Hotrodding the UCD modules

Status
Not open for further replies.
Jump to Latest
#961
A cap question

t. said:
I wish Rubycon ZL/L caps was available in 63v, these are by far the best caps I've tried in decoupling positions bettering any grade of Blackgates in my opinion.
I've been trying a few different caps in a cdp I'm modding and the ZL/A are brilliant! I read somewhere its even advised to remove the plastic jacket to improve the sound which I thought was some kind of joke but I tried it anyway seeing as though it costs nothing😀 the results was actually very surprising:bigeyes: of course its not recommended to do it if the caps are in close contact with any metal objects but this was not a problem in the cdp
Click to expand...


T,
I could find no references to the ZL/L or ZL/A series caps. Did these just come out? Do you have the data sheets or a link?
What I could find is below; 2 ea. 63V and 100V parts. Not bad but probably not the best.
Roger

http://www.rubycon.co.jp/en/tech/v-...t=63&series=YXG&capa=1500&size_d=18&size_l=40

http://www.rubycon.co.jp/en/tech/v-detail.asp?class=1&volt=63&series=ZL&capa=820&size_d=18&size_l=25

http://www.rubycon.co.jp/en/tech/v-...t=100&series=YXG&capa=820&size_d=18&size_l=40

http://www.rubycon.co.jp/en/tech/v-...t=100&series=YXH&capa=820&size_d=18&size_l=40
 
#962
Re: A cap question

sx881663 said:



T,
I could find no references to the ZL/L or ZL/A series caps. Did these just come out? Do you have the data sheets or a link?
What I could find is below; 2 ea. 63V and 100V parts. Not bad but probably not the best.
Roger

http://www.rubycon.co.jp/en/tech/v-...t=63&series=YXG&capa=1500&size_d=18&size_l=40

http://www.rubycon.co.jp/en/tech/v-detail.asp?class=1&volt=63&series=ZL&capa=820&size_d=18&size_l=25

http://www.rubycon.co.jp/en/tech/v-...t=100&series=YXG&capa=820&size_d=18&size_l=40

http://www.rubycon.co.jp/en/tech/v-...t=100&series=YXH&capa=820&size_d=18&size_l=40
Click to expand...

Sorry Roger, that should have been the Rubycon ZL low esr or Rubycon ZA ultra low ESR, I've tried dozens of caps mainly in lower voltage applications both the ZA and ZL are far superior than any including the BC136 IMO.
I've tried the Rubycon YXF which are normal ESR but I didn't like them, they sounded too forward and bright/metalic.
ZA/ZL was fast and uncoloured in the circuits I used them in, I suppose the only downside is that the leads are magnetic
I'll try and find the data sheet, the ZA are best but the voltages are even lower than the ZL's
 
#963
More on caps

T,
I have been using NIC NRSZ in my low voltage apps and have been quite happy with them. I have also used and liked the Panasonic FC's but for low voltage the NIC's are better. They also have other series in their line up that are equally impressive. I too wish for a higher voltage part!
For tube applications years ago I used and really liked the Rubicon photo flash. They were the only ones I found that would last due to temperature in these apps. Of course they were limited to 360 volts or less. I even used them for off line filtering and never had one fail. This was back before the really big value and small size 200 volt caps were around. These were the forerunner of modern low ESR caps for HV applications.
Roger
 
#964
Re: More on caps

sx881663 said:
T,
I have been using NIC NRSZ in my low voltage apps and have been quite happy with them. I have also used and liked the Panasonic FC's but for low voltage the NIC's are better. They also have other series in their line up that are equally impressive. I too wish for a higher voltage part!
For tube applications years ago I used and really liked the Rubicon photo flash. They were the only ones I found that would last due to temperature in these apps. Of course they were limited to 360 volts or less. I even used them for off line filtering and never had one fail. This was back before the really big value and small size 200 volt caps were around. These were the forerunner of modern low ESR caps for HV applications.
Roger
Click to expand...

Roger,
I'll check those NIC's out http://www.rapidelectronics.co.uk/r...CAT_CODE=31198&STK_PROD_CODE=M71426&XPAGENO=1
they are cheap too🙂
I wanted to try Photo flash caps in my tube amp but they are harder to find now, I also need voltages higher than 450V so I've been using mainly high voltage Evox Rifa's.
I've tried 470uf and a bit higher Panasonic FC's for my UCD180 and left them in for a couple of weeks but I didn't like the signature especially the mid and highs, it smoothed everything out and made it sound too artificial/muddy so fitted the original brown Chemicons back in until I can find something else.
Leo
 
#966
I'd love to know where you can buy Rubycon ZL 63v 1500uF. I did a study of the potential power supply caps for my UCD modules and the ZL on paper looks excellent. I'd like to build an array of 5 x 1500uF ZL 63v per rail for the UCD400, this array would have super low ESR and improve the amp dynamic response with the ability of the PS to dump large currents quickly into the amp.

Here is the comparison table I put together.

Make Cap Dia Leng Ripple @ 100Khz Impedence @ 100K
Pana FC 1500 18 35.5 3095 0.030
Pana FC 1800 18 40 3205 0.025
Nichicon UPW 2200 18 40 3205 0.028
Nichicon HE 1500 2400 0.036
Pana TSHA 2200 2000 0.096
Nichicon KG audio 2200 3200 @120KHz
Nichicon KG 1500 25 35 2.65A @129
Nichicon FX 2200 18 40 1600
Rubycon ZL 1500 18 40 3420 0.018
Richey RPE 1200 18 40 2560 0.032
BHC T Network 10,000 8430 0.016
Nichicon PM 1200 18 40 2210 0.025
Nichicon PW 2200 18 40 3200 0.028
Hitano EXR 2200 3600 0.024

So you can see from above, for its capacitance the ripple current & impedence is quite superior, if you parallel 5 of these you get a combined ripple current and impedence significantly better than say the BHC T Network cap (and at a cheaper price).

Please let me know if you find a source for ZL caps.

Regards,
Dean
 
#968
I've started to ponder some of the why this cap sounds that way.

I did a simulation with ltspice, making some vague assumptions about wire length, the inductance of the caps, and the on board ferrites.

What I see is that very low esr caps in place of the 470uF's cause a resonance in the 10-30kHz range. Assuming 1uH of wire plus ferrite inductance. If its way less (and this was just a quick sim in the kitchen with while cooking) then it may cause a peak around switching frequencies, which might be more detrimental.

Again, I'm guessing at the contribution of the on board ferrites.

I suspect that the complaint that when you go to the big FC's the bass gets better but the highs gets colored is a result of this.

It might be interesting to do a supply that is very close to the UcD and short the ferrites, OR replace them with a small value resistor, like 100mohm or so.

There are also a couple of ceramic caps on the board bypassing the main rails. (This is after the 5mohm shunts.)
They can cause a resonance in the ~10Mhz range. More important, it gets very significant as the ESR of the 470uF slot gets low.

I should have some bench data next week if all goes as planned.

Just keep in mind that ultra low ESR may not be the final word on optimal.

Also, it may actually be better to keep the capacitance lower, like 470uF, and go with bigger packages (by using higher voltage ratings) to get the low ESR. Then at least the impedance bump is higher and almost out of band. I'm not convinced this is THEE issue though. As the phase of the impedance may actually be more important.

I can say this. If one intends to do an array of low ESR caps, you may want to pay attention to these affects and think about removing or moving the ferrites, and or replacing them with a low impedance resistor. Keeping the connecting wires very short would likely be only better.

I think in the stock arangement, bruno has protected the user from these affects by using a fairly high ESR cap for bypass, but if you change that....

Clearly more cap and lower ESR is good (except for the high frequency anti-resonance with the on board ceramic bypass caps (I guess 100nF, but its only a guess. They are big ~.180 x .100", X7R or maybe a very low value of NPO. No matter what, they are going to cause a womping anti-resonance if the main bypass electrolytics have very low ESR, but then again, maybe 10Mhz is out of band. Again, all these are guestimates, but the affects are there.

Regards,

Mike
 
#969
Cap discussion

Dean,
Thanks for all the good info, I had no idea the “ZL's” were that good. That they are so close to the “BHC’s” speaks very well for them. The string of “ZL’s” you talk of might be a superior solution if you do a snubber to cancel the lead wire inductive effects.

Mike,
Can you try a variation with the ceramic cap right on the bigger caps pad? The thought comes to mind this might actually cancel the resonance or at least lower it and move it to very high freq. If the parasitic inductance is so low the bypass cap its self is going inductive and increasing in impedance, this might work. This leads toward using a larger value to move its self resonance even lower to be sure of this effect actually happening. Also what about a really stiff snubber like 1 ohm and .1uf? The parts could be surface mount and actually soldered right together for really low inductance.
Roger
 
#970
Re: Cap discussion

sx881663 said:
Dean,
Thanks for all the good info, I had no idea the “ZL's” were that good. That they are so close to the “BHC’s” speaks very well for them. The string of “ZL’s” you talk of might be a superior solution if you do a snubber to cancel the lead wire inductive effects.

Mike,
Can you try a variation with the ceramic cap right on the bigger caps pad? The thought comes to mind this might actually cancel the resonance or at least lower it and move it to very high freq. If the parasitic inductance is so low the bypass cap its self is going inductive and increasing in impedance, this might work. This leads toward using a larger value to move its self resonance even lower to be sure of this effect actually happening. Also what about a really stiff snubber like 1 ohm and .1uf? The parts could be surface mount and actually soldered right together for really low inductance.
Roger
Click to expand...

Roger,

The sims at home right now. I'll check into it.
My gut tells me it will be worse because the on board ceramic has a 5mohm resistance in series with it. That is a snubber if you will.

1 ohm will do nothing when your talking about ESR's in the range of a few milliohms.
My point is the "what sounds best" might be more directed.

I think one might gain insight into lots of caps by just seeing if two is better than one first.

I suspect that there is more than one interaction happening.

Why for example, would lower ESR FC for example, help the bass?
Seems bizzare to me since the impedance of a 680uF cap at 100Hz is 2.3ohms. Clearly the stiffness at 100Hz is comming from the bulk cap off the recifier. Typical values of 10,000uF is going to be the domanant impedance at 100Hz. So, what gives with better bass from lower ESR in a rather small portion of the total bypass. Clearly its not the number of uF's present.

This being a switcher though, its likely subtle differences is the noise on the rails and how they interact with the comparator input. That's why I suspect phase really would matter.

I intend to look closely with a scope next week, but honestly, I doubt I'll see anything at all. My experience with these, "better bass" sound type things is they are rarely measureable.

Regards,

Mike
 
#971
Portlandmike said:
I've started to ponder some of the why this cap sounds that way.



It might be interesting to do a supply that is very close to the UcD and short the ferrites, OR replace them with a small value resistor, like 100mohm or so.


Regards,

Mike
Click to expand...


Hi Mike,

Those ferrites keep a lot of junk from getting out of the UcD module. I actually measured waverforms over the 470uF caps (posted somewhere here) and I also measured the waveform after the ferrites (so at the power supply pins), I did not post those results but I can tell you that the reduction is huge, maybe able to post something sometime in the weekend or so. After the ferrites (so at the supply pins), I see a remaining resonance (high Q) that maybe caused by the ferrites in combination with the main caps (Elna cerafine 63V 10000uF) and the wiring from main caps to UcD. I think those can probably be suppressed by adding a snubber at either the main caps or close to the UcD itself. I have not experimented with a subber in that position yet, that is the reason why I did not post any results yet.

All measurements I did so far are without input signal. I could imagine as you say that the combination of the ferrites and main caps could give an LC filter with a certain Q-factor in the audible frequency range (actually have done some simulations on additional LC filters to be used after an SMPS that indicate such results) this could be an explanation fo why different main caps sound different as the ESR of the main caps would influence that Q-factor as would wiring etc etc. Actually you can easily get multiple resonant peaks. For example one in the audible frequency range determined by the L of the ferrites and the C of the main cap and one at a higher range determined by the parasitic inductance of the main cap (+ wiring) and the ceramics on the UcD board. At least this is what I suspect and some simulations point in this direction but have not confirmed it by doing mods in that direction, such as adding snubbers and trying to find the resonance peak in the audible frequency range (was planning to do that by feeding the UcD with a 1kHz square wave and look at the waverform on the supply rails to see whether there is ringing on the supply rails).

Best regards

Gertjan
 
#972
ghemink said:



Hi Mike,

Those ferrites keep a lot of junk from getting out of the UcD module. I actually measured waverforms over the 470uF caps (posted somewhere here) and I also measured the waveform after the ferrites (so at the power supply pins), I did not post those results but I can tell you that the reduction is huge, maybe able to post something sometime in the weekend or so. After the ferrites (so at the supply pins), I see a remaining resonance (high Q) that maybe caused by the ferrites in combination with the main caps (Elna cerafine 63V 10000uF) and the wiring from main caps to UcD. I think those can probably be suppressed by adding a snubber at either the main caps or close to the UcD itself. I have not experimented with a subber in that position yet, that is the reason why I did not post any results yet.

All measurements I did so far are without input signal. I could imagine as you say that the combination of the ferrites and main caps could give an LC filter with a certain Q-factor in the audible frequency range (actually have done some simulations on additional LC filters to be used after an SMPS that indicate such results) this could be an explanation fo why different main caps sound different as the ESR of the main caps would influence that Q-factor as would wiring etc etc. Actually you can easily get multiple resonant peaks. For example one in the audible frequency range determined by the L of the ferrites and the C of the main cap and one at a higher range determined by the parasitic inductance of the main cap (+ wiring) and the ceramics on the UcD board. At least this is what I suspect and some simulations point in this direction but have not confirmed it by doing mods in that direction, such as adding snubbers and trying to find the resonance peak in the audible frequency range (was planning to do that by feeding the UcD with a 1kHz square wave and look at the waverform on the supply rails to see whether there is ringing on the supply rails).

Best regards

Gertjan
Click to expand...


I had saved some of my measurements on my PC. The atteched picture shows the AC component of the + rail voltage measured over the 470uF cap directly and at the UCD power supply pin. Note that the scales are different other wise you would hardly see the ripple at the power pins.

Best regards

Gertjan
 

Attachments

  • rail.jpg
    rail.jpg
    77.2 KB · Views: 356
#973
ghemink said:



I had saved some of my measurements on my PC. The atteched picture shows the AC component of the + rail voltage measured over the 470uF cap directly and at the UCD power supply pin. Note that the scales are different other wise you would hardly see the ripple at the power pins.

Best regards

Gertjan
Click to expand...


Correction, not over the 470uF cap but over a Panasonic FC 680uF 100V that I put in place of the standard 470uF caps. Actually the waveform (the spiky one) was measured over the 10uF snubber caps. There is a 5mOhm resistor in between that cap and the panasonic FC.

Best regards

Gertjan
 
#977
ghemink said:



I had saved some of my measurements on my PC. The atteched picture shows the AC component of the + rail voltage measured over the 470uF cap directly and at the UCD power supply pin. Note that the scales are different other wise you would hardly see the ripple at the power pins.

Best regards

Gertjan
Click to expand...


Another correction, it seems that the timescale is not correct in that attached figure. It seems that my scope software does not load the x-axis data correctly fro some reason when I load stored data. Anyway, the waverforms are correct 🙂

Gertjan
 
Status
Not open for further replies.
Top Bottom