Universal RGB Amp PCBs - Interest check

Started by BlueBMW, March 15, 2012, 01:00:41 PM

Previous topic - Next topic

BlueBMW

Hello all,  I've done some RGB mods in the past using transistors and NJM2267 chips, and they work fine, but then I was informed of the THS7315 which seems to do wonders in a very small package!  I've done some prototyping using a PC Engine Duo-R and a friend and I have come up with a PCB design to utilize this amp chip.

I'm looking to get these PCBs made up in the next month or two.  What interest would there be in these and does anyone have any design change recommendations?

Also, what sort of price would be considered fair for a pre-built board like this?

Approx 1.125" x 1.375"  - probably a 0.031" thick PCB, but might be 0.0625"
Trimmers are 10K ohms, Q1 is the THS7315, Q2 is an LM1881, all the caps are 0603 0.1uf except c5 which is a 1206 22uf, R1 - R3 are 75 ohm
The sync on green / sync stripper part of the board do not have to be installed / utilized in every application.

I'll be moving the text to the left and adding a mounting hole to the design.






soop

Hello Blue :D

I'd probably be interested in some of these.  Would they be populated?  And how much do you think they'd cost?  As long as it's not loads, I have about 3 White PC Engines that need a bit of RGB love, and maybe a Core II I could mod.

Building my own amp led to failure, and I haven't had the heart to go back to it.

But if it's not populated, I have the parts, though it looks like you intend to use tantalum caps on that one, right?  And I have a bunch of C1815s

Artemio

Interested in a few here as well, if you are willing to do some international shipping or ship to a hotel when I visit the US =P

BlueBMW

I'm doing a little more tweaking / prototyping to make sure I select the best size of trimmer for the application.  Once I've got the best possible values for all the parts, I'll look into having the PCBs made up.  It usually takes about 3 weeks to get boards in.

As far as populated vs unpopulated, I'd do it however people wanted it.  I'll be soldering them by hand, but surface mount on brand new boards / components is not too difficult.  Not everyone will want the sync on green or sync stripper circuits.  Pricing will be as cheap as possible.  Once I find out the total cost etc, I'll figure out the best price for them.

International shipping has never been a problem :)

panzeroceania

what systems would these be used for? PC Engine mainly? I am interested, I want to support work like this. Any idea on price points?

I need to buy/make an RGB amp for my Sega Mark III.

I would buy one though as I need one for my NEC Supergrafx, I already have one for my Super CD2 but it would be nice to have something for the Supergrafx

BlueBMW

Quote from: panzeroceania on April 05, 2012, 12:06:41 PM
what systems would these be used for? PC Engine mainly? I am interested, I want to support work like this. Any idea on price points?

I need to buy/make an RGB amp for my Sega Mark III.

I would buy one though as I need one for my NEC Supergrafx, I already have one for my Super CD2 but it would be nice to have something for the Supergrafx

Primarily I designed these with the NEC consoles in mind, though they could really be used anywhere a signal amplifier is needed.   Aside from RGB, they can also be used on NEC consoles to amplify the S-Video signals that the HuC6260 chip generates (pin 40 for Y) and then strip the composite for nice clean S-Video without an external encoder.

Price point isnt set yet... I'm really hoping to be as cheap as possible though.   As it is, I'm looking at a parts cost of around $10 a board fully populated.  So add in whatever assembly time is worth, shipping / packaging costs and whatever a fair profit is on these, that will be the final price.

viletim

Quote from: BlueBMW on March 15, 2012, 01:00:41 PM
does anyone have any design change recommendations?

Absolutely!

1. Use a THS7314 instead of a THS7315. It has a gain of only 2 which makes it much better suited as a video driver.
2. Remove the potentiometers from the circuit.  They will cause a signal level mismatch between channels and add noise.  The amplitude of the PCE video signal is constant. Measure it!
3. The amplifier is missing a bias network.  The video will swing below zero, outside the linear region and the signal will distort (clipping). In practice, the darker parts of the video will appear black as the details will be lost.
4. The PCE will have a TTL sync signal available, so instead of the expensive LM1881M, you can use a cheap logic gate to buffer the sync signal.
5. Your have silkscreen over your copper pads. Silkscreen will resist solder!
6. You can significantly reduce the noise picked up and radiated by you PCB by filling all the unused space on the green layer with ground connected copper.



BlueBMW

#7
Quote from: viletim on April 11, 2012, 10:15:28 PM
Absolutely!

Thanks for all the advice!

Quote1. Use a THS7314 instead of a THS7315. It has a gain of only 2 which makes it much better suited as a video driver.

The intention of using the higher gain was to make the chip more versatile.  The intention here is to make a chip that can be used on the PCE/TG16 stuff but also be used in other applications needing an RGB amp or similar.

Quote2. Remove the potentiometers from the circuit.  They will cause a signal level mismatch between channels and add noise.  The amplitude of the PCE video signal is constant. Measure it!

I thought about making a way to bridge past the potentiometers if they are not desired in a specific setup.  The idea with this board is to just install the components needed for a specific application.  The pots may not be necessary on the PCE/TG16, but other applications may require them.

Quote3. The amplifier is missing a bias network.  The video will swing below zero, outside the linear region and the signal will distort (clipping). In practice, the darker parts of the video will appear black as the details will be lost.

I consulted the guy who's been helping with the design of this chip concerning this issue.  He said he selected this particular chip because it is self biased and doesn't require a coupling cap.

Quote4. The PCE will have a TTL sync signal available, so instead of the expensive LM1881M, you can use a cheap logic gate to buffer the sync signal.

Again the use of the LM1881M is to make the chip more versatile, and as with all the components, it does not need to be installed in every application.  Though I will look more into this.

Quote5. Your have silkscreen over your copper pads. Silkscreen will resist solder!

The last time I had chips made through this supplier, they did not put silkscreen over any solder pads even if the design had them over pads.  I'll be cleaning up all the silkscreen stuff before final production.

Quote6. You can significantly reduce the noise picked up and radiated by you PCB by filling all the unused space on the green layer with ground connected copper.

An excellent point, one which I will definitely implement!

Thanks again for all the great input!




viletim

Quote from: BlueBMW
I thought about making a way to bridge past the potentiometers if they are not desired in a specific setup.  The idea with this board is to just install the components needed for a specific application.  The pots may not be necessary on the PCE/TG16, but other applications may require them.           

What other applications, exactly?

Off the top of my head, I can think of two consoles that need need a video driver circuit to get the RGB working sensibly. There's the PC Engine and the Nintendo 64. In both cases the amplitude of the signal at the point inside the console is 0.7Vpp. I doubt this is an accident. All you need is a standard (ie. gain of two) driver stage and you're done. No fiddling required. It is better to making something that serves its sole purpose well than some kind of one size fits all device that doesn't really fit anywhere.

The PC Engine's video signal is a little bit more tricky to work with due to the large DC offset, but a properly designed video driver will work on the N64 too with no modification required.

Quote from: BlueBMW
I consulted the guy who's been helping with the design of this chip concerning this issue.  He said he selected this particular chip because it is self biased and doesn't require a coupling cap.   

He's wrong. Read the datasheet.

In short, there are three possible biasing techniques that can deployed against this part.

1. DC coupling - connect the video signal straight into the chip. This works because there's some special 200mV (or thereabouts, I don't have time to check) internal reference which is added to the input. It prevents the signal from banging into the negative rail on the output (which would cause distortion). This method will work with the N64 RGB signal  (those units with the signal available, that is) because it varies between 0v and 0.7v. The PC Engine's video video signal measures between 4.3v (black) and 5v. Hopefully you can see why direct coupling is not an option.

2. AC (Capacitive) coupling with sync tip clamp - This part has inside it what is known as a sync tip clamp. I did a quick search an found a circuit of this arrangement here: Look for Figure 4 (a). This works just fine for a video signal with the sync signal poking through the bottom of it but will distort the bottom of the waveform if you try it on sync-less video (such as what you are dealing with). This is the way you currently have it configured - maybe - you didn't post a circuit diagram.

3. AC (Capacitive) coupling with resistor bias network - this one you want is what you want. Bias the input at about 1V DC or perhaps slightly less to allow maximum output swing without getting near the limits of the output stage or the nasty sync tip clamp. Your bias voltage will be much more critical if you go for the part with higher gain. You probably need capacitive coupling on the output too, otherwise you may exceed the power dissipation for this part when the loads are connected.

Anyway, the datasheet explains it well, read it thoroughly.

soop

Isn't he getting the sync through composite?

thesteve

the "sync tip clamp" used in this chip is a true DC restore circuit, however a pull-up resistor is recommended for sync free signals and will be added.
the value of the resistor will determine the need for output cap coupling, so i will chose to make it un-needed.

thesteve

#11
for best results add a 26Meg to 28Meg resistor from each input pin to pin 4 (+5)

Piratero


Piratero

Quote from: BlueBMW on April 11, 2012, 01:29:37 PM
Quote from: panzeroceania on April 05, 2012, 12:06:41 PM
what systems would these be used for? PC Engine mainly? I am interested, I want to support work like this. Any idea on price points?

I need to buy/make an RGB amp for my Sega Mark III.

I would buy one though as I need one for my NEC Supergrafx, I already have one for my Super CD2 but it would be nice to have something for the Supergrafx

Primarily I designed these with the NEC consoles in mind, though they could really be used anywhere a signal amplifier is needed.   Aside from RGB, they can also be used on NEC consoles to amplify the S-Video signals that the HuC6260 chip generates (pin 40 for Y) and then strip the composite for nice clean S-Video without an external encoder.

Price point isnt set yet... I'm really hoping to be as cheap as possible though.   As it is, I'm looking at a parts cost of around $10 a board fully populated.  So add in whatever assembly time is worth, shipping / packaging costs and whatever a fair profit is on these, that will be the final price.

I've had major trouble with the THS7314 on a PC Engine. It could be my scaler or the fact that I wasn't using an amp on the C-sync signal.

So this amp may not work for everyone.

thesteve

the sync on the PCE is only 0.3V and wont work directly for most things

papa_november

#15
I did not know that the THS7314 wouldn't work with the TurboGrafx without extra external circuitry, that's kind of a bummer. What kind of additional components would be needed to get it to work?

As I understand it, with the THS7314, all you need to do is wire it straight up through three resistors *if* you're using it with an N64. So that means, basically, that any PCB used for the job doesn't need to be much more complex than the SOIC-8 -to-DIP adaptor boards that are already floating around for almost nothing on eBay. So really, there's no need to make this board "universal" if such a simple solution already exists for the N64.

Perhaps a better idea for a PCB design would be to focus on using one of the NJM2267-based circuits floating around, as we already know they work and since building them on a breadboard is a pain in the ass. More important, however, is the opportunity to also incorporate space for a mini-DIN 8 socket on the board and design the whole thing so that it can be soldered in place of the RF modulator on the original PCE/TG-16. Ideally it would look something like this older mod from the outside, but slightly easier to install. You could even repurpose the channel 3/4 switch as a region switch. This would also solve the problem of there being no composite output - just make up a mini DIN-8 to composite a/v cable for those times when composite is necessary.

Also, I wouldn't worry about trying to amplify the bare sync signal from inside the machine. Not when the composite video signal is right there and will work just fine on TVs/XRGBs/any RGB setup that already uses an LM1881.

BlueBMW

Small update to this... weve been working on a complete redesign.  The new board will be tri purpose and designed with the pc engine / turbografx in mind.  It can be assembled to be used as an rgb and sync amp, svideo amp, or component encoder/amp.  It will be 1" x 1.25" and utilize 1206 sized caps and 0603 sized resistors.

Still finalizing some tweaks to it but it is nearly complete and ready to have built.

Comments / suggestions are welcome.






papa_november

So is this based on the older transistor-based design?

It doesn't really matter what design it uses so long as it works. I'd buy one.

micro

I like the idea of having a little amp/buffer board for the PCE. But BlueBMW, I'm wondering why you did fall back on using single transistors?
So I tried to design a little THS7314-based board, including the changes viletim suggested.





With 3.6 MOhm pullup resistors the 7374's input pins should be biased at 0.91 V. The datasheet suggests to set the cutoff frequency of the input high pass filter at about 3 Hz, so that will give us a Cin of 82 nF.

I still have to try out if that circuit really works...

GUTS

There was a guy over that PCEFX boards a couple years ago who had awesome little RGB/S-Video boards made for the Turbo Duo. I bought one off of him for about $25 and it was worth every penny, the picture is beautiful. They look fairly similar to your original design, and were very versatile. I could take a picture if you're interested in what they look like.

RGB32E

Quote from: micro on November 09, 2012, 12:53:56 AM
I like the idea of having a little amp/buffer board for the PCE. But BlueBMW, I'm wondering why you did fall back on using single transistors?
So I tried to design a little THS7314-based board, including the changes viletim suggested.
...
With 3.6 MOhm pullup resistors the 7374's input pins should be biased at 0.91 V. The datasheet suggests to set the cutoff frequency of the input high pass filter at about 3 Hz, so that will give us a Cin of 82 nF.

I still have to try out if that circuit really works...

Looks cool!  I was going to try with a 7374 so that I could use the same device for RGB and CSYNC.  I see you mention both 7374 and 7314!  :o  Also, I'm wondering if using .1uf for Cin and 3.3M ohm pull up resisors wouldn't work right?

Quote from: GUTS on November 09, 2012, 04:01:19 AM
There was a guy over that PCEFX boards a couple years ago who had awesome little RGB/S-Video boards made for the Turbo Duo. I bought one off of him for about $25 and it was worth every penny, the picture is beautiful. They look fairly similar to your original design, and were very versatile. I could take a picture if you're interested in what they look like.

I'd be curious to see pictures of this PCB.  Pictures of both sides (if double sided) would be great!

keropi

#21
we have a similar discussion on assemblergames forum here  , today I got a good picture from my PCE by using 10μF cap in input and 75ohm resistors in output... once there is a good way to amplify c-sync then the picture will be perfect.

first of all here are some photos of the OLD transistor-based amp that uses c-sync , it is amped the same way as RGB
please view them fullscreen 1:1 size

   

and here are the pics with the 7314 that uses composite and has a checkerboard pattern because of this:

     

the pics above are by following this diagram without the pullup parts (for now)



I assume that once an amplified c-sync signal is used with the 7314 , the quality will improve and be at least the same as the old amp...

GUTS

Here's a couple shots of that Turbo Grafx S-Video board, although I removed the S-Video out cable since I only use the RGB. The blue wires are mine, I think I was just routing +5, sync, and RGB to the S-Video pads.






micro

@RGB32E: Sorry, that was a typo, I always meant THS7314 by Texas Instruments.

@keropi: Nice to have a picture but it doesn't look very good (yet).  ;D Without the pull-up resistors you're not using the AC-Bias Input Mode... Also, your input capacitor is way too high, and your output is DC-coupled instead of AC-coupled...?
I'm not unfamiliar with that nasty checkerboard pattern. But it's still possible to use composite video instead of composite sync without having that noise on your screen. Just use good cable in which the video wire is shielded.
Unfortunately my Turbo Duo needs a total cap replacement before I can test out the THS7314....

@GUTS: So your PCB uses a Sony CXA2075. At the RGB Output there are holes for 3x 75 Ohm resistor and 3x 220 µF capacitors but you soldered your RGB wires directly to the RGB output pins?

GUTS

Yeah it's been ages since I messed with it, but if I remember right it looked great coming directly off those pads so I didn't bother adding anything. I think I was out of 220uf capacitors at the time and too lazy to go to radio shack, so I just wired it up for the hell of it to see what it looked like; the picture was bright and beautiful, so I said screw and just put everything back together. I should probably add them at some point, but I was just getting into this stuff back then and didn't have the faintest clue wtf I was doing. I have a faint clue now, so one of these days I'll get around to adding the parts, ha.

Heffa

I would be interested in one or two of these once they are finalised :D

BlueBMW

The design change back to transistors was to make it a dual purpose YUV amp/mixer and RGB amp.  We had some issues with getting the component signal just right, so another redesign is in the works.

Heffa

No worries, I'll gladly wait for the perfect design :)

Sent from my GT-P6800 using Tapatalk 2


keropi

I did the same test with the pull ups on the RGB lines today , with the  info discussed on this assembler forums thread



and I got this result:



it looks the same as the old transistor-based amp to me, now all I need is a good way to build a c-sync amp, the design based on the 74HC04 discussed on the assembler thread does not work for me... using the old amp to amplify c-sync works great and I get awesome results.

micro, I will get the needed resistors/caps today to try your schematic too and see if there is a difference

micro

I finally had the time to replace all my PCE Duo's capacitors. Apparently the leaking caps also destroyed a via on the board which put a 4558 op-amp out of order. After installing a bypass for that broken via (the via was almost black) I finally had full sound again =)

So I could try out the THS7314 and so far I like the result:




Heffa

That's looking great! :)

Sent a long time ago from a Galaxy far far away using Tapatalk HD

Heffa


micro

#32
QuoteAny news on this lately?  :)
In case you're addressing me: Actually, yes :D

I've made some minor changes: The pads for RGB in and out are on the top side only now. There shouldn't be any exposed copper on the bottom side now so you could glue it onto the PCE mainboard directly without causing a short-circuit. Also there's a list on the bottom side showing the value and package of each component in case the schematic isn't in reach ;)





The size of this PCB is 34.5 mm x 18.5 mm.

A few days ago I ordered some PCB's of this design for testing purposes. If it works as intended then you can have one of those PCB's. Or I can give you the files and you can order your own PCB's :)

RGB32E

Looks good!  How about a PCB for a THS7374 for RGBS?

micro

I haven't tried out the THS7374 yet.

It seems the 7374 is very similar to the 7314. But in case of the PC Engine / Turbo Duo you can use composite video for synchronisation and it doesn't have to be buffered because it's already available on the (Turbo Duo's) AV out.

For which console, monitor or video input device do you need the sync signal amped or buffered?

Heffa

#35
micro - I'd gladly buy one of your boards and try it out once you get them.
I'd better stock up on some SMD components though, I have no 0805 in storage  :)

...also, if you are willing to share I'd love to have the files as well for further tinkering...

RGB32E

#36
RGBS output is preferred over RGBCv. 

Here are some devices that require or perform better with CSYNC and not composite video:
1. Extron RGB interfaces -> No picture
2. Extron RGB switches -> No picture
3. GBS-8220 Scaler -> No picture
4. Select Sony PVM/BVM monitors -> No picture
5. NEC Multiscan monitors (XM*/ect) -> No picture as they require RGBS or RGBHV
6. XRGB upscalers -> Better PQ!

What are you using that made you think otherwise?  :o  I've used the THS7374 with a TG16 and it works just fine, haven't tried the 82nF/3.6M AC biasing combination yet (tried 0.1uF/5.3M).

EDIT:
7. Commodore 1084S - http://nfggames.com/forum2/index.php?topic=3387.0
OP ended up switching to CSYNC from the consoles (NeoGeo modified), as the first attempt involved adding extra HW that shouldn't be required (LM1881).

Quote from: micro on May 17, 2013, 06:22:18 AM
I haven't tried out the THS7374 yet.

It seems the 7374 is very similar to the 7314. But in case of the PC Engine / Turbo Duo you can use composite video for synchronisation and it doesn't have to be buffered because it's already available on the (Turbo Duo's) AV out.

For which console, monitor or video input device do you need the sync signal amped or buffered?

micro

Well personally I'm using a Panasonic Plasma TV, a Philips CM8833 Amiga Monitor and sometimes a Sony PVM monitor. They all accept composite video for synchronizing RGB. The Panasonic TV even prefers composite video; that fucker shifts the picture to the left and cuts off a large chunk if fed with RGBS  ;D

But of course I know that composite video can mess up RGB (checkerboard effect) and sound (buzzing) when it's not shielded properly.
So I don't question your love for composite sync. But the question is: For which console do you need a buffer for composite sync? And do all those devices you've listed terminate the composite sync input with 75 Ohm to GND? These are real questions because I haven't dealt with that topic before.  :D

micro

The PCB's have arrived:



Heffa, have you already ordered the SMD parts? I need the SMD parts, too.
I could order all neccesary parts except for the THS7314, which is hard to come by in Germany.

Are you interested?

Heffa

Quote from: micro on June 05, 2013, 07:51:55 AM
The PCB's have arrived:

Heffa, have you already ordered the SMD parts? I need the SMD parts, too.
I could order all neccesary parts except for the THS7314, which is hard to come by in Germany.

Are you interested?
Hey, it's looking good!
I haven't ordered the parts yet, but I have looked for it & found most parts (inc. the THS7314) on eBay.
With the summer coming I don't have that much time, but I'm still very interested in buying one of the PCB's for when I get an evening off :D

Please send me a PM & I'll pay with PayPal if it's ok.

---
Sent a long time ago from a Galaxy far far away using Tapatalk HD