Maker Pro
Maker Pro

Crystal load?

T

Tauno Voipio

I'm ordering a 2-pin crystal (8.86723 MHz) and need to specify the load.

A little help calculating this, please?

http://www.tinyuploads.com/images/yr8Paq.jpg

Thanks!

I'm pretty sure that the circuit does not function
as a crystal oscillator. Is it an own invention?

For a crystal, you should decide if the circuit
needs the series or parallel resonance. There are
both, pretty near each other. For details, get
e.g. the ARRL Handbook and read about crystal
oscillator basics.
 
D

DaveC

I'm pretty sure that the circuit does not function
as a crystal oscillator. Is it an own invention?

I double-checked the board layout; it is as I describe.

The circuit is part of an existing video driver board that outputs ASCII text
to a monochrome CRT monitor. I have substituted a sine wave generator in
place of the crystal and the board works.

The original crystal is gone (missing when I obtained the equipment) else I
would have looked up the numbers on the original.

I'm not asking for evaluation of the circuit's function -- that is confirmed
-- only to help calculate the load on the crystal.

Thanks,
Dave
 
M

Martin Riddle

J

Jamie

DaveC said:
Hmm... maybe this does? ::

http://www.tinyuploads.com/images/X4cOrL.jpeg

Sorry for the crummy first attempt...

Thanks!
I see you're having difficulties in determining the load..

If you are looking for an exact figure, I think you need to use
your signal generator via 100 Ohm R for example and a scope
to measure the drop in the circuit.

First, test the scope probe by measuring the drop after the
R only, to make sure you know the exact cap value in your probe.

Feed the circuit with this signal via the 100 ohm R, measure the
drop. Calculate the load and remove the scope probe load from the
results.

The net results should give you a load that is going to be close
enough.. You may want to operate the circuit for a bit before taking
final values. The logic chips are going to shift a little.

I'm guessing you'll end up with an approximate value that equals 8 pf.

In the capacitor manufacture world, the common practice was to zero
beat a tuned circuit with a fixed frequency. You attach a test subject
to the post which were part of this tuned circuit. You then moved the
calibrated dial which was nothing more than a capacitor, to make it
zero beat again. A scope was used with the X,Y inputs for that nice
lissajous circle or spiral curves.

In any case, this dial would give you the exact capacitance load.

Jamie
 
D

DaveC

Why not hack in an oscillator? It would be dead-on frequency, without buying
a
custom-ground $80 crystal that may not work in that circuit.

If I can find a 4-pin DIP oscillator at 8.86723 M I'd be tempted. But don't
see such an animal...
 
R

rickman

If I can find a 4-pin DIP oscillator at 8.86723 M I'd be tempted. But don't
see such an animal...

I haven't looked at standard frequencies, but you might have better luck
finding that frequency if you drop a significant digit or two. It is
unlikely the original part was specified to 1 ppm, 100 ppm would likely
do the job just fine. Didn't you say this was for a display? It will
probably work fine with just five digits or even four digits of
frequency specified. Is either 8.867 MHz or 8.868 MHz a common value
perhaps? Really anything near 8.87 MHz should do the job.
 
I haven't looked at standard frequencies, but you might have better luck
finding that frequency if you drop a significant digit or two.  It is
unlikely the original part was specified to 1 ppm, 100 ppm would likely
do the job just fine.  Didn't you say this was for a display?  It will
probably work fine with just five digits or even four digits of
frequency specified.  Is either 8.867 MHz or 8.868 MHz a common value
perhaps?  Really anything near 8.87 MHz should do the job.

I just looked at digikey, they list 17.734475MHz and 35.46895MHz
oscillators, but expensive and not in stock

-Lasse
 
D

DaveC

Not if it is a video clock, to a fixed frequency monitor.

I learned a little more about this circuit.

It's a proprietary video card card from a piece of German offset printing
equipment.

I used a sig. gen. in place of the crystal. The circuit generated a video
signal without any data (just sync pulses). The video card is separate from
the system processor, so being on the bench the card has no data to display.

Viewed on a scope, varying the 8.867 MHz frequency doesn't change the video
signal at all.

I'm guessing (with my limited understanding of how video works) that the
8.867 runs the dot clock, basically how fast the dots are shoved out of the
data bus into the video generator IC.

So, how critical *is* this frequency?

Thanks,
Dave (not an EE)
 
D

DaveC

Recently you mentioned a source in Ireland that had crystals of this
frequency. There are also some further source in the UK though a little
more expensive.

I'm therefore surprised you're going for some $50 custom built ones.

Your oscillator circuit is a series resonant kind, where the oscillation
frequency is a few kHz lower than parallel resonance.

I would have thought they would have been sufficiently accurate for your
application as colour systems are generally quite forgiving. If you
want to reduce the frequency of operation then a series inductor might
assist, rather than the series capacitor in your current circuit.

For further info:
http://www.mdpi.com/1424-8220/6/7/746/pdf

Mike Perkins

It's a monochrome system (even more forgiving?).

So, it's series resonance I want?

This is the one in Ireland:

<http://www.donberg.co.uk/catalogue/passive_components/quartz_crystals/8.86723
8mhz.html>

Should this one work in my circuit?

The only reason I was going with the manufacturer is that they can provide
some expertise re. the application, whereas the Irish are only sellers.

I'm no engineer so cannot determine for myself the appropriateness of this
crystal for my application (hence my posting here).

Thanks,
Dave
 
J

Jasen Betts

I learned a little more about this circuit.

It's a proprietary video card card from a piece of German offset printing
equipment.

I used a sig. gen. in place of the crystal. The circuit generated a video
signal without any data (just sync pulses). The video card is separate from
the system processor, so being on the bench the card has no data to display.

Viewed on a scope, varying the 8.867 MHz frequency doesn't change the video
signal at all.

I'm guessing (with my limited understanding of how video works) that the
8.867 runs the dot clock, basically how fast the dots are shoved out of the
data bus into the video generator IC.

So, how critical *is* this frequency?

it's hard to say. it depends on the monitor.
try a 9MHz crystal it's within 2% of the target , you might need to tweak the
vertical and/or horizontal hold adjustemnt on the monitor.

http://www.mouser.com/ProductDetail/ECS/ECS-90-S-4X/?qs=sGAEpiMZZMsBj6bBr9Q9aWDZfF25lWfiN/2lvWpfTc0=
 
D

DaveC

So, how critical *is* this frequency?
it's hard to say. it depends on the monitor.
try a 9MHz crystal it's within 2% of the target , you might need to tweak the
vertical and/or horizontal hold adjustemnt on the monitor.

When I vary the sig. gen. I used to simulate the crystal in this circuit, the
output video signal doesn't vary at all (sync pulse timing stays the same).

Therefore I presume that video timing is determined by the other crystal (6
MHz) that is running the 8085 uP on this antique board.

So I also presume that this 8.867 MHz crystal drives the dot clock (shifting
bus data into the 8275 CRT controller IC).

Is the 8.867 frequency critical at all? Since it doesn't control the video
sync pulses, etc., what effect will it have on the (ASCII text only) video
display?

Thanks,
Dave
 
D

DaveC

So I also presume that this 8.867 MHz crystal drives the dot clock (shifting
bus data into the 8275 CRT controller IC).

Or rather, loading 8-bit bus data into a parallel-to-serial shift register
and then into the 8275 CRT controller.
 
J

Jasen Betts

Or rather, loading 8-bit bus data into a parallel-to-serial shift register
and then into the 8275 CRT controller.

usually the CRTC manges the raster size and does its own
data fetches... I'm not familiar with the 8275, but all
the CRTCs i've encountered had internal counters to
determine the sync rates and sync pulse widths. from the
pixel clock that crystal may be doing something else,
colour-burst perhaps.

Is this circuit used with some sort of genlock - text over video ?
if so then the crystal frequency may be critical.

perhps feed an FM signal into where the crystal goes and
probe around to see where it's going.
 
R

rickman

Not if it is a video clock, to a fixed frequency monitor.

Did you miss the fact that he has specified a common crystal to nearly 1
ppm? They don't make crystals that will hold a frequency to 1 ppm. You
can get units specified to some 10's of ppm along with aging of similar
ranges. To do better you need to temperature compensate and perform
other "magic" in the oscillator.

Why would you think the monitor gives a durn? It may not take the
signal from an 8 MHz crystal, but 1000 ppm off should be no trouble.
 
R

rickman

When I vary the sig. gen. I used to simulate the crystal in this circuit, the
output video signal doesn't vary at all (sync pulse timing stays the same).

Therefore I presume that video timing is determined by the other crystal (6
MHz) that is running the 8085 uP on this antique board.

So I also presume that this 8.867 MHz crystal drives the dot clock (shifting
bus data into the 8275 CRT controller IC).

Is the 8.867 frequency critical at all? Since it doesn't control the video
sync pulses, etc., what effect will it have on the (ASCII text only) video
display?

It is hard to say just how critical that value is. How did you arrive
at this value anyway? A 1 ppm value would be hard to measure.

Anyway, much of it depends on the device that will be displaying the
data. Is this a TV or a computer monitor? A monitor will likely
tolerate a good range of values relative to the number you have
specified. The electronics in the display generator may or may not be
pushed to its timing limits. It would be hard to imagine that upping
the frequency 1000 ppm would cause a problem, but certainly lowering it
is not likely to disturb a good design. Even a 10,000 ppm deviation
shouldn't cause any real problems.

Here are links for parts available from Mouser and Digikey in stock.
Low cost and you can get them within the week.

http://www.mouser.com/ProductDetail/ECS/ECS-85-S-4X/?qs=sGAEpiMZZMsBj6bBr9Q9aQVbz5FLOGxjmYeJ9iHcTC8=

http://www.digikey.com/product-detail/en/9C-8.912MBBK-T/887-1830-1-ND/3585929

The Mouser unit is 8.5 MHz and should work unless there is something
"particular" about the monitor. The Digikey device is 8.912, closer to
the requested frequency, but is high, so there is a small chance it will
over clock some part of the design. It is also a surface mount unit
that might be harder to solder in.

No promises, but this is certainly an easy thing to try.

BTW, the one reason to get a part that is within 100 ppm is because in
the "old" days monitors would have some noticeable affects from power
line noise. If the display is closely synchronized to the power line
frequency this is not such a problem. Today's monitors don't have this
problem.
 
Top