Jump to content
IGNORED

Shape of the output signal on TIA pin


Windless

Recommended Posts

Cortex was a rather simple solution for me, but wht you propose is even simpler :)

 

Wil try this. Thaks you,

Windless.

 

* : (I have two blue pill that I paid $3 a pair and don't use, would need to order the crystal, capacitors, and a coil of undefined value to reproduce the clock found on the schematic of the pal 2600)

Link to comment
Share on other sites

Ok. I'll wait to receive the DIP40 sockets (I don't want to bend or cut pins) to make the test on the PAL-U pin from the 3.5MHz signal.

 

In the meantime, I can start trying to understand the other side of the schematic :

 

1171042552_secamtopalmod-palschmaticblock.thumb.png.08aacd5945a57610ba7564cfc445b76a.png

 

This is the schematic of the PAL 2600 as found on atariage. I took this one because I have a PAL TIA on my REV7 SECAM 2600 board.

 

  • The greyed part are the parts that I think are not relevant to the PAL signal generation.
  • The untouched part are the one I am unsure about
  • The red part is the 4.43MHz signal needed to generate PAL color. this is the one we already discussed.
  • The blue parts seems to be a used to generate a constant adjustable voltage on DEL (CADJ) to adjust color. This is true if 1nF C203 capacitor is just here to clean the OSC signal and has no effect on the blue part, which seems true (but I'm not skilled with electronics).
  • The green part is the part that needs be remade to generate the composite signal (without sound)

 

Blue part

 

Not much to say about this, except that on my board if there are potentiometer to adjust the delay, they are from the TIA (bottom right corner) and have been locked with some welding. The voltage at DEL seems to be a fixed 5.04V. Hopefully this dos not need to be changed.

 

Green part

 

-This seems easy to re-do. CD4050BDR seems to be still available at mouser. But is it the same pat ? The datasheet says it is an hex non-inverter buffer, the FAQ on this forum says it is a video buffer.

-The output goes to J12. I guess this is the composite entry for the RF module,but the RF module on my PAL REV7 is quite different : it uses a MC1373 which has a separate Composite (lum + synch) and Color (+audio) input. Does it mean that if I want to reuse the HF output, I need to no connect R222 and R234 and input the two separated tracks into the MC1373 ? Or maybe I could just add a CD4050 an AV output as described by alex_79 here. Can I do that without disconnecting the existing part of the schematic to preserve the hugly HF SECAM output ?

-I do not understand the part that is let B&W on the bottom right of the schematic. I guess J201/10 in ground (since it is connected to CD4050 pin 8, Vss), but then all the part with coil and the 2N3563 would be a can of modulated powersupply that would create a carrier,  modulate it with AUD, and use this modulated output as a Vcc for the 4050, hence modulating the output to add sound on it ? Anyway with AV output, I can just ignore everything that is inside the purple polygon, add a CD4050 with a clean 0/5V on the its Vss/Vcc ? Or do I need to disconnect the existing components ?

MC1373.pdf

Link to comment
Share on other sites

On my board, instead of a CD4050, there is a T74LS174B1.

Apparently it would be a flip-flop (the datasheet if for t74ls174 no b1), it buffers AND keep he value until next clock signal.

 

An interesting thing is there is no pull-ups before the T74LS174. Are they required only because of the CD5040 ? Are they unneeded when you use some kind of buffer ?

Edited by Windless
Link to comment
Share on other sites

5 hours ago, Windless said:
  • The greyed part are the parts that I think are not relevant to the PAL signal generation.

Correct

5 hours ago, Windless said:
  • The untouched part are the one I am unsure about

This is the modulator for sound.  If you do an AV mod, you'd be removing components from there to disable the FM oscillator.

5 hours ago, Windless said:
  • The red part is the 4.43MHz signal needed to generate PAL color. this is the one we already discussed.

Yep.

5 hours ago, Windless said:
  • The blue parts seems to be a used to generate a constant adjustable voltage on DEL (CADJ) to adjust color. This is true if 1nF C203 capacitor is just here to clean the OSC signal and has no effect on the blue part, which seems true (but I'm not skilled with electronics).

Also correct.  If it's welded in place, it's hopefully just a temporary glue such as Loc-Tite.

5 hours ago, Windless said:
  • The green part is the part that needs be remade to generate the composite signal (without sound)

Yes, though the extent to which it needs to be remade is in question.  It would be nice to see an actual SECAM schematic.

 

The power supply for the 4050 is R231 into R232 and C238 into VDD.  I'm not sure why they decided that it needs reduced power, but it does mean that the pull-up outputs will be somewhat weakened.  The rest of that circuit, as I said earlier, is the sound modulator which has the same power supply and ground as the rest of the system.  You'll notice that pin 7 is also fed from that common power rail.  Whoever did this schematic didn't give much thought into showing the common power and ground rails in an understandable way.  It reminds me of the schematic for my TV:  Save space, above all else.

 

One thing you may be able to do to substantially decrease the rewiring requirements for a composite mod:  Get the UAV board.  I believe it makes all its video and audio connections at the TIA, because rather than rely on the clunky analog components, it processes the digital output directly.  So then all you have to do is get the PAL color output to work.

 

Link to comment
Share on other sites

3 hours ago, Windless said:

On my board, instead of a CD4050, there is a T74LS174B1.

Apparently it would be a flip-flop (the datasheet if for t74ls174 no b1), it buffers AND keep he value until next clock signal.

 

An interesting thing is there is no pull-ups before the T74LS174. Are they required only because of the CD5040 ? Are they unneeded when you use some kind of buffer ?

It's doing a completely different function.  Remember that SECAM uses the "LUM" outputs to generate color.  The flip-flops are part of that circuit that does things like selecting between red and blue for encoding onto a line.  NTSC and PAL use the LUM output to generate shades of each color, so the resistor ladder is a digital-to-analog converter.

 

One thing you'll notice about SECAM output is that it is one brightness, total, with a selection of around 16 colors, just like an EGA adapter.  This is likely why.

 

Link to comment
Share on other sites

2 minutes ago, ChildOfCv said:

It's doing a completely different function.  Remember that SECAM uses the "LUM" outputs to generate color.  The flip-flops are part of that circuit that does things like selecting between red and blue for encoding onto a line.

Do you have a schematic of this part ? Because from what I could see, the LUMx output of the TIA seems to be directly connected to the inputs of the D flip-flops. Maybe the three other inputs/outputs of the hex fliflop serves this purpose ? (the tracks run under some IC, I will ty to find a cheap broken 2600 to remove components and make a complete schematic)

 

23 minutes ago, ChildOfCv said:

The power supply for the 4050 is R231 into R232 and C238 into VDD.

Ho, so maybe I made some wrong assumption and the sound goes to J201.11 and is muliplexed elsewhere ? (by an mc1373 like on my SECAM 2600, even thought its datasheet says it expects color on the same pin as sound, but probably it does not hurt if color is already multiplexed with ULM and SYNCH ?)

 

45 minutes ago, ChildOfCv said:

Get the UAV board.

Is it open source ? Because if I'm going to make a PCB for the clock anyway, It would be nice to make an all-in-one one.

I also, I still have not given up the idea of using a $2 cortex board, which could generate the clock and scan the output from COL, and generate RGB to a SCART. Not sure all this would fit on the 72MHz board, but it would be nice.

Link to comment
Share on other sites

I'd love to see the SECAM schematics for an Atari.  But so far it's apparent that nobody has created them.  I could make it a project if I had one though.

 

Unfortunately, the UAV is not open source.

 

Now, if you want to generate RGB for a SCART interface, then perhaps providing the timing sync is part of what the Cortex board can do for you.  The other part is to take the LUM and COL inputs and generate RGB output.  So that's 4 LUM inputs (one is for sync), 1 COL input, a PAL-S input, a PAL-I output, and some outputs to build analog RGB.  Can your Cortex board handle that?

 

Edit:  Forgot to mention:  The Cortex board is based on 3.3V logic, so make sure the 5V logic from the Atari won't fry it.

Edited by ChildOfCv
Link to comment
Share on other sites

23 minutes ago, Windless said:

Ho, so maybe I made some wrong assumption and the sound goes to J201.11 and is muliplexed elsewhere ?

No, the sound is definitely multiplexed at the location you pointed out.  The point is that the line going to R231 is the main power rail.  You'll notice it also goes to pin 7 of each controller.  It must also connect to VCC somewhere in J201.  The board at J201 does seem to have the voltage regulator and misc stuff.

Link to comment
Share on other sites

7 minutes ago, ChildOfCv said:

So that's 4 LUM inputs (one is for sync), 1 COL input, a PAL-S input, a PAL-I output, and some outputs to build analog RGB.  Can your Cortex board handle that?

It needs to be tested, but I hope that :

-PAL-I can be generated by the timer,

-PAL-S can be put on an interruptable pin, and hopefully since it need to trigger at 900KHz (15 times per VBL) it happens during VBLANK so it won't slow the colour conversion

-Reading the 3 LUM could be just one cycle

-Reading COL would require reading a timer when a signal occures

-Then combining everything and reading a lookup table could take a few cycles

-Generating the SYNC can be done while waiting for COL or while outputing RGB values ?

 

I did not write the code, but it's probably 10-20 cycles total. Since the TIA is 3.54MHz on pal when the bluepill is 72MHz, that *may* work. But it only work if the 1.55% error additionned to the error of the blupill crystal is precise enough for PAL-I

 

21 minutes ago, ChildOfCv said:

Edit:  Forgot to mention:  The Cortex board is based on 3.3V logic, so make sure the 5V logic from the Atari won't fry it.

It has a few 5V tolerant pins, IIRC scart is 0.7V so output will be ok also.

 

Disclaimer : I have very few experience with microcontrollers, and none with analog signals, I may forget lot of things.

Link to comment
Share on other sites

19 minutes ago, ChildOfCv said:

No, the sound is definitely multiplexed at the location you pointed out.  The point is that the line going to R231 is the main power rail.  You'll notice it also goes to pin 7 of each controller.  It must also connect to VCC somewhere in J201.  The board at J201 does seem to have the voltage regulator and misc stuff.

Ok, there is definitively something I don't understand : if the part that are connected to pin 7 of the DB9 port and stady +5V, and pin 8 stay at GND, then I see no way out for the audio signal :

1041453509_audiopath.thumb.png.dfb12622bffa6cae2b524183afb32cbf.png

Is the schematic missing some connection, maybe from the left part of R210 to the RF module ?

Link to comment
Share on other sites

42 minutes ago, ChildOfCv said:

I'd love to see the SECAM schematics for an Atari.  But so far it's apparent that nobody has created them.

Some guy sells 2 dead motherboard for 15€ in france, but he is on the other side of France and do no want to ship them to me :(

Link to comment
Share on other sites

The field service manual has a MUCH more readable schematic.  Plus, that schematic does seem to be inaccurate about mixing the sound in.

 

https://exxoshost.co.uk/atari/mirror/gossuin.be/Service_Manual_ATARI_2600_PAL.pdf for one possible download location.  See exhibit 2-2

 

In that schematic, the audio modulator sits in the upper right corner.  Basically, what it does is it oscillates at a different frequency depending on difference in voltage between VMod and the audio level.  That signal just mixes in with the generated video through C211 (which I can't even find on the reverse-engineered schematic).

Link to comment
Share on other sites

10 hours ago, Windless said:

But it only work if the 1.55% error additionned to the error of the blupill crystal is precise enough for PAL-I

The PAL version of the field manual gives some answer :

 

 

image.thumb.png.9bf4b56122dc3bf8bb00fe33ca45882a.png

The cheapest 8MHz crystal on mouser has a 25ppm precision, we can hope the crystal on bluepills is not more than 400 times that, so +-0.04MHz s really possible

Now the 4.43MHz clock will have to be derived from the 72MHz bluepill. The period of the PWM signal generated by the timer can be adjusted with a one cycle precision. We need 6.5% precision and we have 1/72.000.000 precision

 

Conclusion : precision is not an issue. Output level might be if the clock signal should not oscillate between 0 and 3.3V, but if there is a bluepill anyway, adapting the outputwill still be easyer / cheaper than making a complete clock.

 

Synch with the 900kHz signal from PAL-S can still be an issue thought.

Link to comment
Share on other sites

On 7/20/2019 at 5:41 PM, Windless said:

On my board, instead of a CD4050, there is a T74LS174B1.

 

ok, apparently the T74LS174B1 is not a T74LS174. It seems to be just a buffer like the 4050 :

NewFile2.thumb.png.ad4030db031485eac8b15e23a09fc605.png

 

the T74LS174 as a !MR (master reset ?) but the datasheet says it set output to 0 whatever Qx is, so maybe they are just different components ?

 

I have a 2x12Mpts record (12MB once compressed) if someone wants to have a look.

Link to comment
Share on other sites

I've found he datasheet of the SGS T74L1741B,the exact same one that is on my 2600 here.

 

In the meantime, I also stopped confusing D-flipflops and T-flipflops, and now can tell :

 

  • the T74LS174B1 has the same role as the  CD4050 : it buffers the LUMx signals (only LUM on the SECAM)
  • in addition, it is clocked with a 280ns (this is the 3.54...MHz used by the TIA but it does not share the track with it : the signals comes from a SN74LS00N quadrulple NAND gate - maybe the TIA also gets its clock from there but with a DMM I find a 2kR resistance between T74LS174B1 clock and TIA clock - maybe the clock is delayed to wait for the TIA to output on its LUM pin before the flipflop clocks it ?)
  • !MR (reset) of the T74LS174B1 doesn't seem to be used.

So I apparently could  take the LUM signals from the hex flipflop.

Link to comment
Share on other sites

2 hours ago, Windless said:

maybe the clock is delayed to wait for the TIA to output on its LUM pin before the flipflop clocks it

 

the NAND gate is used to generate a clock signal that is the inverse of the TIA clock :

 

NewFile3.thumb.png.2fb2732b8866b5438af3b4b3f180a6ed.png

 

 

if we look at how the LUM signal rises after the TIA clock,we see it is not completely ready when  the clock rises :

 

NewFile5.thumb.png.db92171f099b00fcdffb224ed34870e3.png

 

 

With respect to the inverted clock from the NAND gate, on the other hand, it is more reliable :

 

NewFile6.thumb.png.f775e33cc88bdaf9e92e32611245dd8b.png

 

Ok, so I have a clan LUM signal I can use, but it is half a (TIA) clock delayed compared to COL and SYNC.

Link to comment
Share on other sites

That NAND gate's output is pretty weak.  Either it has a huge load or it needs to be replaced.  The 4050 has a 3.6V power supply for some reason, but 74LSxxx chips require the full 5V.  So there is no good reason I can think of for its output to be so terrible.

4 hours ago, Windless said:

in addition, it is clocked with a 280ns (this is the 3.54...MHz used by the TIA but it does not share the track with it : the signals comes from a SN74LS00N quadrulple NAND gate - maybe the TIA also gets its clock from there but with a DMM I find a 2kR resistance between T74LS174B1 clock and TIA clock - maybe the clock is delayed to wait for the TIA to output on its LUM pin before the flipflop clocks it ?)

Resistors aren't used for delay (except in combination with capacitors).  Which TIA clock are you talking about?  There's O0 and O2 which it shares with the CPU and RIOT, there's the 900KHz output, and the 4.43MHz input.

 

So they're using a latch instead of a simple buffer then.  I guess they expect the output to change between cycles, and want to stabilize it.

Link to comment
Share on other sites

4 hours ago, ChildOfCv said:

Which TIA clock are you talking about?

Ok, sorry I haven't been clear let me try to fix that :

 

-I first measured with a disconnected 2600 between TIA 3.54The MHz clock input and the clock input of the 74LS174 so i thought they my be connected though a resistor or something. Then I tryed to follow the track from the 74LS174 clock pin,and found it reached the NAND gate

-the weak clock signal,yellow in the fist two picture of my previous message,is the TIA clock pin

-the clean (but a bit overshooting) clock in the last picture is taken at the clock pin of 74ls174 and comes from the output of the NAND gate. I didn't follow the entry of this gate but since the output is same frequency and inverted when compared to the TIA clock, I guess that is where is comes from.

 

4 hours ago, ChildOfCv said:

So they're using a latch instead of a simple buffer then.  I guess they expect the output to change between cycles, and want to stabilize it.

That's what I expected to find, but manually digging through the first 50.000 points I recorded, I didn't see that happen.

Link to comment
Share on other sites

8 minutes ago, ChildOfCv said:

So, to make sure I understand you, are you reading the TIA clock from pin 11, where the oscillator connects?  If so, then I guess that might explain the irregular waveform.

Yes, I did. If you want me o try getting it from the crystal or test something else, jut ask.

 

Thank you for all the help you provide :)

Windless.

Link to comment
Share on other sites

  • 2 months later...

Hi,

 

  Just a quick update to say that:

-I'm done with my other over-timeconsuming occupations, and can now work again on this subject

-I've bought another 2600 : after inspection, it's not working but is SECAM and has the same revision of the same PCB as my other one. Perfect candidat for a complete teardown and PCB scan !

 

Windless.

Link to comment
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

Loading...
  • Recently Browsing   0 members

    • No registered users viewing this page.
×
×
  • Create New...