PSone sync problem

The place for all discussion on gaming hardware
User avatar
VEGETA
Posts: 425
Joined: Mon May 31, 2021 10:40 am

PSone sync problem

Post by VEGETA »

Hello,

I have a working PSone system (verified by video from seller with composite), and I got a SCART cable for it which I verified to work with another PS1 and PS2 systems with OSSC.

However, when I hook it up to the PSone system, OSSC keeps showing sync and not sync with no picture since monitor couldn't get the synced picture in time... it keeps like this but then settles in no sync.

I took my oscilloscope and took the measurements on PCB board itself + SCART pin 20 as seen here: https://slow.pics/c/3swk4P71

in the link you see picture name under it so you can know where I probed.

I still want to do the pure csync mod from pin 156 of gpu to pin 6 of multiout via the 220uf+470R but it ticks me off that this doesn't work despite signals seem ok, and the cable + entire setup works fine for another PS1 and PS2.

what do you think?

could it be that 220uF caps in PSone are bad? but composite video worked well with the seller.

EDIT: "cvbs_at_filter_cap.PNG" should be ignored, i took it from C550 instead of C551.
Last edited by VEGETA on Mon Dec 06, 2021 11:56 pm, edited 1 time in total.
User avatar
VEGETA
Posts: 425
Joined: Mon May 31, 2021 10:40 am

Re: PSone sync problem

Post by VEGETA »

my ossc is from aliexpress, firmware version 0.88. it worked perfectly fine for everything else.

i tried sync options such as h sync and v sync tolerance, sync threshold but none made any improvement.


I noticed that PSone multiout is a bit different than 7500 PS1 multiout, anyone can verify?
User avatar
Syntax
Posts: 1774
Joined: Wed Aug 09, 2017 12:10 am
Location: Australia

Re: PSone sync problem

Post by Syntax »

PAL PS1?
User avatar
VEGETA
Posts: 425
Joined: Mon May 31, 2021 10:40 am

Re: PSone sync problem

Post by VEGETA »

Syntax wrote:PAL PS1?
Yes PAL PSone, board is PM-41. it is the slim PS1 labelled PSone, I am sure you know.

is there a specific problem to PAL ones?
User avatar
Syntax
Posts: 1774
Joined: Wed Aug 09, 2017 12:10 am
Location: Australia

Re: PSone sync problem

Post by Syntax »

Pal PS1 and OSSC is the reason I purchased a Retrotink 5x

You will be hard pressed finding a TV that will accept the Pal refresh rate. Its lower than 50hz

314p 49.76hz from memory
User avatar
VEGETA
Posts: 425
Joined: Mon May 31, 2021 10:40 am

Re: PSone sync problem

Post by VEGETA »

Syntax wrote:Pal PS1 and OSSC is the reason I purchased a Retrotink 5x

You will be hard pressed finding a TV that will accept the Pal refresh rate. Its lower than 50hz

314p 49.76hz from memory
but my other fat PS1 is also PAL (as well as PS2) and it works fine.

the ossc itself doesn't sync to it, not the monitor.

isn't ossc supposed to be compatible with all these refresh rates? and this is PSone not a bizarre unknown console.

I hope someone provides a solution.
User avatar
VEGETA
Posts: 425
Joined: Mon May 31, 2021 10:40 am

Re: PSone sync problem

Post by VEGETA »

I have tried panasonic hack, modifying sync thresholds, but none worked. sometime without any cd inside, it was kinda stable on ossc but couldn't see a thing on my pc monitor. frequency is < 50 hz.

but why my fat pal ps1 works perfectly fine?
User avatar
Syntax
Posts: 1774
Joined: Wed Aug 09, 2017 12:10 am
Location: Australia

Re: PSone sync problem

Post by Syntax »

Fats get the subcarrier and everything else from one crystal, the slims do it over 3 clocks I think.
Could be that?
User avatar
VEGETA
Posts: 425
Joined: Mon May 31, 2021 10:40 am

Re: PSone sync problem

Post by VEGETA »

Syntax wrote:Fats get the subcarrier and everything else from one crystal, the slims do it over 3 clocks I think.
Could be that?
I don't know, but I am using RGB SCART for both. Not composite\s-video so no subcarrier frequency needed for encoding.

However, still, it will be sync-on-luma (or sync-on-composite) for both which has the frequency...

Will this be solved if I did a CSYNC mod for the slim? I mean getting the pure csync from gpu itself then wiring it to after capacitor C551 where the Y signal is (or CVBS signal, dunno). this is how csync mods are for PS1, all via C551.

Routing the signal to that place will be via 470 ohm resistor + 220uF elec. capacitor (usually the same C551 after removal).

what do you think?
User avatar
VEGETA
Posts: 425
Joined: Mon May 31, 2021 10:40 am

Re: PSone sync problem

Post by VEGETA »

I have an important update, please check scope images here: https://slow.pics/c/NAdZ1EYR to each there is the name under it, please check it.

now I have made csync mod, it works. sync is stable now and if not stable I can just use ossc sync threshold a bit and it gets stable enough.

However, there is a new problem which I think it should be easier to fix. Please check the sync pictures above to see how noisy it gets when you load it up with ossc via various methods such as ac coupling and dc coupling.

check the picture of screen here to see the new problem: https://imgur.com/a/stwCO7o and a video showing it here: https://streamable.com/at6dwy and here: https://imgur.com/a/SEXqtyt

I checked the schematic, it should not have anything else in the way really since I desoldered the original cvbs capacitor. but after the capacitor there is a ferrite bead, then the multiout pin which this is soldered to.

so the ferrite bead is in the way, but I did solder it directly at the pin so ferrite beat should be out of the circuit but still same problem.

anyone knows the solution?

notice that this should be a problem regardless of ossc itself since now the sync is kinda stable and monitor accepts it very well. I tried OSSC sync filtering options but none worked.
User avatar
NewSchoolBoxer
Posts: 369
Joined: Fri Jun 21, 2019 2:53 pm
Location: Atlanta, GA

Re: PSone sync problem

Post by NewSchoolBoxer »

Super impressed you have oscilloscope pics. I'm hazy on your exact setup and I don't use PAL but maybe I can help.

What you should do is use the oscilloscope's FFT on the noisy signals and see what frequency range the noise is on. Compare the Total Harmonic Distortion and Signal to Noise Ratio if given to good signal of same type. If high amplitude around 50 Hz then possible noise sources are PlayStation audio crosstalking inside the SCART cable, nearby power line transformer or your console's power source. I specifically had a noisy 9V DC power source for my Super Famicom that was causing very similar scrambling. DC is 0 Hz right but it's generated from a AC source and a low quality or very aged transformer in the block can have too high ripple voltage. Ripple voltage is pure noise and heat to a DC circuit.

Could instead have dying voltage regulator inside the PlayStation but in that case I'd think the console just wouldn't power on.

Is PC monitor what you're feeding OSSC HDMI to or is it CRT monitor you're feeding SCART or VGA? I've never heard of a PC monitor that accepted sync over luma or the ps1 stock composite video as sync. You did csync mod so resolved that potential issue in any case.

Yes, the 220u capacitors could be bad. They typically don't just switch from working to not working. They gradually lose 10% or more of capacitance, the series resistance increases and they lose the ability to handle the rated max voltage. The resistance and capacitance values on the R, G, B lines each form a time constant and that going off spec from aged capacitor can chain react into interference. At 1V analog video and 1V (75 ohm) sync, at least you don't have to worry about overvolting. Not sure if sync line has a capacitor.

You mention the ferrite bead. That exists to suppress high frequency noise. You should keep it in the circuit. I assume that you do not have 220u capacitors in the RGB console since they are already in the cable you bought for it and if you doubled up, effective capacitance becomes out of spec 110u. PS2 comes with capacitors inside it so its ideal RGB cable has none. Edited since I mixed up PS1 and PS2.

Also my official Sony PS1 JP-21 cable does not have a capacitor on the sync line. The ripple could be due to the capacitor you added that is trying to cut off the DC component of the square waves. This is NTSC however.

Yeah, your video and scope pictures show a whole lot of electrical noise. Check if composite video and s-video (if you have cable) look clean or not. If you have a way of unplugging just the R or G or B then can try to test if the noise is stemming from one of the color lines.

Finally, don't have stereo speakers or internet router or some other powerful electronic device on and running right next to the console or television. Speakers if shielded could be fine but at lease rule out the easy things. If you removed the metallic EMI shielding blocks to mod the console then you need to put them back on.
User avatar
VEGETA
Posts: 425
Joined: Mon May 31, 2021 10:40 am

Re: PSone sync problem

Post by VEGETA »

Hello,

here are my comments:

1- I bought the scart cable from buyee. tested it in fat PS1 and fat PS2, it worked perfectly fine for both of them out of the box. cannot open the scart cable without destroying it, but no need since it worked with another PS1.
2- my setup is messy but worked fine for fat PS1 and fat PS2. Also, just tried moving the PSone away with everything, still same issue.
3- Power supply is fine I guess, I measured its ripple and noise and was ok. still, the csync from cpu is without any noise but when you load it, noise appears.
4- PC monitor is having OSSC signal fed into it, no CRT or anything else. OSSC via HDMI, and works well for the other PS1.
5- tried swapping 220uF caps with new through hole ones, still same results. tried hooking csync directly without caps (dc coupling) with and without limiting resistor.. same result.


So right now I am not sure what to do since nothing seem bad, yet the result is very bad. I have composite cable but now it only offers csync, ossc doesn't accept composite nor my monitor. I could hook it up to the component input of the ossc and measure the signal on the scope... however, as said before, scart cable worked fine with the other ps1 so no use doing this test to check it.
User avatar
NewSchoolBoxer
Posts: 369
Joined: Fri Jun 21, 2019 2:53 pm
Location: Atlanta, GA

Re: PSone sync problem

Post by NewSchoolBoxer »

Cool thanks for breaking down the issues for me and checking power supply. I think the root issue isn't from noise but rather from the mod + Scart cable expectation of what it's being given. I made minor edits for more clarity.

Conclusions

I see in your pictures that GPU csync is very clean but raw level is too high. You use resistor + capacitor to lower to ~340mVax and 960mVpp. Fine strategy but most people only use a large resistor to attenuate TTL sync down to 75 ohm level. You may not need the capacitor. I'm sure 960mVpp is too high, noise aside. I see 300mVpp and 350mVax being given at http://forum.arcadecontrols.com/index.p ... c=158601.0. Sorry I only know TTL sync that GPU here outputs that PVMs and computer monitors accept but you're aware that OSSC won't accept TTL on Scart input (nor does default Scart television).

Other issue is GPU using PAL horizontal sync voltage of 15.59 kHz whereas you're using much too high (!) 16.63 kHz. By definition, NTSC is 15.754 KHz and PAL is 15.625 kHz. Analog devices like CRTs are fairly tolerant but digital HDMI can smack down anything out of spec. Like 15.59 kHz is probably acceptable but not 16.63 kHz. See here: https://www.renesas.com/cn/zh/document/ ... nalog-hdtv By contrast, 15.0 kHz is too low. Happens from people using that with Emudriver since they just see "15 kHz" mentioned and roll with it.

In college I used Laplace transform to analyze AC signals. I could get something like 59 Hz output from 60 Hz input due to net impact of resistance, inductance and capacitance. You also have concept of power factor that messes up the phase but, basically, either the electromagnetic interference and/or impedance mismatch and/or R and C values you chose and method of construction are throwing the vertical sync frequency too far off. It follows that some part I just mentioned is adding too much interference.

Noise is already there on the pin you measure. You can easily add noise by running csync cable too close to noisy chips on the console. The Scart video cable isn't expecting csync nor is necessarily using pin 6 composite video. Could be using pin 5 luma. May have an unnecessary sync stripping chip in your case that runs off 5V from pin 10.


Cable Discussion

Let me show you the official Sony SCPH-1050 JP-21 PS1 cable: https://imgur.com/a/1Qn0wo9
I think most people know this but just in case, JP-21 is basically a pin remapping of Scart without some of the odd features that the Japanese market didn't need. I haven't seen JP-21 combine audio and video.
Not extremely helpful for analysis in that it uses composite video as sync but you can see a single 75 ohm resistor for proper video line termination to ground and the RGB 220u capacitors with limit of 6.3V.
I doubt RGB Vpp gets much above 1V.

You fortunately have a scope so can reverse engineer what's in the Buyee Scart cable even if you can't open it without damaging. Figure out which pin it's using for the sync input and output. It's definitely not expecting csync since that doesn't come stock. You're saying cable works on another PS1, does that have cysnc mod too? I need csync to my RGB capture card but I paid for Renesas ISL59885 sync stripper chip inside the cable that uses the luma input pin that gets re-routed to sync output pin. Extron devices need cysnc or sync on green too but I think csync is otherwise overkill unless you're compensating for badly shielded cable.

Mod Discussion

PlayStation 1 and 2 csync mod is pretty niche since most people like me pay $10 for a sync stripper chip without risk of mod going wrong. I found one detailed PS1 mod guide for SCPH-7502 model: https://psx.arthus.net/mods/csync/PSOne ... ur%20X.pdf
Avoid soldering the GPU chip directly which I definitely agree with. Actually has instructions and diagrams very far above typical mod standards. Important parts:

"However because this is a RAW C-Sync Signal I needed to add some internal components before I got a Stable\Clean C-Sync Signal, a resistor and electrolytic capacitor did the trick perfectly...I disabled the V-Sync Signal going to the PSOne's AV Port and replaced it...with C-Sync."

Four X the author is pulling csync from a pad that pulls from pin 14 on A2106R video encoder chip and disabling the vsync via trace cut that goes to PSOne's pin 6 multiout - the composite video pin. I assume this doesn't ruin composite video - just the ability to send useless vsync there.

Instructions show a diagram of series circuit going [video encoder pin 14] -> [75 ohm resistor] -> [(+) 220u capacitor (-)] -> [multiout pin 6]
Author states Large and Slim PS1s have almost identical GUP pin to pin compatibility so mod will work on others. States to confirm pin 156 of PlayStation GPU is indeed RAW C-Sync. Note that this should preserve TTL sync level and not the 75 ohm level you want.

Did you do all this or are certain you did the equivalent correctly on a different PS1 model? I assume you got the polarity on the capacitor right since it basically explodes if you don't. I'm not sure the resistor-capacitor circuit is at all optimal. 75 ohm resistor should be in parallel on a video line but sync could be different. I guess design is fine since author tested but may not be in your 75 ohm case. Could try 75 ohm resistor going to ground and another resistor in series high enough to drop sync to around 300 mV. Then test removing 75 ohm resistor.
User avatar
VEGETA
Posts: 425
Joined: Mon May 31, 2021 10:40 am

Re: PSone sync problem

Post by VEGETA »

Hello,

I pulled Csync from the pads near the video encoder, the big pad which is a bit further. checked with oscilloscope and it is correct.

the author of that guide is confusing the reader by not showing how he derived or see the numbering. for correct numbering see this: http://poofycomputers.com/ps2vga/pics/ps2-pinout.png

schematic is also showing it coming via C551. check here: https://raw.githubusercontent.com/led02 ... iagram.pdf

also he says vsync but vsync doesn't go to the multiout at all. it is actually the composite video labelled as M_VIDEO or something.l

also, the schematic shows a series 75R resistor via a 4-resistor package, this 75R ohm resistor now out of circuit but we already added 470 and 1k resistors in series. I feel like noise has nothing to do with sync level since ossc already accepts it but rather the noise imposed on the sync signal itself right after plugging it in.

I guess OSSC expects sync on scart pin 20, regardless of its source. pin 20 on this scart cable is linked to composite video out pin of the multiout which is now csync. I don't think we should try isolating it then jumper it to S-video Y signal which contains sync... there has to be a better way to use pure csync.

I used 470R and 1k resistor values with and without 220uF but never reached 330mV.

what to do now?
User avatar
Syntax
Posts: 1774
Joined: Wed Aug 09, 2017 12:10 am
Location: Australia

Re: PSone sync problem

Post by Syntax »

What's the 1k resistor for? Did you use it instead of the 470r just to test? If so 1k is way too high to attenuate a 3.3v signal.

470r should give you around 0.45vpp
330r is around 0.61vpp
1k would be 0.23vpp which is below the sync level threshold of 0.3vpp

I use 330r to keep the signal on the hotter side of things, but 470r is fine.

Not sure how you are testing the sync levels but make sure you are not skipping the termination resistor in your receiving equipment.

If testing sync level without the cable plugged into a TV or something add a 75r termination resistor to the circuit.
User avatar
VEGETA
Posts: 425
Joined: Mon May 31, 2021 10:40 am

Re: PSone sync problem

Post by VEGETA »

Syntax wrote:What's the 1k resistor for? Did you use it instead of the 470r just to test? If so 1k is way too high to attenuate a 3.3v signal.

470r should give you around 0.45vpp
330r is around 0.61vpp
1k would be 0.23vpp which is below the sync level threshold of 0.3vpp

I use 330r to keep the signal on the hotter side of things, but 470r is fine.

Not sure how you are testing the sync levels but make sure you are not skipping the termination resistor in your receiving equipment.

If testing sync level without the cable plugged into a TV or something add a 75r termination resistor to the circuit.
yes as a test. but most of the tests are by 470 ohms. none of the options gave the solution, all noisy for some reason.

what do you suggest doing now?

I am really confused why this noise happen when connecting it to ossc.

I plan to put a resistive load tomorrow, like another 470ohm or 75 ohms from pin to ground to see if the noise is due to connecting to ossc or loading in general.
User avatar
Syntax
Posts: 1774
Joined: Wed Aug 09, 2017 12:10 am
Location: Australia

Re: PSone sync problem

Post by Syntax »

Why would you even consider using a 470r termination?

Voltage dividing with resistors of the same value just halves it, so you'd be reading a 1.65vpp signal.

The old dumb posts say use 75r inline and 75r termination and have no idea...
User avatar
VEGETA
Posts: 425
Joined: Mon May 31, 2021 10:40 am

Re: PSone sync problem

Post by VEGETA »

Syntax wrote:Why would you even consider using a 470r termination?

Voltage dividing with resistors of the same value just halves it, so you'd be reading a 1.65vpp signal.

The old dumb posts say use 75r inline and 75r termination and have no idea...
what I meant is using 470 ohm + 220uF capacitor as usual for getting csync but instead of connecting it to ossc, I will get a 75r to ground to act as a load. 75r is the same as any sink device termination resistance.

this test will let me see if the problem is from any load or just the ossc. if the 75R load to ground didn't produce noise then ossc has the issue. in any case it would still be weird since exact same setup using fat ps1 works perfectly fine.

another test to do is disconnect the csync circuit (currently via a wire), and connect S-video Y signal to pin 6 (which is the scart pin 20) to act as sync on luma... and see the result in ossc.

do you have any other suggestions to do?
User avatar
Syntax
Posts: 1774
Joined: Wed Aug 09, 2017 12:10 am
Location: Australia

Re: PSone sync problem

Post by Syntax »

Check that all grounds are linked on either end of the scart cable is all I can think of.
User avatar
VEGETA
Posts: 425
Joined: Mon May 31, 2021 10:40 am

Re: PSone sync problem

Post by VEGETA »

Hello

please check images here:

Power supply: https://slow.pics/c/Pzid6bK8

PSone + fat PS1: https://slow.pics/c/y3ti6i0I < only 2 pics for fat ps1, the rest are slim PSone.

summery:

- power supply seems ok.
- I accidentally burned a fuse, jumpered it and it is ok. will solder new fuse soon.
- no configuration seem to have any effect what so ever on result.
- I feel like fat PS1 images are bizarre... both seem similar despite one is pin20 and the other is color channel.
- tried 1000uF ac coupling cap but didn't work.
- tried doing FFT but it is my first time and I don't know if it will benefit, please check image.
- as you can see from images, the ugly vertical lines of noise are always on the signal when it is above 0v. ALWAYS.


I want to re-solder CVBS cap as before to make it like original state... then test it on a flat tv which supports cvbs signal... will see what happens. seller sent a video showing this to work perfectly fine on a flat TV using CVBS.

Another kinda crazy solution came to my mind which is: use LM358 op-amp to buffer pure csync from CPU, THEN hook the 510R + 220uF cap to multiout pin.

This way it won't be driven directly by CPU pin if this has anything to do with the problem, essentially isolating the entire circuit from output. I only have LM358 but it is 1 MHz which is fast enough for 15 KHz sync signal and slow enough to never get any noise beyond 1 MHz. a better solution which is a schmitt trigger circuit but this is beyond me now, no parts to do it... only LM358.

do you think it could work? this was my out-of-the-box move since all moves within the box failed.

I can only think of scart cable problem specifically for PSone but again, it worked perfectly fine with fat PS1 and fat PS2.
User avatar
NewSchoolBoxer
Posts: 369
Joined: Fri Jun 21, 2019 2:53 pm
Location: Atlanta, GA

Re: PSone sync problem

Post by NewSchoolBoxer »

Thanks Syntax for jumping in. I'm more knowledgeable about general electronics than specifically analog video for video games. Someone deeper down this niche can give you better advice.

That's funny how the guide is inaccurate on vsync going to the composite video pin and in other respects. So you know about 75 ohm to ground to act as load and concept of terminating resistor. I don't have to explain transmission theory or impedance matching. But impedance mismatch from your mod is a possibility and that results in power loss. Video chip soldered to wire, wire soldered to multiout pin and multiout pin to cable and cable to TV is 4 connections where impedance has chance to mismatch.

Seems like good idea to send Y from S-Video to pin 6. If that gives clean signal then you can be confident something is done wrong in the PSOne in terms of soldering joints or electromagnetic interference from different section of console or your added resistor/capacitor that throws off the impedance matching. Per your csync line, you want to be using the same ground reference in all parts of the circuit. Is probably what you're doing but can easily verify with multimeter continuity test.

Using op-amp isn't crazy idea at all. It's a mainstream thing to do to provide electrical isolation and adjust the gain and/or apply a first order filter. You don't see it in mods because it needs positive and negative voltage sources and we only get +5V for free. You could drop the resistor/capacitor and instead use an inverting op-amp with a gain < 1 to attenuate the voltage down to 75 ohm level and add a smoothing capacitor in parallel to Rf. Issue is you also need an inverter before or after. If you don't have that, can do non-inverting op-amp with resistor + capacitor after like you're saying. This site has good explanation with linked circuit simulators: https://ultimateelectronicsbook.com/op- ... amplifier/

Power supply looks good in first pic if the red is the 7.5V DC output with very small ripple voltage. Doesn't rule out the voltage regulator the 7.5V feeds to but if S-Video or else another video source gives good output then all good.

--------------------
FFT Description

FFT in green shows no noticeable interference on 50 Hz or 60 Hz power supply, which is nice to rule out, but you should expand the bandwidth in the chart to 30 kHz or more. What you see on x-axis is frequency and y-axis is power in decibels against a reference voltage. I found this image off of Diligent tutorial: https://ibb.co/cbYHTsC Get graph to look like that.

The function generator makes a bad square wave at 1 kHz as you can see. 1 ms is whole period of rise and fall and 1/ms = 1 kHz. The ripples contribute to the bottom picture of the FFT. You see the highest peak of the fundamental frequency 1 kHz. An ideal square wave only has odd harmonics so you see peaks at 3 kHz, 5 kHz, 7 kHz and 9 kHz. A better square wave with flatter "square" would have those peaks be much lower and scale down more quickly.

A 15 kHz square wave would be expected to show harmonics at 45 kHz, 60 kHz, etc by multiplying fundamental 15 kHz by 1, 3, etc.

Fundamentally, what all this means, is the area under each peak is the power of the signal around that frequency. What is killing your sync is electrical interference around the frequencies it uses. Noise at, say, 1 MHz won't matter and would be invisible here.

What you can tell with an FFT is what frequencies are interfering the most with your signal and use that as a clue to determine the source. You could in addition, add a filter to block out those frequencies but easier said than done when they are inside your bandwidth. If you take the signal's power and divide by the noise's power then you get the Signal to Noise Ratio (SNR). Higher is better. I think composite video runs around 25 and RGB can hit 30.

FFT cuts extremely small slices of your signal so can show interference at frequencies that aren't obvious in the time domain with time on x-axis and voltage on y. You almost always want FFT x-axis to show frequencies on a log scale to be easier to interpret.

Per your pictures, that's bad times but very telling you get noise on the signal without it plugging into your cable. At least if I'm understanding the pictures correctly. Cable could still have an impedance mismatch but that's of much lesser concern. I can estimate the frequency of interference. So about 11 spikes in 20 x 20 us interval = 40 / 11 us = 3.63 repeating period. 1 / period = frequency = 275 kHz. Seems too high so maybe is a fraction of that or sync signal really does operate around higher kHz. FFT would tell you exactly what's going on and then you can lookup what PSOne chip may be operating at the main noise frequency. If it's a chip then can add shunt capacitor network to it to suppress the interference, or I guess to your mod circuit. I don't get fat PS1 pics either.

At least I see sync square wave frequency is fine at 15.59 kHz.
User avatar
VEGETA
Posts: 425
Joined: Mon May 31, 2021 10:40 am

Re: PSone sync problem

Post by VEGETA »

thanks for your explanation.

I will try to do the fft again with more frequencies, my scope is owon vds-1022i.

check the pics I posted, you will find 2 pics for fat ps1. check the name of the picture under it to know what is it for.

putting s-video on pin 6 didn't do well with ossc, didn't recognize the signal at all despite having compatibility with sync on luma.

I am not too familiar with impedance matching, besides doing length + impedance matching for high speed pcb designs. didn't really dig into it much, but from what i see, people do such simple mods always with perfect results.

what annoys me is that there is no noise or ripple anywhere except for that video output part of the circuit, as you saw the psu stuff is good enough. 275 khz maybe the switching frequency of the power supply ic TL594 which is used to regulate 7.8v to 3.6v or so via controlling a pnp transistor. please check schematics to find it out. TL594 pwm frequency is 300 khz. however, after 22uH inductor, 3.5v is without much ripple.

using the opamp? hmmm psone schematic doesn't show any negative rail, plus gpu outputs sync as ttl which has 0v as negative rail. I checked the schematic, even video encoder has "VID Ground" to be the same ground as other stuff... no negative rail, all connected to the negative pin of the input power supply. so it should be ok to connect lm358 to cpu csync as I explained, then use the resistor + cap to ac couple it to multiout. I guess ossc and other devices can tolerate dc coupled sync signals as well.

since connecting s-video didn't work as sync, what should I do? I will do the fft tomorrow but no hopes to solve the issue. as you saw, the sync signal is going there properly but it has those distortions.

all of this is can indicate a faulty PSone and cable but I tested the cable with other ps1 and the seller showed me a video of PSone working with composite.

can't thank you enough for keep going on this with me.
User avatar
NewSchoolBoxer
Posts: 369
Joined: Fri Jun 21, 2019 2:53 pm
Location: Atlanta, GA

Re: PSone sync problem

Post by NewSchoolBoxer »

You're welcome, is all an interesting thought experiment for me. Sorry, I should have spent a minute to look up the LM358. It's a "modern" op-amp that only needs positive voltage and has two op-amps in one chip. It caps at 1.2 MHz which means it's perfect for sync but too low to pass in luma or chroma or composite video.

Can feed the Vcc pin with the +5V PSOne pin or 2x AA or AAA in series or a USB power source, etc. Way you could attenuate TTL sync into 75 ohm sync is wire (+) inputs on both to ground, make the first op-amp have gain of (-1/10) [or other < unity gain] and feed that output pin with a resistor as the (-) input of the other op-amp. Give that op-amp a unity gain of (-1) by making Rin = Rf. Gains of stages multiply so (-1/10) x (-1) = (+1/10) to make a 3.3V TTL level into a 330mV 75 ohm. A little more complicated to make one or both op-amps also work as filters. You would want to see the good sync output from video chip under FFT to know which frequencies to pass besides the obvious ~15 kHz. I'd just guess a low pass filter with cutoff frequency of ~60 kHz would be good.

I found the 40 page user manual for vds-1022i. Its FFT ability is more than what you need. The ic TL594 datasheet, it's capable of "a wide variety of switching converter applications over a frequency range of 1 Hz to 300 kHz" so it could be configured to add 275 kHz noise. Could probe the output after the 22uH inductor into your scope and FFT it but if you've already seen little ripple then is unlikely to be the noise generator that's ruining csync.

That s-video didn't work as sync, uh PSOne s-video still gives good video on television right? Should look at where your mod is sitting near PSOne chips and can move the wiring around. Really ghetto EMI shield is wrapping your wiring in aluminum foil so long as you're careful not to touch any conductors in PSOne or your mod circuit. Could wrap the aluminum foil with masking tape to insulate if you had to. Aluminum tends to be good blocker of high frequency noise while steel is good blocker of low frequency. Hm got any steel plates to wrap around with aluminum foil?

Other possible causes that I doubt are mod bad solder joints or loose connection to ground. More plausible is video-related capacitors in PSOne that need replacement that chain react to mess up sync but you have video footage of PSOne working okay. Could be only composite video works after you reconnect it?

Electromagnetic Interference Discussion

Yeah plenty of people do simple modding and ignore impedance matching and 75 ohm termination until they can't. Tim W. aka viletim of NESRGB fame posted good explanation: viewtopic.php?f=6&t=55948&p=1153713#p1153713

The "I wrote more about it here" science experiment thread he links to I can elaborate on. An AC circuit generates a changing magnetic field around it. This changing magnetic field will induce a voltage around nearby conductors, meaning another circuit that isn't sufficiently shielded. We refer to as "crosstalk" when one wire in same cable interferes with another. Classic examples is 50/60 Hz audio buzz on Scart.

What Tim's doing is using a signal generator aka test pattern generator on the R1 line that induces all the voltage you see on R3. The yellow is the generator on R1 and the teal is the signal generated on R3.

That 2 meters is a "reasonable" cable length and 3.5 meters is "excessive" goes back to the ratio of cable length divided by wavelength. If we go with a 4.43 MHz (PAL) color subcarrier and typical velocity of propagation of (2/3) speed of light then wavelength is ~45 meters. All sources agree that a ratio of 1/4 or greater is bad-bad-bad, you'll 1/10 in some sources and I've seen 1/200 as the lowest possible ratio that is susceptible to signal damage from impedance mismatch or electromagnetic interference. Well, that's a total cable length range you can "get away with it" of 11.25 meters to 4.5 meters to 22.5 centimeters (36.9 ft to 14.8 ft to 8.9 inches).

We're looking at interference at 16x lower 275 kHz = you get 16x more allowable cable distance. All seems well but it's not. The amount of damage is greater the stronger the interfering signal, so apparently the noise sources is extremely strong OR the impedance mismatch is extremely bad or some combination of the two.
User avatar
VEGETA
Posts: 425
Joined: Mon May 31, 2021 10:40 am

Re: PSone sync problem

Post by VEGETA »

hello

please check images here: https://slow.pics/c/lFfWCb9l

what I did:

- used LM324 since it is more solder friendly.
- only one op-amp as voltage follower without feedback resistors. csync from gpu to opamp + then from opamp - to opamp output, from opamp output to multiout directly or via 510R resistor or via 510R resistor + 68nF ac coupling cap as in images.
- results are not as we expected, since opamp output is deformed even without load.
- opamp is powered by 5v and decoupled with 68nF ceramic + 220uF elec (I had it around so I put it).


check and advise if you know better thing to do.
User avatar
Syntax
Posts: 1774
Joined: Wed Aug 09, 2017 12:10 am
Location: Australia

Re: PSone sync problem

Post by Syntax »

Did you ever check your questionable Buyee scart cables grounds/pinout when plugged into the PSONE av multi out?

You are overthinking this issue, posting a plethora of scope reading and going into deep theory wasting time. Just beep the scart cable out with a multi meter.

Its going to be a simple problem but you started removing and modding the system before even checking if composite worked so thats a pain.

I still have my money on a crappy ground, maybe a dirty AV port. OSSC would drop sync constantly if you had no grounds. Unless the PSone is chipped also.
User avatar
VEGETA
Posts: 425
Joined: Mon May 31, 2021 10:40 am

Re: PSone sync problem

Post by VEGETA »

Syntax wrote:Did you ever check your questionable Buyee scart cables grounds/pinout when plugged into the PSONE av multi out?

You are overthinking this issue, posting a plethora of scope reading and going into deep theory wasting time. Just beep the scart cable out with a multi meter.

Its going to be a simple problem but you started removing and modding the system before even checking if composite worked so thats a pain.

I still have my money on a crappy ground, maybe a dirty AV port. OSSC would drop sync constantly if you had no grounds. Unless the PSone is chipped also.
I already wrote that this particular cable tested on fat PS1 and fat PS2 and worked perfectly fine. How can this be an issue in the cable after this?

I have tested continuity of grounds, here is the result:

PS1 PCB: pin 8 which is ground, is connected to ground. as well as pin 1 and 3 which are audio grounds.

SCART cable: only pin 4 (audio ground) and 18 (blanking ground) are connected to ground while blue\red\green\sync (cvbs) grounds (pins 5/13/9/17 respectively) are not. tested the continuity with pcb ground and multiout pin 8.. same result.

However, why fat PS1 worked?! despite these issues??

both PSone and PS1 (and PS2) have modchips installed. didn't affect anything since even PSone displays stuff but has tons of noise as you saw.

Update: I have managed to open the scart cable safely.. here are the pics: https://slow.pics/c/U6xKHLaA

- it has 220uF ac coupling caps on RGB lines.
- color grounds are not connected as mentioned.
- has 180R resistor from pin 16 to pin 8.
- 220uF elec caps seem ok via multimeter check (not so accurate but showed good results).

what do you think?

I always return to idea that this same cable worked with a PS1 fat and also a fat PS2 as is!! I will solder all those grounds together tomorrow and see the results despite reading that they should be left unconnected inside scart cable itself. it is beyond 2:00 Am now and I have a day engineering job.

Please think of the last issue about why it worked on one system not the other.
User avatar
Syntax
Posts: 1774
Joined: Wed Aug 09, 2017 12:10 am
Location: Australia

Re: PSone sync problem

Post by Syntax »

A dirty AV port could cause the bad grounds but still work on a clean PS1, and if the scart cable only wired one of the 3 available grounds on the PSone side then you a relying on that single pin to be clean and conductive.

Ground surface area should equal all signal surface area combined. (or close enough)
User avatar
VEGETA
Posts: 425
Joined: Mon May 31, 2021 10:40 am

Re: PSone sync problem

Post by VEGETA »

Syntax wrote:A dirty AV port could cause the bad grounds but still work on a clean PS1, and if the scart cable only wired one of the 3 available grounds on the PSone side then you a relying on that single pin to be clean and conductive.

Ground surface area should equal all signal surface area combined. (or close enough)

I couldn't get continuity reading from PCB to multiout, maybe my multimeter probes couldn't reach the end inside multiout due to being big. tried cleaning it too.

However, I connected scart cable itself and verified that all pins are reaching from pcb to scart pins. but as mentioned, only pins 4 and 18 are actually connected while the rest are floating. pin 4 and 18 are the only grounds.

so how can the video color and sync signals return if their scart ground pins are not connected? I know maybe grounds are all connected together on ossc but still feels weird that inside scart cable these grounds are floating... not connected to anything at all.

tomorrow I will try connecting them together and to proper scart ground which will directly connect them to pcb ground, effectively solve any grounding problem if existed.
User avatar
NewSchoolBoxer
Posts: 369
Joined: Fri Jun 21, 2019 2:53 pm
Location: Atlanta, GA

Re: PSone sync problem

Post by NewSchoolBoxer »

My understanding of PAL sync is better now. The correct level is 300mV, which is exactly the sync level in composite video and luma that made sense for Scart to use. A little higher or lower is acceptable and I found one forum thread saying a low 150mV will work for OSSC. I don't want to link it since thread was wrong about TTL sync but basically the 3.5V the PSOne outputs is acceptable.

For either sync type, best timing I could find for NTSC is the square wave high voltage period is about 58.9us and the 0V pulse width is about 4.7us with +/- 0.2us tolerance on the pulse. The small amount of time left out of 64us NTSC scanline (PAL is 63.4) could be taken up by the time to raise and lower the voltage. The speed of this is called a slew rate.

I thought you would just setup the LM324 on a breadboard to test with changing resistors and capacitors but you're sending straight csync into it. Like you can do that as you're sending safe voltage and current levels and achieving a form of electrical isolation that you want but I never thought of using an op-amp without setting up a filter or gain or both.

First pic opamp1: Good PSOne TTL sync in red, sync after op-amp in yellow. What we have to consider is the slew rate of the op-amp at 0.5V/us. Going from 3.5V to 0 or 0 to 3.5V therefore takes 7us. That's great for $0.49 but it doesn't work here due to the 4.7us pulse width. That's why you see the voltage only drop ~1.5V - it doesn't have enough time to get to 0 before the spike back up to 3.5V starts. My advice is to add the resistor + capacitor before the op-amp, giving it enough time to go from 0 to 300mV and back down in 0.6us. Probably better to go with 150-225mV that OSSC accepts to get closer to 4.7us. Again, I debate the capacitor but it's okay to use to remove any DC offset since the resistor should be high enough value to keep the square wave intact.

You can get an op-amp with an impressive 100V/us slew rate for $5-6 but that is overkill if the LM324 can suffice.

Second and third pics opamp2/3: I think the 40mV red noise is just from the oscilloscope and is fine and typical. So we're saying the noise here on yellow is without you giving it sync or anything at all. Pure noise response. But PSOne is plugged in and running right and no noise when it is off?

Third and fourth pics opamp4/5: The 0V line is the vertical blanking from being at the end of a frame. Again we see the yellow sync doesn't get to 0V like it needs to for OSSC or TV to trigger on horizontal and vertical sync pulses. We also see the noise on 0V vertical blank which is what we expect from earlier pics.

Seventh, eighth and ninth pics opamp7-9/5: Thanks for doing FFT on noise. That delta you do on 8 is really helpful that shows noise is spaced at 15 kHz increments. In other words, it suggests the fundamental frequency of the noise is ~15 kHz. Indeed, you see the peaks get a little lower as you go from left to right like in the square wave FFT I linked. I would say the first seven peaks are significant and above the "noise floor" so that's about 15-105 kHz bandwidth.

That the horizontal sync has fundamental frequency of ~15 kHz most likely means it's causing its own interference. Unless something else nearby operates at that? It's just a whole lot greater in amplitude than the cable length or signal strength could predict. You can't add a filter that removes what you want. The noise can be self-induced or it can come from reflections from poor impedance matching or soldering joint or not using the same ground reference or a combination. I didn't see a 75 ohm resistor tied to ground in the Scart head, which I definitely saw in official Sony PS1 JP-21 cable and Nintendo's for SNES. I guess you could terminate the sync with 75 ohm resistor in parallel to ground right before the pin the Scart cable connects to. You could do it at more than one connection.

Steel is good low frequency shield like I was saying. Copper is a little worse than aluminum at high frequency. I think copper and steel match potency around 100 kHz. I suppose really ghetto EMI shield is using masking tape on coins and paper clips then wrap that around aluminum foil, then wrap that around your sync line. Don't let any gaps in the shield. Otherwise two rows of steel staples that form a square hold when stack in opposite ends might be okay.
Last edited by NewSchoolBoxer on Sun Dec 12, 2021 5:22 am, edited 1 time in total.
User avatar
NewSchoolBoxer
Posts: 369
Joined: Fri Jun 21, 2019 2:53 pm
Location: Atlanta, GA

Re: PSone sync problem

Post by NewSchoolBoxer »

Post was getting too long. Let me give you mini guide to making an active 2nd order low pass filter with the op-amp.

Op-amp 2nd order low pass filter

In modern no one hand calculates filters. Easiest way without really knowing what you're doing is Sallen-Key topology to form a 2nd order Butterworth filter. The THS7374 modders love uses a 6th order Sallen-Key Butterworth filter made out of transistors. Don't need to be concerned about inverting op-amp, a gain of 1 here is insignificantly greater than 1 and thus no inverter is needed to keep voltage positive.

Go to this website and plug in the cutoff frequency and the R or a C value. I think 75 ohm R is best for impedance matching here but if all you got is 1k ohm or whatever, that'll do: http://www.calculatoredge.com/electroni ... 20pass.htm

You may want 100 kHz cutoff to not degrade the square wave. Need 15.59 kHz way below that and need to protect its harmonics too since they sum to form a square wave. The cutoff frequency is defined to be where 50% of the signal power is wiped out, equivalent to half the square root of 2 of the voltage, which is right at 70% of the original value. Reason is voltage^2 / 2 = power if we're not concerned about current (we aren't). Anyway, should get 30nF for Ca and 15nF for Cb. Nice round numbers. Capacitors in parallel add so can form 15 with a 10 and a 5.

Plug in values at this Japanese site (in English) and use 1k for 1000 ohms, 30n for 30nF, etc. The C1 is Ca and C2 is Cb from before: http://sim.okawa-denshi.jp/en/OPstool.php You should see:

fc = 100035.14623968[Hz]
Q = 0.70710678118655

So 100k kHz is cutoff frequency and half the square root of 2 Q factor. Don't worry about what Q factor means, just that it's the right value for Butterworth filter. Usually can't get that close to the cutoff you want with actual R and C values that exist. Scroll down to Frequency analysis. Can ignore phase since you can change phase sampling anyway in OSSC and CRTs are tolerant. The first graph is the AC sweep that we want to see. In db so 0 means a gain of 1 = signal is unaffected.

The x-axis is a log scale of frequency. The 1E3 is 1 kHz in scientific notation and 10E3 is 10 kHz. The 100 kHz x value gives a -3 y value for 50% power reduction. At X db, formula is 10^(X/10) so -20 db is power reduction of 10^(-20/10) = 99%! Basically, this low pass filter wipes out any and all frequencies at ~250 kHz and above. A higher order will give a steeper decline in wiping out frequencies but can't have too much of a good thing. You need 1 op-amp for every 2 orders.
Post Reply