> Question: is it important to keep the connection of the X short ?
If you want to use floating X, yes.
> I would like to put the two half of the balanced circuit on opposite sided of a monoblock chassis to maximize the use of the heatsinks.
Our PCB (to be available via the forum) will fit a mini-tower monoblock 85mm wide.
This will still allow optimum PCB layout while using two heatsinks for the 4 MOSFETs.
> I am speaking about 20 cm wires.
Then you should considered using the grounded X, which is essnetially 2 separate single-ended F5's in bridge mode, one amplifying the positive signal, the other the negative.
There is little to choose between grounded and floating, so I would not worry too much about it.
Patrick
If you want to use floating X, yes.
> I would like to put the two half of the balanced circuit on opposite sided of a monoblock chassis to maximize the use of the heatsinks.
Our PCB (to be available via the forum) will fit a mini-tower monoblock 85mm wide.
This will still allow optimum PCB layout while using two heatsinks for the 4 MOSFETs.
> I am speaking about 20 cm wires.
Then you should considered using the grounded X, which is essnetially 2 separate single-ended F5's in bridge mode, one amplifying the positive signal, the other the negative.
There is little to choose between grounded and floating, so I would not worry too much about it.
Patrick
That's cute, but where do you stick the transformer ? With all that circuits will look like the monoblock of 2001 Space Odyssey
I'd like to choose a way to go, in order not to waste money. I have to drive my Sonus Faber Grand Piano (4 ohm pain). The source will be a buffalo DAC, so I cannot decide if it's better to go for balanced or for parallel mosfet. Both would probably be the best, but it becomes a monster that I am not sure I am able to deal with alone.
Thanks,
D.
The F5X with 4 MOSFETs and 4A bias will deliver 32Wrms Class A into 4ohm (current limited).
F5X with 8 MOSFETs is not something I have time to look into.
So you are on your own, I am afraid.
Of course you can open a new thread and see who else can give you more advise, including Nelson of course.
Patrick
F5X with 8 MOSFETs is not something I have time to look into.
So you are on your own, I am afraid.
Of course you can open a new thread and see who else can give you more advise, including Nelson of course.
Patrick
I have already a Nelson Pass F5, which I would like to connect to the rear speakers. I want to connect this balanced version to the front speakers, so I would like it to have the same gain as an standard F5.
What is the best way to do that? I imagina than a series resistor plus a resistor to ground, but what could be the best placement and specific values?
Thanks in advance,
Regards
What is the best way to do that? I imagina than a series resistor plus a resistor to ground, but what could be the best placement and specific values?
Thanks in advance,
Regards
The balanced version has a closed loop gain of 6, just as the standard F5.
You get 6dB more output at the speakers not because of the power amp gain, but because of the doubled amplitude of the balanced signal.
Again, this amplifier required a balanced input signal.
The easiest way to adjust the balance between front and rear is via your HT processor / preamp.
Patrick
You get 6dB more output at the speakers not because of the power amp gain, but because of the doubled amplitude of the balanced signal.
Again, this amplifier required a balanced input signal.
The easiest way to adjust the balance between front and rear is via your HT processor / preamp.
Patrick
Hi Patrick, how closely matched must the quad mosfets be? I have enough to bulid 2 stock F5s, will they be good enough for the balance F5, or do I need even tighter matchings between the two channels?
Vds=+/-24v, Ids=+/-1.3A
(vgs=2.000+V)
2SK1530
Set A pairs (used for one channel bal. F5)
0.337/0.341
Set B pairs (used for one channel bal. F5)
0.326/0.326
------------------------------------------------
(vgs=2.000-V)
2SJ201
Set A pairs (used for one channel bal. F5)
0.340/0.341
Set B pairs (used for one channel bal. F5)
0.324/0.326
cheers,
K.
Vds=+/-24v, Ids=+/-1.3A
(vgs=2.000+V)
2SK1530
Set A pairs (used for one channel bal. F5)
0.337/0.341
Set B pairs (used for one channel bal. F5)
0.326/0.326
------------------------------------------------
(vgs=2.000-V)
2SJ201
Set A pairs (used for one channel bal. F5)
0.340/0.341
Set B pairs (used for one channel bal. F5)
0.324/0.326
cheers,
K.
> (vgs=2.000-V)
Meaning that the Vgs at 1.3A is below 2V ??
If you meant :
2SK1530 2.337/2.341 2.326/2.326
2SJ201 2.340/2.341 2.324/2.326
Then the match is good enough, but only at 1.3A.
Ideally you should measure under working conditions, which means Vds 16V, Id 2A.
But I guess someone else did the matching, so it is probably as good as you'll get.
Patrick
Meaning that the Vgs at 1.3A is below 2V ??
If you meant :
2SK1530 2.337/2.341 2.326/2.326
2SJ201 2.340/2.341 2.324/2.326
Then the match is good enough, but only at 1.3A.
Ideally you should measure under working conditions, which means Vds 16V, Id 2A.
But I guess someone else did the matching, so it is probably as good as you'll get.
Patrick
There are no parallel pairs in the balanced version of the F5.
Do we need to match any devices?
The +IN half has a P & an N, these don't need to be N-P matched.
The -IN half also does not need to be P-N matched.
The two halves are each feedback amplifiers and do not need to match N-N nor P-P. It will still work.
I will accept that matching over a range of operating currents will help attenuate distortions, but that requires transconductance matching as well as Vgs matching.
Testing, measuring, matching Id at one current achieves almost nothing as far as I can see. One current means no output. What does distortion minimisation do for zero output?
Do we need to match any devices?
The +IN half has a P & an N, these don't need to be N-P matched.
The -IN half also does not need to be P-N matched.
The two halves are each feedback amplifiers and do not need to match N-N nor P-P. It will still work.
I will accept that matching over a range of operating currents will help attenuate distortions, but that requires transconductance matching as well as Vgs matching.
Testing, measuring, matching Id at one current achieves almost nothing as far as I can see. One current means no output. What does distortion minimisation do for zero output?
Thanks Patrick, but I already knew the reason why gain is doubled (double the output voltage), because of being balanced
. I was wondering about bi-amplifying too, so HT volume control is not an option. The same solution would be to fit a pot, but why placing a pot if I can fit a fixed series and a resistor to ground?
The thing is what values to use: 20k and 20k, 50k and 50k, 100k and 100k?
Now I am realizing, by looking at your schematic labeled "Pass F5 v.1 TOS 17-June-2008" (which I suppose is the latest revision) that you didn't fit input series resistors nor input resistors to ground. What is the explanation for that?
Thanks for all your efforts in this project Patrick, they are greatly appreciated
Best regards,
Regi
> There are no parallel pairs in the balanced version of the F5.
> Do we need to match any devices?
If you want a working amplifier, then there is no need to match. The circuit will work even if all the devices are unmatched.
You do want to match if you want to do open loop gain balancing at all 4 quadrants of the F5X so that you get near-perfect even harmonic cancellation top to bottom, and left to right in open loop, before closing the NFB.
You can, if you know what you are doing, also create an imbalance deliberately to minimise 3rd harmonics between top and bottom, and then kill off the thus increased 2nd harmonics by left to right cancellation. To do the latter, you need a good spectrum analyser.
A bit difficult to explain without writing equations. But I think you get the idea.
Patrick
PS You are correct in saying that you ideally want curve tracer matching and not single point matching.
.
> Do we need to match any devices?
If you want a working amplifier, then there is no need to match. The circuit will work even if all the devices are unmatched.
You do want to match if you want to do open loop gain balancing at all 4 quadrants of the F5X so that you get near-perfect even harmonic cancellation top to bottom, and left to right in open loop, before closing the NFB.
You can, if you know what you are doing, also create an imbalance deliberately to minimise 3rd harmonics between top and bottom, and then kill off the thus increased 2nd harmonics by left to right cancellation. To do the latter, you need a good spectrum analyser.
A bit difficult to explain without writing equations. But I think you get the idea.
Patrick
PS You are correct in saying that you ideally want curve tracer matching and not single point matching.
.
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.