Olafk,
the voltage drop in this circuit is actually a Vgs of the MOSFET at given Id. There is no way that IRFP240 has Vgs=6V (even less 8V that you previously reported) at Id=1.8A. You probably have fake MOSFET (high ripple that your build shows that you have some small, low gm MOSFET in IRFP240 case).
Also, make sure that you measure both values of input DC voltage as well as output DC voltage at the same time (with circuit loaded) because there could be that you measure the voltage sag of loaded against unloaded circuit. There could be another error in your build - cheap chinese fake capacitors, diodes or something else...
This is such a trivial circuit ...
the voltage drop in this circuit is actually a Vgs of the MOSFET at given Id. There is no way that IRFP240 has Vgs=6V (even less 8V that you previously reported) at Id=1.8A. You probably have fake MOSFET (high ripple that your build shows that you have some small, low gm MOSFET in IRFP240 case).
Also, make sure that you measure both values of input DC voltage as well as output DC voltage at the same time (with circuit loaded) because there could be that you measure the voltage sag of loaded against unloaded circuit. There could be another error in your build - cheap chinese fake capacitors, diodes or something else...
This is such a trivial circuit ...
Hi X,
I have to think about it. All parts in the second module (except the capacitors 1000uF / 63V in the block behind the IRFP's (from China)) are supplied by a standard distributor in Germany. I never had fakes from there.
I need a different measurement setup with a load resistor that can dissipate more heat, so I can measure in peace.
During the week I have little time, so it takes a few more days. But I will definitely stay tuned.
regards Olaf
I have to think about it. All parts in the second module (except the capacitors 1000uF / 63V in the block behind the IRFP's (from China)) are supplied by a standard distributor in Germany. I never had fakes from there.
I need a different measurement setup with a load resistor that can dissipate more heat, so I can measure in peace.
During the week I have little time, so it takes a few more days. But I will definitely stay tuned.
regards Olaf
After building a stock Juma cap multiplier and slightly modded version which was just doubling up on the caps in parallel. The modded version seemed to lower the ripple just a tad less over stock.
In the video youll see the dvm read at 0.001mv sometimes it goes to 0.000 for a sec or two. But the amp is on but there is no volume on yet. Then i increase the volume about 1/4 of the way, and dvm measures 0.001 then it goes to 0.002. Then when i adjust the volume to 3/4 of the way the dvm jumps all over the place as high 0.059mv. I observed lower ripple on hf parts of the track compared to highest ripple measurements on parts of the track that had LF parts like bass guitar, bass drum etc etc.
https://www.youtube.com/watch?v=94pSHKJZPDc
I also tested a CLC filter and it measured at a steady 0.008mv with volume turned up during entire test track.
CLC = 11 x 1000uF / 2.2mh / 10 x 1000uF
Amp = Class A amp, 75W into 8 Ohms
Any suggestions on how to improve the juma cap multiplier to get 0.000-0.0001 mv at full volume etc ?
In the video youll see the dvm read at 0.001mv sometimes it goes to 0.000 for a sec or two. But the amp is on but there is no volume on yet. Then i increase the volume about 1/4 of the way, and dvm measures 0.001 then it goes to 0.002. Then when i adjust the volume to 3/4 of the way the dvm jumps all over the place as high 0.059mv. I observed lower ripple on hf parts of the track compared to highest ripple measurements on parts of the track that had LF parts like bass guitar, bass drum etc etc.
https://www.youtube.com/watch?v=94pSHKJZPDc
I also tested a CLC filter and it measured at a steady 0.008mv with volume turned up during entire test track.
CLC = 11 x 1000uF / 2.2mh / 10 x 1000uF
Amp = Class A amp, 75W into 8 Ohms
Any suggestions on how to improve the juma cap multiplier to get 0.000-0.0001 mv at full volume etc ?
Attachments
Last edited:
If I was getting 0.059mV (59uV) ripple off a basic C-Mx power line filter with the amp producing 3/4 volume, I'd consider that a pretty remarkable achievement - a 'full blown' power regulator would be lucky to get anywhere near that - are you sure you don't mean 59mV ripple at 3/4 volume that would still be a pretty good results?
What classA amp are you using that delivers 75W/8R off 20V rails
There are quite a few more sophisticated C -Mx circuits with 'improved performance' but this doesn't automatically give a better sounding amplifier - it's still just an active filter, not some magic bullet, particularly with such a large sized electro cap on the output
- this circuit is ideal for 'playing around with' to get the 'hang' of how these circuits effect the amp's sound/performance (or any other power supply for that matter)
There's a number of circuits and group buys for pcbs here that are based around the IRFP power fets and others that require more specific choice of components ....
On the "Pink Fish Media' website, there's a long running thread called "TeddyRegs" that's full of variations and explanations -
What classA amp are you using that delivers 75W/8R off 20V rails
There are quite a few more sophisticated C -Mx circuits with 'improved performance' but this doesn't automatically give a better sounding amplifier - it's still just an active filter, not some magic bullet, particularly with such a large sized electro cap on the output
- this circuit is ideal for 'playing around with' to get the 'hang' of how these circuits effect the amp's sound/performance (or any other power supply for that matter)
There's a number of circuits and group buys for pcbs here that are based around the IRFP power fets and others that require more specific choice of components ....
On the "Pink Fish Media' website, there's a long running thread called "TeddyRegs" that's full of variations and explanations -
erik777
nothing is clear about your build (PSU and amp - 75W class A?) and measurements. This cap. multiplier gives 50dB or more of PSRR in range wither than audio band and that's tested and proved since Zen v. 3. There are even older Japanese amps that used it successfully.
You might be measuring music signal interjected into power rails or anything else... Without an oscilloscope it's just stabbing in the dark...
nothing is clear about your build (PSU and amp - 75W class A?) and measurements. This cap. multiplier gives 50dB or more of PSRR in range wither than audio band and that's tested and proved since Zen v. 3. There are even older Japanese amps that used it successfully.
You might be measuring music signal interjected into power rails or anything else... Without an oscilloscope it's just stabbing in the dark...
I doubled checked the dvm, and indeed it measures volts/V. I borrowed a relatives dvm as my unit decided to stop working today. At 3/4 volume it measures approx 7-59 mV or 0.007-0.059 V. Some may ask is that peak to peak or ? I dont know what hewlett packard calculates, as i just measure the ripple by way of the ~ feature.
What is interesting is i just tested from PSU - Cap Mul - CLC - AMP
And im showing 1mv steady with volume maxed out. But im getting a 2V drop from my CLC when connected to Cap Mul. But if i subtract the cap mul there is no V loss. Or if i use the cap mul and subtract the CLC, i just get the normal 5V loss by using the cap mul. So something is going on with the cap and CLC, which wasnt expected. But when again when im measuring the ripple, the cap mul + CLC pushed the ripple down to 1mv and sometimes 0.000 V
So to answer your question i have the eXtremeA designed by Bruno Putzeys. Basically the OPS and IPS are powered separately. The test im doing it for the output stage (OPS) only. The IPS is at +/-30V, using a seperate PSU. This amp is 100watts into 8 ohms but dissapates 250watts of heat. So in order lower the floor heater
ive lowered the OPS V , so im at around 70-75 watts.Here is the eXtremeA documention. There has been some robust improvements made to this amp that have not been added to the eXtreme thread.
Its the best sounding amp ive heard to date in the 75-100 watt class.
Here is an exerpt from the eXtreme doc
"Fig 3. The regulated powersupply schematic.
The “power supply” itself has two separate sections. The high-current supply is straightforward with two bridge rectifiers and a few large buffer capacitors for the two output stages. The low-current supply puts out a higher voltage to insure that the input stage has enough voltage to fully modulate the output stage, hence upping efficiency. Although the PSRR (power supply rejection ratio) of the input stage is quite good, regulation does not hurt and we have opted to use a “discrete regulated power supply” which fares much better than conventional voltage regulators, such as the popular L78xx/L79xx. This power supply design effectively operates both the input- and output stages under optimum conditions, which will result in the excellent performance figures this amplifier is capable of. "
This regulated PSU is very well designed with terrific performance. When i added the CLC to both IPS and OPS, the transparency was out of this world in addition to the LF bass the dark details and bass authority took things to another level in reference SQ.
eXtremeA documention
https://drive.google.com/file/d/0B-Pgv5mk9CttX0MxMm50NEY0THc/view?usp=sharing
Last edited:
If I've understood it correctly, you've added the C-Mx after the rail capacitors (the BHC ALS10s?) that are ripple reduction from the diodes and then added another CLC to these OPS rails, yes? And this extra filtering stage for the OPS has achieved an improvement in the sound?
And now you're looking for a better C-Mx to reduce the ripple further? I'm at a loss for words here but I vaguely remember the Levinson's use active filtering (fully regulated supplies) but it was a fully integrated system, not simple at all - maybe this would be more in the direction you might like to explore
And now you're looking for a better C-Mx to reduce the ripple further? I'm at a loss for words here but I vaguely remember the Levinson's use active filtering (fully regulated supplies) but it was a fully integrated system, not simple at all - maybe this would be more in the direction you might like to explore
Erik777,
Looks like the cap multiplier (Cx) worked out well then? 1mV ripple at full volume is excellent and it's more important to get 0mV at low volumes which is where it counts to bring out low level details. I am not understanding the voltage drop. Are you saying with Cx and CLC, there is less voltage drop? That doesn't t make sense if the case. For my CFH9, having the cap multiplier I get 0mV ripple at quiescent current of 400mA. The low level details I could hear was unbelievable and the hiss/hum at the speaker was inaudible with source turned down to zero. Could not tell the amp was on (my ear touching the driver cones), but for the LEDs and heat pouring out of the heat sinks, the speaker appeared to be either off or not connected.
For the cost of a few IRFP's and modest elco caps - this is one of the best bang for the buck upgrades in sound quality. The other nice upgrade is dual mono block power supplies - superb stereo imaging and separation.
Looks like the cap multiplier (Cx) worked out well then? 1mV ripple at full volume is excellent and it's more important to get 0mV at low volumes which is where it counts to bring out low level details. I am not understanding the voltage drop. Are you saying with Cx and CLC, there is less voltage drop? That doesn't t make sense if the case. For my CFH9, having the cap multiplier I get 0mV ripple at quiescent current of 400mA. The low level details I could hear was unbelievable and the hiss/hum at the speaker was inaudible with source turned down to zero. Could not tell the amp was on (my ear touching the driver cones), but for the LEDs and heat pouring out of the heat sinks, the speaker appeared to be either off or not connected.
For the cost of a few IRFP's and modest elco caps - this is one of the best bang for the buck upgrades in sound quality. The other nice upgrade is dual mono block power supplies - superb stereo imaging and separation.
You ask great questions.
So for clarity, about 9 months ago I found that using a CLC filter as described in an earlier post. Really enhanced the SQ of my Amp across the entire audio frequency but was much more noticable in the Mid to LF, which makes sense since linear PSU's dont do as well here. But I had 4 x CLC in my amp. One CLC for IPS and one CLC for OPS. So when XRK demonstrated great results with this very efficient push pull filter called Jumas cX. I watched the thread and had to try it myself. Hence my activity on this post as this cX is totally new to me. Considering the results, size, and the cost, this eX most impresive, provided you can work around the 5V drop. Ill need to take a trip sometime to a friends place who has a audio analyzer, to really understand whats going on. I just added the cX and the CLC together just to experiment to see if anything would show on the DVM.
But if its worth its grain in salt. I did notice that when i doubled up on the low ESR 15 mOhm caps & paralled them, it dropped it down below what my DVM is able to measure.
This test is OPS stage only.
OPS w/ "no" cX= .140 V or 140 millivolt at +/- 16v
With cX =*.001-.000 V* at +/- 10.64v* ** Modded cX, caps paralleled very low esr 15mohms each
w/ cX =*.001-.002 V* at +/- 10.64v* ** Standard cX
Caps used 680uF 50V *and 6800uF 50v 15mohms*ESR each
Note: Modded version is showing slightly better on DVM
Im unable to do IPS with the cX as i need 30V and im unable raise the V to 35V which ill need to offset the 5V loss with this regulated PSU. Hence my interest in the OPS at this time, until I can find a method to raise my PSU to 35V.
I intend to take everything to a friend who has a audio analyzer and scope
so we can see all thats going on.
Im getting an additional -2V droop by using both Cx and CLC together. Which is odd because the CLC is pass through passive filter. Something must be going on between the Cx and CLC, i guess they dont like each other
Your CLC actually has a "R" buried in the L, perhaps circa 0.4R? That might be causing the additional voltage drop. Without that built in R though, the CLC might oscillate like a bell. (has happened to me until I added 10R in parallel with L).
140mV ripple to 0-1mV ripple is quite impressive. So once we have dealt with the voltage drop or are OK with it - the sound quality difference is staggering.
I don't know too many 100w+ amps where I can put my ear on the mid/tweeter cone and cannot tell that the amp is turned on when no music is playing (as is the case with my CFH9).
It's almost like this device lets you gain a virtual 43dB more PSRR in the amp (140mV to 1mV)?
That is impressive performance enhancement for not much money.
140mV ripple to 0-1mV ripple is quite impressive. So once we have dealt with the voltage drop or are OK with it - the sound quality difference is staggering.
I don't know too many 100w+ amps where I can put my ear on the mid/tweeter cone and cannot tell that the amp is turned on when no music is playing (as is the case with my CFH9).
It's almost like this device lets you gain a virtual 43dB more PSRR in the amp (140mV to 1mV)?
That is impressive performance enhancement for not much money.
Last edited:
Im using the Jantzen 2.2mH 18AWG aircore. The DCR is .84 Ohms, it has no magnetic core, which helps prevent oscillation at high current
as a stand only passive filter. Maybe your right, when coupled with Cx all bets are off, so it could be oscillating.
Last edited:
I hesitate to say this but just maybe adding the extra C-L-C is detrimental to the overall sound despite the rather impressive ripple voltages (or absence of them!)
The ALC10 caps are very well known for their remarkable midrange qualities and smooth 'top end' so adding the choke could be costing you something else in the sound department - I'm guessing a bit here but this happened to me awhile ago and I only picked it up when checking the amp with headphones
For the front end extra filter, the K-multiplier has a much lower voltage drop and is a rather simple circuit and does require quite specific transistors depending on the current but are quite readily available - there also some pcbs available, I think.
As the 'front end' supply uses an active regulator, a simple bc multiplier would possibly achieve better results than the bigger power fets with a lower voltage drop
These filters are more about the 'sound' of the supply rather than any specific ripple reduction - these a great discussion about this around the JLH (John Linsey Hood) ripple eater circuit - it's back a couple of years now so maybe it's been archived but just as valid today
Curiously, when adding a high voltage C-Mx for valve circuits, the humble IRF 840 fet does a very similar job of filtering like the IRFP240/9240s, etc for a ridiculously low component count but it can get a bit complicated with the chokes in the supplies - again, some of these old ideas are still extremely useful but have been forgotten in the trend to embrace high gain, high feedback circuits which can be sometimes a real 'pain in the A' regarding sound reproduction
All the best ...
The ALC10 caps are very well known for their remarkable midrange qualities and smooth 'top end' so adding the choke could be costing you something else in the sound department - I'm guessing a bit here but this happened to me awhile ago and I only picked it up when checking the amp with headphones
For the front end extra filter, the K-multiplier has a much lower voltage drop and is a rather simple circuit and does require quite specific transistors depending on the current but are quite readily available - there also some pcbs available, I think.
As the 'front end' supply uses an active regulator, a simple bc multiplier would possibly achieve better results than the bigger power fets with a lower voltage drop
These filters are more about the 'sound' of the supply rather than any specific ripple reduction - these a great discussion about this around the JLH (John Linsey Hood) ripple eater circuit - it's back a couple of years now so maybe it's been archived but just as valid today
Curiously, when adding a high voltage C-Mx for valve circuits, the humble IRF 840 fet does a very similar job of filtering like the IRFP240/9240s, etc for a ridiculously low component count but it can get a bit complicated with the chokes in the supplies - again, some of these old ideas are still extremely useful but have been forgotten in the trend to embrace high gain, high feedback circuits which can be sometimes a real 'pain in the A' regarding sound reproduction
All the best ...
In an effort to raise the V because of the 5V loss from the Cx. I swapped out my 300 VA 15V to a 200VA 22V. The results are amazing....The VA differance is nothing here but the secondary +7V gained is rather interesting.
Measured at the output of the PSU im getting +/- 22.83V, when connected to Cx = +/- 17.53V
I measured again the ripple and at full volume at max. The DVM measured the ripple = .001-.002V
Which is astonishing....
I could hear a definite improvement in the SQ. I suspect the amp was starving for higher V when at 10V.
Still intend to hook up a scope at some point as well.
Measured at the output of the PSU im getting +/- 22.83V, when connected to Cx = +/- 17.53V
I measured again the ripple and at full volume at max. The DVM measured the ripple = .001-.002V
Which is astonishing....
I could hear a definite improvement in the SQ. I suspect the amp was starving for higher V when at 10V.
Still intend to hook up a scope at some point as well.