So wont having a larger bank of smaller capacitors give you the best of both worlds, low ESR & ESL compared to an equivalent uF single large cap? ie. improved bass without muddling up the highs.
One thing is for certain, the most critical element of the power supply for its effect on the sound is the choice of cap for the onboard 470uF decoupler as it is close to the mosfet and supplying most of the current at lower watts.
Regards,
Dean
One thing is for certain, the most critical element of the power supply for its effect on the sound is the choice of cap for the onboard 470uF decoupler as it is close to the mosfet and supplying most of the current at lower watts.
Regards,
Dean
BWRX,
Can pretty well totally agree with all of that, suffite it to say, the difference is extremely audible. This also leads to the point that the bulk storage caps on the UCD modules is less than ideal, since in the given format we're unable to easily achieve the best of both worlds, whereby a healthy variety of caps would be required, in a proper layout where they can actually be effective.
I trust the 700 with it's stripeline configuration would be somewhat of an improvement in this respect.
Dean I agree with you too but as pointed out earlier by the both of us, where's the point of diminished returns given such things as wire parasitics etc. I think you'd certainly be pushing those bounderies, and I would keep a watchful eye out for oscillations, and give very careful thought to how you wire them up.
I think it's a great plan though, hope you get a hold of the caps anyway.
Cheers,
Chris
Can pretty well totally agree with all of that, suffite it to say, the difference is extremely audible. This also leads to the point that the bulk storage caps on the UCD modules is less than ideal, since in the given format we're unable to easily achieve the best of both worlds, whereby a healthy variety of caps would be required, in a proper layout where they can actually be effective.
I trust the 700 with it's stripeline configuration would be somewhat of an improvement in this respect.
Dean I agree with you too but as pointed out earlier by the both of us, where's the point of diminished returns given such things as wire parasitics etc. I think you'd certainly be pushing those bounderies, and I would keep a watchful eye out for oscillations, and give very careful thought to how you wire them up.
I think it's a great plan though, hope you get a hold of the caps anyway.
Cheers,
Chris
I wouldn't like to disapponit you Chris, but I think you are not doing things the right way. You can enjoy the bass, but I couldn't live without tracing such an effect (neither without an oscilloscope). I'm even drawing the PCB for a simple electronically balanced I/O interface with +-20dB adjustable input and output gains and additional selectable 0, -20 or -40dB input attenuation to connect it to PC soundcards and do some accurate (at least down to the accurancy of the card) frequency response, THD and IMD measurements on things such as amplifiers, crossovers, speakers, or individual components... I have been doing it without any balancing, but I'm tired of AC ground loops extending down to the audio band that I'm unable to break just with common-mode filters. It will also include a small 12V (from the PC) to +-18V DC-DC converter as I don't like having dozens of DC adapters 😀
I would do similar measurements if I was you.
Also, a transformer with a grounded or earthed shield is not always a good thing in a class D amplifier. Consider that you have common mode EMI everywhere, so the speaker ground, the power ground and the signal ground are not at the same AC potential. These grounds should met only on the amplifier module and anywhere else (well, you may have a central star ground as carrier residuals are flowing through speaker wires anyway), but mains transformers (not only the ones in the amplifier but also the ones on signal source equipment) suffer from winding capacitances (being toroids an order of magnitude worse than EI). In these circumstances, a transformer shield connected to amplifier ground is just adding more common mode EMI to signal inputs, and if you earth it instead, it will cause earth wiring to act as an antenna (as they don't have low impedance to anywhere above 1Mhz).
Wiring two or more class-D modules to share a signal ground, a power ground and a heatsink or case is much more complex that it may appear, as not only audio gound loops but also RF fully capacitive ground loops arise, and plenty of common mode filters are required to solve that (as the ones used in any piece of digital equipment, for example, last week I dismantled an old laser copier with a very modular layout that had dozens of common mode filters, as most modules had power devices in it and were using a common chasis as heatsink... it was a very nice machine until mice got their way into it
).
I will try to draw some diagrams, as I feel that these issues are no easy to explain with words.
I would do similar measurements if I was you.
Also, a transformer with a grounded or earthed shield is not always a good thing in a class D amplifier. Consider that you have common mode EMI everywhere, so the speaker ground, the power ground and the signal ground are not at the same AC potential. These grounds should met only on the amplifier module and anywhere else (well, you may have a central star ground as carrier residuals are flowing through speaker wires anyway), but mains transformers (not only the ones in the amplifier but also the ones on signal source equipment) suffer from winding capacitances (being toroids an order of magnitude worse than EI). In these circumstances, a transformer shield connected to amplifier ground is just adding more common mode EMI to signal inputs, and if you earth it instead, it will cause earth wiring to act as an antenna (as they don't have low impedance to anywhere above 1Mhz).
Wiring two or more class-D modules to share a signal ground, a power ground and a heatsink or case is much more complex that it may appear, as not only audio gound loops but also RF fully capacitive ground loops arise, and plenty of common mode filters are required to solve that (as the ones used in any piece of digital equipment, for example, last week I dismantled an old laser copier with a very modular layout that had dozens of common mode filters, as most modules had power devices in it and were using a common chasis as heatsink... it was a very nice machine until mice got their way into it
).I will try to draw some diagrams, as I feel that these issues are no easy to explain with words.
Hi Eva,
You may consider this a quick hello. I'll have to give your post a more proper and deserving response when I'm better up to the task, as it's been a rather long night.
For now, I'd sincerely like to request that you email me via the forum so that I discuss a few things with you which are of great interest to me.
Thanks
You may consider this a quick hello. I'll have to give your post a more proper and deserving response when I'm better up to the task, as it's been a rather long night.
For now, I'd sincerely like to request that you email me via the forum so that I discuss a few things with you which are of great interest to me.
Thanks
FYI Eva,
the UcD's do have a very short bypass to ground, 100nF, right at the outputs. Then a 5mohm resistor that does both over current and I suspect damping before the on board electrolytic.
Which by the way to everyone else, that electrolytic has about an INCH of trace before it gets to the outputs.
Also, the heatsink T bar is grounded very short path to the ground plane and thick, low capacitance isulators are used to better decouple the drains from the heatsink.
In a nut shell, Bruno has dotted his i's and crossed his t's on this design per your concerns.
That said, there is no such thing as eleminating high frequency capacitive coupling, only reducing it. Based on Eva's comments, it seems one would be wise NOT to connect two of these to the same hunk of medal. Maybe that's why the "tweaker" that designed the CIaudio amps made them monoblocks. Likely a more significant advantage than minimizing pumping. Who knows for sure? Maybe they just look cooler!
When looking at supplies like for a class D, or a high bandwidth digital system, its more like designing an eliptical RLC filter than just simply tossing in parts with low ESR or making them arbitrarily big capacitance or super low ESR. ESR is clearly part of the equation, not something ideally zero.
The final goal having many compromises, or optimizations if you will, like keeping hf current on the board, and being low impedance to the switches over as wide a band as possible, minimizing impedance peaks, rather than trying to have one really low impedance over a narrow range of frequencies. For example, a really low ESR on board electrolyitic can be shown in simulation to actually cause a boosting of the overall impedance seen by the switches at the upper end of the audio band. (This seems to be in line with Chris and others listening experiences that the highs suffer from lower ESR, although not conclusive, because it has little to no affect on the impedance seen by the switches at low frequencies compared to the notible sonic affect on the percieved bass increase.)
Might mention that Class D are capable of "downconverting" supply anomalies from around the switching frequency. Thus, impedance peaks or dips, at or around the switching frequency can be seen as "in the audio band" affects.
Since there is feedback though, all these affects don't matter!
😉 One could conclude that the goal is to minimize the "maximum" impedance at and above the lowest switching frequency to be expected to a bandwidth of say 10 to 100X that frequency (2MHz to 20Mhz.)
Of all these caps tweaks, the one you play the easiest with w/o much problem, is the main bulk cap, just becuase its decoupled so well from the rest of the circuit by the ferrites. The focus of this IMO would be to minimize potential negative signals getting from the xfmr to the Ucd, and providing ample charge reserves to minimize pumping and supply load current to the load.
My experience to date, if you have a late model UcD, is to leave the board alone, with the exception of shorting the AC coupling caps if you can, and spend your tweaking on the power supply using a transformer with split secondaries and dual bridges. This had far far more sonic affect IMO then playing with lower ESR electrolytics on the board, which IMO was not a positive change over stock.
If you start tweaking on the "on board caps, look at the whole supply management system, thats 6 ferrites, two electrolyitics, to shunt resistors, and 3 ceramic bypass caps. Estimate there parasitics, and see where its at. Then play with values that make it better in some way. Or, you could just "tweak" your circuit and listen. My conclusions after guessing at some values was that lower ESR bypass caps on the board were NOT a clear advantage in the whole audio band. Out of band, it has even more potential negative affect. They caused a potential (potential,becuase I was guessing at the inductances) peak in the upper end of the audio band. (Note: The wierd part is it had very little difference at the lower end of the audio band, since the bulk bypass dominated this.
Also don't forget these affects might already be reduced to a level where secondary affects, like cap microphonics, are more domanant than the idealized lumped parameter models.
I have a dead UcD to take apart, I will try to find the time to actually measure the on board inductors.
Eva, I"d guess you have a clue to what the ESL is for a 16mm diameter electrolytic. Can you give me an estimated value?
I guess I could try to measure that too, but would save some time.
Regards,
Mike
the UcD's do have a very short bypass to ground, 100nF, right at the outputs. Then a 5mohm resistor that does both over current and I suspect damping before the on board electrolytic.
Which by the way to everyone else, that electrolytic has about an INCH of trace before it gets to the outputs.
Also, the heatsink T bar is grounded very short path to the ground plane and thick, low capacitance isulators are used to better decouple the drains from the heatsink.
In a nut shell, Bruno has dotted his i's and crossed his t's on this design per your concerns.
That said, there is no such thing as eleminating high frequency capacitive coupling, only reducing it. Based on Eva's comments, it seems one would be wise NOT to connect two of these to the same hunk of medal. Maybe that's why the "tweaker" that designed the CIaudio amps made them monoblocks. Likely a more significant advantage than minimizing pumping. Who knows for sure? Maybe they just look cooler!
When looking at supplies like for a class D, or a high bandwidth digital system, its more like designing an eliptical RLC filter than just simply tossing in parts with low ESR or making them arbitrarily big capacitance or super low ESR. ESR is clearly part of the equation, not something ideally zero.
The final goal having many compromises, or optimizations if you will, like keeping hf current on the board, and being low impedance to the switches over as wide a band as possible, minimizing impedance peaks, rather than trying to have one really low impedance over a narrow range of frequencies. For example, a really low ESR on board electrolyitic can be shown in simulation to actually cause a boosting of the overall impedance seen by the switches at the upper end of the audio band. (This seems to be in line with Chris and others listening experiences that the highs suffer from lower ESR, although not conclusive, because it has little to no affect on the impedance seen by the switches at low frequencies compared to the notible sonic affect on the percieved bass increase.)
Might mention that Class D are capable of "downconverting" supply anomalies from around the switching frequency. Thus, impedance peaks or dips, at or around the switching frequency can be seen as "in the audio band" affects.
Since there is feedback though, all these affects don't matter!
😉 One could conclude that the goal is to minimize the "maximum" impedance at and above the lowest switching frequency to be expected to a bandwidth of say 10 to 100X that frequency (2MHz to 20Mhz.)
Of all these caps tweaks, the one you play the easiest with w/o much problem, is the main bulk cap, just becuase its decoupled so well from the rest of the circuit by the ferrites. The focus of this IMO would be to minimize potential negative signals getting from the xfmr to the Ucd, and providing ample charge reserves to minimize pumping and supply load current to the load.
My experience to date, if you have a late model UcD, is to leave the board alone, with the exception of shorting the AC coupling caps if you can, and spend your tweaking on the power supply using a transformer with split secondaries and dual bridges. This had far far more sonic affect IMO then playing with lower ESR electrolytics on the board, which IMO was not a positive change over stock.
If you start tweaking on the "on board caps, look at the whole supply management system, thats 6 ferrites, two electrolyitics, to shunt resistors, and 3 ceramic bypass caps. Estimate there parasitics, and see where its at. Then play with values that make it better in some way. Or, you could just "tweak" your circuit and listen. My conclusions after guessing at some values was that lower ESR bypass caps on the board were NOT a clear advantage in the whole audio band. Out of band, it has even more potential negative affect. They caused a potential (potential,becuase I was guessing at the inductances) peak in the upper end of the audio band. (Note: The wierd part is it had very little difference at the lower end of the audio band, since the bulk bypass dominated this.
Also don't forget these affects might already be reduced to a level where secondary affects, like cap microphonics, are more domanant than the idealized lumped parameter models.
I have a dead UcD to take apart, I will try to find the time to actually measure the on board inductors.
Eva, I"d guess you have a clue to what the ESL is for a 16mm diameter electrolytic. Can you give me an estimated value?
I guess I could try to measure that too, but would save some time.
Regards,
Mike
Chris,
In light of Eva's comments, I wonder if your HF degradation was due to using 18mm caps rather than 16mm, thus coupling HF grunge on the shield into the cap shield. (I'm assuming the 180's are like the 400's in that a 18mm will fit, but it almost touches the T bar.
Did you ever try a 16mm FC?
Regards,
Mike
In light of Eva's comments, I wonder if your HF degradation was due to using 18mm caps rather than 16mm, thus coupling HF grunge on the shield into the cap shield. (I'm assuming the 180's are like the 400's in that a 18mm will fit, but it almost touches the T bar.
Did you ever try a 16mm FC?
Regards,
Mike
Eva, Mike, Chris,
Great posts. Thank you for your scientific insights. There is a lot more complexity to power supply & decoupling of the UCD than appears on the surface.
From these last posts the following conclusions can be drawn:
- the UCD modules are already optimised for decoupling
- You need to be experienced & have the relevant test equipment to adjust the decoupling
- The UCD has the experience of one of the most gifted designers, Bruno, built into it already.
I'll stick to more conventional power supply capacitors (the T-Network BHC caps) and leave the decoupling alone on the UCD modules. As I also like the neutral sound of the OPA2134 I wont bother switching out to the AD8620 due to the poor reports. The only tweaks I'll do is shorting the signal coupling cap and swapping the opamp regulator cap for a panasonic FM part.
Eva, I look forward to your information about common mode filters for UCD amps sharing the same chassis. Do you have any thoughts about tweaking the output filter cap?
Regards,
Dean
Great posts. Thank you for your scientific insights. There is a lot more complexity to power supply & decoupling of the UCD than appears on the surface.
From these last posts the following conclusions can be drawn:
- the UCD modules are already optimised for decoupling
- You need to be experienced & have the relevant test equipment to adjust the decoupling
- The UCD has the experience of one of the most gifted designers, Bruno, built into it already.
I'll stick to more conventional power supply capacitors (the T-Network BHC caps) and leave the decoupling alone on the UCD modules. As I also like the neutral sound of the OPA2134 I wont bother switching out to the AD8620 due to the poor reports. The only tweaks I'll do is shorting the signal coupling cap and swapping the opamp regulator cap for a panasonic FM part.
Eva, I look forward to your information about common mode filters for UCD amps sharing the same chassis. Do you have any thoughts about tweaking the output filter cap?
Regards,
Dean
deandob said:
Dean,
I didn't intend to make people think they shouldn't try stuff. My limited testing of bypass caps didn't ever best what was stock, that's my experience and opinion to date. I found that the stuff off the board was much more important. Others have there own findings, and I'd be the last to tell anyone not to try anything (non-distructive) and listen. I don't think low ESR is the most important spec, to be honest, mostly based on my listening experiences. The bench would tell me it does matter, or at best, inconclusive. (did that today too.) I'd admit, it seems like a good spec to optimize( low ESR) , but listening is my final answer. If someone comes up with a zero ESR cap or a 100mohm ESR cap, and it sounds great, I might just give it a try!
Here is a curious example that exemplifies audio to me.
Chris, (who is a listener by default, and one I'd trust since changes he's recommended and predicted the results of have come true when I've done them.) finds that a different cap (happening to have lower ESR and ~20% more capacitance, has an affect on BASS. Significant affect none the less. (Measurable low freq, I would not be suprised to find the difference is not measurable, been there done that, don' t know why.)
Why would that be, when any affect that cap has is going to be in the mid range at best. Say 30mohms, 680uF, that's a zero (or flattening of the impedance) at 7.8kHz. Thus, below that, its impedance is increasing 6dB per octave and above, its flat til ESL kicks in.
WHY WHY WHY, would this make a hill of beans difference in the BASS?
The main supply cap on Chris's system is more like 20,000uF per rail I think. So an extra 200uF is like nothing, and at sub 500Hz, I assure you that the on board inductor is as good as not there compared to the cap impedance.
I don't have a clue to why it has better bass. Worse highs... maybe, at least there is a difference in the supply impedance at and above 8kHz, although one would think different in a good way... but ....... its mechanism is not understood.
Maybe its the phase of the impedance, and sounds better when the current and voltage are more in phase? Don't know why, but that's one difference that would explain it. Its a big comparator, so maybe delays in the voltage screw up the next transition? Don't know.
maybe its nothing to do with ESR but its cap construction, vibration sensitivity, absortion....
Maybe its HF stuff, ringing, noise... at and around the switching frequency that's causing "down converted errors?
I guess my point in this is to for sure undue any thinking that I know what's going on.
I did bother to do some simulations, and the results were inconclusive. Didn't see anything that I'd think would matter in the supply impedance besides a slight potential peaking at ~7kHz and 12 Mhz when you use lower ESR bypass caps.
Maybe that's it, but for sure there range of what is concidered bass, under 1k for sure, is unchanged.
One note to those inclined. Adding a second cap of the same type as stock seems only to help. I'd put it on the bottom and not worry about an extra 10mm of lenth since there already is lots of trace inductance.
Mike
p.s. I agree with Eva. Someone really needs to measure the LF response to see if its different with the big FC's.
I tried the 470ufd FC once. Liked it better than the 680ud FC. A little less bass but the mid's and hi's were still not right. Also the FC caps seem to kill the soundstage. Not as deep, wide or 3D.
Thanks Mike.
I guess its a combination of good engineering and tweaking to taste.
OT - As I'm on holidays I have time to build those audio projects, I have just completed a dual TDA1451a NOS DAC and it sounds much sweeter than the EMU 1820M soundcard I've been using.
Next is to complete the UCD amp to replace an ageing mosfet AB amp - I'll try with minimum tweaking to start with then adjust to taste later.
Regards,
Dean
I guess its a combination of good engineering and tweaking to taste.
OT - As I'm on holidays I have time to build those audio projects, I have just completed a dual TDA1451a NOS DAC and it sounds much sweeter than the EMU 1820M soundcard I've been using.
Next is to complete the UCD amp to replace an ageing mosfet AB amp - I'll try with minimum tweaking to start with then adjust to taste later.
Regards,
Dean
Still OT:
As this is the hottest UCD thread 😉
IF you need a preamp for your UCD amp or another amp (or think that your passive pre has not enough testosterone ) please check out this digitally controlled preamp from our forum-mates from twistedpearaudio.com : the "kookaburra "
Here' the link to the starting thread:
http://www.diyaudio.com/forums/showthread.php?s=&threadid=64215
I only built it last night but my limited listening impressions are very satisfactory: clean, fast and detailed sound, with good F extension and punchy midbas and bass. All for US$62 😎
It drives very well my UCD400.
It can even be configured to work on balanced mode, using two kits.
They are also planning a multichannel preamp for HT, for those interested in that.
End OT
M.
As this is the hottest UCD thread 😉
IF you need a preamp for your UCD amp or another amp (or think that your passive pre has not enough testosterone ) please check out this digitally controlled preamp from our forum-mates from twistedpearaudio.com : the "kookaburra "
Here' the link to the starting thread:
http://www.diyaudio.com/forums/showthread.php?s=&threadid=64215
I only built it last night but my limited listening impressions are very satisfactory: clean, fast and detailed sound, with good F extension and punchy midbas and bass. All for US$62 😎
It drives very well my UCD400.
It can even be configured to work on balanced mode, using two kits.
They are also planning a multichannel preamp for HT, for those interested in that.
End OT
M.
Portlandmike said:
Portlandmike said:
Mike,
Are you aware that the CIAudio "tweakers" claimed to have a significant hand in the design process of the UCD? I'm not sure they'd appreciate you calling them tweakers. Furthermore if they were interested in revealing their "tweaks" they'd have already done so. You can always email them I'm sure.
That said it doesn't take a genius to know that true monoblocks will always be superior. Certainly it will inherently be more forgiving to such a poor layout.
I decided I'd prefer to opt for better components in a stereo setup with a good layout than to use cheap industrial ones in a dual mono config, I haven't yet regretted that decision.
I think it's a stretch to be implying that less prominent high's brought on by a change of caps would be considered as degradation, which I'd consider to be something entirely different.
If I had a scope I'd use it to do some real experimenting with and give an exact answer as to why it makes such a difference.
I guarantee the results would be measurable, start with an SPL meter and go from there.
While I may be a "listener" by default, I don't attempt to ignore the usefullness of measuring equipment, and make no claim that the difference can't be properly observed, and explained, without having to turn to black magic as a result.
Cheers,
Chris
classd4sure said:
Chris,
No assult was intended by calling you a "lister by default". Sorry if it came accross that way.
I don't think Ric would mind. I think he'd admit he'll do something if it sounds better even if he can't quite explain it. Besides, didn't you call him a "tweaker" first 🙂
I concure with your value judgement on stereo vs mono, as I just today recommended the same to a good friend.
Okay, you like the FC's and what they do.
I've just read several comments that concure with my preference that they are not an improvement worth doing in various peoples configurations. I assumed, shame on me, that you just liked the bass improvement more than what the negative "coloration" I think you called it, was percieved to be in the highs.
I'm attempting to get minds working together to try and find a mechanism that explains the improved bass. Theories anyone? (Besided psycho acoustic please!)
I find it hard to believe ESR is doing it unless its the affect of the ESR at the switching frequency. Any theories on how a lower ESR cap would improve bass in the position of bypass? Love to hear from anyone on that.
Look forward to your measurements. You guess more than a 100mDB difference anywhere? Think they'd need to be measured acoustically, or do you think the difference could be measured electrically? Maybe just into a loudspeaker load and measure it acoustically to take into account possible vibration related issue.
Mike
Hi Mike,
Not that I'm in love with the FC's, I just require that my speakers do a little work in the low end, and the FC's do that, while attenuating the high's some. That doesn't mean that they've degraded however. They are crystal clear and fluent, hardly degraded. I have that filter cap which you made fun of to thank for that. In stock configuation the microdetail which offers 3D holographic sound is overly inflated with respect to the meager bass offered. It seems without the ideal cap being found, we have a choice between good bass or good 3d.
If we can't find the one ideal cap, then we need a variety of less ideal caps, for which the pcb design doesn't really allow.
Mike2 stated the FC's do have more high's with a smaller value, which corresponds with my theory. A smaller cap will give more prominent high's, while a bigger cap will give more prominent lows.
He also stated the FC has less 3D.. that's one thing I don't like about them, and I found that as being less prominent high's in the microdetail range, that overlooks the coloration of the highs aspect, but yeah they aren't the most neutral.
I've no doubt with the right equipment this would be easily observable, but you won't be seeing any such measurements from me.
Anyway I would like a cap with more 3D but I'm not willing to give up the bass for it. That just gives too much of a synthesized/fake sound, more like a sound effect than a good system..... pretend hi-fi.
This is a capable module and well designed as we can all agree, but you shouldn't _need_ to go bi-amped just to get some decent bass out of it.
I heard the BC136 was a good cap .... two people who tried them said they didn't like it. So it goes.
Not that I'm in love with the FC's, I just require that my speakers do a little work in the low end, and the FC's do that, while attenuating the high's some. That doesn't mean that they've degraded however. They are crystal clear and fluent, hardly degraded. I have that filter cap which you made fun of to thank for that. In stock configuation the microdetail which offers 3D holographic sound is overly inflated with respect to the meager bass offered. It seems without the ideal cap being found, we have a choice between good bass or good 3d.
If we can't find the one ideal cap, then we need a variety of less ideal caps, for which the pcb design doesn't really allow.
Mike2 stated the FC's do have more high's with a smaller value, which corresponds with my theory. A smaller cap will give more prominent high's, while a bigger cap will give more prominent lows.
He also stated the FC has less 3D.. that's one thing I don't like about them, and I found that as being less prominent high's in the microdetail range, that overlooks the coloration of the highs aspect, but yeah they aren't the most neutral.
I've no doubt with the right equipment this would be easily observable, but you won't be seeing any such measurements from me.
Anyway I would like a cap with more 3D but I'm not willing to give up the bass for it. That just gives too much of a synthesized/fake sound, more like a sound effect than a good system..... pretend hi-fi.
This is a capable module and well designed as we can all agree, but you shouldn't _need_ to go bi-amped just to get some decent bass out of it.
I heard the BC136 was a good cap .... two people who tried them said they didn't like it. So it goes.
classd4sure said:
Chris,
Interesting comments to ponder. Have you tried the stock caps since you did your dual bridge and new supply caps?
I know you did at one point.
Right now, with the stock modules I'm exstatic about the bass.
Its not...... can't find the right word, thump your chest bass, but its very full and deep and musical, and shakes the house. Typannies for example are right there. Acoustic bass sounds live in my living room. I've heard better in different rooms, but never from these speakers. (JBL 4435's)
At any rate, its the best I've had. That said, I didn't change anything on the board and in the other chassis, mono blocks, I they were not as good in any dimension.
I can only attribute it to 3 or maybe 4 things.
1)IE vs Toriod
2)Dual bridge split secondaries vs single bridge center tap
3)Different Mains caps (Denon branded, so who knows) (Oh, yeah, and they had a 100uF first, in parallel with the big 12mF caps, but that shouldn't be it.)
4)And perhaps a well damped chassis, which I know many doubt matters.
Maybe five.
5) Didin't connect input ground to anything at all. (I did this because being a stereo amp, I didn't want to seperate connections to ground. Amp ground is defined by the center tap of the power supply caps.
I have some smallish 63V 470uF FC's. Perhaps that's worth a try. Bruno has mentioned that often caps of one type will sound good at one end of the size spectrum and it can be either big or little and just depends on the type.
So anyways, the 64$ question is why does that FC change the Bass?
Mike
Let me know if you want those parts.
Portlandmike said:
For me also a substancial difference to the original grounding scheme from Hypex. Grounding it at the center tap of the power caps worked out to be the best for me(stereo amp. with 1 toroid and 2 bridges, 4 caps.). Input grounds are floating, from chassis/frame, and only connected through a 100nF cap.(each input ground) to ground at the input cinch parts.
Been busy with a lot of caps. also but was quite disappointed in the outcome. Used a reference model(UCD400amps.) to compare it to after every change and was quite disillusioned of what the mind can do to your sense of improvement of the mods.
Wiring layout and power supply caps. made the biggest differences for me and those were mainly audible as increasing the soundstage.
Bgt said:
Bert,
If your not tired of trying, I suspec two caps similar to the stock caps, (higher ESR) might actually be an improvement all around, if you can fit the second pair on the bottom.
My latest theory is this should be better. That being keeping the phase of the impedance as flat and close to 0 degrees as possible is better. That's the only metric that begins to explain what people hear, although I only can guess as to why. I'll explain my guess if two are better than one 😉
Best Regards,
Mike
Daul mono, yes!
Dual mono is a must. I had one 500 watt Plitron and two sets of diode bridges and two sets of caps.....also a separate transformer and regulators for each channels front end. I have always experienced that mono amps sound better but only expected very little as the modules are so efficient and I had the separate diodes and 10,000 uf Jensens on each rail of each module.....well, I was very wrong!!!!....hooked up another 500 watt Plitron....made a serious difference! Images were another 5-6 inches farther outside the speakers and now have my 22 feet wide soundstage back. Also more dynamic and vivid (once the thing burned in a bit....about 4 hours....at first kind of recessed). I immediately went to my website and removed the one transformer stereo amp and replaced it with a dual mono job....I will only sell dual mono...it is that important. I can only guess it was the ground being common that made the audible difference....by the way, the front end still had common ground through the common regulated supply.
As far as caps go....I pesonally think it is much more complex than impedance. For instance, the ZA, ZL caps from Rubycon have very low impedance. However, Rubycon makes a type with even lower impedance....well, Kyle at Reference Audio Mods tried them and says they sound sucky (he loves and pushes the ZA, ZLs). You could get 10 caps from several manufacturers that had very similar impedance specs and you would get 10 different sounds. Even the Cerafine, Starget, and Silmic caps from Elna all have a very noticable sonic difference....same with the various Blackgates, etc.
You must know that the 10uf cap that is in parallel with the bigger ones makes a difference as well. You guys seem scared because of what some engineer said about ringing, etc. Bruno eventually told someone to try some different ones there....go on, try it.....also there is the bootstrap cap....does it make a difference?...you will never know till you try.
All the parts on the front end as well as the output make a noticeable difference...some here have tried a different op amp...a few have current sourced it (there are various ways to current source...and some sound way better). Hardly anyone has regulated the front end and how many of you have tried to replace the resistors on the front end?...do you really think that a 2 cent surface mount resistor is the best sonically?.....actually, in my tests all brands of surface mount resistors sound different....there is lots to play with here. Then of course there is using a discrete front end or even running a high gain balanced preamp directly into the differential demodulator circuit bypassing the op amps on board. How about battery supplies for the front end....or even the whole amp....
Breath and think before replying...show me some respect...please.
Dual mono is a must. I had one 500 watt Plitron and two sets of diode bridges and two sets of caps.....also a separate transformer and regulators for each channels front end. I have always experienced that mono amps sound better but only expected very little as the modules are so efficient and I had the separate diodes and 10,000 uf Jensens on each rail of each module.....well, I was very wrong!!!!....hooked up another 500 watt Plitron....made a serious difference! Images were another 5-6 inches farther outside the speakers and now have my 22 feet wide soundstage back. Also more dynamic and vivid (once the thing burned in a bit....about 4 hours....at first kind of recessed). I immediately went to my website and removed the one transformer stereo amp and replaced it with a dual mono job....I will only sell dual mono...it is that important. I can only guess it was the ground being common that made the audible difference....by the way, the front end still had common ground through the common regulated supply.
As far as caps go....I pesonally think it is much more complex than impedance. For instance, the ZA, ZL caps from Rubycon have very low impedance. However, Rubycon makes a type with even lower impedance....well, Kyle at Reference Audio Mods tried them and says they sound sucky (he loves and pushes the ZA, ZLs). You could get 10 caps from several manufacturers that had very similar impedance specs and you would get 10 different sounds. Even the Cerafine, Starget, and Silmic caps from Elna all have a very noticable sonic difference....same with the various Blackgates, etc.
You must know that the 10uf cap that is in parallel with the bigger ones makes a difference as well. You guys seem scared because of what some engineer said about ringing, etc. Bruno eventually told someone to try some different ones there....go on, try it.....also there is the bootstrap cap....does it make a difference?...you will never know till you try.
All the parts on the front end as well as the output make a noticeable difference...some here have tried a different op amp...a few have current sourced it (there are various ways to current source...and some sound way better). Hardly anyone has regulated the front end and how many of you have tried to replace the resistors on the front end?...do you really think that a 2 cent surface mount resistor is the best sonically?.....actually, in my tests all brands of surface mount resistors sound different....there is lots to play with here. Then of course there is using a discrete front end or even running a high gain balanced preamp directly into the differential demodulator circuit bypassing the op amps on board. How about battery supplies for the front end....or even the whole amp....
Breath and think before replying...show me some respect...please.
Re: Daul mono, yes!
Ric,
So glad to hear your comments. I'd like to think all will be respectful. I for one very much would like your insights and findings.
I've got some surface mount resistors on order but Mouser apparently takes orders w/o stock and who knows if they will ever come. Got any brands or types you'd recommend. I've found resistors do make a difference too.
by the way, maybe you can pull some strings to get hypex to stop potting the modulator. I blew one up by pure foolishness, and bothered to scrap it all off. Somehow I doubt that those same 2 cent surface mount resistors and caps in the feedback loop are a positive thing! Love to change those out, but experience shows that is some fine expoxy
Would you recommend the Jensen 4 poles over say Cerifines?
I want to upgrade the no-name old bones caps.
I first did monoblocks in some old bones chasssis, the problem is the toriods are center tapped. I found a stereo chassis with dual bridges and only two supply caps was better, but no one could argue that a monoblock might be better. My conclusion is I need to get a split secondary toriod or its not worth the effort.
Its a plan I have, but first I want to try a discrete amp to drive the board and bypass the op amp. Not sure it is best left on board. Seems like putting this "preamp" away from the switching noise would be a feature. What I'm actually hot on trying is just a very simple no NFB diff pair with a discrete output buffer. By engineering metrics, it should suck with lots of low order distortion 😉
Thanks for the tip on the ZL/ZA's. I've grown tired of trying things that don't seem to improve things.
Also, the tip on trying different current sources. I personally like highly degenerated BJT current sources.
Warm Regards,
Mike
Ric Schultz said:
Ric,
So glad to hear your comments. I'd like to think all will be respectful. I for one very much would like your insights and findings.
I've got some surface mount resistors on order but Mouser apparently takes orders w/o stock and who knows if they will ever come. Got any brands or types you'd recommend. I've found resistors do make a difference too.
by the way, maybe you can pull some strings to get hypex to stop potting the modulator. I blew one up by pure foolishness, and bothered to scrap it all off. Somehow I doubt that those same 2 cent surface mount resistors and caps in the feedback loop are a positive thing! Love to change those out, but experience shows that is some fine expoxy
Would you recommend the Jensen 4 poles over say Cerifines?
I want to upgrade the no-name old bones caps.
I first did monoblocks in some old bones chasssis, the problem is the toriods are center tapped. I found a stereo chassis with dual bridges and only two supply caps was better, but no one could argue that a monoblock might be better. My conclusion is I need to get a split secondary toriod or its not worth the effort.
Its a plan I have, but first I want to try a discrete amp to drive the board and bypass the op amp. Not sure it is best left on board. Seems like putting this "preamp" away from the switching noise would be a feature. What I'm actually hot on trying is just a very simple no NFB diff pair with a discrete output buffer. By engineering metrics, it should suck with lots of low order distortion 😉
Thanks for the tip on the ZL/ZA's. I've grown tired of trying things that don't seem to improve things.
Also, the tip on trying different current sources. I personally like highly degenerated BJT current sources.
Warm Regards,
Mike
Re: Daul mono, yes!
In the situation you described with one tx and dual bridges and cap banks, a common ground isn't necessary is it? You could reference ground at the center of your cap banks couldn't you? Anyway, I am doing dual mono too, so it isn't critical, just curious.
I'd still really love to see a drawing of one of the better grounding schemes some use. I get lost in the semantics of the descriptions of grounding schemes. My day job is not engineering related (not in the way you would think anyway) and I have no formal electronics training.
Ric Schultz said:
In the situation you described with one tx and dual bridges and cap banks, a common ground isn't necessary is it? You could reference ground at the center of your cap banks couldn't you? Anyway, I am doing dual mono too, so it isn't critical, just curious.
I'd still really love to see a drawing of one of the better grounding schemes some use. I get lost in the semantics of the descriptions of grounding schemes. My day job is not engineering related (not in the way you would think anyway) and I have no formal electronics training.
- Status
- Not open for further replies.