New RGB amplifier mod for early nintendo 64 NTSC with RGB driver and OPAMPs

Started by RyleFury, June 07, 2009, 03:44:43 AM

Previous topic - Next topic

RyleFury

Hello everyone,
I tested several rgb amplifiers on internet with transistors which display poor results with nintendo 64, especially with the contrast/luminosity ratio. However,  I didn't test the Viletim's mod. Since the output of U4 is at 0,7Vpp without cable with the 180 ohms impedance, I thought set an impedance adapter with 75 ohm in output.  I have old RGB switcher (TEA5115) which adapt the impedance and x2 the input, I tested with this circuit and the result is perfect.

I also conceived a small circuit with 3 OPAMPs which do the same thing as TEA5115. However, I didn't test this circuit and it don't convert top synchros (I don't know if this is a problem with rgb signals). The pixel's timer is approximately 13Mhz, therefore I recommend OPAMPs with a 25Mhz range and a slew rate with 100V/µs minimum.

TEA5115 is obsolete and hard to find. OPA4354 is a quad OPAMPs and his price is 5€ on Radiospares or Farnell.

Here the diagram :





If you test report the results please.  ;)

RyleFury

Is nobody interested  ? ??? This mod works really great and the quality is far better than the mods with transistors since the voltage level and the impedance are perfectly respected. Anyway, I ordered triple OPAMPs (AD8013) with a 1000V/us slew rate to test without TEA5115. The circuit with OPAMPs is simple to realize, although the OPAMPs themselves are not always easy to find.

There is the possibility to make a similar circuit with the nes with RP2C03B graphic chip. The objective would be to find the white level in input to adapt it in output at 0,7V and to adjust resistors.

Link83

I think you have done great work :)  Its just unfortunate that its hard to find the TEA5115 for a reasonable price  :( and the OPAMP is untested as yet.

Really great work though, and please dont be put put off by the lack of replies - these forums dont always get the quick repsonses that other forums do, but I know from personal experience that theres alot of people out there who really appreciate the info in these forums but never post or reply to the threads  ;) I still get PM's about threads from years ago which never even got one response at the time.

papa_november

Is the quad op-amp available in through-hole? The datasheets I can find only mention surface-mount.

Also, would these chips work for RGB on the Turbografx/PC-Engine family as well?

RyleFury

The quad op-amp is surface-mount only, but there are a lot of op-amps which have a 25Mhz range and a slew rate with 100V/µs minimum. Of course, this mod could be compatible with every device. But we need to know the input level and adjust the resistors to obtain the proper level in output.
With an oscilloscope, this is possible to know this level knowing the white level (highest level of intensity = 0,7V).
An another solution would be to place a potentiometer to adjust the intensity level as you want.
The circuit in theory works, but I prefer to test it to make sure that really works.  ;)

RyleFury

I tested the ampli with op-amps. I had issues but I found problems. I used AD8013 triple op-amp, this is a very good quality ampli for video applications. I use input signals on the VDC-NUS directly, which are good clamped (blanking level at about 0V). However, my first issue was that without capacitor in serial in input, the op-amp don't amplify the signals correctly.  I placed 10µF capacitors, but there is a second issue. After the capacitor and before the IN+ on op-amp, the signal have an offset O_o. And this offset is greater if the power supply of op-amp increase  :-\

When the ampli-op is supply with 5V, the offset in output is greater that 4V, and the signals are x2 with an 1,4V amplitude max. The signal exceed 5V and is satured. I don't know the reason and this disturb me.  ???

When the ampli-op is supply with 12V, the offset in output is about 8V. Therefore, the signal don't exceed 12V and isn't satured. This is very important to place capacitors in serial on R,G,B signals to lower the offset to 0V in output.

Therefore, with 10µF caps in input, 220µF caps in output and a 12V power supply, the circuit works perfectly. However, if someone understand why there is an big offset, please enlighten me.

The good new is that the RGB Gamecube cable is perfectly compatible with this circuit and this is easy to find for 5€ or less. The bad new is the 12V power supply to find. On Nintendo 64, there not a problem. On Famicom, there is 10V and this is to test. On PC Engine, I don't know.

I found out all ppu chips send 0,7vpp signals without circuit in back. Therefore this circuit is compatible with all devices without modification.

Here circuit diagram :



Here compared signals :




viletim

RyleFury,

You are completely ignoring the "lage signal" characteristics of your circuit. To the extent that you are connecting 8V DC into your television's video input!!

The reason it doesn't work when directly coupled is that you have chosen an op-amp which doesn't have rail to rail inputs and outputs swing (here we're concerned about swing to -ve rail). The DC level goes wild when you attach a capacitor is series with the signal because you haven't provided any path for the bias current to flow.

Unless you want to add some extra parts, I suggest you find a more suitable op-amp.

RyleFury

I understand. It's dangerous to apply 8 V DC at the TV input if the caps are absent.  :-X

Although the op-amp isn't rail-to-rail, if there wasn't DC offset, the range with 5V (5V-1,2V for output range max) supply would be enough, and the input range isn't in negative amplitudes or more than -0,2V. When I place no capacitor, the signal in output is odd and isn't like input signal. It don't follow. For the bias current of cap, I didn't paid attention, thanks. I will try to remove the offset in this case !

This Op-amp is specially designed for this kind of applications  :-\

I do again tests to make work this circuit with 5V supply without offset or a little. I hope the circuit will not become a factory ^^'

viletim

Quote from: RyleFury on July 14, 2009, 10:57:44 PM
This Op-amp is specially designed for this kind of applications  :-\

...whilst being powered from a dual power supply.

You are simply hitting the limit of the output swing. To use this part with only a +5v source you will have to:

- Capacitively couple the input. (you did this)
- Add a DC reference voltage through a high value resistor to the positive input. This'll make sure the signal is within the input common mode voltage range.
- Intorduce another reference voltage, this time to the negative input to prevent the output from clipping on the positive rail.
- Capacitively couple the output to prevent the extra DC voltage flowing through the TV.



RyleFury

Ok, I understand all now. I already saw this kind of circuit.  I must add 2 resistors per Op-amp to realize that. But with a rail-to-rail Op-amp, I saw that the negative rail can go down to -0,2V, not less, the same as PMOS topologie. The rail-to-rail Op-amp allow to go up to max supply or more but not less that -0,2V of the min supply.

I will add an offset in my circuit, six resistors are cheap.

Edit :

I lowered the offset voltage without reference voltage. The impedance at the input of Op-amp is very high and the answer is this add a high offset. I thought that if I lowered this impedance, the offset could move. I tested with 12V and 5V, and with a 10K resistor lied to the ground, the offset in output is lowered to 1,4V and the Op-amps can be supply with 5V without saturation ! The problem is, the output isn't perfectly faithful with the input.

After I tested with a 220k resistor lied to the ground. In this case, the offset in output is to 2,3V ! But the problem with the 10k resistor is resolved and the output signal is identical to the input signal.

Therefore, the offset voltage can be adjusted with the input resistor and 220K appear be very good to me. I will modify the diagram.

Edit 2 :

The negative trail is limited to 1,2V minimum. I didn't think the two rails could be limited. I understand all issues now, but the present diagram works in the range of Op-amp :

Input min = 1,15V (limit)
Input max = 1,85V
Output min = 2,3V
Output max = 3,7V

The circuit would work with rail to rail Op-amps as viletim said. In this case, the 2 capacitors and the 220k resistor must be removed. The best example as I said above is OPA4354 quad Op-Amp. This component has all required conditions for this application.

kendrick

A reminder. As a best practice, if you're adding new content rather than fixing spelling or correcting an error of fact, you should add a new post instead of editing an existing one. A new post makes it clear that there's something new to read in the message order, and adds new entries in the RSS feeds. If you're worried that you're not getting enough responses, it may be because you've edited a post that's too old.

RyleFury

Very well, I  take advantage of this to add a new information. I noticed the input capacitors are bad biased. This isn't horrible but to guarantee the duration of capacitors and the reliability of the circuit, the positive side must be reversed toward the op-amp edge . I will change that.

I received OPA4354 op-amps, I will make the circuit with this op-amp and I will do a diagram. Because the problem is AD8013 is difficult to find. OPA4354 can be found on farnell, though the delivery service is expensive for a particular. The circuit will be easier to make with OPA4354. I didn't think but the op-amps are very different with the swing range caracteristic in input and output. Therefore, the circuit would be different with each op-amp because of offset setting...

Cyrix

i cannot find omamps for 25MHz and cannot find tea5115...
i do  not now much theory about  electronics so could you please tell me can i do anything with TEA2014A, because i could buy it here.

Cyrix

hard :)
i acn find only NJM4558L.
can i do the job with this?
thanx anyway

RyleFury

Hello,

TEA5115 is really hard to find because this component is obsolete. Sorry but TEA2014A will not operate with the circuit.

There is TEA5114 too (maybe easier to find) which works the same as TEA5115 without sync amplification signal (not important here).

NJM4558L is too slow for this circuit (1V/us slew rate). Moreover, according to the op-amp used, the caracteristics change and the resistors value must be modified (or removed for the input resistor) and for the capacitors too.

I wanted to test with OPA4354 but nobody seemed to be interested so I abandonned. Because I wanted to post that to help interested people...

However, if you succeed to find a fast op-amp and easy to find (slew rate > 100V/µs or even > 1000V/µs), I will make a diagram with pleasure.  ;)

There is always AD8013 which works great with this circuit but it remains expensive and hard to find.

EDIT :
On ebay, there are TEA5115, TEA5114 and AD8013  ;)

venchia3

Yes you are absolutely right, TEA5115 are present on ebay and I'm going to purchase one as well...they are not too pricey too...like 7 euros ;-)

I can't wait to try you circuit with this IC :D

cheers from Italy ^^

Cyrix

i bought tea5115 and i'm going to make an amp. tell me on pin 7 on ic what should i put, which resistor and which cap? it's not written what should it be strictly so on what it depends or it does not matter which value is putted? thanx anyway.

RyleFury

Hello Cyrix,

The value of the capacitor and the resistor isn't a problem as much as it corresponds at the range :
Resistor : 45 to 100 ohms (75 ohms per example as the others)
Capacitor : 10 to 470uF

These components allow to smooth the 12V supply, because the voltage must be very accurate for a good precision.
Therefore the value of the components doesn't change much of things in this range.

I forgot to precise, the capacitor have to support 15V minimum !

venchia3

Hello,

I have another question for you! ^^

Ok, I've an original nintendo GameCube RGB PAL Scart cable. The capacitor I have to remove for Red, Green and Blue signal are not connected directly to the scart pin, but the are soldered on a pcb board.

in this link:

http://www.mmmonkey.co.uk/console/nintendo/pal-snes-rgb.htm


It just says instead of removing the 3 capacitor, just join the 2 legs of capacitor together with a blob of solder, so they will not work. Is that a good working solution? Or is better to remove them and use wires to connect everything?

RyleFury

Do as on the photo but remove the capacitors before, or cut them.

venchia3

So, it's better first to remove capacitors and then link the 2 points right? Or cut them away...but I think is easier to remove them desoldering them and then connect with a piece of wire ^^

How did you do RyleFury?

RyleFury

The purpose here is to replace capacitors by wires. You have several solutions. The easiest solution is to cut pins of capacitors and join the cutted pins together as on the photo.

To prevent losses, the best solution would be to desolder the 3 wires (Red, Green, Blue) from the PCB input and connect them directly to the scart connector  instead of wires from the PCB output.

venchia3

Yes you are right! I didn't thought of desoldering the wires instead of the capacitors. So maybe is really better to take the 3 wires, desolder them from the pcb and connect directly to peritel ^^ thanks!

RyleFury


venchia3

I officially love you RyleFury!  :o ;D ;D

And want to give you credit for all the work you did, and are doing for the community! I asked you so many things and you were always patient and kind to explain me everything and now you helped me and the other with the schema of the stripboard circuit!

Thank you so much! I hope all the community can give you credit for the support, you really deserve it!

Grazie mille!!! (It's italian, just means thank you so much!)  :D :D :D