Polystyrene Capacitor

[metrologist]

What type of capacitor should I order for 0.1uF?

DK does not list polystyrene caps, and the schematic shows polarized caps, but I don't think they are.

For this: http://www.rt66.com/~shera/QST_GPS.pdf

[Gyro]

I can't see any critical capacitors in that design. No idea why he's listed them all as polystyrene - most of them are supply decoupling capacitors where you can simply use 100nF ceramic. You might want to make the input coupling caps (C1 and C2) and the VCXO tuning voltage filter cap (C15) basic metalized film types, but even that's debatable. I can't see any need to go off searching for big polystyrene types.

P.S. Looking at the PCB picture he certainly hasn't used Polystyrene, they look like mylar.

[metrologist]

I was going to order these:

http://www.digikey.com/scripts/DkSearch/dksus.dll?Detail&itemSeq=234318732&uq=636367337004725280

[Gyro]

It may be one of my script blockers, but that link brings up a  search result with 6,297,942  hits for me!

[metrologist]

It's a Polyester, Polyethylene Terephthalate (PET), Metallized - Stacked, 495-5055-1-ND

The last item I'm looking for is the 24mhz oscillator. The board you can buy looks like it's 8-pin DIP.

http://www.a-aengineering.com/gps.htm

[Gyro]

The capacitors sound (...look) fine.

Regarding the oscillator, this looks like the datasheet... http://www.datasheet4u.com/pdf/ECS-300C-pdf/559359 No idea where to find one - you might have to look at adapting something (non fitting) on an adaptor board.


Edit: It looks like it puts out 24MHz and a divided down 6MHZ. You could substitute an off the shelf 24MHz Xtal oscillator and a divide by 4 dual flip-flop divider (74HC74).

[GreggD]

My understanding is that there are no more polystyrene  caps. Polystyrene film is no longer in production.

[JanJansen]

You can use polypropylene instead.

Reichelt has those polystyrene under the name styroflex, up to 10n not 100n.
https://www.reichelt.de/Styroflexkondensatoren/2/index.html?ACTION=2&LA=2&GROUPID=3158

[FrankBuss]

Looks like a complicated circuit. I have one of these with 10 MHz output:

http://cgi.ebay.de/261842721138

Works very well, even with the antenna inside on the window ledge. Then you would need only a 74LS73 for 5 MHz output.

[metrologist]

Not too bad, but it will be more complicated without that special oscillator. I was thinking of implementing a /4 chip.

http://ww1.microchip.com/downloads/en/DeviceDoc/sy10-100el33-l.pdf

[TimFox]

Polystyrene is slightly better than polypropylene, but the low service temperature (melting point) of polystyrene has rendered it all but obsolete.  For any reasonable purpose, polypropylene is the modern solution, especially at 100 nF, for critical applications.  However, it is difficult to find polypropylene at low values, below 500 pF.

[schmitt trigger]

I'm with Gyro on this one, Polysterene or other film capacitors are not the best choice for supply decoupling.

A ceramic MLLCC is an order of of magnitude better in this respect. And cheaper too.

Film caps have many excellent attributes. PLL loop filters, Sallen-Key filters, those applications.

[JanJansen]

However, it is difficult to find polypropylene at low values, below 500 pF.

down to 33p here :
https://www.reichelt.de/FKP-2/2/index.html?ACTION=2&LA=2&GROUPID=7164;SID=94WXnE26wQATcAAMHeUe00018a1c02fc43c4a7508f26be8025a98

[Gyro]

Not too bad, but it will be more complicated without that special oscillator. I was thinking of implementing a /4 chip.

http://ww1.microchip.com/downloads/en/DeviceDoc/sy10-100el33-l.pdf

Ouch! Be very careful using something like that (4GHz toggle) for something as slow as 24MHz. Ultra high frequency 'prescaler' type dividers can be very sensitive to minimum slew rates associated with slow clocks. They also need very good multiple ceramic cap sizes supply decoupling. Also take a look at the output logic levels - it's an ECL device. A 74HC74 would be far less overkill (as in suited to the job) and would actually work.

[edavid]

Polystyrene is an error in the parts list.

It says "C1-C12, C15 0.1 µF, 50 V polystyrene (DK P4593)".

But, if you go to DigiKey and search for P4593, you find:

Panasonic Electronic Components
ECQ-B1H104JF
0.1µF ±5% Film Capacitor 50V Polyester Radial

https://www.digikey.com/product-detail/en/panasonic-electronic-components/ECQ-B1H104JF/P4593-ND/136800

Of course I agree that they are not the best choice for the decoupling caps, but they are good enough, and the author probably wanted to simplify the BOM.

[floobydust]

Yes a mistake in the bill of materials, the PCB pictures show Asian polyester film caps.

Film caps are not best for decoupling at these frequencies (i.e. 24MHz osc) and I would use ceramic disc caps. But polyester is OK.

Polystyrene film caps are still in production and preferred for signal use in audio filters and low leakage use.
Mouser sells Xicon through-hole polystyrene film capacitors and many surplus outlets have them, if anyone searching and reads this thread.

[metrologist]

Not too bad, but it will be more complicated without that special oscillator. I was thinking of implementing a /4 chip.

http://ww1.microchip.com/downloads/en/DeviceDoc/sy10-100el33-l.pdf

Ouch! Be very careful using something like that (4GHz toggle) for something as slow as 24MHz. Ultra high frequency 'prescaler' type dividers can be very sensitive to minimum slew rates associated with slow clocks. They also need very good multiple ceramic cap sizes supply decoupling. Also take a look at the output logic levels - it's an ECL device. A 74HC74 would be far less overkill (as in suited to the job) and would actually work.

Well, I was just looking for a divider and that is about all I could find, so will use a 74HC74 instead. Thanks!

[Gyro]

If you get it in SO14 pkg and use an smd 24MHz oscillator, then you might be able to hand wire a bit of protoboard that would bodge onto the DIP8 footprint without looking too silly. Time for some ingenuity. 

[metrologist]

I was going to glue it between the legs of the TH osc on the back side of the board and wire it up. It's going to look silly anyway, but nobody will see it under the cover.

[Gyro]

That sounds like it would work well. Don't forget supply decoupling capacitors (ceramic, 10-100nF) on both the oscillator and the HC74 supply pins - you could probably solder both to the oscillator can to act as a groundplane.

[edpalmer42]

My apologies to Metrologist.  When I suggested Shera's GPSDO, I didn't realize that it used such an oddball oscillator chip. 

Let that be a lesson to designers everywhere.  Don't use weird chips or single source chips unless there's a really good reason for it!  In this case, the alternative would be a regular oscillator and a divide by 4 circuit.  Much better!

Ed

[metrologist]

That's OK. I have 5 oscillators to build into GPSDO's, if I don't put one into my HP5385A, or just make a stand-alone reference. At least I will learn something more than just soldering up some chips.

[Gyro]

When I suggested Shera's GPSDO, I didn't realize that it used such an oddball oscillator chip. 

A very odd choice. He either already had some in a drawer, or he didn't expect his project to have such longevity / popularity. It's a bit wierd when you have people selling PCBs which are footprinted for unobtainable parts. I've looked at that design several times and never noticed it either.

[metrologist]

Now that I've gotten into it, I saw on the time-nuts mailing list someone looking for this part, or suggesting a divider or just a standalone 6MHz xtal. I was wondering of the mcu needed to be sync'd?

p.s. I realize my posts about this whole endeavor are a mess...thanks for bearing with me 

[edpalmer42]

Now that I've gotten into it, I saw on the time-nuts mailing list someone looking for this part, or suggesting a divider or just a standalone 6MHz xtal. I was wondering of the mcu needed to be sync'd?

p.s. I realize my posts about this whole endeavor are a mess...thanks for bearing with me 

The way it's phrased in the article, the 6 MHz output is simply a convenient source of a clock for the processor.

Interestingly, it is desirable to have the frequency of U7 drift slightly rather than being synchronized with the VCXO. A slight random drift averages out the ±1 count ambiguity that is inherent in any pulse-counting device. My measurements indicate that the simple phase-measuring circuit I use is consistently accurate to 2 or 3 ns (for a 30-second measurement), while without drift, the resolution would be limited to 42 ns. The $5 crystal oscillator module drifts adequately and also provides a divided 6 MHz output to clock the CPU.

The 'slight drift' he's talking about appears to be in the 24 MHz signal rather than the 6 MHz processor clock.

Ed