help understanding logic gate failure

[algorithm]

so I have an old drum machine. it uses a 7 segment display with a bunch of buttons on a matrix for editing parameters/#. uses a z80 cpu.

Im having a problem where the display wont turn on. leds light up like it boots, but unable to make changes. seems like the cpu is freezing maybe but Im guessing.

strangely if I clip a logic probe on the q101 + line at 4.7v (2n3904) it makes the display light up partially and seems to get closer to functioning.

Ive seen 2 different results on 2 CD4042BE side by side that I think are responsible. on pin 10 when its closest to functioning presumably pin 10 was high, but 1 CD4042 is low on pin 10 if the display is blank. I should add that at all times 1.9v is going to each segments display. Im not sure if it was more when both pin 10s were high.

I do get a clock input pulse on logic probe on both 4042.

So Im wondering how can I know if 1 of those logic gates is bad. I tried looking at the datasheet but It was all over my head basically.

Theres also an OS eprom. I doubt thats the problem but I could try to rewrite that.

Heres the service manual

https://archive.org/details/sequential-circuits-420-tom-service-id-10946


Any ideas on what to look at is appreciated. I am always guessing but have nobody to ask in reality. I have an oscilloscope, esr meter, logic probe, dmms, can test Hz, so hopefully whatever the culprit is I am equipped to check.

I tested the caps and they seemed to be fine.

tested q101 and it was good out of circuit too. part of me wondered if the 5v line wasnt enough to turn on the 2n3904? but I think of so many possibilities in uncertainty.

[wn1fju]

The CD4042 is just a quad latch.  That means there are four inputs and four output pairs (Q and Q').  The Q basically follows what is on the input line when the clock strobe hits.  So you can do a couple of quick checks with a pair of scope probes.  First, make sure that if an output Q is high, then its complement Q' is low (and vice versa).  Do this for all four pairs.  Next, put one probe on an input line, the other on its corresponding output Q, and move the drum machine controls around so that hopefully the input data to the chip changes.  See if the output follows.  Make sure the logic levels are correct for CMOS.  You don't want voltages that are too far away from 0 or Vcc.

If it were me, and I had any doubts about the chip, I would simply put a new one in.  They can be found (in the USA) from almost every well-known distributor for about 50 or 60 cents. 

[algorithm]

awesome. thank you so much.

I will report back with results.

that helps a lot knowing about the complementary difference. will have to retest but think I saw both pairs the same. definitely will replace if so. thanks again.

[Kim Christensen]

If you haven't done this already:
The very first thing I would check are the supply voltages on TP1 - TP5 relative to TP0. Are they within tolerance and without ripple?
Also, I'd make sure the RESET line (Pin 26 of U142) goes and stays high after power on and that U142 has a 4Mhz clock on pin 6.

[algorithm]

I will check that next. thank you greatly.

I know Im getting proper voltage but not sure about ripple. if there is ripple what would that point to? because I replaced all caps with nichicon properly specced and tested before install on capacitance and esr.

I checked the outputs on the cd4042be's and most all lows were blank (signal level between 1 and 0) no clock pulse. stuck hi.

not sure what that means but I guess I need to retrace some paths.

[Kim Christensen]

I know Im getting proper voltage but not sure about ripple. if there is ripple what would that point to? because I replaced all caps with nichicon properly specced and tested before install on capacitance and esr.

If you replaced the power supply filter capacitors (C162, C163, C158 C161) then one other thing that could cause high ripple would be if one of the diodes (D123 - D126) was open. I've even seen diodes that check OK on diode test, but drop several volts when any appreciable current is passed through them so check the ripple on the filter caps with a scope... The frequency should be 2x the line frequency (100-120Hz). If it's 50-60Hz, then you've got a bad diode. (Ignore this rule for C158 since that's fed from a halfwave rectifier)

I checked the outputs on the cd4042be's and most all lows were blank (signal level between 1 and 0) no clock pulse. stuck hi.

The no clock pulse (I assume you mean pin5 on U116 & 117) would mean that the Z80 CPU is not addressing the display at all. This signal comes via an IO address decoder IC (U137) and some other logic (U138 pin 3) and will not go active unless the Z80 CPU accesses this IO address. The CPU could be running and just not accessing this address because of some other fault elsewhere. Or the decoder logic (U137, U138, U139) could have a fault. There should be regular pulses on U137-11, U138-1, etc, if the CPU is trying to load data into the cd4042be's and multiplex the display.
Another easy check is to scope the CPU data-buss and address-buss. There should be lots of activity there. If not, then the CPU isn't running at all..

[algorithm]

thank you very much for the informative answer. I got a chance to test the cpu reset pin as well as halt, int, wait, rfsh, m1 and the weird symbol pin (6).

wait= lo pulse (logic "0" with pulse (low duty cycle))
int= hi
halt= hi
rfsh= lo pulse
m1= lo pulse
pin 6= lo pulse

on power reset pin was hi, sometimes would blink but wasnt sure if that was my shaky hand. i counted 15 or 20 seconds of reset pin being lit hi, then starting getting occasional pulses.

sometimes part of the display comes on like quarters but some chunks missing. usually wont light. when it did light display it lit every led except 1 on the long row. when before only 1 led was lit. like it inverted the operation? lol i dunno.

between the resistor array and q101 I get 0.4v. also on 1 leg on 1 side of resistor array I get 0.00 while every other leg on that side has 0.2.

I have run around this to my best ability but by the time I ask for help Im nearly burnt out so excuse my stupidity please.

sometimes q101 has less than 0.3v coming out to the resistor array. I dont think thats right but no idea whats causing it.

part of me is worried if theres some break on via or between layers. but yeah... I can test all those diodes and see whats going through. should be easy to find if they all arent bad lol.

thanks again.

kinda hard to find a picture of the pcb but it has foil traces which I like the look and era of but wow do I not like working on.

[Kim Christensen]

wait= lo pulse (logic "0" with pulse (low duty cycle))

This makes me wonder about your logic probe or maybe the 5V supply voltage/decoupling... The wait input (pin 24 of U142) should always be high because it is directly tied to the +5V supply rail! (+5VD)

int= hi
halt= hi
rfsh= lo pulse
m1= lo pulse

These seem Ok to me.

pin 6= lo pulse

Pin 6 is the CPU clock input. It should be a 4Mhz 50% duty cycle logic level squarewave. From what I remember of logic probes, both the H & L lights should be lit and the pulse light should flash.

on power reset pin was hi, sometimes would blink but wasnt sure if that was my shaky hand. i counted 15 or 20 seconds of reset pin being lit hi, then starting getting occasional pulses.

This pin should only go low for a few hundred milliseconds after power on and then stay high until power off.
But your logic probe readings on the +5VD line (via the CPU wait pin) make me wonder about the stability of the +5VD rail or the functionality of your logic probe. At this point, you really need to look at the +5VD rail with an oscilloscope. A voltmeter may show a low voltage, but it won't show if you have a decoupling problem (Try measuring the +5VD at one of the logic chips close to the CPU or directly on the CPU itself). You will see various decoupling caps scattered around the PCB (C167 on the CPU, C141 on the 8Mhz oscillator, C145 on U135, etc). These are usually ceramic, so they don't fail as often as electrolytics, but it's still a possibility. You can get weird glitches if a few of these have failed open.

Conclusion:
If you don't have an oscilloscope, check the +5VD rail with a voltmeter close to the CPU. If that's OK, then look at the CPU clock circuitry because something seems off with that. The crystal might be bad or U134/U135. If you have another source of a logic level squarewave, double check your logic probe on that. ( both the H & L lights should be light and the pulse light should flash)
Randomly replacing the 5V rail decoupling caps would be a pain, but without a scope I don't know how you would test for those types of problems.

[algorithm]

so I havent got a chance to do much. I should explain my working conditions are terrible. even looking at documentation is a struggle with load times even between pages. I have very little space to test or service. having to solder on a table with a hole in the middle without good fume extraction. its nightmarish.

so I desoldered d121 and d122.

when I did that I powered it on to see what voltage was there. the u161 center pin was outputting 8.8v seemingly to both diodes. both diodes dropped 0.3v of input. but when the other side in on d121 122? that 8.8v on the middle pin of u161 reads 0.6v. d122 reads 4.7v input and has 4.6v out or something. d121 reads 0.6v in and 0.00 out.

Im thinking 4.7v is the only off thing so wanted to see if I isolated part of the circuit if I could get 5v exactly like everywhere  else.

is that just the circuits design to suit voltage drop on lines or? I still dont get why what I see going in is totally different if the rest of the circuit isnt there.

I do have an oscilloscope but I am not great at using it.
but I will test what you suggested next on scope or logic probe. wish me luck lol and thanks again. your 2 cents means a lot.

[Kim Christensen]

even looking at documentation is a struggle with load times even between pages

You can right click a page and save as a JPG on your PC.
Another option is to download the entire manual as PDF and then never worry about load times again.

so I desoldered d121 and d122.
when I did that I powered it on to see what voltage was there.

Ooh. That's a bad idea. Especially removing D121. The problem is that U161's pin2 is it's ground reference pin. If that is left floating, that means it's output (pin 3) will rise above 5.7V relative to ground and possibly damage whatever it is powering. The reason D121 is in the circuit is to raise the output of U161 by 0.7V to compensate for the voltage drop across D122 & D127. I hope you also had D127 removed and just forgot to mention it. Otherwise everything powered by the +Vnv line (Looks like the RAM chips) could have been damaged by overvoltage.

[algorithm]

well thats part of my problem, no pc, just an obsolete phone and tablet that are painful to use and slow.

hmm, I dont think I did remove that diode. hopefully no damage but might be. well since I was seeing 8.8 on ground of u161 (7805) I replaced it. still isnt fixed but did change the fault so a bit closer hopefully. I will try redoing all the tests Ive done. will report back if I see any differences.