The majority of the amps I've designed (and built) have used an output zobel cap which is similar to this:
Arcotronics | Passives | Capacitors | Plastic Film | Metallised 105°C |R82EC3100DQ70J
This component type has always had the desired effect, with no failures.
But lately, I've been wondering if a 100nF 100V 1206 X7R ceramic capacitor will do the job just as well.
Does anyone have any experience using an X7R ceramic in an output zobel? Any comments?
Arcotronics | Passives | Capacitors | Plastic Film | Metallised 105°C |R82EC3100DQ70J
This component type has always had the desired effect, with no failures.
But lately, I've been wondering if a 100nF 100V 1206 X7R ceramic capacitor will do the job just as well.
Does anyone have any experience using an X7R ceramic in an output zobel? Any comments?
I don't know about using ceramics as the change in value with temperature might affect the working of a zobel. I have seen mylar (greencaps) used and they work fine in this application. The metallised polyester is probably overkill.
the Zobel C needs low esl and esr, construction/case sized to survive full output swing/full damping R current at MHz
X7R C have significant voltage coefficient - you may want 2-4x V rating to keep C modulation within reason - although exact RC time constant isn't too important and the audio frequency Z is high enough for the distortion not to matter there's no real need to build in a nonlinear load
they will also be somewhat piezoelectric and actually emit sound
many inexpensive film types should work fine
X7R C have significant voltage coefficient - you may want 2-4x V rating to keep C modulation within reason - although exact RC time constant isn't too important and the audio frequency Z is high enough for the distortion not to matter there's no real need to build in a nonlinear load
they will also be somewhat piezoelectric and actually emit sound
many inexpensive film types should work fine
May be you need to build an in-stable amp or load the amp with difficult load to see the zobel in action. For normal-working amp it is difficult to hear the difference. But when I can put the zobel off-board, I prefer to use those green resistors usually used in speaker crossover, and an Orange Drop capacitor. For on-board, I use smaller devices such as carbon-film or VTM resistor, and an MKT (those orange Philips caps that I can find in abundance in old TV/monitor PCBs)
I've seen 0.1µF 250V Mylar caps fail in a 200W amplifier zobel network.
Caps have to be de-rated for high power, high frequency AC use.
Caps have to be de-rated for high power, high frequency AC use.
I've used Kemet CKR06 (moulded ceramic, 50v), MIL Spec (MIL-PRF-39014) with a value of 0.1uF, and a metal film resistor of 10R in my LM3875 Gainclone. I haven't had any issues with colouration in the sound output.
Chris.
Chris.
I ear that ceramic caps would made bad contact and crispis sound, on a previ. sansui.
I have a Yamaha AX-500 and one problem with overvolt on the bias from my amp, it should be stable to 10 mv , but I only obtain an endless volt increment and it does not seem to stop. I don't know what cause the oscillation , I be crazy with this matter. Before I fail with an cross accidentaly betwen base and collector and the power npn transistor burned and others. I changed all diodes and a lot of transistors, some capacitors electrolitic and polyester but I must ceramics capacitors to revise, 50 volts most of them, an range 5pf - 0,022 uF.
The question is, could one o some capacitors alter frequency or oscillateand increment volt bias, I think what these task is transistor exclusive work.
What type of variation could cause capacitors ceramic, mylmar and polyester, on an amplifier?
The BIAS is some type of grid who regulate to derive energy for a range power to transistor, one cathode type who do any type of absortion. It is regulate with a variable resistor from 1 omh to 2,7K or more, in my case when less omh more down the BIAS. When the bias is more to 10 mV the transistors heats in idle mode.
I have a Yamaha AX-500 and one problem with overvolt on the bias from my amp, it should be stable to 10 mv , but I only obtain an endless volt increment and it does not seem to stop. I don't know what cause the oscillation , I be crazy with this matter. Before I fail with an cross accidentaly betwen base and collector and the power npn transistor burned and others. I changed all diodes and a lot of transistors, some capacitors electrolitic and polyester but I must ceramics capacitors to revise, 50 volts most of them, an range 5pf - 0,022 uF.
The question is, could one o some capacitors alter frequency or oscillateand increment volt bias, I think what these task is transistor exclusive work.
What type of variation could cause capacitors ceramic, mylmar and polyester, on an amplifier?
The BIAS is some type of grid who regulate to derive energy for a range power to transistor, one cathode type who do any type of absortion. It is regulate with a variable resistor from 1 omh to 2,7K or more, in my case when less omh more down the BIAS. When the bias is more to 10 mV the transistors heats in idle mode.
Last edited:
My 100W/channel amp apparently has problems with the RLC caps dying if they are not MKT-P (metalized polyester and paper) caps. After an oscillation problem cooked the resistors in the RLC I decided I should replace the caps too, couldn't find the so called MKT-P caps (the originals were Phillips). Instead I used some of these Rifa X2 metalized paper caps Haven't had any problems with them, and they look really nice too 😉
Tony.
Tony.
How do you know that? I have always found zobel caps to be highly audible and would not use anything which is not good enough in direct signal path. Ceramics? No, thanks.
"What about a class X2, 0.1uF, 275Vac polypropylene capacitor? "
They might be OK. I would have to see the data sheet for them. I look for how much the cap is rated at at 20Khz, some 250V film caps are only rated at 8V RMS at 20Khz. An X or Y rated cap is designed for 50hz/60hz, I don't even know if their data sheets show high frequency data.
They might be OK. I would have to see the data sheet for them. I look for how much the cap is rated at at 20Khz, some 250V film caps are only rated at 8V RMS at 20Khz. An X or Y rated cap is designed for 50hz/60hz, I don't even know if their data sheets show high frequency data.
Here is a link to the PME271's (X2) I'm using ---> Download | Datasheet Archive
and a screen grab:
It doesn't give any stats on voltage rating at higher frequencies, but considering that the graph is showing the attenuation in the Mhz range I'd say that for audio freq's it shouldn't be a problem??
Tony.
edit: feeling a bit silly as just realised this graph is in fact showing the self resonant freq of the caps and probably has no relevance to what DJK said 😉
and a screen grab:
It doesn't give any stats on voltage rating at higher frequencies, but considering that the graph is showing the attenuation in the Mhz range I'd say that for audio freq's it shouldn't be a problem??
Tony.
edit: feeling a bit silly as just realised this graph is in fact showing the self resonant freq of the caps and probably has no relevance to what DJK said 😉
Attachments
Last edited:
Ideal whould be if the Zobel is optimised for the loudspeaker and cable used. The amp would see close to a resistor at high frequencies then. This is impractical for a comercial amp that should be stable under various conditions but for DIY it´s an easy modification.
Dissipation factor of 1.3% at 1Khz would seem to be a problem.
http://www.illinoiscapacitor.com/ic_search/seriesDocuments/PHC series.pdf
These would seem to be 0.06% at 1Khz for values less than 5µF.
http://www.illinoiscapacitor.com/ic_search/seriesDocuments/PHC series.pdf
These would seem to be 0.06% at 1Khz for values less than 5µF.
Hi djk, thanks for that. I'd wondered about the df, but polyesters I'd looked at seemed to be similar. I hadn't looked at polyprop mainly because I was going with something similar to what was already installed.
In which area would the high DF cause a problem? Heating (and subsequent failure) when higher freq's are present due to the relatively high DF at 1Khz (assuming it gets worse with rising freq)??
The implementation of the zobel/RLC network in my amp is as per the attached schematic. The only stuff passing through the cap (I would hope) would in fact be very high frequencies, if these are not present then I guess it is not doing very much. running the value of 6.8 ohms and 150nF through a low pass calculator (ignoring the L) gives an FC of 156Khz....
In this case is it likely that the cap will only have a problem if there *is* stuff present that the RLC is supposed to be removing?
As you can see I don't have a firm grasp on function of this cap. I'd always just regarded it as a shunt to earth of high freq "noise" present at the output, either passed through the amp, or coming back in from the speaker leads. I guess in that case, a cap that performs well at HF is a desirable (mandatory) thing, I had never questioned the parts specified by the original designer!
Tony.
In which area would the high DF cause a problem? Heating (and subsequent failure) when higher freq's are present due to the relatively high DF at 1Khz (assuming it gets worse with rising freq)??
The implementation of the zobel/RLC network in my amp is as per the attached schematic. The only stuff passing through the cap (I would hope) would in fact be very high frequencies, if these are not present then I guess it is not doing very much. running the value of 6.8 ohms and 150nF through a low pass calculator (ignoring the L) gives an FC of 156Khz....
In this case is it likely that the cap will only have a problem if there *is* stuff present that the RLC is supposed to be removing?
As you can see I don't have a firm grasp on function of this cap. I'd always just regarded it as a shunt to earth of high freq "noise" present at the output, either passed through the amp, or coming back in from the speaker leads. I guess in that case, a cap that performs well at HF is a desirable (mandatory) thing, I had never questioned the parts specified by the original designer!
Tony.
Attachments
Bearing in mind that we are considering RF power here, a cost effective AC suppressor type, 150 or 100n X2 275V AC has remained the preferred device for medium power amps locally. Obviously, high power gear may need scaling to suit and low power will be fine with 150V X2.
These are similar to MKT construction so scaled, DC rated types as described by JCX can and have been widely used commercially without many problems for decades.
These are similar to MKT construction so scaled, DC rated types as described by JCX can and have been widely used commercially without many problems for decades.
Last edited:
Doug Self did play around with zobel networks a bit in his book (I think this was in the 5th ed, could also be in an earlier one but not sure) and says they're for suppressing parasitic RF oscillations in the output stage. At least he found evidence for oscillations when he took his zobel out of circuit.
As for the OP's question about using an X7R - do you really want a wailing cap? As jcx says, they're piezoelectric.
The RC shunt portion of the amp Zobel is for feedback stability. It provides a load for the OPT and driver stage at light currents and reduces peaking in the loop gain an octave or so before unity gain point. Ideally it needs to be mounted near the output devices and return the HF impedance to the PCB where the HF rail decoupling caps also join. Under normal conditions hardly any current flows through that shunt path because the signal is band limited. If the amp oscillates, that is considered a fault condition and the Zobel resistor will take the excess dissipation. So the cap requirements are simply that it can see the full rail voltage swing and must be low esr out to at least 2 MHz. Normal metalized DC film caps would be ideal. Line X caps tend come in large case sizes for the job and may not be lowest esr at HF. Generally small case size and narrow lead spacing is proportional to HF performance.
Thanks Ian, Abrax and infinia 🙂 The comment about ideally being mounted near the output devices gives me some food for thought too.
Tony.
Tony.
- Status
- Not open for further replies.