Portlandmike said:
Maybe this is a dumb question, but is it really that important to have them flush against the PCB?
I'm looking for an upgrade (I'm a bit suspicious of the Panasonic FC because of some of the feedback on the sound), but it seems the 'better ones' are usually too big which would make them very close/touching the heatsink and that can't be good. Fitting them on the back would be much easier. Also, it shouldn't be to hard to fit them on the back with leads just about as long as fitting them on the top of the pcb, taking the thickness of the pcb and of th smd components into account...
Hans.
Re: Output impedance
Mike,
This is a valid point. All circuits like to see an optimum load. With cable capacitance acting as a very low impedance at high frequency the driving circuit can become unstable to the point of ringing or even oscillating. If you look at most opamp spec sheets they recommend an isolating resistor on the output. Most of the time this will be from 10 to 50 ohms but with typical high impedance preamp circuits I can see 500 ohms being better suited. I typically use 30 ohms for opamp circuits as a good compromise. This requirement will be true with any cable driving circuit that has overall negative feedback. The best thing is to get around this problem with a no feedback unity gain buffer. Then you can use very low value resistors and have really low source impedance.
The objection to using a resistor isolator this way is that it raises the source impedance and causes cable sensitivity. This can range from practically no effect to rather severe HF roll off with its dynamic compression and loss of life like sound. At this point you must find a set of cables that will work and sound ok. These high impedance circuits really limit your choice of cables! This is putting a band-aid on the problem and the best results are far from optimum. Best to fix the problem first!
I can speak with some authority on this as I designed a cable driver circuit for Audience that had 3 ohms output impedance. This was a discrete component buffer that was capable of driving 5 watts into 75 ohms. The effect was greatly reduced cable dependency and the ability to drive low inductance cables that sound so much better. In all cases the buffer’s effect was from very positive to little effect but in every case an improvement.
Roger
Portlandmike said:
Mike,
This is a valid point. All circuits like to see an optimum load. With cable capacitance acting as a very low impedance at high frequency the driving circuit can become unstable to the point of ringing or even oscillating. If you look at most opamp spec sheets they recommend an isolating resistor on the output. Most of the time this will be from 10 to 50 ohms but with typical high impedance preamp circuits I can see 500 ohms being better suited. I typically use 30 ohms for opamp circuits as a good compromise. This requirement will be true with any cable driving circuit that has overall negative feedback. The best thing is to get around this problem with a no feedback unity gain buffer. Then you can use very low value resistors and have really low source impedance.
The objection to using a resistor isolator this way is that it raises the source impedance and causes cable sensitivity. This can range from practically no effect to rather severe HF roll off with its dynamic compression and loss of life like sound. At this point you must find a set of cables that will work and sound ok. These high impedance circuits really limit your choice of cables! This is putting a band-aid on the problem and the best results are far from optimum. Best to fix the problem first!
I can speak with some authority on this as I designed a cable driver circuit for Audience that had 3 ohms output impedance. This was a discrete component buffer that was capable of driving 5 watts into 75 ohms. The effect was greatly reduced cable dependency and the ability to drive low inductance cables that sound so much better. In all cases the buffer’s effect was from very positive to little effect but in every case an improvement.
Roger
Re: Re: Output impedance
Hi Roger,
It wasn't the preamps output impedance that was the issue though, but I hear you loud and clear 🙂, as I used it before with out AC coupling into an amp with AC coupling, same cables. Its a pretty stiff output impedance, off hand, a few ohms. I started to always put a 50~100ohm resistor in series with the output just becuase audio equipment in general ignores reflections and cable termination and doesn't terminate the other end of the cable. (as you well know from your project!)
That wasn't enough to tame the coupling cap affect though.
What suprised me was increasing this made it so much better with AC coupling. The preamp bandwidth is about 10Mhz and zero NFB before I limit the bandwidth, and the series resistor is most all that is required. It can drive some low impedances, but no where near what you did! Two transistors, three resistors, THD below 110dB! But..... it has a Vbe offset 🙁
I''ve pondered a complementary darlington version that simulates quite well and can have zero offset (trimmed.) but that's going to happed after my single ended to diff output UcD op amp eleminator 🙂
btw, I'm buying a HP4195A and HP41951A impedance analyzer for work, so maybe I'll be posting some stuff in the not so distant future! 🙂
Mike
sx881663 said:
Hi Roger,
It wasn't the preamps output impedance that was the issue though, but I hear you loud and clear 🙂, as I used it before with out AC coupling into an amp with AC coupling, same cables. Its a pretty stiff output impedance, off hand, a few ohms. I started to always put a 50~100ohm resistor in series with the output just becuase audio equipment in general ignores reflections and cable termination and doesn't terminate the other end of the cable. (as you well know from your project!)
That wasn't enough to tame the coupling cap affect though.
What suprised me was increasing this made it so much better with AC coupling. The preamp bandwidth is about 10Mhz and zero NFB before I limit the bandwidth, and the series resistor is most all that is required. It can drive some low impedances, but no where near what you did! Two transistors, three resistors, THD below 110dB! But..... it has a Vbe offset 🙁
I''ve pondered a complementary darlington version that simulates quite well and can have zero offset (trimmed.) but that's going to happed after my single ended to diff output UcD op amp eleminator 🙂
btw, I'm buying a HP4195A and HP41951A impedance analyzer for work, so maybe I'll be posting some stuff in the not so distant future! 🙂
Mike
Cheap 4 pole caps?
Hey guys,
Just ran accross this in the digikey catalog:
http://dkc3.digikey.com/PDF/T061/1039.pdf
Looks suspect of perhaps being a 4 pole.
Look at the "NEW! 5 pin terminal" figure.
Not a bad cap with its ESR either, so this might be my default.
I'm thinking the 80V, 18000uF 23mohm one!
Not the cheapest, but if it is a 4 pole, a bargin at $22!
Mike
Hey guys,
Just ran accross this in the digikey catalog:
http://dkc3.digikey.com/PDF/T061/1039.pdf
Looks suspect of perhaps being a 4 pole.
Look at the "NEW! 5 pin terminal" figure.
Not a bad cap with its ESR either, so this might be my default.
I'm thinking the 80V, 18000uF 23mohm one!
Not the cheapest, but if it is a 4 pole, a bargin at $22!
Mike
Panasonics T-UP Caps
Hello Mike,
I'm afraid that the panasonic T-UP caps are convertional caps with more points to connect to just to get a better mechanical connection. See web page below.
http://www.panasonic.com/industrial/components/pdf/brochure_power_supply.pdf
Regards,
Greg
Hello Mike,
I'm afraid that the panasonic T-UP caps are convertional caps with more points to connect to just to get a better mechanical connection. See web page below.
http://www.panasonic.com/industrial/components/pdf/brochure_power_supply.pdf
Regards,
Greg
Re: Panasonics T-UP Caps
Hi Greg,
Did you look at the new 5 pole versions.
You still may be right in that internally the two connections likely tie to the film at the same point, unlike the Jensens and Tcaps that tie in at different places.
I wonder why the 5 pin versions, with two pins for both pos and neg, cost almost twice the price? i.e. ~$12 for 4 pin nad 22 for 5 pin?
Also, and hey, again, it would be almost to good to be true, but they call the pins (in the digikey catalog) Pos, and pos isolate and neg and neg isolate. Curious.
Mike
p.s. If I missed something in the link, sorry, all I saw was terninal T-UP's. The p/n that are 5 ternimal are called ECE-P vs the 4 pole refered to in the DS as ECE-T.
starkeyg said:
Hi Greg,
Did you look at the new 5 pole versions.
You still may be right in that internally the two connections likely tie to the film at the same point, unlike the Jensens and Tcaps that tie in at different places.
I wonder why the 5 pin versions, with two pins for both pos and neg, cost almost twice the price? i.e. ~$12 for 4 pin nad 22 for 5 pin?
Also, and hey, again, it would be almost to good to be true, but they call the pins (in the digikey catalog) Pos, and pos isolate and neg and neg isolate. Curious.
Mike
p.s. If I missed something in the link, sorry, all I saw was terninal T-UP's. The p/n that are 5 ternimal are called ECE-P vs the 4 pole refered to in the DS as ECE-T.
Tweaking UCD400
Hi Mike,
I'm told those Jensens are about $25 each but well worth it. I guess you need 2 of them, per channel. So how did the LM6172 finally work out? I was thinking of buying a kit from Kevin DIYCables and putting it together. Then once done, do any mauds in stages. Looks like these Lima caps might be worth checking out.
Ray
Hi Mike,
I'm told those Jensens are about $25 each but well worth it. I guess you need 2 of them, per channel. So how did the LM6172 finally work out? I was thinking of buying a kit from Kevin DIYCables and putting it together. Then once done, do any mauds in stages. Looks like these Lima caps might be worth checking out.
Ray
Re: Tweaking UCD400
t,
I'll let you know if they are just two connections or have some benifit. I'll find some way to test them. They have decent enough HF specs that I'll try them on a budget.
Hi Ray,
If you know a place to get Jensens for $25 bucks, I'm all over it, but the best I've seen is North of $50, and that for only 10,000uF, 80V.
My curiosity is peaked on the 5 pins, there going to be in the lab amp for work so I can check them out 🙂
As for the 6172's, I haven't got to them yet. I still need to get some decent coupling caps.
Mike
btw,
I'm looking at one of these two trannies for monoblocks:
http://cgi.ebay.com/45-45V-700VA-To...QitemZ7540251656QQcategoryZ4660QQcmdZViewItem
http://cgi.ebay.com/80V-CT-800VA-To...QitemZ7612499695QQcategoryZ4660QQcmdZViewItem
I'll again, try them in the lab amp first, single trannie, stereo, and see how they work and let everyone know. Don't hold your breath though. I don't move fast.
t. said:
ray bronk said:
t,
I'll let you know if they are just two connections or have some benifit. I'll find some way to test them. They have decent enough HF specs that I'll try them on a budget.
Hi Ray,
If you know a place to get Jensens for $25 bucks, I'm all over it, but the best I've seen is North of $50, and that for only 10,000uF, 80V.
My curiosity is peaked on the 5 pins, there going to be in the lab amp for work so I can check them out 🙂
As for the 6172's, I haven't got to them yet. I still need to get some decent coupling caps.
Mike
btw,
I'm looking at one of these two trannies for monoblocks:
http://cgi.ebay.com/45-45V-700VA-To...QitemZ7540251656QQcategoryZ4660QQcmdZViewItem
http://cgi.ebay.com/80V-CT-800VA-To...QitemZ7612499695QQcategoryZ4660QQcmdZViewItem
I'll again, try them in the lab amp first, single trannie, stereo, and see how they work and let everyone know. Don't hold your breath though. I don't move fast.
No
Hi,
@Mike
"1) You want to pass 120Hz current spike to get power to the amp.
2) You want to attenuate HF stuff that the transformer is coupling or is created by the rectifiers.
3) You want to present the lowest possible impedance to the amp in relation to its PS inputs."
1) No, You don´t want any spike to get to the amp. 100/120Hz Spikes are hum, i.e. noise. Just simple and clean DC...nothing else is our wish
2) high-frequent noise generated by the rectifier should be treated right there...at the pins of the rectifier, using a RC-snubber, or at least some C-snubber.
3) Yes and thats the reason to use a low ohmic and inductive layout. The circuit likes to be powered by a voltage source -->ideally 0Ohms, 0mH output impedance. The Pi-Filter has the advantage that the relatively small first cap and the inductance filter spikes much better than just a cap. The stress on the rectifier and therefore the generated noise is lower because of lower loading currents. The cap at the PS-output serves as the power reservoire and the low impedance drive point for the circuit, therefore it can be built with big low ESR caps. Of course the PCB layout has to be a low ohmic and low inductive one.
"relitive to its ~1uH inductor on a UcD." Don´t know what You mean??? I´m not familiar with the actual implementation of the UcD (and sorry, no, having a low speed and quite costly internet connection I won´t read all 119 pages of just this thread to be updated
)
So could You just explain which inductor You mean??? A circuit schemativ would be very helpful too. I´ve got this one, which is only showing the amp´s circuit (guess its the UcD180)
jauu
Calvin
Hi,
@Mike
"1) You want to pass 120Hz current spike to get power to the amp.
2) You want to attenuate HF stuff that the transformer is coupling or is created by the rectifiers.
3) You want to present the lowest possible impedance to the amp in relation to its PS inputs."
1) No, You don´t want any spike to get to the amp. 100/120Hz Spikes are hum, i.e. noise. Just simple and clean DC...nothing else is our wish
2) high-frequent noise generated by the rectifier should be treated right there...at the pins of the rectifier, using a RC-snubber, or at least some C-snubber.
3) Yes and thats the reason to use a low ohmic and inductive layout. The circuit likes to be powered by a voltage source -->ideally 0Ohms, 0mH output impedance. The Pi-Filter has the advantage that the relatively small first cap and the inductance filter spikes much better than just a cap. The stress on the rectifier and therefore the generated noise is lower because of lower loading currents. The cap at the PS-output serves as the power reservoire and the low impedance drive point for the circuit, therefore it can be built with big low ESR caps. Of course the PCB layout has to be a low ohmic and low inductive one.
"relitive to its ~1uH inductor on a UcD." Don´t know what You mean??? I´m not familiar with the actual implementation of the UcD (and sorry, no, having a low speed and quite costly internet connection I won´t read all 119 pages of just this thread to be updated
)So could You just explain which inductor You mean??? A circuit schemativ would be very helpful too. I´ve got this one, which is only showing the amp´s circuit (guess its the UcD180)
jauu
Calvin
classd4sure said:
Bypassing them with film caps can give nasty ringing, I saw it on a scope when tried for fun to bypass an FC with an WIMA MKP2! I think I even posted the results somewhere here. So be careful.
On paper, the rubycon ZL 63V caps look very good as well, may try them if I can get them somewhere.
Gertjan
Hans L said:
Hello Hans,
Having seen on a scope how nasty the signal over those caps looks, I would for sure want to keep the wiring as short as possible, so mounting them flush on the PCB would be preferred. For the same reason, I also like the wima MKP2 output caps as they are small enough to lush mount them at the top and bottom of the PCB.
Gertjan
Re: Re: layout
Hi Mike,
Exactly because of the point you make above, I started experimenting with an SMPS.
1. SMPS has smaller caps after the rectifier, so smaller peak currents.
2. The SMPS regulates, so output voltage could be more stable while those peak currents can be kept lower
3. One could still add more caps at the output of the SMPS to avoid pumping Iin such case, the feedback network in the SMPS may need to be adapted to avoid oscillations.
I would not do a zener based regulator, why not use a capacitance multiplier (if you do not want to use an SMPS), this could reduce peak currents as well, you will burn less power and need less heat sinking for the regulater (in comparison with a zener based approach). You would still have to add quite some capacitance after that capacitance multiplier to avoid pumping, same as you would need with a zener based approach.
Best regards
Gertjan
Portlandmike said:
Hi Mike,
Exactly because of the point you make above, I started experimenting with an SMPS.
1. SMPS has smaller caps after the rectifier, so smaller peak currents.
2. The SMPS regulates, so output voltage could be more stable while those peak currents can be kept lower
3. One could still add more caps at the output of the SMPS to avoid pumping Iin such case, the feedback network in the SMPS may need to be adapted to avoid oscillations.
I would not do a zener based regulator, why not use a capacitance multiplier (if you do not want to use an SMPS), this could reduce peak currents as well, you will burn less power and need less heat sinking for the regulater (in comparison with a zener based approach). You would still have to add quite some capacitance after that capacitance multiplier to avoid pumping, same as you would need with a zener based approach.
Best regards
Gertjan
ghemink said:
Rubycon ZL's are my favourite cap, my modded cdp is filled with them, much better than Panasonic FC's
I like them better than the BC136 and even BlackgatesThe 63v ones are hard to find, i'd love to stick some in my UCD's
t. said:
Rubycon ZL's are my favourite cap, my modded cdp is filled with them, much better than Panasonic FC's
The 63v ones are hard to find, i'd love to stick some in my UCD's
If I can't find a source for them on the internet, I may have to go to Akihabara to try to find them (a 1 hr train ride from where I live), may not be easy to find them but I have to admit, I've never looked for them until I heard of them at this forum.
Gertjan
ghemink said:
Well good luck with finding them, I've been unlucky so far, the 50v ones are easier but I wouldn't even want to risk them in my UCD180 which has around 45/46v dc
ghemink said:
Well good luck with finding them, I've been unlucky so far, the 50v ones are easier but I wouldn't even want to risk them in my UCD180 which has around 45/46v dc
Hi all,
Are these Lima caps to be used as the two filter caps on the board, as well as replacing the .068 output cap? Any ideas of the 63V or above variety where can I get them? Thanks.
It seems that these two areas might be the place to start any mods. Thanks.
Ray
Are these Lima caps to be used as the two filter caps on the board, as well as replacing the .068 output cap? Any ideas of the 63V or above variety where can I get them? Thanks.
It seems that these two areas might be the place to start any mods. Thanks.
Ray
cap ratings
I have no problem running caps very near their rated voltage. For about 5 years I had mono amps with 50V caps and 47V rails...no problem ever. Using 22,000 uf Cerafine and some other brand of 10,000 uf for the front end.
Are Lima caps from Peru? How about those German Vunder caps! he he.
I have no problem running caps very near their rated voltage. For about 5 years I had mono amps with 50V caps and 47V rails...no problem ever. Using 22,000 uf Cerafine and some other brand of 10,000 uf for the front end.
Are Lima caps from Peru? How about those German Vunder caps! he he.
- Status
- Not open for further replies.