JLH1969 MOSFET Class A/AB amp

gogowatch

2SK1058 has a much lower point of inflection.

Cheers,

jason2511

可惜看不懂啊

millwood

here is a picture of my jlh1969m.

it runs off 40v unregulated power supply, a pair of FQA27N25 (together with irf510/540, my favoriate mosfets. they are connected to the board via those red wires), 2n5401, 2200uf Nichcon output capacitor. because of the gate capacitance of the FQA (2450pf), I used a pair of gate stoppers.

this is the same board that i posted at diyaudio a while ago, with different output devices.

Sound? Beautiful. I have a 3-car garage (close to 100 square meters) and thiis amplifier, with a piece of crap RCA bookshelf speaker, can fill the entire garage and then some more. Definitively louder than what's required for normal listening.

millwood

gogowatch, you are absolutely right about the negative tempco thing between lateral mosfets and verticial (V- or D-) mosfet.

here is the same picture from FQA27N25's datasheet.

let's say that your amp starts at 25C degrees, Vgs=5v, and Id =400ma (point O on the graph) and Vds=20v (JLH1969 on a 40v power supply). power dissipated off the device is 400ma*20v=8w.

As the amp heats up to 150C degrees, if you maintain the same Vgs of 5v, your Id will be 2amp (point A on the chart), or a 5x increase. and your power dissipation will be 2amp*20v=40w. that requires some seirous cooling!

in order to maintain the same Id=400ma, you will have to reduce Vgs to about 4.5v (point B on the chart).

That is why you will need a Vgs multiplier when you are using V-mosfet power devices.

enjoy.

millwood

no a/c hum, and the sound is just very spacious.

good clipping behavior and no fatigue.

on power on, there is no thump. However, there is a split second of inrush current.

Just beautiful sound.

gogowatch

Thanks Millwood for the explanation about the thermal stability when using V type Mosfet.  

Could you show us more about the Vgs multiplier arrangement suitable to this circuit?

BTW, I will try your suggestions on #119 when I can get access to my stuff in the office tomorrow.

Cheers,

millwood

gogowatch:

for the jlh, you wouldn't need a Vgs multiplier as the circuit does thermal tracking itself - if you want, I can show you that analysis was done.

A typical Vgs multiplier is just like a typical Vbe multiplier, execpt that you use a mosfet rather than a BJT.

Yeah, please let us know know how your experiment goes.

BTW, you go to the office to work on your amplifier? you have the greatest job in the world.

millwood

more pictures,

the first is a shot of scope sreeen (a B&K oscilloscope). Because of the slow shutter, you are seeing multiple traces here. Y: 5v/div and X: .5ms/div. The scope probe is at the speaker terminal, after the output capacitor. The signal comes from an old Denon DVD player (dvd-2800), playing a CaiQing's CD. no sign of any oscillation.

the 2nd picture is a shot of the same scope screen, with flash on.

the 3rd picture is a screen shot with the DVD player in pause. Y: 5mv/div, and X: 0.5ms/div. very clear, no hum of any kind.

enjoy.

millwood

btw, the amp was within inches from a open frame 250khz switching mode power supply.

millwood

gogowatch, here is Rod Elliot's take on this whole debate of lateral mosfet vs. vertical mosfet. it has lots of interesting reference.

http://sound.westhost.com/articles/hexfet.htm

enjoy.

millwood

here is a picture of my jlh1969m.

it has been running straight for several hours. the mosfets are cooled by a small pentium class heatsink and remain lukewarm. the heatsink does have a 12v fan and it is cold where i live.

gogowatch

BTW, you go to the office to work on your amplifier? you have the greatest job in the world.

Yes, I do.  Just because ............... I'm the boss of a small software house, .  And, the most influential family member does not be there all the time.  I've put my breadboard stuff in a drawer and simple close the drawer after measuring and testing.  Very neat indeed.

Yeah, please let us know know how your experiment goes.

Today will be a busy day and will continue the experiment in the evening.

for the jlh, you wouldn't need a Vgs multiplier as the circuit does thermal tracking itself - if you want, I can show you that analysis was done.

How stupid I am!!  The BJT verson does not need a Vbe multiplier too.  But I do want to know the analysis.  I'm a sudden audio DIYer, a new comer to analogue circuit and want to know more.  I even don't know how to use an op-amp several months ago!


Cheers,

gogowatch

hello millwood,

I'm happy to tell you that your suggestion to add gate stopper works great.  I've added two 100 ohm resistors as gate stoppers and have the scope trace attached showing no more high frequency oscillation.  Thanks.

The Id current is graduately going up from ~100mA.  As you can see in the second picture, I attached a tiny heat sink to the MOSFETs.  Is this the reason of the ramping up of the Id?

I just got a E-MU 0404 sound card and will measure the amp performance in a couple of days.

And also, do you have some more links for the discussion of thermal stability of JLH topology, thanks.

Cheers,

Biglee_163

设备不错啊!

millwood

that's a wonderful set-up for a software company.

congratulations.

the best thread is the one at diyaudio.com. there was extensive discussion on if this particular setup is thermally stable.

you can do simulations to measure it as well: that's why in some of the schematics I had for jlh1969m, you will see a "voltage source" in the source of the mosfet.

thermal run-away is a real risk if you are running at high voltage (over 50v). but other than that the circuit is quite stable by itself.

please let us know how it works out for you.

gogowatch

My experimental JLH1969M amp is pretty stable after I mounted the MOSFET on a Pentium II Slot 1 CPU heat sink as shown in the picture.  The idle current is now 550mA.  I'll try to tune it up to Class A in the coming days.  Since I'm using unregulated power supply, hum is pretty serious.

Cheers,

millwood

that's so typical of many set-up I have. breadboard + cpu heatsinks.

way to go.

hum has never been my issue with unregulated power supply - the only regulated power supply I have is an open frame lm2596-based switching mode power supply. and I have no hum, not just with jlh but other amps I have used either.

can you put the probe on the base of the input transistor and see if you get any hum? and then the output, before the capacitor?

老牛288043

这里一片鸟语花香，就是一句都看不懂。

gogowatch

hello millwood,

My unregulated power supply, with 10000uF cap, produced lots of hum.  Picture 1 and 2 are the traces using the unregulated PSU.

Picture 1: X=0.2ms/div
- Upper trace is base of the input transistor 10mv/div
- Lower trace is the output before capacitor 100mv/div

Picture 2: X=0.2ms/div
- Upper trace is base of the input transistor 10mv/div
- Lower trace is +ve rail 100mv/div

Hum at output is 200mv pp.  PSU ripple is 300mv pp.  Idle current is 500mA.

I tried to use switching power supply.  I stacked two IBM notebook PSUs to give 32V.  The setup gave traces in Picture 3, 4.  Idle current was tuned to 1A.  No more hum then and only high frequency switch noise was seen.  The MOSFET case temperature was 72 degree Celcius.

The setup sounds great!  Thanks for your sharing.

I'll proceed to do the measuring task after I solve the hum problem.

BTW, I think I know why the circuit is thermally stable now.

Cheers,

millwood

let's me try my set-up and let you know how it goes.