How sync worked in old monitors?

[Mr.Madguy]

I want to try to find solution of my problem from another end. My problem - why CRTC controllers had exactly those settings, they had? I'm not familiar with old TVs and monitors. I've found old EGA monitor's schematic. It's sync circuit is based on TDA1180. I've dug into it's spreadsheet, but there isn't any detailed explanation there.

What I understand - is that it has feedback-based flyback frequency control. It makes sure, that flyback frequency = sync frequency. But it's not enough. Syncing frequency only still allows distorsions due to loss of sync with graphical data. Sync should also control picture alignment.

So, my question is - how picture alignment is performed? Is it controlled by position of sync signal only or also by it's witdh? I understand, that cutting sync signal early can cause flyback cycle to be also cut. But how about making sync pulse wider? Does it matter? In order for beam to be always synced with graphical data, it should always return to exactly the same position. How this position is detemined? Is it fixed and only controlled by WHEN sync is issued? Or may be witdh of pulse controls how long flyback is and therefore wider pulse moves beam further to left? In this case line starting position would be detemined by both WHEN it's fired and HOW WIDE pulse is. How alignment between lines is achived then? Via frequency only? Wouldn't picture float in this case?

 

[Sheiladad76]

That is an excellent observation. You are correct that frequency alone isn't enough; phase and timing are what actually 'anchor' the image to the screen.

 

[Mr.Madguy]

That is an excellent observation. You are correct that frequency alone isn't enough; phase and timing are what actually 'anchor' the image to the screen.

Can you be more specific? I understand, that there are things, that are considered to be so obvious, that one wouldn't even be able to find any explanation around whole internet. But I need more detailed explanation.

What I know for now. There is capacitor, that is charged during normal scanning. When sync signal is fired - it discharges through transformer. Around 100V bell curve signal is sent to CRT's deflection coil. It allows quick flyback. But it's not enough. Beam should return to exactly the same position every scanline. Therefore either both timing and frequency of this signal should be 100% accurate, or it's position should be determined by either when sync signal starts or ends. For example beam is held in start position till sync ends. In this case picture alignment is defined by sync end. Another possible scenario - flyback is cut early due to short sync signal and therefore picture is offset to right.

 

[Mr.Madguy]

Ah. I'm wrong. Detailed explanation is given here. It also explains, why wrong HSYNC/VSYNC frequences can possibly damage monitor. Lower sync frequences mean higher flyback voltages involved, so at some point they can go beyond components' allowed rates. But it only applicable to PLL-less monitors. PLL monitors just lose sync.

So, overall idea, as I understand it: PLL tries to keep frequency and phase of flyback pulses in sync with sync pulses. Right? Hold control is used to adjust frequency of oscillator and Phase control is used to offset pulses vs sync ones. Right? Therefore width of sync pulses doesn't matter. Right? It only matters in terms of keeping front and back porches standard. Because some VGA-to-TV converter can insert color burst pulses to that "free" space. Right?

 

[orac81]

Its worth going back to the original Analog CRT, and understand why/how it was done originally.
In old TVs vertical synchronisation was achieved with a simple diode, capacitor and resistor circuit, the "freewheel diode":

Flyback diode - Wikipedia

en.wikipedia.org

The verical sweep oscillator generates a sawtooth at near but not syncronised to the TV frequency. The diode circuit compares the tranmitted sync pulse to the sawtooth. If it is late, it generates a (say) -ve voltage, if early a +ve voltage. This voltage alters the frequency of the sweep gen, forming a feedback that locks the signal.

The original circuits/methods are very simple, and I think the early computer monitors used similar circuits for sync.

When young I was given a pile of 1960s "Radio Constructor" magazines. One of the features was "In your workshop" (with Smithy and Dick!) which covered TV and Radio repair with some humour. You can read thier take on this in Dec 1966 RC:

https://www.worldradiohistory.com/Radio ... gazine.htm

 

[daver2]

Those magazines took me back to my youth!

I (too) was given a number of these magazines by various relatives involved in TV and audio electronics.

I learnt a lot of basic electronics from these - along with Everyday Electronics.

I know what I am going to be doing this afternoon - reading a selection of them!

Dave

 

[orac81]

Yes it nostalgic, you could build a device that didn't need to login to xyz every week to work..

 

[g4ugm]

Those magazines took me back to my youth!

I (too) was given a number of these magazines by various relatives involved in TV and audio electronics.

I learnt a lot of basic electronics from these - along with Everyday Electronics.

I know what I am going to be doing this afternoon - reading a selection of them!

Dave

I loved those magazines! Nice light reading after Wireless World. In the 1970s, on payday I would go and buy Wireless World and go to Thorton's Chocolate Cabin and get a qtr of chocolate fudge and a qtr of coffee creams. Trouble was every time I read Wireless World I get a yearning for chocolate fudge....

... oh and as for sync I think some circuits used edge triggering, so the width of the pulse isn't so important. If you look at the TDA1180 data sheet you will see it does "phase comparisons" so you lock the waveforms together. If you lock the phase you lock the frequency as well, allowing the picture to be steady.

I remember on old TVs frame hold was always dodgy and could require regular tweaking, and I remember at the time of the Apollo missions to them moon the required reliability being compared to not adjusting the vertical hold on a large number of TVs for the duration of the mission....

 

[orac81]

I think at the start of Apollo you would be more likely to have ECC81/12AT7's rather than TDA1180's in your TV, you could probably fry an egg on the back of your TV, which was the main reason for the drift, and why TV servicing was profitable.

 

[Mr.Madguy]

TDA4851's data sheet has more info.

As I understand:
1) PLL1 keeps "middle" of sawtooth in sync with HSYNC signal.
2) PLL2 keeps flyback in sync with "middle" of sawtooth + applies flyback phase offset via delaying horizontal output pulse.

What is interesting here - flyback can actually begin BEFORE HSYNC's leading edge. I.e. it's not about syncing beginning of flyback with leading edge of HSYNC, as I expected. As flyback pulse has bell shape, it means syncing beam passing center of screen position during flyback with middle of sync pulse. I.e. at the end WIDTH OF SYNC PULSE MATTERS.

 

[Mr.Madguy]

Errr. Still don't understand. According to description, these chips control flyback only. They generate flyback control pulses.

Pins 2 and 3 on this picture for example:

So, they don't output their oscillator's sawtooth. But how final sawtooth is generated then?

 

[Mr.Madguy]

So, my current understanding of situation is following:

I'm not sure, if flyback duty factor should be standard or deviation is compensated by HPhase.

 

[Mr.Madguy]

Oh! I'm stupid! Distance between "forward" and "backward" screen centers is always equal to 1/2 forward + 1/2 backward beam movement distance, i.e. it's always equal to 1/2 of cycle!