So this is what I've been playing around with for the last months, I think it might be ready for the vultures to tear it apart now ;-D hehe. I've been listening to it for some time now and I'm pleased with it.. sure it could be better with ccs and stuff but as usual, I'm using vicor modules with 4wire connection so, add it if you like, If I would, they would be something simple like a led and a resistor + bd139. And also I really want a low component count. Idle current is adjusted to 300mA or higher, hmm I've used somewhat lower too and It seems to work pretty ok at that too..
ps. add another couple of output transistors if you feel like it, but at 28v I'm not sure what it would be good for..
ps. add another couple of output transistors if you feel like it, but at 28v I'm not sure what it would be good for..
Attachments
I know spice is not to be trusted very much but as comparison or just for fun, here is the fourier analysis in multisim10.. My old hp thd+n analyzer stops at 0.01% or something like that so thats not of much use here. so there is probably quite much space to add emitter resistor s here and there if you'd like.. hmm also a tip I got was the transistors zvn3310, zvp3310 and 06 transistors, but I hav'nt got around trying them yet.
Attachments
I guess it could be connected to the output ,tried that too and did'nt like it as much, then the 1k resistor was put there instead, at first it was like 33ohm and then 100 and 150 and so on , ended up with 1k . maybe 470ohm would be nice to. also, I guess I wanted a short loop there, not letting the output transistors disturb how the vas is working , letting that happen in the input stage alone. hmm maybe that's wrong I'm not a 100% sure.
Hmm the reason why I put that connection there in the first place was some vague idea that the vas transistors should see a similar voltage, hmm but that was at 03: o clock something at night so I'm not sure exacly what went on In my mind then..
Hmm the reason why I put that connection there in the first place was some vague idea that the vas transistors should see a similar voltage, hmm but that was at 03: o clock something at night so I'm not sure exacly what went on In my mind then..
Hi nikwal,
There are a few genius designers who believe that if an amp uses MOSFET outputs, then using only JFET and MOSFETS throughout the design can reduce the odd harmonic content, which they see as important for sweet, no fatigue sound. There is data to support this claim, as well as data showing that although bipolar designs may have higher odd harmonic distortion, the total distortion can be lower using higher feedback. JFETS and MOSFETs have lower gain than bipolars, and the all FET amps take advantage of lower odd harmonic distortion to use less global feedback while achieving adequate gain from a few gain stages. The Ayre V3 is one example.
I favor using JFET inputs and then all bipolar circuits and outputs, using the higher BJT gain to support local in-stage feedback. Similar the Ayre and the Nelson Pass Super Symm amps, I find using both positive and negative outputs to drive the speaker seems to remove ground noise - including AC power line crud. You've probably studied the V3, but it might be of interest to other readers.
There are a few genius designers who believe that if an amp uses MOSFET outputs, then using only JFET and MOSFETS throughout the design can reduce the odd harmonic content, which they see as important for sweet, no fatigue sound. There is data to support this claim, as well as data showing that although bipolar designs may have higher odd harmonic distortion, the total distortion can be lower using higher feedback. JFETS and MOSFETs have lower gain than bipolars, and the all FET amps take advantage of lower odd harmonic distortion to use less global feedback while achieving adequate gain from a few gain stages. The Ayre V3 is one example.
I favor using JFET inputs and then all bipolar circuits and outputs, using the higher BJT gain to support local in-stage feedback. Similar the Ayre and the Nelson Pass Super Symm amps, I find using both positive and negative outputs to drive the speaker seems to remove ground noise - including AC power line crud. You've probably studied the V3, but it might be of interest to other readers.
Attachments
Zetex Zvn3306 Zvp3306 Zvn3310 Zvp3310
Should we use MOSFET for audio amplifier inputs?
I say YES!
They can work great both in VAS and Differential Input. Even as Complementary Diff Input Pairs.
And I have created several circuits that spice just as well as BJT.
LineSource mentions, the harmonics content can be different. True!
This makes it very interesting to experiment with MOSFET inputs/VAS.
But I find no reason to use IRF TO-220 HEXFET for input stage.
If design one bipolar input power amplifier, I would, for example,
not use MJE15030, MJE15031 for line livel input if I can use 2SC2240, 2SA970 etc.
If I did, I would not feel very good about it.
There are good TO-92 MOSFET from ZETEX. Among other manufacturers.
Better suited for small signals.
I imagine these can be a bit more 'complementary', too.
ZVN3306, ZVP3306 and ZVN3310, ZVP3310 for example.
And some other similar small signal MOSFET.
They are not difficult to find. These are worldwide distributors of ZETEX transistors [ http://www.zetex.com/ ]
http://www.digikey.com/
http://www.farnell.com/
I know Nelson Pass has used ZVN3306, ZVP3306
Lineup Selection Guide of audio transistors
LineSource said:
Should we use MOSFET for audio amplifier inputs?
I say YES!
They can work great both in VAS and Differential Input. Even as Complementary Diff Input Pairs.
And I have created several circuits that spice just as well as BJT.
LineSource mentions, the harmonics content can be different. True!
This makes it very interesting to experiment with MOSFET inputs/VAS.But I find no reason to use IRF TO-220 HEXFET for input stage.
If design one bipolar input power amplifier, I would, for example,
not use MJE15030, MJE15031 for line livel input if I can use 2SC2240, 2SA970 etc.
If I did, I would not feel very good about it.
There are good TO-92 MOSFET from ZETEX. Among other manufacturers.
Better suited for small signals.
I imagine these can be a bit more 'complementary', too.
ZVN3306, ZVP3306 and ZVN3310, ZVP3310 for example.
And some other similar small signal MOSFET.
They are not difficult to find. These are worldwide distributors of ZETEX transistors [ http://www.zetex.com/ ]
http://www.digikey.com/
http://www.farnell.com/
I know Nelson Pass
has used ZVN3306, ZVP3306Lineup Selection Guide of audio transistors
lumanauw: not sure, maybe both at some time ;-), I vote on back most of the time.
LineSource: Hmm good point, I'll have to study this closer. No did'nt see an ayre schematic before as I can remember hmm ,so no i've got something to look at in the bath today.
Got A LOT of irf's here now so.. will try them, at a later time hehe.. also to220 dissipate quite a lot of heat, i wonder how they will perform If put on a cooler and run a bit harder, I guess transconductance will be higher (good or bad?)..
maybe if using zvp/n it could be a good idea to clamp them together with some heat conductor between?..
LineSource: Hmm good point, I'll have to study this closer. No did'nt see an ayre schematic before as I can remember hmm ,so no i've got something to look at in the bath today
.Got A LOT of irf's here now so.. will try them, at a later time hehe.. also to220 dissipate quite a lot of heat, i wonder how they will perform If put on a cooler and run a bit harder, I guess transconductance will be higher (good or bad?)..
maybe if using zvp/n it could be a good idea to clamp them together with some heat conductor between?..
Are we on Laterals now Wondered why you wanted to get rid of the Vbe multiplier. The IRF's do however need thermal compensation.
The laterals are superb for audio - no doubt about that - but I would want to bias them a bit more accurately than using a two diode volt drop.
I dont think the unequal resistances seen by the input stage is a problem. The FET's draw no gate current.
Regards Karl
Wondered why you wanted to get rid of the Vbe multiplier. The IRF's do however need thermal compensation. The laterals are superb for audio - no doubt about that - but I would want to bias them a bit more accurately than using a two diode volt drop.
I dont think the unequal resistances seen by the input stage is a problem. The FET's draw no gate current.
Regards Karl
Mooly said:Are we on Laterals now
The laterals are superb for audio - no doubt about that - but I would want to bias them a bit more accurately than using a two diode volt drop.
I dont think the unequal resistances seen by the input stage is a problem. The FET's draw no gate current.
Regards Karl
Like transistors diodes have positive temperature coefficients.
Or just the drivers.. theyre really there just to Try to bump slewrate from high to stupidly high. I'm thinking of adding another transistor(of the other type) to the vbe multiplier and a resistor, it looks a bit lonely there. also then comes a nice source junction to feed somewhere or be fed ;-)..
even if the temperature works I'm sure one diode(per transistor) wont be enough for this gate voltage. also adjustment of quicent current will become somewhat strange.
Simpler still. The accumulative harmonic content is less than .00045% @ 50 watt, phase shift at 20 kHz only 1 degree. Bias is stable at about 250 mA at 27 deg centigrade and 228 mA at 50 deg centigrade.
Gain is 29 dB and perfectly flat to 800kHz, slew must be okay
I will knock it together and see what it sounds like, probably not too shabby
Gain is 29 dB and perfectly flat to 800kHz, slew must be okay
I will knock it together and see what it sounds like, probably not too shabby
Attachments
Hello
I think you should think about something better to set the bias of your output mos... The Vgs on of these devices is not really reliable. Other than that it's just the super classic amp topology... Look at this one from France, very good sound indeed, simple schematic but the cascode on the VAS is a nice addition
http://pagesperso-orange.fr/jm.plantefeve/nrds2.jpg
I think you should think about something better to set the bias of your output mos... The Vgs on of these devices is not really reliable. Other than that it's just the super classic amp topology... Look at this one from France, very good sound indeed, simple schematic but the cascode on the VAS is a nice addition
http://pagesperso-orange.fr/jm.plantefeve/nrds2.jpg
nikwal said:
Hi,
A brief tutorial on how to do this in MS-10? For those (me!) too lazy to read the manual, please.
I trust the THD figures that MS gives, as long as the models you use are good ones. I have verified the accuracy on one amp - if anything, the THD given by MS was higher than the actual amount.
Nico Ras said:Simpler still. The accumulative harmonic content is less than .00045% @ 50 watt, phase shift at 20 kHz only 1 degree. Bias is stable at about 250 mA at 27 deg centigrade and 228 mA at 50 deg centigrade.
Gain is 29 dB and perfectly flat to 800kHz, slew must be okay
I will knock it together and see what it sounds like, probably not too shabby
Hi Nico,
Brilliant, as usual.
Been happy with your headphone amp? I haven't gotten around to building it yet, but soon...
MultiSim. How I make my FFT.
You setup your circuit.
With 1 kHz AC Voltage source input. With some level.
Then you right click on the net point where you want to do FFT.
Properties. Usual at the output.
Note down the Net Number of this point.
Analysis -> Fourier Analysis
You need to configure parameters.
Like freq (1kHz default). How many harmonics(default 9).
You need to click one tab, to specify output Net Number.
There is a list of all net nodes.
Now, you need to spicy output format.
I use Graph + Decibel scale.
After this you click down at the left: Run analysis
When you have big capacitors in your circuit,
you can sometimes need to delay the start of Analysis.
Until your capacitors are charged.
I usually run my FFT between Time: 0.8 - 1.0 seconds. (length = 0.2 sec with 5000 iterations)
This means circuit has got 0.8 sec initially to charge any electrolytic caps
and settle the voltages at different part of my circuit.
Good luck!
You setup your circuit.
With 1 kHz AC Voltage source input. With some level.
Then you right click on the net point where you want to do FFT.
Properties. Usual at the output.
Note down the Net Number of this point.
Analysis -> Fourier Analysis
You need to configure parameters.
Like freq (1kHz default). How many harmonics(default 9).
You need to click one tab, to specify output Net Number.
There is a list of all net nodes.
Now, you need to spicy output format.
I use Graph + Decibel scale.
After this you click down at the left: Run analysis
When you have big capacitors in your circuit,
you can sometimes need to delay the start of Analysis.
Until your capacitors are charged.
I usually run my FFT between Time: 0.8 - 1.0 seconds. (length = 0.2 sec with 5000 iterations)
This means circuit has got 0.8 sec initially to charge any electrolytic caps
and settle the voltages at different part of my circuit.
Good luck!
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.