# Are there any gotchas to using Safty Capacitors (X1/X2 box style) in the construction of a filter?

I guess this is just not done for whatever reason. I hit Google and even after over 20 mins could not really find any information on point.

I've got a fair amount of .47uF and .68uF X1/X2 Metallized Polyester Film(I've got data sheets on a couple of them, these are all box) safety caps.

I want to use them to create a DAC reconstruction filter. I have not yet decided on the topology though. I'm thinking a 3 to 5 pole LC combined with a 3 pole Sallen Key active. I mean the caps would be perfect, high ratings, good D, not ceramic(The ceramic NP0 would be too small).

If they were not rated, I would not think twice, but I'm just wanting to make sure they won't cause problems based on their construction. Was there something done to enable them to be rated that would cause problems. The datasheets does not say, nor kind I find/see any complaints.

Ops forgot a data sheet:Okaya PA Series

NOTE: I added the x-capacitor tag, but I still feel the safety-capacitor tag would be better, or maybe a link. I think those Y caps feel discriminated against..:)

• What does the datasheet say about inductivity? Is it in the acceptable range for your filter? – Janka Jun 18 at 19:27
• Don't see it. I've linked the datasheet. The caps I have are that series. Not the newer PA-LE on their website. Max freq for the DAC will be around 100Khz and I was going to amplify after the filter(why amplify junk?) – GB - AE7OO Jun 18 at 19:36
• "safety capacitors" are either X-rated or Y-rated capacitors. There's a tag for x-capacitors, you could use that. – marcelm Jun 18 at 21:30
• Then I think a synonym might be a good idea. I did not think to search for x-. – GB - AE7OO Jun 18 at 21:32
• You might find that you can't get close tolerance devices in X rated capacitors (like asking for +/- 0.01mm tolerances on a hammer), but other than that they are the ideal low/mid frequency filter cap – Neil_UK Jun 19 at 5:03

" Was there something done to enable them to be rated that would cause problems." No. "Are there any gotchas?" Other than possibly enough inductance to affect performance, No.

Safety rated capacitors are more robust than equivalent film or ceramic unrated parts. They also are almost always larger and more expensive, especially in a low-voltage application. But other than an ESL (Equivalent Series Inductance) that is much higher than that of a 25 V or 50 V part, I don't think there are other issues.

• These are all pulls of one type or another, so cost does not apply here. As far as ESL goes, I'm going to be hooking them up to a 2 or 3 9mH inductors so it will just help... :) I'll have to see, but unless I hear a show stopper in the next couple of hours, I'll accept your answer and build the circuit. And if these things kill it, I've got some PE cap[s(I would rather save these for audio usage though.) – GB - AE7OO Jun 18 at 20:12

Plastic film caps of all kinds are the closest to an ideal cap in this size( except motor start-run caps ( which are designed for slightly higher ESR).

There are subtle differences but PP X-Y caps are about the best (exc. TEFLON \$) for the parameters you mention and also the good choice for small value S&H , FOR touch screen reference caps and low leakage for Integrators.

Inductance will add only about 0.5nH/mm for short L/D ratios. Internal foil can be found by SRF frequency tests but I suspect they’ll will outperform MOST 150 MHz GBW Op Amps operating at filters well below 10MHz.

The reason I mentioned that is that, you tried to make an active notch LPF with high Q at the DAC spectral peak like Q=100 at 100kHz for Class D,E filters you need >150MHz of BW. But passive filters no problem. Just look for the Ripple current rating. which comes into play at high power levels. THe Dissipation Factor is a percentage of the power transferred and filtered out by the filter (losses). Since L presents a high impedance, your can export your Falstad Model into the main time domain site ( export text...import text and then add ESR and scope average power on that ESR computed automatically= by selecting the scope properties for Power , Avg, min,max on that added non-ideal cap and inductor.

I highly recommend you model it in http://www.falstad.com/circuit/e-filt-lopass-l.html and choose from >circuits> high order passive filters and choose the one you want and the order and then add DCR,ESR if you want realism for you source RdsOn and speaker load R1+R2//L = 8 Ohms for example as that may affect the filter.

They are also the best caps for crossover filters in speakers and reactive offline 5W non-isolated step-down transformers.

• I guess it would not have hurt to add the DAC application which is not audio. This is part of the excitation circuit for my LCR meter, I'm working on in my coupious spare time. – GB - AE7OO Jun 18 at 20:54
• Ah . OK the only gotcha on PP caps I found after testing 5k of them with my LCR meter is you have to screen for 1% tolerance when you buy 10%. I put them in film containers for each 1% increment. I hope you are using a CC sine wave to measure impedance and phase shift. For that I might not use a DAC but rather a true sine wave Oscillator into a voltage controlled Howland converter CC. At 120Hz, 1k, 10k and 100kHz. .. maybe not 1 MHz unless you have 100MHz GBW Op Amps. Rail to rail. – Sunnyskyguy EE75 Jun 18 at 21:00
• I'm running DAC -> filter -> modified howland CC. I just need from about 120Khz and down. I've got a 1.8Mhz+ VNA with I-V head I built a couple of years ago for RF. I'm using these STM32F3 DACs at 1Msps, even without filtering they look decent on my Kikusui 5060. I'd be willing to put them on the air after a simple RC low pass. I may just do a 5 pole LC(which is what my VNA had originally) and call it a day. I want them 0 and 180 to give me more ability. And I may be running at less than 60Khz if the SDADCs barf at the 120Khz. – GB - AE7OO Jun 18 at 21:15
• That’ll work. We used phase frequency PLL’s for Eddy current R&D for a product in the 70’s It had 50 ppm resolution and calibrated within 0.1 degree and 0.01% for amplitude with a test coupon. It was all analog. With quadrature servos using Bourns Sin-Cos pots driven by Swiss gear motors for servo calibration on every test. It could detect a molecular strain in Monel Steel tubing well before a leak of heavy water or a pin hole created by 10k atmospheres of pressure as a 0 and 90 degree phase shift. With a moving inductive coil inside the miles of tubing guided by our automated Robot – Sunnyskyguy EE75 Jun 18 at 21:29
• Yeah, I had seen a nice circuit where the Howland was put into the feedback loop of the controling OpAmp. Add in a digipot and it works. The DAC controls the freq and the Howland controls the levels(I need to add a DC bias circuit in there too). The hardest part I'm having is which way to go;Auto-balancing bridge or some form of I-V. It's a toss up. – GB - AE7OO Jun 18 at 21:50