Nonfunctional Gigatron.

Information about getting support on Gigatron builds
DevilledEggs
Posts: 6
Joined: 23 Nov 2021, 23:24

Nonfunctional Gigatron.

Post by DevilledEggs »

I know this is a rare case when it comes to support but, I have exercised all of the options and this is the only remaining one; I emailed support/Walter a while ago, but no luck persists. I assembled my Gigatron during the pandemic and from the time I assembled it, it has not worked. I have looked through every trace and test point, desoldered anything that seems off and resoldered the components after they were proven functional but, It does not work. When I plug it in, I get a power light, that is stable but, the Blinkenlights will light up a random sequence of lights and just hold it. (E.g. The first and third lights will light up and that is it). When unplugged and plugged back in, it shows a different random sequence. I also feel the need to address the top of the board on the picture. For some reason the label on the ROM chip just smeared on my finger when I touched it. Also, in checking this chips, I color coded them for better understanding... That explains the dots on them. Im giving this information to make this debug as quick and painless for this forum as possible. If anyone could help, that is much appreciated. -DevilledEggs :) .
Attachments
Another random sequence. Again, it just picks a random sequence and keeps it on until you plug in the board. No video signal if you plug it in either.
Another random sequence. Again, it just picks a random sequence and keeps it on until you plug in the board. No video signal if you plug it in either.
nooooooo].JPG (54.87 KiB) Viewed 112 times
The board plugged in getting a power light and it just so happened that the random sequence of light landed on 0000.
The board plugged in getting a power light and it just so happened that the random sequence of light landed on 0000.
0-02-01-2a57ee08fd550f8245bff6ec960174c17cfd49bc2c3993d99e2f4cb46f644b0c_938b6f22eff0f656.jpg (236.01 KiB) Viewed 112 times
The board, off. It was awkwardly placed on my desk in this photo, touching a screwdriver bit. The board was checked after the photo was taken, it is ok.
The board, off. It was awkwardly placed on my desk in this photo, touching a screwdriver bit. The board was checked after the photo was taken, it is ok.
0-02-01-20af6e42b834eebd7d9d0643c84b7e29cca0a24f0b273c691b2ea0b3fdf833d2_6e32d432a2cc33a7.jpg (230.35 KiB) Viewed 112 times
Sugarplum
Posts: 43
Joined: 30 Sep 2020, 22:19

Re: Nonfunctional Gigatron.

Post by Sugarplum »

If you have a scope, one of the first things I'd check is the clock output. Another thing to check would be the power supply, maybe try another one with a higher mA rating.
DevilledEggs
Posts: 6
Joined: 23 Nov 2021, 23:24

Re: Nonfunctional Gigatron.

Post by DevilledEggs »

Sugarplum wrote: 24 Nov 2021, 03:27 If you have a scope, one of the first things I'd check is the clock output. Another thing to check would be the power supply, maybe try another one with a higher mA rating.
Ok, I will check. That sounds like a good idea! I have a scope, but I don't know if it works after my basement flooded, we'll see I guess. And no, the Gigatron was not in the water along with the oscilloscope, it was in my cabinet upstairs ;)
(Edit: The scope does not work, I just ordered a cheap one off of Amazon.)
at67
Posts: 431
Joined: 14 May 2018, 08:29

Re: Nonfunctional Gigatron.

Post by at67 »

It's painful but have you checked for shorts/open circuits on every chip using a multi-meter whilst the board is not powered?

Have you measured the voltage across a random selection of the TTL chips? (VCC should be 4.75v to 5.25v)

It's hard to tell, but it looks like you have used HCT as your TTL family, are ALL the TTL chips HCT or are any of them from a different family?

What speed is your ROM and RAM, (ROM looks to be 100ns on closer inspection https://pdf.dzsc.com/88889/23473.pdf), so what about RAM? It needs to be 70ns max.

Have you verified the direction of every diode? Especially in the instruction decoder.

Is R2 in your clock generator 1k or 10k, once again it's hard to tell. (It should be 1k, brown black red).
DevilledEggs
Posts: 6
Joined: 23 Nov 2021, 23:24

Re: Nonfunctional Gigatron.

Post by DevilledEggs »

at67 wrote: 24 Nov 2021, 12:51 It's painful but have you checked for shorts/open circuits on every chip using a multi-meter whilst the board is not powered?

Have you measured the voltage across a random selection of the TTL chips? (VCC should be 4.75v to 5.25v)

It's hard to tell, but it looks like you have used HCT as your TTL family, are ALL the TTL chips HCT or are any of them from a different family?

What speed is your ROM and RAM, (ROM looks to be 100ns on closer inspection https://pdf.dzsc.com/88889/23473.pdf), so what about RAM? It needs to be 70ns max.

Have you verified the direction of every diode? Especially in the instruction decoder.

Is R2 in your clock generator 1k or 10k, once again it's hard to tell. (It should be 1k, brown black red).
The ttl chips are whatever came in the kit I bought from Walter/Gigatron.io. Im pretty sure the diodes are correctly soldered, but I will have to check a bit later as I am in my lunch break writing this. As for the voltage in the TTL chips, it is confirmed good, and I checked all the traces this morning, they are fine. Thanks for the feedback, I will have to check it.
DevilledEggs
Posts: 6
Joined: 23 Nov 2021, 23:24

Re: Nonfunctional Gigatron.

Post by DevilledEggs »

DevilledEggs wrote: 24 Nov 2021, 15:48
at67 wrote: 24 Nov 2021, 12:51 It's painful but have you checked for shorts/open circuits on every chip using a multi-meter whilst the board is not powered?

Have you measured the voltage across a random selection of the TTL chips? (VCC should be 4.75v to 5.25v)

It's hard to tell, but it looks like you have used HCT as your TTL family, are ALL the TTL chips HCT or are any of them from a different family?

What speed is your ROM and RAM, (ROM looks to be 100ns on closer inspection https://pdf.dzsc.com/88889/23473.pdf), so what about RAM? It needs to be 70ns max.

Have you verified the direction of every diode? Especially in the instruction decoder.

Is R2 in your clock generator 1k or 10k, once again it's hard to tell. (It should be 1k, brown black red).
The ttl chips are whatever came in the kit I bought from Walter/Gigatron.io. Im pretty sure the diodes are correctly soldered, but I will have to check a bit later as I am in my lunch break writing this. As for the voltage in the TTL chips, it is confirmed good, and I checked all the traces this morning, they are fine. Thanks for the feedback, I will have to check it.
After a little bit of research and testing, all of the chips are HCT chips. The RAM chip is less than 70ns, (https://www.farnell.com/datasheets/1674430.pdf), and the ROM chip is good (https://static6.arrow.com/aropdfconvers ... 8148st.pdf). The R2 in teh clock generator is 1k, brown black red. And, the voltage between VCC (On the TTL chips) is about 5v. As there is not much documentation in the manual and such about the diodes, I have soldered them in the same way they are soldered in the other photos of board I can find.
DevilledEggs
Posts: 6
Joined: 23 Nov 2021, 23:24

Re: Nonfunctional Gigatron.

Post by DevilledEggs »

DevilledEggs wrote: 24 Nov 2021, 11:55
Sugarplum wrote: 24 Nov 2021, 03:27 If you have a scope, one of the first things I'd check is the clock output. Another thing to check would be the power supply, maybe try another one with a higher mA rating.
Ok, I will check. That sounds like a good idea! I have a scope, but I don't know if it works after my basement flooded, we'll see I guess. And no, the Gigatron was not in the water along with the oscilloscope, it was in my cabinet upstairs ;)
(Edit: The scope does not work, I just ordered a cheap one off of Amazon.)
I found out you can use an arduino and make a simple oscilloscope with processing IDE and I tried it with my signal generator and it seems to be pretty accurate, out of curiosity what should the waveform look like? Should it be a square wave, a sine wave? My common sense and good bit of electrical knowledge says it should be a square wave but I dont know.
(Another Update) The clock output is a perfect square wave, and it looks crisp. So that isn't the culprit.
Sugarplum
Posts: 43
Joined: 30 Sep 2020, 22:19

Re: Nonfunctional Gigatron.

Post by Sugarplum »

DevilledEggs wrote: 24 Nov 2021, 22:46 I found out you can use an arduino and make a simple oscilloscope with processing IDE and I tried it with my signal generator and it seems to be pretty accurate, out of curiosity what should the waveform look like? Should it be a square wave, a sine wave? My common sense and good bit of electrical knowledge says it should be a square wave but I dont know.
(Another Update) The clock output is a perfect square wave, and it looks crisp. So that isn't the culprit.
It would be a rough square wave (with ripple) and around 6.25 Mhz. I'd take that measurement from the outermost (logically, not physically) output of the inverter chip near the crystal or from the clock input of the program counter or a register. Maybe check at each chip that takes the clock signal.

Now that sounds a tad suspicious, that the signal is crisp. While that is a good sign, that could also mean that other things are not working. I mean, if it is perfect with no ripple, then maybe it is not reaching the program counter (or there is another reason it might not be advancing) or some other chip that needs it (like the Accumulator). The clock transitions of the devices that use the clock are expected to add ripple to things such as the clock (and power rails, hence all the tiny capacitors).

I guess you could test the lowest bit of the program counter (at the lowest-ranking chip). You should expect roughly a 3.125 Mhz signal there (except during jumps). And you could check the clock at that point too on the clock-in pin of the program counter chips. If that is as expected, then maybe check the clock signal at every register too, particularly the Accumulator, X, and Y. Since the lights are stuck, the Accumulator would likely be more relevant.

Checking to make sure the power supervisor is installed the correct way would make sense, that its flat edge is turned the same way as the silkscreen on the board indicates.

On diodes, the instructions seem clear to me. You bend the ends with the cathode band and put the other end in the holes with the circle drawn. The bent end from the cathode marking goes into the unmarked square pads.

"The diodes D3 and D4 lay flat.
The stripe on the diode should match the marking on the
board. Diodes D9~D36 are standing up. So for these, only
bend the pin on the side that has the black ring. That pin
goes into the square padded hole, the other in the round
padded hole. In short: the diodes are on the left with the
black ring on top, the wires are on the right."
DevilledEggs
Posts: 6
Joined: 23 Nov 2021, 23:24

Re: Nonfunctional Gigatron.

Post by DevilledEggs »

Sugarplum wrote: 25 Nov 2021, 12:27
DevilledEggs wrote: 24 Nov 2021, 22:46 I found out you can use an arduino and make a simple oscilloscope with processing IDE and I tried it with my signal generator and it seems to be pretty accurate, out of curiosity what should the waveform look like? Should it be a square wave, a sine wave? My common sense and good bit of electrical knowledge says it should be a square wave but I dont know.
(Another Update) The clock output is a perfect square wave, and it looks crisp. So that isn't the culprit.
It would be a rough square wave (with ripple) and around 6.25 Mhz. I'd take that measurement from the outermost (logically, not physically) output of the inverter chip near the crystal or from the clock input of the program counter or a register. Maybe check at each chip that takes the clock signal.

Now that sounds a tad suspicious, that the signal is crisp. While that is a good sign, that could also mean that other things are not working. I mean, if it is perfect with no ripple, then maybe it is not reaching the program counter (or there is another reason it might not be advancing) or some other chip that needs it (like the Accumulator). The clock transitions of the devices that use the clock are expected to add ripple to things such as the clock (and power rails, hence all the tiny capacitors).

I guess you could test the lowest bit of the program counter (at the lowest-ranking chip). You should expect roughly a 3.125 Mhz signal there (except during jumps). And you could check the clock at that point too on the clock-in pin of the program counter chips. If that is as expected, then maybe check the clock signal at every register too, particularly the Accumulator, X, and Y. Since the lights are stuck, the Accumulator would likely be more relevant.

Checking to make sure the power supervisor is installed the correct way would make sense, that its flat edge is turned the same way as the silkscreen on the board indicates.

On diodes, the instructions seem clear to me. You bend the ends with the cathode band and put the other end in the holes with the circle drawn. The bent end from the cathode marking goes into the unmarked square pads.

"The diodes D3 and D4 lay flat.
The stripe on the diode should match the marking on the
board. Diodes D9~D36 are standing up. So for these, only
bend the pin on the side that has the black ring. That pin
goes into the square padded hole, the other in the round
padded hole. In short: the diodes are on the left with the
black ring on top, the wires are on the right."
After testing everything on this list, I realized my electrical knowledge starts to degrade at around this point :). So, here are my findings and I really hope someone on here can decode them, as I cant. Sometimes, the signals coming in from the clock are proper and correct, other times, they make no sense and come through on the oscilloscope (Which I tested) as white noise basically. On the low side of the program counter, the clock signal is the same. Sometimes good, other times just white noise. The Accumulator's readout is good only very rarely but, I accidentally touched my function generator's output lead to one of the pins on the low end of the program counter and it sort of sprung to life, very quickly flashing the test pattern, but not very accurately as it seemed to skip some of the lights as it went by. as the signal speed set on the function generator was much higher, the lights flashed faster. This leads me to believe it might be something with the clock or the crystal. As for X and Y, the waveform is basically what is should be, except the sporadic breaks it likes to take for seemingly no reason. And yes, the instructions on the diodes turned out to be simple, and they are soldered correctly. As for the power supervisor, it too is in the correct orientation. I shout also mention that the lights will change sometimes, when the board is on, but very rarely, and in no particular pattern. I am not a professional when it comes to electronics as I am just a guy in his basement making stuff.
Attachments
The sporadic signal coming from the accumulators clock input.
The sporadic signal coming from the accumulators clock input.
0-02-01-84955e1395f40751a95b917869dd2ad6db3c1bbf5d371f0807bc31889f5f42f7_7e20a2efb06aaa86.jpg (149.18 KiB) Viewed 47 times
The signal coming from the clock.
The signal coming from the clock.
0-02-01-9ed56b2b750a6084249586e5d23b1f1abc1975a0a7fd5c06ecc728b1e5f21b88_94d76985f9fea0d9.jpg (155.96 KiB) Viewed 47 times
One of the better signals from the program counter chip's low end clock input.
One of the better signals from the program counter chip's low end clock input.
0-02-01-56cb096324091e7f2f3f6ee8eb474e197dd1c551d76c0ab682113fc52e47964c_1e002184cba656b7.jpg (153.09 KiB) Viewed 47 times
Sugarplum
Posts: 43
Joined: 30 Sep 2020, 22:19

Re: Nonfunctional Gigatron.

Post by Sugarplum »

Hmm. The first couple of tests look close to how I'd expect.

One that would be interesting to know would be to look at the low bit of the lowest nybble of the program counter output (pin 14 of the '161 I think), not its clock, but the data. A counter chip, if allowed, will increment. The fastest changing pin of the outputs would be bit 0. That would make 1 transition per clock cycle. So if you check that one, it should be half the speed of the clock (except during jumps since the code could take it to nearly any place).
Post Reply