IRFP240 IRFP9240 design

[Steve Conner]

JM, nice find! Those 5 pin dual devices are cool.

As far as I know, all the lateral FET dice for Magnatec, Semelab, all brands except Hitachi are proudly made right here in Scotland. They'll put any brand you want on the package if you buy enough.

[J M Fahey]

*Now* I understand why the recommended cooling fluid by the semiconductor ovens (for the operators, not the chips) is good old "uisquebaugh"
I learnt about them in Interesting device for an output stage.

[Alex R]

I have a few of the Class D labelled items hanging around. Yep, gate and drain reversed compared to the switching mosfets but that's easy to do on this board. I will report back...

[KevinOConnor]

Hi Guys

The Toshiba TO-3P equivalents of the old Hitachi mosfets are sold by Digikey: 2SJ201YF and 2SK1530YF. Others like 2SJ162 and 2SK1058 should also be available.

One of the attributes of these older mosfets is their very low transconductance. This contributed to their linearity and results in a low damping factor. The latter makes them the "tube-like" choice in guitar amps built by Tubeworks et al in the 1980s.

The N/P packages are generally optimized for switching apps, like a class-D output stage. You could use them linearly but control over thermal runaway is needed.

Some of the negative comments about the original circuit - with or without original mosfets - are a bit on the doom-and-gloom side. The circuit is definitely the bare minimum of what will work - unless you consider the Lin circuit to be even more utilitarian. There are some simple improvements that can be made that cost nearly nothing given modern semi prices, while increasing performance and removing much of the unit-to-unit inconsistencies. This does not mean that such changes are absolutely required. You may like the sound of the amp and that is what truly matters.

I replaced blown H&H power stages with both mosfet and bipolar circuits and the result was always that the owner thought the amp sounded better than it ever had. One sound company owner even brought his working H&Hs over for conversion after the blown unit had been modified.

It seems the more primitive a circuit is, the more people will charge for it and others will pay for it. There are still a lot of quasi-comp amps sold for big bucks simply because they are shrouded in black aluminum and voodoo.

Have fun
Kevin O'Connor

[Giaime]

So this works by continuously sensing current passed by output devices and supplying correct bias at continuously varying levels? At every point on the signal wave?

Yes. It does not work with traditional thermal compensation, it just samples the actual instantaneous power device current and with analog algebra (it has a multiplier in it) derives the correct bias point. No risk of thermal runaway (because bias current is strictly fixed), no need for trimmers, it even includes overcurrent protection.

The only drawbacks I see:
- stability has to be checked, after all it is another amplifier with a negative feedback INSIDE the main amplifier negative feedback. But this is common practice in analog engineering, so if you know how to design an amp you can surely get the LT1166 as stable as you will.
- doesn't work well with bipolar transistors due to much lower turn-on voltage needed (thus limited power supply voltage to the chip). There are solutions to this, explained in the datasheet and appnotes.

[J M Fahey]

This info should help Alexr turn his sow's ear into silk.
Should check pinout though.
Lin you say?
I *do*use Lin type amps today, with only N switching type devices to boot.
I *should* be charging big bucks then !!!
May God (or at least some of my customers) hear you !!!
I even house them in black (painted) aluminum
Only problem is, I had no budget left for Voodoo and Mojo
Oh well.

[lowell]

That schematic reminds me of a Trace Elliot TA100 amp that's been puzzling me for a while. Gonna get back to it here soon. I cannot seem to get the bias voltage to cooperate. The p-channel MOSFET gate doesn't have a negative bias voltage. I know this probably doesn't serve to help this thread, just chiming in cause it's quite similar.

[J M Fahey]

What does

The p-channel MOSFET gate doesn't have a negative bias voltage.

mean?
You must measure straight gate to source; if you measure relative to ground you won't get an accurate result.

[lowell]

Oh really? What if I'm measuring before the gate resistor? So the gate resistor is between my meter and the MOSFETS gate.

[Alan0354]

First time I look at the schematic, have not read through the whole thread. One problem is the two MOSFETs are bias by the voltage across R10 which is controlled by current through the second differential stage. That is a very bad way to set up the idle current through the MOSFET. You don't know the turn on voltage, there is no control on how much idle current and you can burn the FETs out if one has turn on voltage particularly low. The Vgs drift quite a bit with MOSFET, even you adjust and set the idle current, it change with temperature. There is no current limiting resistors between the two source and you can pop the FETs easily if shorted. There is no short circuit protection.

Regarding to switching FETs, there is no problem of using switching FETs for analog amplifiers.

Second, I can't read the number of the BJTs, looks like MPSA something. This is a DC coupled circuit. If you use individual transistors ( not match pairs). Even though it is closed loop feedback like an opamp, there is an output offset and they better use a DC blocking cap at the output to the speaker.

There is only one dominant pole compensation using two of the 47pF cap to form the miller cap on the base collector junction of the second differential pair. I wonder what is the phase and stability margin of this opamp with reactive speaker load.

This is actually a working circuit used in guitar amp?!!! There is no thoughts into this design. Remind me the day when I was designing BJT IC that I just whipped out a simple opamp inside the IC to perform some non critical function and never can be access by the external world, I could strip it to the bare bone, no worry about distortion or anything. But at least I would put some current limiting resistor between the two FETs!!!!

[J M Fahey]

This schematic, as is, has been used by tons of makers, from HH to Marshall to Trace Elliot to ... you name it ........ but ..... with the proper transistors !!!
In fact it's the Hitachi factory suggested circuit for their MosFets.
Their very special characteristics make them work in such a simple circuit.
Now, if you substitute Darlingtons (it can be done) you have to add thermally tracking bias and emitter resistors , and, with switching MosFets , well, they are meant for switching !! you can use them for (Linear) audio but biasing becomes critical and tracking even more important.
No short protection is acceptable in a closed system: combo (without extension jacks) or powered boxes.

[Alan0354]

This schematic, as is, has been used by tons of makers, from HH to Marshall to Trace Elliot to ... you name it ........ but ..... with the proper transistors !!!
In fact it's the Hitachi factory suggested circuit for their MosFets.
Their very special characteristics make them work in such a simple circuit.
Now, if you substitute Darlingtons (it can be done) you have to add thermally tracking bias and emitter resistors , and, with switching MosFets , well, they are meant for switching !! you can use them for (Linear) audio but biasing becomes critical and tracking even more important.
No short protection is acceptable in a closed system: combo (without extension jacks) or powered boxes.

I just calling it out as it is. MOSFET bias is know to vary quite a bit from part to part and drift with temperature. Yes it does self regulate to some extend but I don't think you can have that well a control on the idle current.

[Alan0354]

This schematic, as is, has been used by tons of makers, from HH to Marshall to Trace Elliot to ... you name it ........ but ..... with the proper transistors !!!
In fact it's the Hitachi factory suggested circuit for their MosFets.
Their very special characteristics make them work in such a simple circuit.
Now, if you substitute Darlingtons (it can be done) you have to add thermally tracking bias and emitter resistors , and, with switching MosFets , well, they are meant for switching !! you can use them for (Linear) audio but biasing becomes critical and tracking even more important.
No short protection is acceptable in a closed system: combo (without extension jacks) or powered boxes.

I just calling it out as it is. MOSFET bias is know to vary quite a bit from part to part and drift with temperature. Yes it does self regulate to some extend but I don't think you can have that well a control on the idle current.

Don't give too much credit to reference circuit. Seen too many less than optimal reference circuits. I even personally went to Linear Technology to demonstrate one of their IC has power up latch up problem and you should see their face turning red. And I proved to Analog Devices one of their ADC had major missing code under Nyquist condition. I worked in Exar Integrate circuit manufacturer before, I know the caliber of their product engineers. Another case of Linear technology switch converter IC, one engineer at work copy exactly the reference circuit and it was not stable. It oscillate and cause slanted lines on the TV screen, I had to go in and fix the oscillation. the closed loop feedback was not stable. I had to re tame the circuit.

But in defense to the product engineer, their quality has improved drastically in the past 10 years. this is because as IC and application of transistors become more complicated and critical, people are more and more rely on the reference circuit, the caliber of product engineer has gone up quite a bit.

[Steve Conner]

Are we talking about the circuit shown in the first post of this thread? It's a sack of sh!t. Here's the original 1970s Hitachi circuit that it was based on:
A Paul Kemble web page - Hitachi Fet designs.

Note that: 1) There's a trimpot for adjusting the bias and 2) It specifies lateral MOSFETs for the reasons discussed in this thread.

For more information on audio use of VMOS devices read this: Using HEXFETs in High Fidelity Audio

[Steve Conner]

Are we talking about the circuit shown in the first post of this thread? It's a sack of sh!t. Here's the original 1970s Hitachi circuit that it was based on:
A Paul Kemble web page - Hitachi Fet designs.

Note that: 1) There's a trimpot for adjusting the bias and 2) It specifies lateral MOSFETs for the reasons discussed in this thread.

For more information on audio use of VMOS devices read this: Using HEXFETs in High Fidelity Audio