Lets talk about 'recaping ' for a second...
In the process of recaping most amps and preamps, you'll have a variety of low value electrolytics.
Say under a few uf.
Other than the issues of physical size and cost.
Is it feasible to replace them with foil/film caps ?
Are there any sort of guide lines for this ?
Am I safe in assuming the same can be done for any bipolar electros in the circuit ?
Thanks.......
In the process of recaping most amps and preamps, you'll have a variety of low value electrolytics.
Say under a few uf.
Other than the issues of physical size and cost.
Is it feasible to replace them with foil/film caps ?
Are there any sort of guide lines for this ?
Am I safe in assuming the same can be done for any bipolar electros in the circuit ?
Thanks.......
Yes...I do this all the time. Typically anything up to 3.3 uF is easy and relatively small if you can use 63 Volt cap or less.
Akitika Preamp and Power Amp Kits
Akitika Preamp and Power Amp Kits
All film caps will be non-polarized and that removes concerns about compatibility in audio applications. The main problem is obviously size and though you may be able shoe-horn the caps into odd spaces, they can act as antennae for power supply generated EMR and also RFI. Adding shielding is always possible but at what labour cost? Some amps and large cap. values are never going to work out so I wouldn't speculate on the chances of success without a good look at the schematic and the PCBs first.
The smallest film caps are metallized polyester in MKT or Greencap form. These are OK and better than generic electros but the best common types are based on polypropylene film/foil type MKF (huge), or metallized film MKP type (v.large). As a general rule, anything over 1uF in MKF is too big as is anything over 3.3uF in MKP and maybe 10uF MKT.
I have to use a scope or sensitive audio mV meter to see what happens and troubleshoot this process. It's usually experimental, tedious and expensive work as you can't be sure of the problems until you inspect and start fitting parts. I also found people weren't keen on paying for this because often, the results weren't that spectacular, since many electros were still fine.
Having said that, polyester caps with higher voltage ratings like 100V+ can perform very well. If, for example you recapped a preamp full of old 2.2uF coupling caps with MKTs, you will still get significant improvement but just watch the size:noise:cost ratios.
The smallest film caps are metallized polyester in MKT or Greencap form. These are OK and better than generic electros but the best common types are based on polypropylene film/foil type MKF (huge), or metallized film MKP type (v.large). As a general rule, anything over 1uF in MKF is too big as is anything over 3.3uF in MKP and maybe 10uF MKT.
I have to use a scope or sensitive audio mV meter to see what happens and troubleshoot this process. It's usually experimental, tedious and expensive work as you can't be sure of the problems until you inspect and start fitting parts. I also found people weren't keen on paying for this because often, the results weren't that spectacular, since many electros were still fine.
Having said that, polyester caps with higher voltage ratings like 100V+ can perform very well. If, for example you recapped a preamp full of old 2.2uF coupling caps with MKTs, you will still get significant improvement but just watch the size:noise:cost ratios.
Film cap is always A LOT bigger than electrolytics. I am looking at my 1uF and 4.7uF metalized polypro as I type this. There is no way in hell I can fit that into any electrolytic footprint of that value!!!
I would be a lot more concern you try to force a square peg into a round hole, making long wire leads. Like Ian said, you can pickup noise, have excess lead inductance and all other bad things.
Also, cap is critical for distortion IF you use it for high pass or low pass filter or AC signal coupling where distortion at close to the pole or zero frequency can by significant due to voltage coef of the cap. If it is for power supply filtering, put smaller ceramic cap in parallel will do you a lot more good. eg. if it is a 100uF, parallel a 2.2uF ceramic or even 1uF ceramic. This is to keep the loss at high frequency loss down.
Even for the power amp negative feedback gain setting resistors, I use 220uF non polar electrolytic to provide AC ground . I use 20K feedback and 1K gain set resistor to get closed loop gain of 21. I eliminate the DC offset by AC ground the 1K so the DC gain is unity. The high pass is 0.7Hz. So, at 20Hz ( lowest audio you can hear) the cap is a short circuit no matter what cap you use. I just parallel a 0.1uF film cap over the 220uF for high frequency.
It's all about the circuit. Look at the schematic, determine which one you have to modify, don't just change for the sake of changing.
I would be a lot more concern you try to force a square peg into a round hole, making long wire leads. Like Ian said, you can pickup noise, have excess lead inductance and all other bad things.
Also, cap is critical for distortion IF you use it for high pass or low pass filter or AC signal coupling where distortion at close to the pole or zero frequency can by significant due to voltage coef of the cap. If it is for power supply filtering, put smaller ceramic cap in parallel will do you a lot more good. eg. if it is a 100uF, parallel a 2.2uF ceramic or even 1uF ceramic. This is to keep the loss at high frequency loss down.
Even for the power amp negative feedback gain setting resistors, I use 220uF non polar electrolytic to provide AC ground . I use 20K feedback and 1K gain set resistor to get closed loop gain of 21. I eliminate the DC offset by AC ground the 1K so the DC gain is unity. The high pass is 0.7Hz. So, at 20Hz ( lowest audio you can hear) the cap is a short circuit no matter what cap you use. I just parallel a 0.1uF film cap over the 220uF for high frequency.
It's all about the circuit. Look at the schematic, determine which one you have to modify, don't just change for the sake of changing.
Last edited:
This may cause an HF parallel resonance, which will significantly raise the HF impedance.Alan0354 said:
Replacing a small electrolytic with a film cap is rarely useful, and sometimes unhelpful. If the cap has the right value then it won't have much signal voltage across it so it can't add very much distortion. In some cases the designer may be relying on a little ESR to dampen a resonance - a film cap will have much smaller ESR so will no longer do this. A physically large cap can pick up hum and interference, as others have said.
As I often say, in order to improve a circuit you need to understand it better than the original designer. If not, making it sound different is easy - making it genuinely better is harder.
Parallel a smaller value cap is to dampen the resonance. This is well documented in modern bypass circuits. This is standard technique like using 10uF, then 0.1u and for RF, 330p 0402 cap in parallel to make sure low impedance all the way to RF.
This is what I do on both guitar amps and my Acurus. I did a lot of modifications and up the bias of the Acurus including putting it quite a few 0.1uF ceramic bypass caps. I cannot say whether it helped the sound with the 0.1uF caps. BUT with the modifications, the sound of the amp improved significantly.
I modify one side of the amp first, then I switch the speaker left and right, the improved sound followed the side of the modified amp, not the speaker. So I know it's real, not psychosomatic.
No, adding a small cap in parallel can create a resonance. RF engineers know what they are doing, and probably have the test equipment to look for such problems. Most audio designers do not.
As I said, making it sound different is easy.
As I said, making it sound different is easy.
you need to do the measurements, the ear is unreliable tool except for golden eared gurus, are you planning that route as a career move? network analyzers are useful
No, this is a standard practice in the industry you parallel 0.1uF with 4.7uf or 10uF. I am an RF engineer, I know.
In Microwave circuit, not only I parallel 10uf with 0.1 and 220p. My last cap in parallel is a distributed cap using power and ground plane to form a parallel plate capacitor. I design the layer stackup so I have 5mil dielectric between the power and ground plane. This capacitor is so low parasitic that it works for circuit way into GHz. I do it for all the RF transistors and amplifiers.
then you should know better,
i learned a early lesson as an RF guy multilayer caps come in two sizes and their resonance is not only worse at the larger power sizes but is related to their SMD mounted orientation. RF should end at the 1st bypass
i learned a early lesson as an RF guy multilayer caps come in two sizes and their resonance is not only worse at the larger power sizes but is related to their SMD mounted orientation. RF should end at the 1st bypass
Last edited:
I disagree, audio amp is for listening. That is the ultimate test. You listen with your ears, not listening with the spectrum analyzer. This is where I think a lot of people missing the moon.
I was a working engineer for 30 years, and I am still contracting with a company to do R&d work in designing circuits that cannot supposed to be done. I use plenty of network analyzer and spectrum analyzer. You can measure to the cows come home, but the end of the day, it's the sound. I take the good sounding amp any time of the day over a perfect amp on the spectrum analyzer.
I worked to 5GHz, never saw that. Ask any working engineer, it's standard to pair 0.1uF with 4.7uF for normal(non RF). Look at any circuits. Never have that failed me once. I work with nV and pA electronics that more sensitive beyond any audio circuit has to deal with. Never have I seen problem in the last 30 years.
Worms, so many worms to be unleashed.....
Alrighty.
I'm learning, and very appreciative of what your all telling me. 🙂
Specific question.
Circuit I'm looking at is for a receiver.
The preamp section is fed into the amp section via a series 4.7uf electro that is bypassed with a 0.047uf (noted as being a "mylar cap").
Would I be correct in thinking that replacing that combo with a 4.7uf foil/film could be a reasonable spot for upgrade ?
(Assuming that I can get said replacement to fit without excessive leads).
Other can of worms I'm going to open is . . . . . low esr caps for recaping.
Now, I know this is very subjective, and every man and his dog has their pet cap....
Normally I've used 'standard cost effective' low esr 105 deg caps for my recaps.
Sometimes I will use Pana FC's if requested.
So far every unit I've done, the customers report back about much much improved sound across the board.
New research has shown me that panasonic have also got the fm range which has even lower esr (and somewhat regarded as a decent 'audio' cap) but, they have now got the FR range which is the replacement for the FM range....
Would anyone care to comment on their thoughts of FC/FM/FR vs foil/film...
Alrighty.
I'm learning, and very appreciative of what your all telling me. 🙂
Specific question.
Circuit I'm looking at is for a receiver.
The preamp section is fed into the amp section via a series 4.7uf electro that is bypassed with a 0.047uf (noted as being a "mylar cap").
Would I be correct in thinking that replacing that combo with a 4.7uf foil/film could be a reasonable spot for upgrade ?
(Assuming that I can get said replacement to fit without excessive leads).
Other can of worms I'm going to open is . . . . . low esr caps for recaping.
Now, I know this is very subjective, and every man and his dog has their pet cap....
Normally I've used 'standard cost effective' low esr 105 deg caps for my recaps.
Sometimes I will use Pana FC's if requested.
So far every unit I've done, the customers report back about much much improved sound across the board.
New research has shown me that panasonic have also got the fm range which has even lower esr (and somewhat regarded as a decent 'audio' cap) but, they have now got the FR range which is the replacement for the FM range....
Would anyone care to comment on their thoughts of FC/FM/FR vs foil/film...
Attachments
Calculation tells me that 4.7uF feeding 11k has an LF rolloff at 3Hz. However, the transistor input is in parallel with the 11k, so 4.7uF might not be big enough. If it is big enough then changing it for a 4.7uF film cap will make no difference. If it is not big enough then a film cap may be an improvement, but so would a bigger value electrolytic. The bigger electrolytic would probably be physically smaller than the film cap.
I mostly agree with DF96. I just want to add what I learn from AndrewT that in the whole signal chain, you want ONE signal coupling cap to do the band limit to say 10Hz, that's the one that you absolutely use film cap.......to be more specific.....metalized polypropylene cap to achieve the lowest distortion at the lowest break frequency( 10Hz in this case). Then the rest of the coupling, use as big a cap to make it as low a frequency as possible, use electrolytic cap.
In this case, 4.7uF give 3Hz as DF96 said, if this is the band limiting stage, you don't want to use electrolytic cap, you get distortion at 3 to 10Hz. BUT if this is not the band limit stage, you want to make it to break below 1Hz, more like 0.5Hz or so. In another words, I'll try to find a 22uF that can fit into the pcb holes.
As you can see, they use 0.047uF film with low esr in parallel. That's the exact reason I talked about paralleling caps. Inside the caps, the higher the value, the more layers your have, more parasitic inductance you have. Particular electrolytic, they literally roll two layers in a roll. part of the roll is long way from the leg that comes out. Inductance is proportion to the length/width. If you have a roll, length >>width and create high series inductance. this will create a resonance peak at higher frequency. That's when the parallel cap with smaller value comes in.
The smaller cap has less inductance, it takes over right before the resonance frequency of the big cap. and keep the impedance low. Then when reaching the resonance of the smaller cap(say 0.1uF), you want to have another smaller(330p surface mount) to cover that.....and so on and so on.
Here, the designer must be thinking about keeping the impedance low at higher frequency. I would have done the same thing.
Now!!! This parallel cap business is assuming that you have good layout with ground plane to provide low inductance ground. People in audio seems to be against ground plane. If you start adding ground trace as return side of the filter cap. That CAN act as inductance and you open a new can of worm and see resonance/distortion and oscillation. Then in turn you blame the cap!!! I can even see the low impedance of the small parallel cap can create sharper resonance peak. BUT this is the designer's fault, don't blame on the poor cap!!!
Layout of the pcb is everything. I don't want to start another can of worm. I do all pcb with ground plane and treat them as high frequency circuit to avoid parasitic of the trace. But that's just me. I was the manager of engineering. But I layout all my pcb and some of my engineer's pcb on the part of the most sensitive circuit. The battle is won and lost on the pcb layout.
In this case, 4.7uF give 3Hz as DF96 said, if this is the band limiting stage, you don't want to use electrolytic cap, you get distortion at 3 to 10Hz. BUT if this is not the band limit stage, you want to make it to break below 1Hz, more like 0.5Hz or so. In another words, I'll try to find a 22uF that can fit into the pcb holes.
As you can see, they use 0.047uF film with low esr in parallel. That's the exact reason I talked about paralleling caps. Inside the caps, the higher the value, the more layers your have, more parasitic inductance you have. Particular electrolytic, they literally roll two layers in a roll. part of the roll is long way from the leg that comes out. Inductance is proportion to the length/width. If you have a roll, length >>width and create high series inductance. this will create a resonance peak at higher frequency. That's when the parallel cap with smaller value comes in.
The smaller cap has less inductance, it takes over right before the resonance frequency of the big cap. and keep the impedance low. Then when reaching the resonance of the smaller cap(say 0.1uF), you want to have another smaller(330p surface mount) to cover that.....and so on and so on.
Here, the designer must be thinking about keeping the impedance low at higher frequency. I would have done the same thing.
Now!!! This parallel cap business is assuming that you have good layout with ground plane to provide low inductance ground. People in audio seems to be against ground plane. If you start adding ground trace as return side of the filter cap. That CAN act as inductance and you open a new can of worm and see resonance/distortion and oscillation. Then in turn you blame the cap!!! I can even see the low impedance of the small parallel cap can create sharper resonance peak. BUT this is the designer's fault, don't blame on the poor cap!!!
Layout of the pcb is everything. I don't want to start another can of worm. I do all pcb with ground plane and treat them as high frequency circuit to avoid parasitic of the trace. But that's just me. I was the manager of engineering. But I layout all my pcb and some of my engineer's pcb on the part of the most sensitive circuit. The battle is won and lost on the pcb layout.
Last edited:
Given that the PCB in question is not up for redesigning and a technician is restricted to the traces, holes, spacings and whatever copper fills are available there now, we should be pragmatic. This isn't DIY so rather than shooting for SOTA performance, I think this is simply a case of best selection and fit of replacement caps.
From my own experience, customers may expect you to turn their consumer gear into high-end performance equipment but they won't be impressed by parts tacked on or squeezed awkwardly into small areas in the attempt. Most receiver boards are tightly packed, single layer, cheap paper substrate so I think we have to ease up on the expectations of how much can be achieved.
From my own experience, customers may expect you to turn their consumer gear into high-end performance equipment but they won't be impressed by parts tacked on or squeezed awkwardly into small areas in the attempt. Most receiver boards are tightly packed, single layer, cheap paper substrate so I think we have to ease up on the expectations of how much can be achieved.
Hmmm ok.
This is turning into a 2 thread, thread...
I'm trying to learn how to improve customer results, yet I'm also looking at best results for my own receiver..
Sounds like for normal customer recaps I'd be best of staying with 'like for like' replacement using whatever flavor of low esr cap is in vogue.
And yet.... For my personal baby, I'm willing to go further 😉
This is a nakamichi re-2 which I believe is based of a nakamichi receiver 2.
Ian is right in that mine is indeed (vomit) bakelite boards.
As is pretty much 99% of what I work on.
But that was the norm for 80's japanese consumer items .
Having a quick look at the schematic, I've found another series cap.
2.2uf from volume control......
I'll try to edit the relevant parts of the schematic and post those.
System here won't like the 8mg pdf's 🙂
EDIT - Added pics of main audio section of RE-2 and power supply of Receiver 2 (Supposed to be the same ?)
Sorry for the bad images, there the best I have..
This is turning into a 2 thread, thread...
I'm trying to learn how to improve customer results, yet I'm also looking at best results for my own receiver..
Sounds like for normal customer recaps I'd be best of staying with 'like for like' replacement using whatever flavor of low esr cap is in vogue.
And yet.... For my personal baby, I'm willing to go further 😉
This is a nakamichi re-2 which I believe is based of a nakamichi receiver 2.
Ian is right in that mine is indeed (vomit) bakelite boards.
As is pretty much 99% of what I work on.
But that was the norm for 80's japanese consumer items .
Having a quick look at the schematic, I've found another series cap.
2.2uf from volume control......
I'll try to edit the relevant parts of the schematic and post those.
System here won't like the 8mg pdf's 🙂
EDIT - Added pics of main audio section of RE-2 and power supply of Receiver 2 (Supposed to be the same ?)
Sorry for the bad images, there the best I have..
Attachments
Last edited:
- Status
- Not open for further replies.