It's exactly as it should be. The important thing is: do you hear the noise/hum through the speakers ? You shouldn't. And if you do - fix you wiring.
If you want to significantly improve the ripple figure, go active - I had great results with capacitance multiplier:
An externally hosted image should be here but it was not working when we last tested it.
Digi-Key sells them
for the n channel - IXFN180N10 IXYS | IXFN180N10-ND | DigiKey
for the p channel - IXTN170P10P IXYS | IXTN170P10P-ND | DigiKey
If I recall correctly, there were other options with lower Ciss, but the price jumped up quite a bit.
I prototyped the IXFN180n10 in an Aleph style amp and the sonics were amazing - best amp I've ever heard.
Last edited:
Back, briefly, to the shorted toroid discussion: Because there are two out (actually I mean IN phase) of phase toroid (from current in loop perspective) in this layout there will be no current if both transformers have exactly the same primary turns. It's still not a great layout idea, the original one here, that is. Turn one of the cores over and you'd likely see smoke in a minute. There'd be no way to miss it. 
Last edited:
Before making any changes, I checked the voltage drop across the 0R51 resistors.
After the F5 was on for 2 hours, the voltage drop is still sitting there at 0.69v.
That's without the thermistors
Zen,
I 1/2 'ed the value of the source resistors for Q1 and Q2.
So R1 = 10 // 10 and R2 = 10 // 10
This made a significant improvement to the dynamics and bass.
The F.B. resistors were keep at the original values of 100 // 100 and 100 // 100
The trimmers became overly sensitive, so I'll need to re-adjust the value of the drain resistors for Q1 and Q2.
The mid range sounds very dynamic - but there is still not enough bass.
.
Juma,
There is no hiss or hum.
If music is Not Playing, I can put my ear right up to the tweeter and can't tell if the amp is on or off.
Its only when the Cello starts that I can tell if the amp is on.
However, when music is playing, I can hear a hit of harshness in the high frequencies.
Don't know if that's caused by the 90mVpp ripple in the voltage rails, or digital glare from the CD source.
In terms of voltage regulation - I was thinking of a compromise.
How about voltage regulation for the voltage gain stage and leave the power supply passive for the current stage ?
Or is it best to have the same amount of ripple on the rails for both the voltage and current stage ?
.
Last edited:
That's how it should be.
The only cure (known to me) for that is a use of more linear devices at the output - 3 pairs (per channel) of 2sk2013/2sj313 biased at 400ma per MOSFET transform the F5 into an amp of a higher class. As long as you stick with industrial type of switching MOSFETs the sound artifacts like above mentioned will be present.
When I tested the use of k2013/j313 I had at hand a toroid with slightly higher secondary voltages and I used the capacitance multiplier to steal 4V from PS and it did show some sonical improvement so I kept it...
Juma, would you still maintain that your version of the F5 with K2013/J313 is the best yet in your repetoire?
I haven't read up on Cascode voltage amplifier stages, however,
looking at Turbo version V3, the voltage dividers and the voltage drop across Vbe,
set up voltage regulation for the Q1 and Q2 jFET's.
Unfortunately, I don't think the Cascode set up in Turbo V3,
will provide enough current to effectively charge and discharge Ciss = 12.6nF.
.
BTW: A typo that may be on F5 Turbo V3.
Should R26 read 10K ?
Last edited:
Zen and Juma,
You are right, the Ciss = 12.6nF of the monster MOSFET's needs more drive current.
So I've just tripled up Q1 and Q2 - Huge improvement !!!
With the tripling up of Q1 and Q2, there was a significant reduction in harshness.
Now Mozart Piano Concerto No.20* is starting to sound good.
Rather than a follower stage driving these monster MOSFET's,
I'm wondering if a darlington or complementary pair will work with Q1 and Q2.
Where a jFET, with a lower gm, is used to drive a Bi Polar transistor.
A second option, is there a MOSFET's with a transconductance around 20 mS ?
Does that exist ?
Correction - N.P. does discuss paralleling Q1 and Q2 jFET's.
This can be found on page 14 of the Turbo version.
* Mozart Piano Concerto No. 20
Lief Ove Andsnes
Norwegian Chamber Orchestra
EMI 50999 5 00281 2 2
.
Your right, for a IRFP240 Ciss ~ 1200 puFF
However, Juma said he had great results with 2sk2013's which have Ciss ~ 170 puFF.
No doubt - the sonics of the monster MOSFET's has improved with more drive current.
So for Q1 and Q2 - I'm wondering if a jFET ==> BJT darlington or complementary pair work.
Or is this old fashioned ?
.
that will need some major redo
it was made , off course , but that will be different amp
in case that you have virtues of Jfet input (and believe me , there are some virtues ) , best way is to introduce follower stage ( mosfet or bjt , but mosfet as first choice ) with , say , 10 times current of input stage
easy peasy ...... or little more complicated - whatever you choose
it was made , off course , but that will be different amp
in case that you have virtues of Jfet input (and believe me , there are some virtues ) , best way is to introduce follower stage ( mosfet or bjt , but mosfet as first choice ) with , say , 10 times current of input stage
easy peasy ...... or little more complicated - whatever you choose
Zen,
To point out - in an Aleph style amp, a single j177 or 2N5087 had plenty of juice to drive a IXFN 180n10 monster.
The selection between the jFET and PNP was simply a matter of sonic taste.
Previously, I used a H/K 7600 CD player ==> modded Yaqin CD3 tube buffer for a digital front end to test the F5.
Not good enough - It seems the F5 is a very transparent amp.
In terms of audio systems, a Piano Concerto will quickly separate the sheep from the goats
so I had to improve the digital front end.
Now the source is an H/K 7600 used as a transport ==> Modded Xindak DAC-5 ==> Roksan Caspian preamp.
With this change - more harshness was removed.
This may be a contraversal comment - but I've found most wire wound ceramic power resistors to sound harsh.
There are exceptions like the Mills or Mundorf.
But I've found metal oxide resistors to sound quite good.
So replacing the wire wound 0R51's with metal oxide - will probably reduce harshness some more.
Very Importantly - the F5 is a DC coupled amp.
The first tube buffer I used with it was a Musical Fidelity v3 - which is a hybrid buffer.
The transistors are not biased properly - until the tube warms up (15 seconds).
So Vout hits one of the DC rails until the buffer warms up - yikes !
In other words, using this tube buffer with the F5 is not recommended.
I just tried the MF v3 buffer because its Zout is so low.
.
To point out - in an Aleph style amp, a single j177 or 2N5087 had plenty of juice to drive a IXFN 180n10 monster.
The selection between the jFET and PNP was simply a matter of sonic taste.
Previously, I used a H/K 7600 CD player ==> modded Yaqin CD3 tube buffer for a digital front end to test the F5.
Not good enough - It seems the F5 is a very transparent amp.
In terms of audio systems, a Piano Concerto will quickly separate the sheep from the goats
so I had to improve the digital front end.
Now the source is an H/K 7600 used as a transport ==> Modded Xindak DAC-5 ==> Roksan Caspian preamp.
With this change - more harshness was removed.
This may be a contraversal comment - but I've found most wire wound ceramic power resistors to sound harsh.
There are exceptions like the Mills or Mundorf.
But I've found metal oxide resistors to sound quite good.
So replacing the wire wound 0R51's with metal oxide - will probably reduce harshness some more.
Very Importantly - the F5 is a DC coupled amp.
The first tube buffer I used with it was a Musical Fidelity v3 - which is a hybrid buffer.
The transistors are not biased properly - until the tube warms up (15 seconds).
So Vout hits one of the DC rails until the buffer warms up - yikes !
In other words, using this tube buffer with the F5 is not recommended.
I just tried the MF v3 buffer because its Zout is so low.
.
whatever you choose - poison is yours 
even if I can't agree that one tiny Jfet (or bjt , with same Iq ) is able to properly drive that brick on output
while yuo at it , you can try something in line with this - I didn't calculate source resistors for buffer stage , but I had on mind (as I already wrote ) 10 times greater current than in input stage
even if I can't agree that one tiny Jfet (or bjt , with same Iq ) is able to properly drive that brick on output
while yuo at it , you can try something in line with this - I didn't calculate source resistors for buffer stage , but I had on mind (as I already wrote ) 10 times greater current than in input stage
