CPU address line voltage too low?

[pspring]

No picture or sound, sometimes steady tone or very garbled pic.

I shorted the power switch for now.

Power supply is 12.2v, 4.9v, -5.1v.

Sound IC pin 7(output) is normally quite so BIOS seems to be loading most of the time.

 

CPU seems to be partially running. Reset cap uF measured ok. Using an oscilloscope I found the CPU address line signals to be 3.3v p-p. Shouldn't they be 5v p-p like other CPU lines? CPU Vcc is 4.9v.

 

Some of the 4116 IC's where nearly too hot to touch so I just removed them all, will probably replace them with 5v, to check if one or more were loading down the Z80. It made no difference.

 

I'm an electronic/software technician but have no experience with CV. Unit is in fair shape and untouched, till now. Doing a lot of reading here. Wondering what makes sense to do next. Appreciate any help.

[pspring]

ACK! I misread the schematic! Video ram ain't connected to the Z80.

 

Maybe I'll hold off removing more chips till someone pipes in on 3.3v on the CPU address pins.

[ChildOfCv]

Output voltage on the Z80 is allowed to dip as low as 2.4V for a high. This is still TTL compatible, which defines voltages from 2.1+ as definitely high. The 74LS chips typically have high outputs around 3.2V. So to answer your first question, 3.3 is perfectly serviceable.

 

Do you get the "insert a cartridge before turning it on" screen when you turn it on without a cartridge?

[pspring]

Ok thanks! Good to hear 3.3v is ok. I was chasing a non-problem. Never done that before... lol

 

The insert cartridge message is not shown. The most video I've even gotten was some vertical white bars. Video is mostly a squiggly snowy mess. Both channel 3 and 4 were tried. I don't have Frogger to easily check the audio but have tried waiting 15secs then press 1 to start Donkey Kong but get no output on sound chip pin 7.

 

I've read that the 2 ram chips can cause black screen / dead unit so I'm trying to figure out how to test them. From reading here it seems like getting sound working is a good first step?

 

I'm also going to do the video line out mod to bypass the rf module for now, in case that ain't working.

 

The -5v does not go through the switch, that I have jumpered. Is it ok to just plug the power supply into the wall or does that cause issues?

[ChildOfCv]

If you're not sure about power usage, you should probably unplug the system when it doesn't need power. By the way, I assume you've already checked the capacitors for leakage? Rather than bridging the +5V across the switch, connect your meter across in amps mode, to see how many it's pulling.

 

If you have an Arduino or similar board that is 5V-tolerant, you can use it to check the SRAM chips from the expansion port. It's easiest if you have a 60-pin card edge connector.

 

1) Connect BUSRQ to ground.

2) Connect A15 to ground, and A14 and A13 to +5.

3) Connect MREQ to ground.

4) Connect A9-A0 to the Arduino, along with WR. If you have enough pins left, you can either connect all the D lines or just a subset to test.

5) Now write a program to loop through SRAM memory writing values then reading them back.

Set the address, and data, set WR low. Do a NOP instruction. Set WR high. Next.

 

Leave WR high, set the address, do a NOP instruction, read the data and compare with expected value. Next.

[+nanochess]

Clean with alcohol the crystal area and surrounding chips. Maybe you'll need to change the U22 74LS04.

 

If after cleaning you don't see improvements it's very probable the video chip is failing, the video is always rock steady in a good chip even if it shows trash.

[pspring]

Soldered all the chips back in.

Caps on main board measured a bit over 10uF(with cheap meter).

Measured current draw with no cartridge or controllers connected:

5v = 460mA

-5 = 9mA

12v = 225mA

Hooked up composite video and line level audio per these instructions. https://www.benheck.com/bens-colecovision-composite-video-mod/

Still no video or audio on TV. Scope also shows no activity at either output.

 

Don't have Arduino but can maybe dig up an STK600 or some other demo board. Or maybe I can figure out how to use my Coco 2 or TI99/4a!? Might be quicker to replace the ram chips.

 

We have an old DIP memory programmer at work. I check if it can check RAM.

[pspring]

Cleaned with distilled alcohol around crystal and supporting chips.

 

AUD_CLK and VID_CLK are signwave. I think this correct? CPU_CLK is square.

[ChildOfCv]

Ben did some unnecessary engineering there...

 

Composite video can be read off of pin 6 on the RF to PCB connector row. Sound can be read from pin 2, both with respect to ground.

 

If you scope the 9928's pins 35, 36, and 38 (B-Y, Y, R-Y), triggering on Y's low pulse, you should see waveforms matching the 9928 technical manual on page 5-13. See http://www.bitsavers.org/components/ti/TMS9900/TMS9918A_TMS9928A_TMS9929A_Video_Display_Processors_Data_Manual_Nov82.pdf

 

Note that the manual is wrong about B-Y back porch voltage sync. It DOES do that on the 9928, to create the color burst signal. Just not on R-Y.

[pspring]

Thanks! I'm slowly learning pieces. : )

 

No output on 35, 36, 38. Level is steady 2-3v.

Found the Coleco troubleshooting guide so I poked around with the scope comparing it to the pictures in the guide and found no NMI. Not sure how it works.

 

VDP CSW(couldn't get to trigger after reset release). Followed flow chart to where U9:17-24 are good(data) but couldn't fond where it is supposed to pick up on sheet 6...

[ChildOfCv]

Hmmm... someone who knows the TMS9928A better should chime in on this, but just after reset, according to the manual, it may not be outputting anything. Hmmm... on second thought, it probably DOES need to at least output sync signals, or the TV would just run wild. Well anyway, just after reset, it's at least blank until the CPU programs it. So unfortunately, I can't tell you whether this definitely is or isn't a problem.

 

The INT output on the 9928 is triggered during retrace, so if there's no retrace there will be no INT either. The data bus (17-24) is common to all of the CV--SRAM, CPU, sound, ROM, inputs, expansion, etc. The VRAM is connected to pins 25-32 and 1-11.

 

Definitely check that the clock input is 10.738MHz though. As you might have noticed, it uses an LC circuit to get close, but relies on the 3.58MHz clock to keep it honest.

Edited January 14, 2019 by ChildOfCv

[pspring]

Thanks for the help! I brought the CV board to work where I have much better equipment. Both clocks were spot on.

 

Maybe the 9928 is not getting initialized. I read there are registers needing setup at boot time, one of them controlling INT.

 

Hooked up the mixed signal scope(16 digital + 2 analog) and tried to trigger on U6 addr x01 or x02 pattern while watching the read and write lines controlling the VDP. Neither line went low. The only output line that wasn't always high was the CTRL_EN_1(pin 11). So I went home and swapped U5 and U6 thinking bad decoder. But, no joy... Unless maybe both are bad(I only had time to check U6).

[ChildOfCv]

Did you watch for activity on CS pin on the ROM? The first thing the Z80 does after reset is start running instructions at address 0000. CV placed the ROM at that location, for obvious reasons.

[pspring]

More scope poking revealed that my CV is "flaky". LOL

 

Pressing reset will eventually make different pieces partially operate like white bars on the TV or a steady tone. Things that didnt seem to be working(in previous posts) will work if I press the reset enough times. It's like it wants to start but hangs up part way while booting. I think I've seen all chip selects work. Seem like RAM? Kinda hard to debug without a data analyzer. : )

[ChildOfCv]

Interesting.

 

Well, since it begins its life by making function calls, it's important to have a working stack. And guess where the stack is? In RAM. So yeah I'd look for a way to test the RAM, or if you happen to have working chips already, try putting them in.