RAS-300 Mosfet Amp.

[gnalbant]

Originally posted by PigletsDad
I agree with Andy L Francis and peranders, this is not a good design. In particular,
the BD139 and BD140 are 80V transistors. Running the circuits on +/-55V, the transistors will see much more than this for signals close to full power, and will fail prematurely.

Fixed biasing with diodes is a lottery and can give under or over bias, depending on the particular batch of output devices. I expect the designer made one prototype, and that was OK, so he thought no more about it. Adjustable bias with a pot is a must for these output transistors.

To get anything like 300W you need more output device pairs. He claims 380W rms into 4 Ohms. Now a quick calculation shows that that means delivering peak voltages of +/- 55V, equal to his rail voltage without any drop across the transistors! These voltages require nearly 14A of current, but each transistor is only good for 8A maximum, taking them very close to saturation, and giving a vast voltage drop.

The designer may be a nice guy, but his circuit is simply incapable of delivering anything like the performance he claims, and will probably just blow the BD139/B140 up the first time a big signal arrives.

BD139/BD140 pair is not recommended, MJE340-MJE350 should be used for high voltage option, as Nico stated in his website.

He has also a version with a adjustable bias option. The diodes can be replaced by resistors also as he stated.

With full power the amp gives 180W rms into 8 ohms and 250 rms into 4 ohms. The 380W is music power and it is strictly determined by the capability of your power supply. You can find this in his website, and usually 100W rms is quite enough even for large rooms, if you have appropriate speakers.

I have a friend who is quite experienced in high-end systems. He is replacing his Vincent high-end amp with RAS250 using in total 100,000uF capacitors, 35 amp rectifiers and 1 KVA transformer. Gods must be crazy, so he is

I am not a defender or the lawyer of Nico Ras, but I suggest his amps to the musicians especially, not to the technical design gurus. Mozart's music may not be technically very good, but when it is played it is the sound of God, as his opponent Salieri says

Anyway, any technical critics is wellcome, however, please keep in mind that there some people very happy about his designs, there are many amps built around world. If you don't like him or his designs, just leave it.

[Nico Ras]

Originally posted by Andy L. Francis
That design is NOT so elegant.
First, it's not Nico Ras's design as he claims, it's a Hitachi AppNote amplifier - badly modified.
Then it won't give 240Wrms/8Ohm - 380Wrms/4Ohm. Nico, what about Ohm's law?!
Then the VAS current is 250uA! I't really not enough!
Then R3 causes decreasing of PSRR. Use CCS here, or a rail RC filtering in the + rail is needed.
A cap may needed in parlel with R8. How does the square-wave response looks now?
BD139/140 has low Vce0. Sorry, this is a foolish error on the circuit again.
D1-D2 makes me crying too. A trimmpot is needed here for bias adjustment. Now your amplifier is a true B, or even C class amplifier. That's crazy.
Anyway an input capacitor is always needed in a DC-coupled amp, and I'd add some protection in the OP stage (zeners on gates).

First and last I recommend NOT to build this amplifier.
And I'd ask the designer to redesign it. Now it's a bad joke, not a proper amp design.

I hope I wasn't offending. Sorry then.
But hey I hate bad amplifier designs!

Hi Andy,

You make some interesting comments. The things you say I do not claim.

The introduction to this particular amp on my web is:
"The circuit diagram shows a single channel of my simple MOSFET power amplifier. For stereo applications two identical units would be required." In other words this simple joke, should be easy for you to follow.

Yes the design is based on data provided by Hitachi from around the nineties if I remember correctly.

There are few new unique topologies in amplifier design, maybe you could share a few of yours.

I do have an input capacitor, it is called C1 it is typically 10 uF.

If you you measure the quescient current through the output devices it is around 200 mA, maybe you should verify this before speculating. Class B is zero bias, not AB. The diodes used are fine here and provides stable bias.

If you look at the 2SK1058 and 2SJ162 you will notice that these devices have gate protection zeners included in the die, you actually don't need to add it.

BD139/140 has a particularly nice sound to me, the Vceo is adequite for the 100 watt amplifier as shown on my web-page (see the safe operating area of this particular transistor (dc conditions) then you will realize that they are up to the job.

It is not necessary to run high current through the VAS since you need to overcome gate capacitance of the MOSFETS. You may also notice that at low current these transistors are particularly linear.

In the higher power version I use of two pairs of lateral mosfets and also use MJE340/350 and in place of MPSA56 I recommend using MPSA92 or 2SA970.

My four amps are permanently powered and they have been operating for the past 17 years non stop without failing once, this is not a too shabby MTBF. I am aware of literally hundred that has been built over the past 15 or so years.

Even for such a bad joke it sounds pretty nice to most people who have heard it and many who built it and thanked me for sharing it.

Furthermore nobody is forced in any way to build this amp, it is freely available to anyone, I will assist anyone who wants to construct it, I ask nothing in return.


Kind regards to you all

Nico

[sakis]

you here mr ras
best regards from athens greece

[Nico Ras]

Originally posted by sakis
you here mr ras
best regards from athens greece

Hello Sakis nice to hear from you, Greeting from Durban. Have you completed your special project - how did it go?

[Nico Ras]

Originally posted by peranders
I wouldn't call it "badly designed". I wonder though why Nico has so little current in the VAS stage? The drive capability for the mosfets will be very low meaning a slow amp.

The amp will also run in class B (possibly) since the D1 and D2 is not enough and certainly not trimmable. The Gate threshold voltage is normally aroung 2 volts. This means that the two diodes must produce 4-5 volts at least compared to 1,4 volts right now. It's worth mentioning that the gate voltage can be down to zero volts so in this case you'll have trouble with too much idle current.

See the original in the attached file.

Hi Peranders,

I cannot I agree with you. Even looking at the original Hitachi circuit you attached. The VAS is running at about 8mA in this circuit. If you leave the 1K pot as 1K then the quescient current of the mosfets are about 3 Amps and you are running them in class A because the gates are sitting at about 4.6V and -3.1V respectively.

Adjusting the bias for about 250 mA means this resistor must be trimmed to about 70 ohms. So using the two diodes will provide a much more stable bias at varying supply voltage than the resistor would.

Since the gate capacitance is only 500 pF, I see no reason running the VAS at 8 mA what is the purpose. If I use the diodes as bias I can run the VAS at a less than 500 uA and the mosfets will remain biased properly at about 250 mA quescient current.

Running the VAS at 8 mA decreases linearity and increases harmonic content by an order of magnitude over running at 300 uA also the linearity improves.

My circuit basically replaces the current source with a current mirror, sets the VAS current to 350uA reduces the harmonic content by a further 6dB while retaining a quescient current of 250 mA, I cannot see how you could miss this. This amplifier is definitely biased class AB and with the values I show measures the lowest harmonic content.

The response is ruler flat beyond 100kHz Open loop gain is 100 dB. What is bad about this design. Why would this amp be slow, it is not. Full power at 100 kHz, calculate the slew, it is quite adequate.

[Nico Ras]

Originally posted by PigletsDad
I agree with Andy L Francis and peranders, this is not a good design. In particular,
the BD139 and BD140 are 80V transistors. Running the circuits on +/-55V, the transistors will see much more than this for signals close to full power, and will fail prematurely.

Fixed biasing with diodes is a lottery and can give under or over bias, depending on the particular batch of output devices. I expect the designer made one prototype, and that was OK, so he thought no more about it. Adjustable bias with a pot is a must for these output transistors.

To get anything like 300W you need more output device pairs. He claims 380W rms into 4 Ohms. Now a quick calculation shows that that means delivering peak voltages of +/- 55V, equal to his rail voltage without any drop across the transistors! These voltages require nearly 14A of current, but each transistor is only good for 8A maximum, taking them very close to saturation, and giving a vast voltage drop.

The designer may be a nice guy, but his circuit is simply incapable of delivering anything like the performance he claims, and will probably just blow the BD139/B140 up the first time a big signal arrives.

Hi Pigletsdad,

you are wrong in the statement that using diodes is a lottery. Could you consider that the voltage developed across a forward biased diode will be pretty stable?

The designer made more than one prototype, in fact there has been several built that I know of and I know of four that has been running flawlessly for about +10 years continually. I do not believe in a power switch.

Kindly explain what in your opinion would would cause >80V from developing across the BDs.

Also explain why resistor bias is a must for these devices, where do you get your information from kindly explain these facts.

Furthermore, besides yourself who claims 380 W rms, do you know what rms means. I did not make any such claim, see my web site, I am talking about 240 w into 4 ohms.

Also 8A at 55V will destroy the mosfet, it is rated at 125 watt, why would you want to do this.

The statement of a vast voltage drop across the output devices with high current, I do not understand could you explain this. Mosfets are like variable resistances when they turn on their resistance decreases to milli ohms, at 8 or whatever amps you refer time milli ohms does not constitute a vast volt drop the voltage across the source resistance would be higher than across the mosfet.

Firstly I don't think you have any idea what I claimed, secondly you have never built this amplifier so you have absolutely zero experience or any fact to base you statements on, it is purely speculative.

I think this forum is great for opinions but when you make statements then also make sure that you can back your claims. Unfortunately you cannot.

Someone asked me if he can build this amp with an output of 450 watts into 4 ohms, I told him that I have never done this but he could try by using stable rails of around +-60V and at least 8 pairs of output devices.

Kind regards

Nico

[Nico Ras]

Hi to all,

this was an interesting exercise, unfortunately I only got time to respond very late. Thanks to you guys who contributed and those who did not, no love lost.

You did make me revisit a very old design and although there are room for improvement and make it much more complex, I am confident that it should remain unchanged as a simple to build unit.


The improvements in my later upgraded designs which has not been published does include current mirrors and CCS in first stage LTP and changed the input pair to MAT02 resulting in a notable improvement. I can recommend this to the old users of this amp and if you would like PCBs done this as usual would only be a pleasure to my friends.

Gohkan you are at the beginning of this great journey into the unknown, I would be pleased to send you the diagrams and PCBs for the new improved RAS100II we Christian it the Carma


Kindest regards


Nico

[gnalbant]

Thank you Nico

It is a great pleasure to work with you.

Any design, any idea, anything comes from you is wellcome..

Best regards.

[peranders]

Originally posted by Nico Ras


Hi Pigletsdad,

you are wrong in the statement that using diodes is a lottery. Could you consider that the voltage developed across a forward biased diode will be pretty stable?

The designer made more than one prototype, in fact there has been several built that I know of and I know of four that has been running flawlessly for about +10 years continually. I do not believe in a power switch.

I have had 100 pcs 2SJ50 and equally many 2SK135. The Vgsth goes from almost zero up to 2.2 volts so using a fixed bias value is not a good idea in the long run. In my case you were fource to have 0-6 Volts, adjustable. and also temperature stabilised. My first mosfet amp had to stable temperatures, one luke warm and one hot and the bias element was only a pot.

An another comment: If you have a large voltage swing on your BD139/140 (or any other model for that matter) you will get heavily increased distortion with the voltage swing is near Vce max. Max 25-30 V peak would be advisable in your case.

[Nico Ras]

Originally posted by peranders
I have had 100 pcs 2SJ50 and equally many 2SK135. The Vgsth goes from almost zero up to 2.2 volts so using a fixed bias value is not a good idea in the long run. In my case you were fource to have 0-6 Volts, adjustable. and also temperature stabilised. My first mosfet amp had to stable temperatures, one luke warm and one hot and the bias element was only a pot.

An another comment: If you have a large voltage swing on your BD139/140 (or any other model for that matter) you will get heavily increased distortion with the voltage swing is near Vce max. Max 25-30 V peak would be advisable in your case.

Hi Peranders,

Thank for your valued comment. What you say here is true and for this amplifier it is valid also as soon as the you reach the point where the MOSFETs start gain compressing, there didtortion suddenly rises exponentially.

I have never used these MOSFETs and will write your practical research in my design book for future reference.

It may be silly, but I come from an old engineering background which is 1976 B/C. (before computers) and then we used to keep design books, a black folio sized book for design notes in.

Everything technical I read or heard or done I write in this book. I have black books from 31 years ago 273 in all, and useful comments like you made.

Kind regards

Nico