sx881663 said:
Roger,
Good points. This is the way to do it if you have a monoblock. If you are planning for two amps, you need to tie at some point of compromise.
FYI, on the 0.1uF on the output windings. From simulations I found that a pure cap accross the secondary to seriously complicate matters, and I would not do that unless you have either a scope, or a way to measure the leakage inductance of the transformer, and then I'd still add an additonal snubber RC.
with no cap, the ring frequency is set by the leakage inductance and parasitic capacitance of the diode and windings.
Adding a cap only loweres the frequency and raises the Q.
Then again, put a spot for it on the board. Maybe it sounds good! I wouldn't be the last person to be suprised.
If you have a spot for series RC, then you also have a spot for a C 🙂
Best Regards,
Mike
Snubber circuit for free!
Mike,
Did you include a reasonable amount of resistance for the secondary winding in your simulation? In actual practice the .1 uf works very well. I have used up to 2.2uf for low voltage transformers and have been impressed with the lower noise floor. Most of the better manufactures do this as a mater of course in their better products.
Roger
Mike,
Did you include a reasonable amount of resistance for the secondary winding in your simulation? In actual practice the .1 uf works very well. I have used up to 2.2uf for low voltage transformers and have been impressed with the lower noise floor. Most of the better manufactures do this as a mater of course in their better products.
Roger
Portlandmike said:
What I noticed with the added gain the wiring in the amps. cabinet played a big role in my crosstalk values. If I sqeezed all the wires in 1 bundle it would destroy my soundstage/depth completely. The soundstage was barely better than 2 mono amps. I used the grounding scheme which I could find on the Hypex site.
Doing the ground like I have now improved the soundstage so much that I was stunned what a difference this could make.
I later got 2 new UCD400 modules and noticed that with normal gain the "capacitive coupling"was so small that it hardly made any difference.
But this showed that whatever you do with 2 very high gain amps. in 1 cabinet the grounding scheme is crucial.
It took me a long time to optimize the soundstage to what it is now. What happens also is when you have a wide sounstage the music is very direct/present. I use a passive preamp and don't use an earthed mains wire in any cabinet. So there are no
groundloops.
PS with the Coldamps, which I got later, it was even worse because the the suggested wiring took the minus of speaker out from the zero point at the PSU instead of off the amps. PCB.
Changing the minus speaker out from PSU to the zero connector at the Coldamps PCB improved the soundstage to what it should be, extremely wide/transparent.
Hi Mike:
Pi filter: is it better to put the inductor between my two Sikorel or between last Sikorel and module's own PS cap?
Dear Ric:
You probably use the same Plitron 500VA TX as I do. Do you use a snubber between secondaries? (no scope here 🙁 ) .
If so, what values?please
Kind regards
M
...and some of them even work! 😀
Pi filter: is it better to put the inductor between my two Sikorel or between last Sikorel and module's own PS cap?
Dear Ric:
You probably use the same Plitron 500VA TX as I do. Do you use a snubber between secondaries? (no scope here 🙁 ) .
If so, what values?please
Kind regards
M
Re: PCB question
Thanks Roger,
I used to have two separate grounds running to the amp modules but I was advised not to do it
I'll also try fitting a 100nf cap and also try a cap/resistor in series across the secondary winding, as always I appreciate any advice fom you guys🙂
sx881663 said:
Thanks Roger,
I used to have two separate grounds running to the amp modules but I was advised not to do it
I'll also try fitting a 100nf cap and also try a cap/resistor in series across the secondary winding, as always I appreciate any advice fom you guys🙂
PCB update
I think this a lot better, the more options the better. Where are the big holes for connecting wires in and out? You should probably cover up to 16ga. wire size. Wish I was close by and could help out.
Roger
t. said:
I think this a lot better, the more options the better. Where are the big holes for connecting wires in and out? You should probably cover up to 16ga. wire size. Wish I was close by and could help out.
Roger
Re: PCB update
Thank you Roger😎
I fitted mounting holes for spade sockets but I can easily add holes for thick gauge wire if that would be better?
sx881663 said:
Thank you Roger😎
I fitted mounting holes for spade sockets but I can easily add holes for thick gauge wire if that would be better?
Re: Snubber circuit for free!
I did have some small, I think it was 100mohm, resistance in the winding. Unless the resistance was on the order of 50 ohms, it didn't totally damp it out. Done right, you could barely see any affect of the ring. Done sub optimally, the ring would go on for say half a dozen cycles. Done pure capacitively, it went on for the whole cycle.
One thing that is unclear though is what the driving impedance of a transformer is. With lots of wiring.... it could be almost anything at high frequencies. That's why I concluded that an X cap on the primary would be best.
I for one will not argue techical superiourity over what sounds good though.
One could argue that getting the resonace down is in deed all that is required, since the supply caps are very low impedances before there ESL kicks in.
The data I took for leakage and modeling the trasformer was a 500VA 35V Avel transformer I had laying around. The bridge was a Diodes Inc, 50V 50A bridge.
I did measure the affect too, but it was very hard to see, since you could only trigger becuase of the big 60Hz, that and turns out there was lots of noise on the lines! Suprised?
I didn't spend much time on measuring it, but simply by the width of the trace I got some idea. I think to do it right you need a high pass to minimize the 50V AC signal. Eva mentioned that putting a small resistor in series with the center tap was a good way to do this measurement. I was doing dual bridge split secondaries, so that didn't help.
One could wire it center tapped, but again, my diodes were 50V, so that would be a bit beyond what I dare do, i.e. 100V on a 50V diode.
REgards,
Mike
Were you using shielded input wires?
I don't run high gain, and I do use the signal ground to sheild with input wires. If you not doing that, then it might help.
I'm also running passive preamp as of this weekend. Haven't listened long enough for a solid comment, but the only thing I might be noticing is a harshness at high volumes in the highs.
Very interesting though, thanks for sharing it.
Mike
Roger,sx881663 said:
I did have some small, I think it was 100mohm, resistance in the winding. Unless the resistance was on the order of 50 ohms, it didn't totally damp it out. Done right, you could barely see any affect of the ring. Done sub optimally, the ring would go on for say half a dozen cycles. Done pure capacitively, it went on for the whole cycle.
One thing that is unclear though is what the driving impedance of a transformer is. With lots of wiring.... it could be almost anything at high frequencies. That's why I concluded that an X cap on the primary would be best.
I for one will not argue techical superiourity over what sounds good though.
One could argue that getting the resonace down is in deed all that is required, since the supply caps are very low impedances before there ESL kicks in.
The data I took for leakage and modeling the trasformer was a 500VA 35V Avel transformer I had laying around. The bridge was a Diodes Inc, 50V 50A bridge.
I did measure the affect too, but it was very hard to see, since you could only trigger becuase of the big 60Hz, that and turns out there was lots of noise on the lines! Suprised?
I didn't spend much time on measuring it, but simply by the width of the trace I got some idea. I think to do it right you need a high pass to minimize the 50V AC signal. Eva mentioned that putting a small resistor in series with the center tap was a good way to do this measurement. I was doing dual bridge split secondaries, so that didn't help.
One could wire it center tapped, but again, my diodes were 50V, so that would be a bit beyond what I dare do, i.e. 100V on a 50V diode.
REgards,
Mike
Bgt said:
Were you using shielded input wires?
I don't run high gain, and I do use the signal ground to sheild with input wires. If you not doing that, then it might help.
I'm also running passive preamp as of this weekend. Haven't listened long enough for a solid comment, but the only thing I might be noticing is a harshness at high volumes in the highs.
Very interesting though, thanks for sharing it.
Mike
maxlorenz said:Hi Mike:
...and some of them even work! 😀
Pi filter: is it better to put the inductor between my two Sikorel or between last Sikorel and module's own PS cap?
Dear Ric:
You probably use the same Plitron 500VA TX as I do. Do you use a snubber between secondaries? (no scope here 🙁 ) .
If so, what values?please
Kind regards
M
Mauricio,
I think pi filters for the primary of the transformer.
What you are doing is different.
Don't put it between the PS and the amp board. There is one there that is pretty optimal I think.
I'm not sure about even putting it between your caps, but this is what's going on.
When you add an inductor between them, you cause a resonance that will be at a low frequency, of coarse depending on the size of the inductor, but low in that those are big caps.
The reason to do this might be to try to gain a lower impedance at the lowest frequencies compared to just two caps.
The penalty is a peak above that dip.
Try it. I'd add some resistance.
Mike
t. said:
t,
Yup, bigger holes for you diodes too. Really to big a hole is not much a problem I think.
If you want to go to the fringe, try adding a slits in the main traces so as to force all the electrons under the cap pads.
Does that make sense?
I've noted this has a very measurable affect at high frequencies, especially on switchers. NOt sure it will help with big inductive caps, but it will not hurt.
Mike
Portlandmike said:
Thanks Mike,
Slits in the main traces😕 I don't quite understand, could you please explain or give me a clue🙂
On the subject Mike would it be worth fitting something like 10uH inductors between the two caps on the rails?
Re: Re: Snubber circuit for free!
Yes, look here:
http://www.diyaudio.com/forums/showthread.php?s=&threadid=65223&highlight=
Portlandmike said:
Yes, look here:
http://www.diyaudio.com/forums/showthread.php?s=&threadid=65223&highlight=
Re: Re: Re: Snubber circuit for free!
The slits.... they would be vertical slits, going up, or comming down from the cap pads, that cut the trace. The incomming signal (power) will then need to flow very close to the pads.
For example, upper right 6,800uF cap, negative terminal, add a slit comming down from the "squarish" copper chunk to cut the rest of the trace.
Traces are not zero impedance, and this addresses this.
6 oz copper is better than 2oz, which is better than 1oz.
So if your trace has parasitic impedance, make it work for you.
Not sure I'd recommend an inductor between them, but put pads in if you like. Cutting a trace is no big deal.
10uH will give a impedance null at ~600hz (with 6,800uF) and a peak at twice that. If you go down that path, you will likely want a resistor to tame the peak. Also, think more like 1mH, which gives a 60 Hz dip, and 120hz peak, a bad choice, but stiffer at deep bass frequencies, or a .25mH which would give a null at 120Hz and a peak at 240Hz.
I don't know if this is good or not. I looked at it with sims a bit, but adding more cap decreases the LF impedance and has no peak! Maybe best played with before you cut a board for it.
Also, and sorry for not mentioning this earlier, you should add pads for say a 100uF cap and a small film, like 100nF in parallel with the big caps and to the left of the big caps.
Best Regards,
Mike
Bgt,
Why the resistor in series with the shield?
Did that help?
Are you connecting the sheild to the input ground?
I'd maybe put the resistor at the other end of the shield, or at least try it. You want your shield to be low impedance.
If that resistor is very low value, its likely fine though.
Sorry, I've lost track of what you have. Is it a dual bridge split secondary for a stereo amp?
Just two bridges, or four?
Your results are very interesting and good clues to what matters.
Regards,
Mike
t. said:
The slits.... they would be vertical slits, going up, or comming down from the cap pads, that cut the trace. The incomming signal (power) will then need to flow very close to the pads.
For example, upper right 6,800uF cap, negative terminal, add a slit comming down from the "squarish" copper chunk to cut the rest of the trace.
Traces are not zero impedance, and this addresses this.
6 oz copper is better than 2oz, which is better than 1oz.
So if your trace has parasitic impedance, make it work for you.
Not sure I'd recommend an inductor between them, but put pads in if you like. Cutting a trace is no big deal.
10uH will give a impedance null at ~600hz (with 6,800uF) and a peak at twice that. If you go down that path, you will likely want a resistor to tame the peak. Also, think more like 1mH, which gives a 60 Hz dip, and 120hz peak, a bad choice, but stiffer at deep bass frequencies, or a .25mH which would give a null at 120Hz and a peak at 240Hz.
I don't know if this is good or not. I looked at it with sims a bit, but adding more cap decreases the LF impedance and has no peak! Maybe best played with before you cut a board for it.
Also, and sorry for not mentioning this earlier, you should add pads for say a 100uF cap and a small film, like 100nF in parallel with the big caps and to the left of the big caps.
Best Regards,
Mike
Bgt said:
Bgt,
Why the resistor in series with the shield?
Did that help?
Are you connecting the sheild to the input ground?
I'd maybe put the resistor at the other end of the shield, or at least try it. You want your shield to be low impedance.
If that resistor is very low value, its likely fine though.
Sorry, I've lost track of what you have. Is it a dual bridge split secondary for a stereo amp?
Just two bridges, or four?
Your results are very interesting and good clues to what matters.
Regards,
Mike
Re: Re: Re: Re: Snubber circuit for free!
I've probably got this totally wrong so please excuse me😀
I can change anything easy so thats no problem, just quickly done the + side first so if its not right I can change it.
Portlandmike said:
I've probably got this totally wrong so please excuse me😀
I can change anything easy so thats no problem, just quickly done the + side first so if its not right I can change it.
Re: Re: Re: Re: Snubber circuit for free!
the resistors in the shield take the humm(groundloop) out. You may put lower values there also. Remember I have the middle of the 4 caps. to ground. This is also my middle leg of the toroid. I use 1 rectifier bridge for each channel but the ground is the 0V of the PSU where the 0V of the power amps are connected 2.
It all comes down to having no humm, no capacitive influences(wire) and great depth. I have spent weeks on end trying out how to get the soundstage wider and wider(improve crosstalk figures). It sounded flat, still nice but not OK enough. And thats with 1 toroid with 2 windings, 2 rects., 4 caps.
PS tried the resistor on the other end of the input cable but it distroyed my crosstalk.
Made the input cable more sensitive to cap. coupling.
Bert
Mike,Portlandmike said:
the resistors in the shield take the humm(groundloop) out. You may put lower values there also. Remember I have the middle of the 4 caps. to ground. This is also my middle leg of the toroid. I use 1 rectifier bridge for each channel but the ground is the 0V of the PSU where the 0V of the power amps are connected 2.
It all comes down to having no humm, no capacitive influences(wire) and great depth. I have spent weeks on end trying out how to get the soundstage wider and wider(improve crosstalk figures). It sounded flat, still nice but not OK enough. And thats with 1 toroid with 2 windings, 2 rects., 4 caps.
PS tried the resistor on the other end of the input cable but it distroyed my crosstalk.
Made the input cable more sensitive to cap. coupling.
Bert
Bruno gave you some good advice on why that resistor is less than optimal and how to do it right. You'd gain tremendously in soundstage by following that advice, and breaking your groundloop elsewhere, such as by floating the supply.
This resistor thing will surely decrease audible hum if you choose to accept a ground loop, but it will reduce the soundstage /depth of field and everything all at the same time.
It's like using a dirty shoelace as a bandaide.
I'd say heed Bruno's advice and try to experiment with breaking the groundloop via other means while keeping the signal path as pure as possible, with the input stage referenced to source ground, or even floating as a last resort.
Mike made that same mistake and noticed vast improvements after changing it. 🙂
This resistor thing will surely decrease audible hum if you choose to accept a ground loop, but it will reduce the soundstage /depth of field and everything all at the same time.
It's like using a dirty shoelace as a bandaide.
I'd say heed Bruno's advice and try to experiment with breaking the groundloop via other means while keeping the signal path as pure as possible, with the input stage referenced to source ground, or even floating as a last resort.
Mike made that same mistake and noticed vast improvements after changing it. 🙂
- Status
- Not open for further replies.