Who else is building a Beeb from scratch?

[maniacminer]

Well, I've finally put together one board out of three and I've got serious stability problems and the Econet just returns "Line jammed" I'm wondering if my stash of "new" 6502A ICs is fake. I bought them off eBay a long time ago and tested them on the 3-Chip Plus - but it dawned on me that they are probably 1MHz parts and that's why they are reported by the chip tester and 3-Chip Plus as good, but crap out after a while on the Beeb making it like an electronic box o' frogs...

It looks reasonable, I haven't tried solvent on it apart from the freezer spray - the IC runs hot.

[!FOZ!]

Well, I've finally put together one board out of three and I've got serious stability problems and the Econet just returns "Line jammed" I'm wondering if my stash of "new" 6502A ICs is fake. I bought them off eBay a long time ago and tested them on the 3-Chip Plus - but it dawned on me that they are probably 1MHz parts and that's why they are reported by the chip tester and 3-Chip Plus as good, but crap out after a while on the Beeb making it like an electronic box o' frogs...PXL_20240503_211213274.jpg

I am not sure if your stability problems are the same, but early in my build I was having random slow downs and some crashes. It turned out for me that C25 was missing. It is also missing from BOM. I added that and it fixed my issue and the analogue port issue. Here is the specific post from this thread viewtopic.php?p=403172#p403172. Good luck!

[!FOZ!]

Well, I've finally put together one board out of three and I've got serious stability problems and the Econet just returns "Line jammed" I'm wondering if my stash of "new" 6502A ICs is fake. I bought them off eBay a long time ago and tested them on the 3-Chip Plus - but it dawned on me that they are probably 1MHz parts and that's why they are reported by the chip tester and 3-Chip Plus as good, but crap out after a while on the Beeb making it like an electronic box o' frogs...PXL_20240503_211213274.jpg

I am also using UMC 6502, UM6502A. However my chip looks a little different as I have the UMC logo printed.

[maniacminer]

Yeah, C25 missing will make the ADC continuously poll the end of conversion CB1 on IC3 (system 6522 VIA) will be continuously interrupting the CPU as a result. This will cause slow downs, but shouldn't cause crashing, that's likely to come from another source, most likely the 1MHz<->2MHz switching, catching the edge of a flip-flop and sending the CPU off into the bracken. I guess the original BOM missed out C25 as it's very expensive and would be one of the parts transferred from an existing board.

I've got one working board so far, fully tested. The second board is giving me problems. A blank screen with a cursor, I can type commands and change mode and set the background colour, but I can't see any text. I'm assuming it's something to do with the 81LS95's or it's activation, I've got to find some time to check all my soldering...

I found one problem so far

I get wrap around like this eventually

My stash of 81LS95 board pulls in sockets

[maniacminer]

LOL I need to not drink coffee when putting ICs into sockets... can you spot my further two errors, one prevented MODE7 from working, the other prevented other modes from displaying anything but the cursor, until I scrolled...

I wonder why I wasn't getting any MODE7

I wonder why I was only seeing the top 1kB of MODE0, surprised there was no magic smoke released...

All the boards work, I gave them a good run in Cybertron Mission and Starship Command for an hour or so and all is well.

Three working boards - I need serial bits to fully test though

[maniacminer]

The RGB2HDMI won't give me a stable MODE7 on these new PCBs.

I noticed an issue in the original BOM for the board where C48 is specified as 220pF but the Service Manual gives it as 270pF. I'll pop mine out and change it later, to overcome this, I replaced IC37 with a 74F04 and the problem went away (for the most part)

If increasing the capacitance of C48 to 270pF improves the stability, I will update the PDF. I have an X7R ceramic 220pF/50V at this location.

[hoglet]

The RGB2HDMI won't give me a stable MODE7 on these new PCBs.

I noticed an issue in the original BOM for the board where C48 is specified as 220pF but the Service Manual gives it as 270pF. I'll pop mine out and change it later, to overcome this, I replaced IC37 with a 74F04 and the problem went away (for the most part)

Is IC40 a 74S04?

If increasing the capacitance of C48 to 270pF improves the stability, I will update the PDF. I have an X7R ceramic 220pF/50V at this location.

Generally we found that reducing the value of C48 improves the symmetry of the 6MHz clock:
https://github.com/hoglet67/RGBtoHDMI/w ... dification

You should try and get a scope shot of the 6MHz clock so you can see what effect changing the capacitor has. Because of the asymmetry it can be difficult to get the scope to trigger consistently. The shape repeats every 3 cycles. So you might need to play with the hold-off control.

Dave

[maniacminer]

Yeah, I was unsure whether increasing or decreasing the value would have the effect, so taking an idea of yours, put C48 in a socket. My IC40 is a 74F00 at the moment, I can swap it out for a 74S00 (even the 74LS00 seems to work identically) The scope looks very nasty, storing a long sequence I can see the widths varying but by much more than another Beeb I have to hand. I use a simple program to display a pattern that I find is very difficult to get a zero error on.

Code: Select all

   10MODE7:VDU23;8202;0;0;0;
   20FORX%=&7C00TO&7FFFSTEP4
   30!X%=&3E2A3E2A
   40NEXT
   50REPEAT UNTIL GET=32
   60MODE7

[hoglet]

The scope looks very nasty, storing a long sequence I can see the widths varying but by much more than another Beeb I have to hand.

If you scope the 6MHz clock (pin 19 of the SAA5050), set the horizontal timebase to 50ns/division and then adjust hold-off you should get a stable trigger showing three complete cycles. A typical hold-off value would be between somewhere between 0.5us and 1.0us.

Post a phoo of what you see...

If you can't get a stable trigger, using hold-off then something else is amiss.

The circuit will be very sensitive to the TTL logic families used for IC37/38/40/43. The design will have been hand-tweaked for LS/LS/S/S respectively. Any deviation from this will likely require further tweaking, but it's impossible to do this blind. A stable scope shot is essential.

Dave

[maniacminer]

The scope looks very nasty, storing a long sequence I can see the widths varying but by much more than another Beeb I have to hand.

If you scope the 6MHz clock (pin 19 of the SAA5050), set the horizontal timebase to 50ns/division and then adjust hold-off you should get a stable trigger showing three complete cycles. A typical hold-off value would be between somewhere between 0.5us and 1.0us.

Post a phoo of what you see...

If you can't get a stable trigger, using hold-off then something else is amiss.

The circuit will be very sensitive to the TTL logic families used for IC37/38/40/43. The design will have been hand-tweaked for LS/LS/S/S respectively. Any deviation from this will likely require further tweaking, but it's impossible to do this blind. A stable scope shot is essential.

Dave

OK, two problems. Firstly, the SAA5050 is misbehaving when it heats up, so bin for that. Secondly, the "auto-calibrate" mode on the RGB2HDMI ignores good settings and then gives me a shimmery/shaky MODE7 - I can see good settings drift by, I'm thinking - "why didn't it just stop and use that one?" the clock signal with no capacitor was wild, with 100pF it had weird lines through the text at various points, with 150pF there was one less line, 220pF reasonably stable, 270pF one line and 330pF no line and the clock looks good, a repeated triple vs. random strings of seven or eight pulses with different intervals. When I try to manually calibrate the RGB2HDMI I am unable to find any setting where the image is stable, despite seeing a stable image when running an auto-calibration - do I have access to exactly the same parameters from the sampling menu, or are other things at play?

Lots of repeats with the same waveform (memory scope)

Battered 330pF capacitor - been in and out a lot...

BTW:I'm not grounding the probe at the tip, hence the wibbly/wobbly bits.

[hoglet]

When I try to manually calibrate the RGB2HDMI I am unable to find any setting where the image is stable, despite seeing a stable image when running an auto-calibration - do I have access to exactly the same parameters from the sampling menu, or are other things at play?

That clock doesn't look too bad to me...

You should have access to everything in the sampling menu (i.e. there are no hidden calibration settings).

What does the INFO menu show for calibration detail after then auto-calibration?

You can take a screen shot of the menu by pressing up and down at the same time.

Dave

[maniacminer]

This is the final frame of the video I took of the calibration sequence. The 4's are shimmering.

[!FOZ!]

LOL I need to not drink coffee when putting ICs into sockets... can you spot my further two errors, one prevented MODE7 from working, the other prevented other modes from displaying anything but the cursor, until I scrolled...

There's Something Not Right.jpgThere's Something In The Wrong Place.jpg


All the boards work, I gave them a good run in Cybertron Mission and Starship Command for an hour or so and all is well.
PXL_20240506_223928334.jpg

Wow that is such a great picture! Congratulations - fine specimens indeed!

[!FOZ!]

The RGB2HDMI won't give me a stable MODE7 on these new PCBs.

This post https://www.stardot.org.uk/forums/viewt ... 56#p401456 includes a pic of my MODE7 using RGB2HDMI. I followed the BOM specs as I was not aware of a difference in the manual. I don't mention this in anyway to tease as such, just in case it helps narrow things down.

[maniacminer]

The RGB2HDMI won't give me a stable MODE7 on these new PCBs.

This post https://www.stardot.org.uk/forums/viewt ... 56#p401456 includes a pic of my MODE7 using RGB2HDMI. I followed the BOM specs as I was not aware of a difference in the manual. I don't mention this in anyway to tease as such, just in case it helps narrow things down.

Yeah, thanks, but I'm not having a problem with those characters, only the number '4'

I found out what it was in the end, it's firmware related. There's a bug in the way the interlacing -> progressive is calculated and this makes the the '4' character have a shimmering line in it. The reason why some of the calibration settings looked good as the came by was the "Advanced Motion" seems to be disabled during this phase and then re-enabled afterwards. Setting the Teletext de-interlacing to "Weave" made the problem disappear.

[maniacminer]

Wow that is such a great picture! Congratulations - fine specimens indeed!

Thanks did you figure out what my mistakes were? (it's almost like a BBC Micro User Group Xmas Quiz)

[!FOZ!]

Wow that is such a great picture! Congratulations - fine specimens indeed!

Thanks did you figure out what my mistakes were? (it's almost like a BBC Micro User Group Xmas Quiz)

Hmmmm well I am no expert, but studying your pics vs my board, in that post I can see IC15 and IC6 and of course the 6202 missing? I know enough to know IC6 is key to the video display and have picked up from reading past threads the cursor is managed elsewhere...

[!FOZ!]

The RGB2HDMI won't give me a stable MODE7 on these new PCBs.

This post https://www.stardot.org.uk/forums/viewt ... 56#p401456 includes a pic of my MODE7 using RGB2HDMI. I followed the BOM specs as I was not aware of a difference in the manual. I don't mention this in anyway to tease as such, just in case it helps narrow things down.

Yeah, thanks, but I'm not having a problem with those characters, only the number '4'

I found out what it was in the end, it's firmware related. There's a bug in the way the interlacing -> progressive is calculated and this makes the the '4' character have a shimmering line in it. The reason why some of the calibration settings looked good as the came by was the "Advanced Motion" seems to be disabled during this phase and then re-enabled afterwards. Setting the Teletext de-interlacing to "Weave" made the problem disappear.

Wow keen eye sir. I must admit I have always been puzzled by the calibration logic, watching it go through its cycles and seeing what looks like a good result, only to see it land on a lesser one. Never really questioned it tbh, mostly because I know my video circuitry in my older Beebs does also need an update - and I've not much clue on the inner workings of amazing RGB2HDMI device.

[maniacminer]

Hmmmm well I am no expert, but studying your pics vs my board, in that post I can see IC15 and IC6 and of course the 6202 missing? I know enough to know IC6 is key to the video display and have picked up from reading past threads the cursor is managed elsewhere...

IC15 pin#11 isn't able to make contact with the socket as it doesn't have an insert. That pin clocks the data from the memory to the SAA5050 hence no MODE7.

IC39 should be a 74LS283 and I stuck in a 74LS138 and that was causing the empty screen until the very last 1k byte of RAM then repeats appeared all over the screen brain had gone for a Burton