diyAudio

diyAudio (http://www.diyaudio.com/forums/)
-   Solid State (http://www.diyaudio.com/forums/solid-state/)
-   -   mosfet amp questions ... (http://www.diyaudio.com/forums/solid-state/58810-mosfet-amp-questions.html)

Ahmad_tbp 9th June 2005 01:48 PM

mosfet amp questions ...
 
hi
here s the mosfet amp i m tryin to build : http://users.swing.be/edwinpaij/ampli_mosfet_simple.htm
i have 3 questions ...
1- i m tryin to use irfp250 instead of irfp240 ... is there any problem ... ?
2- can i use 4 output transistors in array instead of 2 transistors (to gain more power) ?
3- can i build two seprated of this amp and brigde their output together (to double the output power) ???
thnx ...

richie00boy 9th June 2005 02:13 PM

Yes. The amp is not really suited to either. Because there is not much in the way of thermal compensation it may just keep heating up. You may be lucky and get away with it but the design is not inherently safe.

The IRF devices are listed in the parts list alongside a bunch of lateral (audio) devices which are safe. I wonder if the author has really properly tested the design with vertical (switching, e.g. IRF) devices.

The amp will probably only cope with one pair of vertical output devices because vertical devices have much higher input capacitance = harder to drive.

You can bridge any amp as long as you only run double the usual impedance.

BobEllis 9th June 2005 02:40 PM

1 - from the datesheet, the IRFP250 looks like two IRFP240 chips in a single package. Much higher gate-source capacitance. It does not have a compliment. Stick with the 240.

2 - Adding more transistors won't add more power, per se, but it will enable you to get more power reliably into lower impedances. To raise the power output you need to increase the rail voltage, which often drives you to more output devices for reliability

3 - that design is bridgeable in theory. If your output stage and power supply are up to the task, you can get four times the output power of each amp by itself. Bridging presents an apparent load to each amplifier that is half its actual value. So each half of the amp "sees" a 4 ohm load if you use an 8 ohm speaker. This means more current and higher dissipation.

A problem with the design as it stands is it needs a more robust output stage to reliably operate bridged at high power (or even normally) I would use at least 3 pairs of output devices, if not four or more, depending on your heat sink. Smaller heat sinks require more output devices. Duplicate R12-15 as many times as you have output pairs.

Note that the stereo version on that page has inadequate heat sinking. If not for the fan it would die, and that arrangement wastes a lot of the fan's cooling capability. The monoblock is better for heat sink size, but be sure to permit adequate airflow.

Another issue is thermal stability. You'd be better off with a Vgs multiplier (see http://passdiy.com/pdf/A75p1.pdf for an example and discussion) with the transistor mounted on the main heat sink. With just a pot to set bias you leave yourself open for thermal runaway.

You would probably be better off building one of Rod Elliot's designs http://sound.westhost.com/projects-1.htm. or a Leach amp http://users.ece.gatech.edu/~mleach/lowtim/ . Ready made boards make the projects a lot easier and more likely to be successful. I don't have experience with Rod's amps, but there are plenty of folks here who have built them and love them. I am happy with my Leach amps.

Ahmad_tbp 9th June 2005 03:31 PM

hi & thnx
my goddd !!! i searched all the country for irfp240 & i couldnt find it :( ... somebody told me that the irfp250 will work for it ! ,, but it seems the irfp250 & irfp9240 is not compliment as u said , so the amp will not work at all ?????!!!
and i have not any problem for heatsink , i wanna use it w a larg heat sink and fan colling system for high duty use , ...
i m tryin to build the leach super amp for another use , and i want this amp as a bridgable amp to use as a guitar amp ... and also another single one to drivin my sub ... any suggestion ?

richie00boy 9th June 2005 03:40 PM

If you can get the IRFP250 ok then look at both of Quasi's n-channel amps or Lars Clausen's Zeta.

Nelson Pass 9th June 2005 06:35 PM

Quote:

Originally posted by richie00boy
Yes. The amp is not really suited to either. Because there is not much in the way of thermal compensation it may just keep heating up. You may be lucky and get away with it but the design is not inherently safe.

The IRF devices are listed in the parts list alongside a bunch of lateral (audio) devices which are safe. I wonder if the author has really properly tested the design with vertical (switching, e.g. IRF) devices.

The amp will probably only cope with one pair of vertical output devices because vertical devices have much higher input capacitance = harder to drive.

While I would prefer to see a Vgs multiplier for a bias voltage
source, it is not a given that the bias will be unstable, as long
as adequate sinking is provided, the power supply is fairly
constant, and the bias is adjusted after warm-up. Larger
values of Source resistance will improve the bias stability.

As to the relative capacitance of Lateral vs Vertical, this has
been hashed out in these threads at length - the specified
capacitance of the two types often shows the Lateral as
actually having greater capacitances. It is the linearity of
the capacitance which becomes the issue in any comparison.

IRF250's will probably work as well or better, and I observe that
for every design there is an optimum number of paralleled
devices, and often the number is greater than 1.

amplifierguru 9th June 2005 11:26 PM

Given that this amplifier is intended to be nice but simple - why throw out the baby with the bathwater in applying IR mosfets to a non-tempco compensated resistive bias network. A simple and cheap thermistor across it and positioned carefully could do the trick.

Greg

tlf9999 10th June 2005 12:56 AM

Quote:

Originally posted by Nelson Pass
It is the linearity of
the capacitance which becomes the issue in any comparison.


Nelson, is it the linearity (straightness of the capacitance curve) or general slope of such a curve that matters?

For example, of two devies, one with a constant capacitance (perfectly linear), and another with a perfectly sloped capacitance (perfectly linear two), which one should we prefer, assuming everything else equal?

Nelson Pass 10th June 2005 01:20 AM

Any character of the device which is not constant over the
operating area gives rise to distortion. The variation in
capacitance with voltage creates distortion at high frequencies
not seen at low frequencies. Of course we would prefer a
constant value, but if we have to accept variation, it's a lot
easier to put up with a gentle variation, which is one reason
we say that Mosfets like voltage: the capacitance and
capacitance variation is less when there is greater voltage
across the device.

lumanauw 10th June 2005 02:44 AM

Quote:

which is one reason we say that Mosfets like voltage: the capacitance and capacitance variation is less when there is greater voltage across the device.
Mosfets prefers higher working voltage (Vgs I assume)
Do bipolars works better in higher voltages too? Or in the contrary bipolars prefers low operating voltage (the cascoding in VAS for example)?


All times are GMT. The time now is 07:44 PM.


vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2014 DragonByte Technologies Ltd.
Copyright 1999-2014 diyAudio


Content Relevant URLs by vBSEO 3.3.2