The problem is that when I try to measure with a frequency counter the
vertical sync output on PIN 3 of the LM1818 I obtain bad values (in order of
Khz not in Hz as it should be.Usually it should be about 60Hz)
To say that I check the circuit and is OK as well as my frequency counter
that works fine when I measure, for example, the clock of the various CPU.
Any suggestions?I'm going crazy...
P.S.
I use the VIDEO Sync (pin P of solder side of the jamma connector) as
COMPOSITE VIDEO IN (pin 2 of the LM1881)
Have you looked at the output of your sync separator chip on a 'scope to
verify that *only* field sync pulses are appearing at the pin that you are
connecting to your counter ? Is the power supply clean ? No hash or high
frequency ripple that might be appearing on the output ? Have you tried a
simple R-C low pass filter between the chip output and the frequency counter
input ? (pure off the top of my head guess, try 10k series resistor followed
by 1000pF shunt capacitor). Does your counter have appropriate sensitivity
for the level coming out of the chip ? Too little level (under-drive) or too
much level (overload) can cause counters to produce erratic and erroneous
readings. Also remember that a field sync pulse train is *highly*
asymmetric, being a series of short pulses with long gaps between. Can your
frequency counter handle that sort of waveform, and interpret it as a
'frequency' ? If the counter has a "time" function on it, you may do better
to try using this, and just work out the apparent field frequency from the
time between pulses.
Have you tried measuring the odd / even output to at least see if that gets
you in the right ballpark ? If it is an interlaced sync pulse train, then
the vertical rate should be twice the odd / even rate.
You probably need to do a bit more research as to what exactly the problem
is, before reporting back so that we can see if we can help you figure how
to make it work as you want it to.
Arfa
Do you have a scope? Can you verify the sync and burst gate outputs?
LM1881s are fairly forgiving of poor quality / wrong level video. Did
you try measuring the odd/even pulse? It should be 30 Hz (OK,
29.97002616 for the purists).
G²
The LM1881 is notorious for giving undesired output sync waveforms.
They were used in the old (pre-GPS) days for deriving a reference
frequency from TV signals, and caused enormous grief to many.
We had such a frequency reference system using a widely published
schematic, and it would not provide the correct output frequency due
to an extra pulse in the train. Eventually we switched to the Gennum
4981sync separator and the problem disappeared. YMMV.
I don't have a scope, sorry..
Is the power supply clean ? No hash or high
> frequency ripple that might be appearing on the output ?
Ground and power pins of the LM1881 are directly connected to the ground and
+5 Volt
of my JAMMA test rig.
Have you tried a
> simple R-C low pass filter between the chip output and the frequency
> counter input ? (pure off the top of my head guess, try 10k series
> resistor followed by 1000pF shunt capacitor).
No, I didn't.Is it useful?
>Does your counter have appropriate sensitivity
> for the level coming out of the chip ? Too little level (under-drive) or
> too much level (overload) can cause counters to produce erratic and
> erroneous readings. Also remember that a field sync pulse train is
> *highly* asymmetric, being a series of short pulses with long gaps
> between. Can your frequency counter handle that sort of waveform, and
> interpret it as a 'frequency' ? If the counter has a "time" function on
> it, you may do better to try using this, and just work out the apparent
> field frequency from the time between pulses.
I use this counter:
>
> Have you tried measuring the odd / even output to at least see if that
> gets you in the right ballpark ? If it is an interlaced sync pulse train,
> then the vertical rate should be twice the odd / even rate.
Yes, on some PCBs I have tested I obtain a value of 30 Hz on the odd / even
output.Surely I ALWAYS obtain the value of 15.675 KHz (PAL signal) on pin 1
of the LM1881.
But is the Vsync and refresh rate the same thing?I'm speaking about arcade
PCBs...
Without a 'scope, you are at a huge disadvantage for trying to figure the
foibles of this sort of circuit, where potentially complex waveforms
predominate ...
>
> Is the power supply clean ? No hash or high
>> frequency ripple that might be appearing on the output ?
>
> Ground and power pins of the LM1881 are directly connected to the ground
> and +5 Volt
> of my JAMMA test rig.
>
At the very least, there should be an additional decoupling cap of say 47 to
100uF, situated close to the supply pin of the sync separator chip, if you
are going to use a remote supply, connected to other electronics within the
system. This is important.
>
> Have you tried a
>> simple R-C low pass filter between the chip output and the frequency
>> counter input ? (pure off the top of my head guess, try 10k series
>> resistor followed by 1000pF shunt capacitor).
>
> No, I didn't.Is it useful?
It is if there is any residual HF noise on the signal that you are trying to
measure. It would normally be considered good practice to have a simple low
pass filter on such a feed.
>
> >Does your counter have appropriate sensitivity
>> for the level coming out of the chip ? Too little level (under-drive) or
>> too much level (overload) can cause counters to produce erratic and
>> erroneous readings. Also remember that a field sync pulse train is
>> *highly* asymmetric, being a series of short pulses with long gaps
>> between. Can your frequency counter handle that sort of waveform, and
>> interpret it as a 'frequency' ? If the counter has a "time" function on
>> it, you may do better to try using this, and just work out the apparent
>> field frequency from the time between pulses.
>
> I use this counter:
>
> http://cgi.ebay.it/0Hz-to-50Mhz-Frequency-Counter-Module-Kit-LCD-display-/140353600315?cmd=ViewItem&pt=LH_DefaultDomain_0&hash=item20adb9fb3b#ht_1974wt_939
>
>
>
>
>>
>> Have you tried measuring the odd / even output to at least see if that
>> gets you in the right ballpark ? If it is an interlaced sync pulse train,
>> then the vertical rate should be twice the odd / even rate.
>
> Yes, on some PCBs I have tested I obtain a value of 30 Hz on the odd /
> even output.Surely I ALWAYS obtain the value of 15.675 KHz (PAL signal) on
> pin 1 of the LM1881.
> But is the Vsync and refresh rate the same thing?I'm speaking about arcade
> PCBs...
Yes, I understand that JAMMA is arcade stuff. As to whether vertical sync
rate and refresh rate are the same thing, it's a bit more complex than that.
It depends on whether you are talking interlaced or non interlaced. If you
see a stable 30Hz on the odd / even pin, then chances are it's a 30Hz
interlaced system, which is sort of equivalent to a 60Hz refresh rate
Arfa
They'll deal with level variances but they fall apart when the LF
response of the video is bad. You need _good_ coupling caps in the
entire video chain. That's going to be very difficult to correct
without a scope. That lousy .1uF input to pin 2 cap is much better
when it's 1uF and the 680K resistor on pin 6 works better at 470K. I
was _not_ using this with consumer gear but in a broadcast/post
production environment where the video was generally excellent.
G²
Not sure if that is a response to my post ...
We had tried the 1881 on a variety of signal sources with the same
issues presenting. The 4981 worked fine on all those sources. Go
figure.
I tried a 1uf capacitor on pin 2 and nothing has changed (I didn't obtain
anymore the value of 15.575 KHz on pin 1..).A friend of mine told me that he
can measure correctly Vsync of 60 Hz with a LM1881 with exactly the same
circuit on NATIONAL datasheet (the same I'm trying).
So,I can conclude, it's a issue of mine frequency counter...
P.S.
My friend uses a DMM which can measure frequency up to 20 MHz, probably it
is better (more sensible..) than my "assembled from kit" counter
You cannot make that assumption without a scope which unfortunately
you don't have. I'm only guessing but I bet your video has bad low
frequency tilt which shows of as the 'DC' of the back porch wandering
all over the place. Cause: bad capacitors somewhere in the video
chain. IIRC the Gennum chip includes clamps to guarantee sampling at
50% of the sync pulse meaning it's more tolerant of poor video. The
scope would tell all.
G²
G�
Absolutely agreed on all points
Arfa
G�
Ok, thanks.Sadly the video (R,G,B ground + SYNC) is from arcade videogame
pcb, so I assume they have all poor video signal???
I 'd wan to try the Gennum GS4981 but it seems unfindable...
?? have you tried their website? Google "gennum gs4981" and first
hit. Then go to the "buy from distributor" or "get free sample"
(sounds better) link at
http://www.gennum.com/buy/sales-network
Ok, thanks.I will try with the GS4981 instead of LM1881.
But, are you sure I will solve my problem with this IC?
Not at all. But (a) it did for us, and (b) it won't cost a fortune to
try - especially if you can get a "sample".
Be aware though that although the DS1881 is a drop-in replacment for
the LM1881, the GS4981 has a different pinout.
To get the sample is it necessary to have a company (it's a field required)?
>
> Be aware though that although the DS1881 is a drop-in replacment for
> the LM1881, the GS4981 has a different pinout.
Yes, I see the pinout from there:
http://www.datasheetcatalog.org/datasheet/GennumCorporation/mXrrsxs.pdf
but it changes only in PIN 7 which is horizontal sync output instead of
ODD/EVEN circuit.
So, I have to use the same circuit (two 0.1 uf capacitors and a 680 KOhm
resistor) like in the LM1881?
You should read the data sheet and not just look at the pinout. They
explain why you might want smaller vs larger coupling caps and do show
the same resistors as the LM1881 -- logical if you're trying to make a
drop-in replacement. For your purposes the 4981 vs 4881 in not
important, just a reassignment of the odd/even pin which you're not
interested in.
G²