replacements?

 

 

Edited December 18, 2019 by jschaap26

[wierd_w]

Dunno about best..

 

Amazon has a listing.

https://www.amazon.com/TMS9918ANL-TMS9918A-TMS9918NL-TMS9918-DIP-40/dp/B07X7LHC7D/ref=sr_1_6?keywords=tms9918a&qid=1576653914&s=industrial&sr=1-6

 

Aliexpress has them even cheaper.

https://www.aliexpress.com/item/32993234455.html?src=google&src=google&albch=shopping&acnt=494-037-6276&isdl=y&slnk=&plac=&mtctp=&albbt=Google_7_shopping&aff_platform=google&aff_short_key=UneMJZVf&&albagn=888888&albcp=1582410664&albag=59754279756&trgt=539263010115&crea=en32993234455&netw=u&device=c&gclid=EAIaIQobChMItMi68NW-5gIVhcDACh0KkwmnEAkYASABEgJCXvD_BwE&gclsrc=aw.ds