Which (THT)C0G caps to choose (Disc vs. MLCC)?

[DL3CE]

Hello,

I have been watching the eevBlog and also reading the forum for some time, and I finally went to ask something now.

I'm getting back into doing some RF experimenting (mostly in the amateur shortwave region, i.E. 1-50 MHz, maybe up to the 144MHz band in the near future) and after being fooled by using questionable ceramic caps from junk sales etc. which turned out to be Z5U or similar high tolerance, high loss stuff, I want to get an assortment of quality-brand C0G/NP0 caps. I'm looking for 1 pF up to a few nanofarads in THT. I'm well aware that SMT is far superior in most cases and I'm not scared at all of using it, but I like to do experimental builds on veroboard (of course not the stripboard variant and I often use the, being granted, quite expensive, boards with the groundplane on it) or using that "Manhattan" technique. If I transfer that to permanent builds, I will most likely use SMT parts on a professionally manufactured PCB .. but I will order those specifically for that project then.

For values above a few tens to hundreds of picofarads there isn't really an alternative to using MLCCs (they don't make that ridiciously large discs anymore) but for lower ones you can still choose. By instinct I would think the MLCCs are better in parts of parasitics since they are essentially a large count of smaller caps in parallel, but did I overlook something here? Are the old disc types maybe just going obsolete?

The caps will be used in resonant circuits quite often, so ... low loss and low parasitics are essential.

Thanks,
Christian

[DaJMasta]

The lead length of through hole parts is a real risk in trying to use non-SMT stuff, but it's certainly something possible in those frequency bands.  You can get good dielectric, low value, through hole caps fairly easily through major distributors, but they are considerably pricer (maybe 50-100 for the price of 500-1000 NP0/C0G).  Large discs are still available in high voltage, low capacity versions for power supply filtering and such, but since the small ones do well, are easy to use, are cheaper to make, and are widely available... it's definitely a more standard form factor these days.

SMT parts can certainly be used on protoboards, though, if they're small enough to just go between single pads, so while 0805s are a bit large (though workable in most situations), 0603s may be better for fitting between pads.  One nice thing about them is that because the parts cost is so low with SMT passives, you can usually afford to use them in a prototype and then just use more in the final version with almost no extra price.

[floobydust]

If you are pushing transmit power through the capacitor, then dielectric heating and breakdown voltage makes larger through-hole parts more desirable. Although mica is very good but expensive. Mica SMT does exist from CDE.

You can get large size disk caps in through-hole, just the few nH lead inductance is their disadvantage I believe. Depends on your voltages.
Or is this for the RX section?

[DL3CE]

The lead length of through hole parts is a real risk in trying to use non-SMT stuff, but it's certainly something possible in those frequency bands. 

It was done in commercial gear for decades, even with much larger components and lead lengths in tube gear (being granted, the impedances there are much higher so the influence of parasitic inductances is reduced proportionally). We are not talking UHF here, I don't think the 2-3mm lead length is going to be very critical.

SMT parts can certainly be used on protoboards, though, if they're small enough to just go between single pads, so while 0805s are a bit large (though workable in most situations), 0603s may be better for fitting between pads.  One nice thing about them is that because the parts cost is so low with SMT passives, you can usually afford to use them in a prototype and then just use more in the final version with almost no extra price.

I'm aware of that, but I just don't like that. I'm somewhat finicky about the looks, even it's a protoboard. That may be a bit irrational, but it is like I am. You are having  a point there about the cost, since the leaded stuff can be easily 5-10 times the price of the SMT stuff. But I'm going to accept that.

If you are pushing transmit power through the capacitor, then dielectric heating and breakdown voltage makes larger through-hole parts more desirable. Although mica is very good but expensive. Mica SMT does exist from CDE.

My current experiments are just going to be some RX stuff and QRP transmitters, so high-power components aren't required yet. I wouldn't have those "in stock" anyway, but order them specifically when I need them.

Until now I was using ceramic discs made by Hitano, which are available in C0G to me up to 330pF / 100V. Those are getting already quite large (12.5mm diameter) at the highest capacities. Everything above 330pF is Y5P or Z5U.
Also available to me are the "Goldmax 300" Series by Kemet, which are C0G MLCCs in leaded cases (if I look at the form factor though, I wouldn't be surprised if they used the SMT components, stuck some leads on and then dipped them.)
Maybe the "Skycap" SR-Series from AVX, too.

I just want to know if I can straight go for a assortment of MLCCs or if disc types have some advantages here and there. I think MLCC cracking shouldn't be that much of a deal with leaded components since the leads can bend and take the twisting forces at least partly?

Thanks,
Christian

Edit: Just clarifying the AVX product designator.

[T3sl4co1l]

Uhhhh... lemme see here...

I've got:

- 0805 SMT chip, in standard values from 1pF to 1nF (C0G) and 1nF to several uF (X7R), in 1-2.2-4.7 or finer steps
- And other 1206 and 1210s in the uF's
- 10pF to 680pF, C0G or Y5P disc (various)

I'd have to look up what all I'd ordered in the past, but the ceramic chips happen to be TDK C2012C0G1H101J and so on.

The 220pF C0G disc is Murata DEA1X3A221JA2B (you can pick from the same series if you like).

Also, 100pF (MLCC-on-leads style) Vishay K101J15C0GF53L2 (the print is hard to read, though -- if you're throwing lots of caps together in a parts bin, I recommend not buying multiple from the same family, you'll never tell them apart ).

MLCC-on-leads kind are very common and cheap, but classic ceramic disc kinds are still quite plentiful and cheap.  Poke around the usual distributors and sort by price and availability (qty in stock) to make sure you're not buying random ripoff stuff.

Tim

[DL3CE]

I think I will order an assortment of the KEMET MLCC ones from a reliable distributor. They aren't that cheap but I think they are worth their money. It doesn't seem that there is something about the MLCCs which makes them a big No-No, as long as they are C0G.

[bd139]

I’ve built a fair few receivers up to 2m band.

I use Vishay or Suntan MLCC C0G capacitors purhased from Rapid (Conrad!) or Farnell. They are both rock solid parts. I use TH. Even at 140MHz because you’re not likely looking at terribly repeatable designs you can get away fine with those parts. I tend to just wire the parts together ugly style at 50Mhz+.

The TH parts are just SMD parts with leads added and dipped so you get a tiny bit of extra capacitance and some lead inductance. This isn’t really a problem in practice until about 200-300MHz. Most designs need tweaking physically at 70MHz+ anyway. Inductors are a far larger problem!

As for part selection I don’t stock many: 10, 22, 47 in each series (x1, x10, x100 pF). It’s actually better if you parallel 2 or more anyway as it reduces RF heating and increases stability. One thing to note is that the tempco of the above capacitors is the opposite of most toroidal ferrites so there is some additional stability gained here.

Also get a massive bag of X7R 100nF parts for decoupling. I get through tens of these a month.

TX filters are the only exception to this. You need something that has a higher voltage capability. I use a selection of 2.5% silver mica caps (47,100,220pF). Those were expensive but they don’t blow up.

[DL3CE]

Tnx fr info

I just realized that the distributor I was looking at doesn't even supply a full E6-series of the KEMET ones  . So I'm still looking for alternatives...