13 replies to this topic

[1FF8]

I phrased the subject that way to make it clear that the following is *NOT* a complaint about the Harmony.

I rescued an old 7800 a couple of weeks ago, and apparently people have been keeping the old machines alive long after I mothballed my old 5200 (which must have been more than 20 years ago). Is there an initiation ceremony or oath or something?

Anyway, I sprayed no-residue contact cleaner in the port and ordered a Harmony for it, which came a couple of days ago. It works every time with the one cartridge I have that fits it (2600 Berzerk), but the harmony can't always come up. It probably got to the menu 60% of the time. I looked closely into the cartridge port and it was filthy enough that I feel bad to have ever subjected a cartridge to it without looking, and with the case off you can see the machine has generally not had much love for a long time. My guess is the Harmony is working perfectly and poor contacts are why it doesn't always come up--it must be more sensitive to poor connections than the Berzerk cartridge. Is that true? The symptoms vary--sometimes I get rolling stripes on the screen, sometimes just a blank dark screen (and some sound, IIRC). Are those plausible failure modes for a Harmony with bad contacts? If worst comes to worst, is the connector replaceable like the old Nintendo one?

I steel wool'd off the RF cable corrosion, which fixed the "short" in the video cable I cobbled together (it didn't come with one, but fortunately had a power supply with the wierd plug), and used an NES cleaning kit to get some ugly black gunk out of the cartridge connector. I think the Harmony is coming up a larger percentage of the time (80%, maybe, unless it's wishful thinking), but still sometimes needs a second or third try. I believe I can get more gunk out of the cartridge port with another cleaning pass or two, and that may fix the remaining issues, but I'd be happy for advice if anyone has any. I'm currently using a couple of strips of and old credit card cut to fit the openings, which works OK, but I was tempted to remove the RF shield and see if I could figure out how to reversibly remove the plastic protector so I could get at the contacts better.

I obviously have nothing to test it in 7800 mode with yet.

F8

[batari]

When did you buy your Harmony? If you have one of the first 50 (preproduction) units from 2009, these sometimes don't boot right (85-90%.) The only solution is to just try again, and it will eventually work.

If you have anything bought Dec. 2009 or later, these had a slight hardware revision to fix the startup issue above.

it could be the contacts, or could be the console. A pencil eraser works very well at cleaning the contacts on Harmony.

If the console - Harmony has its own CPU, which is far faster than anything in the 7800. It should (and is designed to) boot up and start running its code well before a normally-working 7800's chips are up and running. But in certain extreme cases, such as where capacitors in the 7800 are dried up, it could "beat" Harmony in the race to start up. So I'd either just keep trying until it starts, or try another console if you want it to work every time.

[1FF8]

batari, on Thu Jan 19, 2012 7:24 PM, said:

When did you buy your Harmony?

It arrived two days ago--your turnaround is pretty fast. :-) I didn't even know it existed until a few weeks ago.

Quote

If the console - Harmony has its own CPU, which is far faster than anything in the 7800.

Sure--it's ARM based, right?

Quote

It should (and is designed to) boot up and start running its code well before a normally-working 7800's chips are up and running. But in certain extreme cases, such as where capacitors in the 7800 are dried up, it could "beat" Harmony in the race to start up. So I'd either just keep trying until it starts, or try another console if you want it to work every time.

Hmm. Judging from the stuff inside the case, the machine hasn't had much love for a long time, so maybe the electrolytics (that's where I guess the issue lies, from some vague memory about old hardware) haven't been fired up for a long time. I'm surprised to hear they can go out of spec enough to boot faster than the Harmony but still function correctly.

Hmm...do old electrolytics get restored to some extent with use? In any event, after cleaning the contacts the the failure rate is tolerable. Once it's up, the Harmony is marvelous.

Hmm. Sometimes it doesn't read paddle button input, come to think of it, though I never noticed a problem with games. Detection bug, or console problem?

That said, I guess I don't *know* it functions correctly, though it works fine with Berzerk and everything I've thrown on the Harmony except one rom. That one fails in Stella too, so it's not the 7800's fault.

F8

[1FF8]

As an amusing addendum, being the naive sort that I am I did a quick scan of the board schematic. I counted 57 capacitors, and I'm sure that is an undercount as I wasn't careful. Granted that some are probably ceramic or mica or something else durable, recapping wouldn't be fun unless there is a known small number of electrolytic and/or paper capacitors that could cause the boot problem. Known to someone accessible, that is--I certainly don't know enough to identify them.

Go ahead and laugh at me for even asking myself whether it was feasible. :-)

[batari]

It would just be two caps, I'd think, that could cause issues - the one on the reset circuit, and the big power cap.

To illustrate the issue that I think it might be, the 7800's reset circuit is a resistor and electrolytic cap, and a properly functioning cap should give at least 20 ms after powerup before the 6507 starts going. The Harmony's reset circuit should get its CPU going in under a millisecond but the CPU itself needs to run bootloader code before it's ready, but this should be done in less than 5 ms.

So essentially, a dried up cap on the 7800's reset circuit could cause the 7800 to win the race but perhaps still start normally. Perhaps a dried up power cap could cause issues as well. It should be easy enough to determine from the schematics which cap(s) these are.

[1FF8]

batari, on Fri Jan 20, 2012 1:30 PM, said:

To illustrate the issue that I think it might be, the 7800's reset circuit is a resistor and electrolytic cap, and a properly functioning cap should give at least 20 ms after powerup before the 6507 starts going. The Harmony's reset circuit should get its CPU going in under a millisecond but the CPU itself needs to run bootloader code before it's ready, but this should be done in less than 5 ms.

So essentially, a dried up cap on the 7800's reset circuit could cause the 7800 to win the race but perhaps still start normally.

That makes some sense--thanks for the explanation. I suppose that the fundamental issue is that until the ARM boots it's firmware the cartridge doesn't look like ROM to the 7800, so it does something like it would do if it were started without a cartridge, or with a cartridge with a bad chip. Also, I suppose the exact symptom might depend on just how much of the firmware the ARM managed to run, which would explain why the symptoms vary from one run to the next.

What it sounds like you're saying is that the reset capacitor has to charge before the 6502 (wasn't it a 6502C on this one, not the 2600's 6507?--inquiring minds want to know) starts, and so if the cap is dead the remaining capacitance might be so low as to give too little delay for the ARM to get a head start. I can sort of picture how that would work in my head--the reset switch would discharge the capacitor, and presumably the cap is there in the first place to provide enough time for the chips to completely power down rather than the momentary switch amounting to a power glitch (or glitches, if the switch bounces a lot). If that's so, then it would probably be better for the machine if it were replaced even aside from the Harmony issues.

If it's a reasonable sized component, then perhaps not too hard to replace. I'd thought of doing an A/V mod, if I'm still willing to tackle that after some research then this would be much simpler. I'm assuming everything from back then has through-holes, not surface-mount. I've never messed with surface mount so it's probably beyond my soldering skills without risking the whole board.

I might give it a try, on the theory that if I can't do it I might as well forget trying to get decent video off the board. Besides, part of the agenda for messing with these retro machines is to provide a fun way for my boy to learn some electronics.

Quote

Perhaps a dried up power cap could cause issues as well. It should be easy enough to determine from the schematics which cap(s) these are.

Yeah, that might not be too bad. I initially thought it might be one power capacitor, since those die on a lot of stuff, but I argued myself into thinking that probably wasn't it. The Harmony gets it's power from the board, so my argument was that if it were a power capacitor it would affect the 7800 and the Harmony equally--they start at the same time, whenever the capacitor is charged and the power comes up. I'd be happy to be educated on why that's not true, though, since I'm just guessing. I suppose if the ARM and the 6502 start at radically different voltages as the power comes up, then the voltage edge profile could make a difference, but my blind guess was that would be too small an effect to matter. But then, I don't know how many ms the cap could take to charge--I guess that would be easy to estimate from the voltage and capacitance, though. I'd just have to look up the values and scribble on a paper for a second, since I'm very rusty.

Hmm. I have a vague mental picture that those two problems could interact--perhaps a bad power cap could create a ramp-up edge profile that would magnify problems with the reset capacitor. I'm not sure if that's right, but it feels like it could be.

Edit: looks like the reset capacitor is C41, 0.001 uF. I wonder if Atari was so nice as to silkscreen the component numbers on the board? I'll have to remove the RF shield to see.

F8

Edited by 1FF8, Fri Jan 20, 2012 2:11 PM.

[batari]

OK, so I am looking at the schematics and the details of the reset line are about as clear as mud - I must have been thinking of the 2600s, most of which have a simple RC and a toggle switch.

The 7800 has a couple of levels of logic attached to the reset line and a pushbutton power switch running through a pair of SR latches. I've not analyzed the circuit fully but there are quite a few caps here, but most are unpolarized and small, so I assume ceramic. The only exception is C30 - which doesn't state the size, but merely that it's polarized and thus possibly electrolytic.

As for a dried up power cap, my thought was that Harmony's CPU core is only 1.8v and could start to run well before power is up on the 2600, but a dried up cap could change this delay. Might not matter, but a power cap is wise to at least check for other reasons.

[GroovyBee]

Don't forget that you have a bit of extra time when the processor comes out of reset on the 7800 due to its BIOS getting up and running before it starts probing the cart slot.

[1FF8]

batari, on Fri Jan 20, 2012 2:45 PM, said:

OK, so I am looking at the schematics and the details of the reset line are about as clear as mud - I must have been thinking of the 2600s, most of which have a simple RC and a toggle switch.

The 7800 has a couple of levels of logic attached to the reset line and a pushbutton power switch running through a pair of SR latches.

Hmm, we might be looking at different areas of the schematic. I was looking at the A-B 4-5 area in the lower left part of the scan here:

http://www.atariage....TSC_LL_Low.html

which looked like a simple 0.001 uF debounce cap (C41) and a momentary reset switch attached to the PB0 line of the 6532. But it occurs to me that you probably weren't talking about the reset switch, but rather the reset signal line connected to pin 34 of the 6532, pin 11 of U12, pin 40 of the 6502, and pin 1 of U11. I don't know enough to even guess what that's about.

Keep in mind that I don't actually know what I'm talking about here, I'm just trying to fake it. :-)

Quote

I've not analyzed the circuit fully but there are quite a few caps here, but most are unpolarized and small, so I assume ceramic. The only exception is C30 - which doesn't state the size, but merely that it's polarized and thus possibly electrolytic.

Hmm...I guess S1 is the power switch? Pause, select, reset, and the difficulty switches are elsewhere, so I don't know what else it could be, though I don't understand why it is where it is. So C30 is debouncing the switch and maybe doing some filtering? And I guess C62 is the power supply cap.

OK, I removed the RF shield. Atari didn't see fit to silkscreen very many capacitor labels for us. C62 is pretty easy to identify, since it's the biggest cap on the board (outside the shield) and clearly labeled 2200 uF. C30 is probably the 220 uF electrolytic nearby under the shield. I'd think 220 uF is way too big for just debouncing, so I suppose it is conditioning the 5V supply (looks like it might be wired across the 5V regulator). Anyway, those are the only two electrolytics I see on the board. I don't know enough to recognize if any others are failure-prone types though.

Quote

As for a dried up power cap, my thought was that Harmony's CPU core is only 1.8v and could start to run well before power is up on the 2600, but a dried up cap could change this delay. Might not matter, but a power cap is wise to at least check for other reasons.

If the Harmony can run on that low a voltage, then I guess maybe I follow. If the cap is bad the power ramps up more quickly, so there is less time between the cartridge getting 1.8V and booting up and the 6502 getting whatever it wants (looks like 5V from the schematic) and booting.

Replacing the electrolytics seems logical and easy, though I'd probably worry needlessly about doing it since I haven't had to solder on other people's boards much. OTOH they are isolated by themselves, so I probably wouldn't hurt much replacing them unless I managed to delaminate the pads by sheer incompetence. *Theoretically*, it would be easy....

Or, since what little electronics I knew is so rusty, all that could be sheer nonsense.

F8

Edited by 1FF8, Sat Jan 21, 2012 1:27 AM.

[Mitch]

batari, on Fri Jan 20, 2012 2:45 PM, said:

OK, so I am looking at the schematics and the details of the reset line are about as clear as mud - I must have been thinking of the 2600s, most of which have a simple RC and a toggle switch.

The 7800 has a couple of levels of logic attached to the reset line and a pushbutton power switch running through a pair of SR latches. I've not analyzed the circuit fully but there are quite a few caps here, but most are unpolarized and small, so I assume ceramic. The only exception is C30 - which doesn't state the size, but merely that it's polarized and thus possibly electrolytic.

As for a dried up power cap, my thought was that Harmony's CPU core is only 1.8v and could start to run well before power is up on the 2600, but a dried up cap could change this delay. Might not matter, but a power cap is wise to at least check for other reasons.

There are only three electrolytic caps on the 7800 motherboard; C22, C61 and C62. I don't know why the schematic lists C30 as polarized, I've never seen anything other than a plain old ceramic cap in that location.

Mitch

[1FF8]

Mitch, on Sat Jan 21, 2012 1:27 AM, said:

There are only three electrolytic caps on the 7800 motherboard; C22, C61 and C62. I don't know why the schematic lists C30 as polarized, I've never seen anything other than a plain old ceramic cap in that location.

OK, I'm apparently an idiot, or just too tired to see straight. Yeah, tired--that's my story, and I'm sticking too it. I didn't see either C22 or C61 on my cursory scan of the schematic, nor did I notice the *clearly labeled* C22 sitting next to...is that the Maria chip? Yah. Time for bed, I think. I'm supposed to get up for a Linux conference tomorrow anyway. I'm going to leave the tabs on the RF shield straightened for now so I can easily remove it as often as necessary to follow along with the home game.

F8

[1FF8]

I have to get on the road back to the conference, but it occurred to me that there are probably longer-lived substitutes for electrolytics, and for a one-off thing I don't have to save every buck. What are the alternatives? It looks like maybe polymer capacitors?

F8

[1FF8]

After doing some reading, I'm not sure I can just drop in polymer caps in place of the electrolytics--apparently they're likely to have a significantly lower resistance, and I don't know enough to decide what I can get away with. A resistor in series would probably work, but would end up looking pretty bad on a board that isn't meant for it. So instead, I've found some electrolytics rated for 20,000 hours:

http://www.mouser.co...647-UPX1C222MHD

That's about ten times the rating of cheap stuff, which is what I suspect Atari used originally (this *is* Atari Corp, not Atari Inc), so I expect I can't ask for anything more. Anyone see any reason not to go with those? I repeat that I'm just guessing here.

There is a small supply problem, since they have the 2200 uF power cap in stock but not the 220's, but I might be able to find another source.

F8

[Harbinger]

batari, on Thu Jan 19, 2012 7:24 PM, said:

When did you buy your Harmony? If you have one of the first 50 (preproduction) units from 2009, these sometimes don't boot right (85-90%.) The only solution is to just try again, and it will eventually work.

If you have anything bought Dec. 2009 or later, these had a slight hardware revision to fix the startup issue above.

it could be the contacts, or could be the console. A pencil eraser works very well at cleaning the contacts on Harmony.

If the console - Harmony has its own CPU, which is far faster than anything in the 7800. It should (and is designed to) boot up and start running its code well before a normally-working 7800's chips are up and running. But in certain extreme cases, such as where capacitors in the 7800 are dried up, it could "beat" Harmony in the race to start up. So I'd either just keep trying until it starts, or try another console if you want it to work every time.

I'm not sure exactly when I purchased mine--is there a way to tell if it was made after 12/2009 to know if it has that bootup fix integrated? Mine occasionally is fidgety on bootup and I'd love to get it fixed (if possible) or know what the fix is if it's something I could do myself. I'm handy that way.