EEVblog Electronics Community Forum
Electronics => Projects, Designs, and Technical Stuff => Topic started by: caius on September 24, 2017, 10:32:15 am
-
Hi all,
I'm designing an RGB VIDEO DAC formed by two 74LS374 and three resistor ladders (one for each color)
(https://s26.postimg.org/drihb6n51/circuit1.jpg) (https://postimg.org/image/drihb6n51/)
On the output of each resisitor ladder I can measure :
(https://s26.postimg.org/e5jtas78l/image.jpg) (https://postimg.org/image/e5jtas78l/)
The correct signal should be this:
(https://s26.postimg.org/3wvxip8kl/image.jpg) (https://postimg.org/image/3wvxip8kl/)
I tried to amplify the signal with an NPN transistor :
(https://s26.postimg.org/iimy6y5d1/circuit3.jpg) (https://postimg.org/image/iimy6y5d1/)
But colors on screen are now too strong, they are "bleeding"
How can I amplify the signal to the correct level (which is TTL, so 3-5V and not 0-0.7VPP)?I was thinking to use 74HCT374 instead on LS family or perhaps some OP-AMPs (LM358, LM318).It can work?Any advice will be appreciated.
Thanks in advance
-
Actually the emitter followers you are using to buffer the
ladder do not V amplify, their effective G =~ 1.
A clean simple way is to use a Video OpAmp, say a quad,
that would handle all 3 signals + mix if you needed to.
What is your dot clk frequency ?
Norton type OpAmps one possibility http://www.ti.com/product/LMH6714 (http://www.ti.com/product/LMH6714)
There are even V OpAmps that can handle video BWs.
Start with http://www.ti.com/amplifier-circuit/special-function/video/products.html (http://www.ti.com/amplifier-circuit/special-function/video/products.html)
Note your board layout has to be clean and designed to handle the high freq signals
you are dealing with.
Regards, Dana.
-
LSTTL output voltages are very poorly defined (i.e. anything >2.4V is within specification) and will vary depending the state of other outputs. Put your scope on the LSTTL outputs, to assess if that is an issue. Modern CMOS outputs are more dependable and reach the supply rails.
"Strong colour" could mean many things; you need to disambiguate.
"Bleeding" could be due to many things. Poor circuit layout would be a classic.
-
Change you LS374 to an HC374/HCT374, or better an AC374/ACT374.
At the Emitter output of your NPN transistors, add a 75 ohm series resistor before you drive your output color. Without a load on the video out, the video should be 1.5vp-p. When you connect it to a monitor which has a 75 ohm load, the video level will drop to 0.7vp-p.
At the Base add a 2k pull up to VCC. and a 2k pull-down to GND. (You can adjust these 2 to bias the transistors better into their optimum range and change the video brightness) Don't start the video down at 0v, more like 0.2v, otherwise, then dark colors might smear.
And PLEASE, if you are breadboarding this, or have yet to build a pcb, PLEASE change the LS374 to an HC574, or HCT574, or AC 574, or HCT574. (You will thank me for this one...)
-
Many thanks to all your replies.
Actually it's not a design from scratch but I'm reproducing a custom chip used on arcade PCB:
(https://s26.postimg.org/6gnfifnol/custom.jpg) (https://postimg.org/image/6gnfifnol/)
I removed the epoxy but all parts on it have no marking but looking at other similar designs I could identify the two bigger ICs in a 74LS374.As you can see from pinout I figured out the 74LS374 takes 15 bits of color palette on inputs then outputs are connected to a resistor ladder (with typical values from 4.7K Ohm of the LSB, 2.2K Ohm, 1K Ohm, 470 Ohm till to 220 Ohm of the MSB).The output of each resistor ladder (one for each color) is directly tied to the pins of the custom ( pin 3 is RED, pin is GREEN, PIN 5 is BLUE, this is the final stage since the three outputs of the custom are tied to the connector edge of the PCB along with composite SYNC, I use a monitor with a 75 ohm load in order to display the video output ) but they are also tied to the the little unknown chip which I was unable to identify (I first tought it could be a darlington array or a resistor array).
The strange thing is that if I remove the unknown IC I always get better results on the RGB outputs than my reproduction, I can't understand the reason and what is actually the function of this IC.
-
Change you LS374 to an HC374/HCT374, or better an AC374/ACT374.
OK, will do
At the Emitter output of your NPN transistors, add a 75 ohm series resistor before you drive your output color. Without a load on the video out, the video should be 1.5vp-p. When you connect it to a monitor which has a 75 ohm load, the video level will drop to 0.7vp-p.
At the Base add a 2k pull up to VCC. and a 2k pull-down to GND. (You can adjust these 2 to bias the transistors better into their optimum range and change the video brightness) Don't start the video down at 0v, more like 0.2v, otherwise, then dark colors might smear.
I tried to use a BC47 with collector tied to VCC, emitter to GND via a 100 Ohm resistor and 2K resistors pulld-down/pull-up on the base, image is no more bleeding but it's too dark (also tried different values with no improvement).I get better result when I use only a 10K pull-down on the base (so no voltage divider..).The 75Ohm in series on the emitter makes the image blurrier so I can omit it.
And PLEASE, if you are breadboarding this, or have yet to build a pcb, PLEASE change the LS374 to an HC574, or HCT574, or AC 574, or HCT574. (You will thank me for this one...)
Pardon, where is the difference?I always thought 374 and 574 were functionally identical.
-
And PLEASE, if you are breadboarding this, or have yet to build a pcb, PLEASE change the LS374 to an HC574, or HCT574, or AC 574, or HCT574. (You will thank me for this one...)
Pardon, where is the difference?I always thought 374 and 574 were functionally identical.
Signals route from from left to right with a 574, right old rats nest with a 374.
-
Signals route from from left to right with a 574, right old rats nest with a 374.
Yes, I know the pinout is different so it 's only a matter to make connection easier (also when routing the schematics to a PCB), I presume.
-
what game is this?
-
This custom is present an some game from Seibu/TAD Corporation manufacturer : Raiden, Blood Bros, Sky Smasher and few other.Embeding RGB DAC in a SIL package IC was a common practice back in '80-'90.There are other similar customs like the Taito 'TC0070RB' used on Rainbow Islands:
(https://wiki.arcadeotaku.com/images/3/32/RI8.jpg)
-
TC0070 rings a bell, i may have already reverse engineered one.
have to check my older harddrives.
Macro did it too.
http://www.macros-arcade.com/Taito-RGB.html (http://www.macros-arcade.com/Taito-RGB.html)
-
TC0070 rings a bell, i may have already reverse engineered one.
have to check my older harddrives.
It has been already reverse-engineered from a UK guy (you perhaps..), it's another 15-bit DAC :
http://www.macros-arcade.com/Taito-RGB.html (http://www.macros-arcade.com/Taito-RGB.html)
Did you reversed other customs?Did you work in arcade world?I have done some of them :
http://www.jammarcade.net/toaplan-hk-1000-reproduction/ (http://www.jammarcade.net/toaplan-hk-1000-reproduction/)
http://www.jammarcade.net/r-type-repair-log-a-k-a-custom-kna6034201-reproduction/ (http://www.jammarcade.net/r-type-repair-log-a-k-a-custom-kna6034201-reproduction/)
-
i'v done a few,
i recently netlisted the "custom" for capcom commando and the Rygar video counter from a couple of bootlegs.
not had time to convert the netlists into schematics yet though.
-
Ah, yes, I remember of you, you left a comment on a post of mine on JAMMArcade.
-
yes, about creating a site section.
8)
btw, why not use small smd to try to keep the replacements close to original size - or smaller ?
a lot of replacement stuff i see is huge relative to the old parts!
-
Infact I use always SMD devices for all my reproduction projects :)
-
I think the problem is not TTL vs CMOS. Although using a modern CMOS logic with high current symmetric drive output capability would improve the performance, what used to work should work, right? I guess the unknown chip is a quad NPN transistor like 2N2222s. Like what you did they are connected as emitter followers to convert high-ish resistor impedance to a low 75 ohm impedance for video. Most video monitors take 700mv peak-to-peak or 1v peak-to-peak with a 300mv sync tip. Remember there is a 0.6v Vbe drop at the emitter follower, and you don't want to drive it too hard or too weak to lose linearity. So you need some kind of Thievning network between the resistor network and the emitter follower. You can use Excel spreadsheet to calculate the voltage you need at the output (and at the base of the transistors) and use LTSPICE to verify the design.
-
I think the problem is not TTL vs CMOS. Although using a modern CMOS logic with high current symmetric drive output capability would improve the performance, what used to work should work, right? I guess the unknown chip is a quad NPN transistor like 2N2222s. Like what you did they are connected as emitter followers to convert high-ish resistor impedance to a low 75 ohm impedance for video. Most video monitors take 700mv peak-to-peak or 1v peak-to-peak with a 300mv sync tip. Remember there is a 0.6v Vbe drop at the emitter follower, and you don't want to drive it too hard or too weak to lose linearity. So you need some kind of Thievning network between the resistor network and the emitter follower. You can use Excel spreadsheet to calculate the voltage you need at the output (and at the base of the transistors) and use LTSPICE to verify the design.
I switched to 74HC374 (actually 74HC574, thanks for tip!), this improved a lot the video output.I don't think the unknown chip is a quad transistors or better, if it's so, it's not connected as emitter follower because the output of each resistor ladder is directly connected to the RGB pins of the custom so it's not a transistor which drives each color.But each output of the resistor ladder is also connected to this unknown chip via a 100 Ohm resistor.If I remove the unknown chip, the colors are still present, this is the proof that they are not driven by it (otherwise I will get no colors on screen)
This is the pinout/connection of the unknown chip:
pin1 tied to pin 11 thru a trace on PCB
pin2 VCC
pin3 to pin3 (RED color) of custom via 100 Ohm resistor
pin4 VCC
pin5 to pin4 (GREEN color) of custom via 100 Ohm resistor
pin6 VCC
pin7 to pin5 (BLUE color) of custom via 100 Ohm resistor
pin8 GND
pin9 not connected
pin10 GND
pin11 tied to pin 1 thru a trace on PCB
pin12 to pin 1 (/OE) of both 74HC374
pin13 not connected
pin14 GND
pin15 GND
pin16 VCC
-
video blanking,
probably a 367
-
I don't know if it's really a 74LS367.I used the auto-recognize function of my programmer but they could find any match.Then I remember that RGB signals are coming from the resistor ladders to the pins of the custom (but three pins of this unknown chip are tied to the three RGB pins of the customs as well)
-
auto-recognise is not reliable,
i have several devices that do that.
try selecting ls367 manually.
-
I metered this unknown chip and then a 74LS367, I don't get same measurements.On the unknown chip I can recognize the typical voltage drop of a junction (600-700mV) but on different points.I'm starting to thing it's a custom chip too.
Anyway, it seems I can omit it from my replacement, without it I get pretty same result of original custom.
-
it may not be LS
-
Sure, a CMOS one, you mean.I don't know, I will try first a prototype omitting this IC.Then maybe in later revisions I can try to implement it (if it's really a 367 ot other TTL).Thanks for you suggestions.
-
I think the problem is not TTL vs CMOS. Although using a modern CMOS logic with high current symmetric drive output capability would improve the performance, what used to work should work, right? I guess the unknown chip is a quad NPN transistor like 2N2222s. Like what you did they are connected as emitter followers to convert high-ish resistor impedance to a low 75 ohm impedance for video. Most video monitors take 700mv peak-to-peak or 1v peak-to-peak with a 300mv sync tip. Remember there is a 0.6v Vbe drop at the emitter follower, and you don't want to drive it too hard or too weak to lose linearity. So you need some kind of Thievning network between the resistor network and the emitter follower. You can use Excel spreadsheet to calculate the voltage you need at the output (and at the base of the transistors) and use LTSPICE to verify the design.
I switched to 74HC374 (actually 74HC574, thanks for tip!), this improved a lot the video output.I don't think the unknown chip is a quad transistors or better, if it's so, it's not connected as emitter follower because the output of each resistor ladder is directly connected to the RGB pins of the custom so it's not a transistor which drives each color.But each output of the resistor ladder is also connected to this unknown chip via a 100 Ohm resistor.If I remove the unknown chip, the colors are still present, this is the proof that they are not driven by it (otherwise I will get no colors on screen)
This is the pinout/connection of the unknown chip:
pin1 tied to pin 11 thru a trace on PCB
pin2 VCC
pin3 to pin3 (RED color) of custom via 100 Ohm resistor
pin4 VCC
pin5 to pin4 (GREEN color) of custom via 100 Ohm resistor
pin6 VCC
pin7 to pin5 (BLUE color) of custom via 100 Ohm resistor
pin8 GND
pin9 not connected
pin10 GND
pin11 tied to pin 1 thru a trace on PCB
pin12 to pin 1 (/OE) of both 74HC374
pin13 not connected
pin14 GND
pin15 GND
pin16 VCC
In some arcade game machines, the OE is important, especially if you have more than 1 resistor DAC unit driving the CRT. It is used for a custom sprite overlay's enable/disable from a separate bank of graphic roms/processor.
-
It seems the unknown IC is a 74HC365 since other boards use a different revision of the custom (marked as 'UEC-51') which has no unknown chip but a 74LS365 on game PCB in the nearby.
-
We were both close but wrong, it's a 74HC368!
The one I tried to identify for first first was bad in a output of a gate (pin 7), indeed I got not BLUE color on screen.So I removed another one from a good custom and I was able to identify it as a 74HC368.Now I have to review my schematics in order to implement it.Thanks to all (especially to 'stj')