ZEN working

I finally got all the parts and made my PCB's for my first amplifier project, the pass ZEN. It worked first time without anything going wrong whatsoever. I have a few people on the forum to thank for their help, namely vince and grey. I took greys advice and made the board both ZEN and ZEN revisited compatible so I can substitute values at any time without replacing the board (thanks for the tip). I am presently using the Infineon BUZ350 mosfets for both the CCS and gain stages but I am going to replace them with lower voltage devices (IRF150) in the gain stage to see if I can hear any difference, as lower voltage = less distortion (as is my understanding). I hope to put all of this stuff on a web site soon (-ie- components, PCB layout, findings etc.) but in the mean time I have some questions.

1) How hot is too hot for the heatsinks? The mosfets are rated at 175 deg C junction temperature. Without my pedestal fan pointing at the sinks they get hot very quickly and seem to stabilize after about 5-10 minutes so I can touch them for 2-3 seconds without burning myself. NB: I am using the lower powered original ZEN design at present.

2) I noticed the heat rejection from the CCS mosfet is much less than the gain stage mosfet, does anyone know roughly what the 2 different Power dissipations are for these. I need it for some rough calcs for my case that I am building. At present its a big mess mounted on a sheet of MDF on my lounge room floor.

Thanks once again for your help guys so far this has been a most rewarding project.

Congrats on the amp.
Me? I like being able to leave my hand on the heatsinks for as long as I want. This led me down a long, winding path trying to find big enough heatsinks to passively cool an Aleph 2 (power dissipation=300W/channel). Eventually, I arrived at the water-cooled scheme described in the Water Cooled Audio Circuitry thread. I may at some time in the future go with passive heat sinks, but at present I'm enjoying being able to put the heat elsewhere. (Summer in the southeast is a bitch...not just hot--lotsa places get as hot or hotter--but hot and *humid*...yuk.)
That said, Nelson ran the Alephs hot. Check the manuals, but I think it was on the order of 120-130 degrees F.
Just use as much heatsinking as you can get your hands on, then add fans if it's still too hot. No gain device ever croaked from being too cool...


P.S.: Happy listening.
The Aleph 2 and the Aleph 4 are both 100W amps, but since the Aleph 4 has two channels in the same chassis, they run the bias current a bit lower in order to keep the poor thing from frying. As a consequence, the 4 will current limit sooner into low impedance loads. That's why I decided to build 2's, as I've got ribbons and planar drivers with low (albeit purely resistive) impedance.
I am considering building a pair of 1.2s, just for the fun of it--but with increased bias current, and possibly fewer output devices per bank (primarily to reduce the capacitance of the output stage). But that will be a month or two away, as I've got a full plate at the moment.
Yes, the power dissipation in the Alephs is equally balanced between the current source and the gain device.

Mosfets living dangerously

Thanks for the responses,

I am still curious to know how much heat is dissipated by the CCS and gain stage mosfets (individually). I am using the original ZEN design with a 0.33 Ohm resistor, hence 2 amp quiscent current arrives at the drain of Q1.

I tried the IRFP150's as the gain devices I'm not sure I could hear the difference although I don't think I am allowing myself to listen to the music properly as I am too concerned about the well being of my fets.

If the sink to fet interface has been done properly should the case of the fet be the same temperature as the sink? This is certainly not the case with mine at the moment. Although heat is evenly distributed to all areas of the sink the case of the fet is bloody hot (while the sinks with fan assist are luke warm).


I looked over the aleph 2 manual which nominated 120 - 130 F as you said. Is this the sink temperature, the fet case temperature or the device junction temperature.


I have owned both the Aleph 1.2 and Aleph 2's and can say that the 1.2's sound better, possibly due to a larger power supply. A possible solution to the problem of gate capacitance could be increasing the bias current of the differential pair, which is I believe is 20ma, and adjusting the drain resistor value. If heat becomes a problem cascoding the differential could be a solution as long as stability is maintained, with possible sonic benefits.
I agree with Nelson Pass that using a current mirror on the differential will not sound as good despite lower distortion numbers.
If you're running a 34V rail & 2A current, the current source and the gain device will split the power dissipation evenly at about 34W per device--about 70W for the pair. The device will always be hotter than the surface it's resting on; mechanical connections are inefficient at transferring heat--at least they are after you add in insulators etc. (Which is one reason I went to copper bars soldered to copper pipes for my heat sinks. The mechanical connection I started with--aluminum L bolted to aluminum water channel--wasn't transferring heat as well as I wanted.) The 120-130 degrees Nelson quotes is for the heat sink. The devices themselves are going to be hotter.
One thing you can do to increase efficiency of heat transfer is to mount the transistors on separate heatsinks and omit the mica insulators. Note that you will have to *electrically* insulate the heat sinks from one another, and from ground, as they will be at the potential of the case of the transistor (usually the drain).
I considered building the 1.2 (actually, I sat down and went over all of them...), but decided not to because it current limited too soon. For a number of reasons, I decided to go with the 2. At this point, I've got enough spare gain devices (and my water-based heat sink seems capable of handling an extra 600W) that I'm going to build another pair of 2s so that I can experiment with the difference between the way they image vs. the way my tube circuit images (see my listening notes elsewhere--I won't take up space repeating them here). It's only costing about another $75 or so...so why not? Anyway, I'm within spitting distance of getting done with a new phono stage, then I'll build another pair of 2s, then I'm going to try out a few ideas I've got for a line stage. And I'm *still* fiddling with my crossover. At my current pace, I will be ready to start on a pair of 1.2s later this summer or early fall. I've already begun to accumulate parts, but I'm not making a big effort yet. Amongst other things, I'll have to order in a new batch of output devices, probably TO-247 case so as to dissipate heat better, especially if I follow through with my threat to increase the current in the output stage.
Your suggestion about increasing the current in the input differential is interesting, but it might be necessary to fiddle with some of the other circuit values to compensate; in particular, the drain resistor for Q1 might need to be decreased.
Although I've never experimented sound-wise with cascode vs. non-cascode, I did do a lot of work along the lines of resistors vs. current sources for differential circuits for my tube amp and definitely agree that resistors sound better. Since the upper device in a cascode is a current source (some people like to call 'em active loads), I see no reason not to be suspicious of them. Current mirrors are nothing but dual current sources and are just as suspicious in my book. To me, all gain devices are guilty until proven innocent. Unfortunately, we need them...

Current Sources

In experimenting with current sources I have found that they work best if they are connected to a regulated rail ie. the B+ line in the Aleph amplifiers.
Resistors(tail of the differential)if seem to work best if they are attached to a lower or higher B+ supply(in the case of Aleph amplifiers).Apparently this improves the performance of the differential amplifier.(More reach?)
This trick was used by a number of manufacturers years ago before the use of the current source became common.
I might be an interesting experiment to replace the current source of the aleph with a resistor and attach it to a 100v
regulated line.

Yeah, we were talking about this in another thread (I forget which one). I (grudgingly) use a current source under the input differential in my tube amp circuit. I've considered using a resistor and a really deep negative rail to accomplish the same thing, but have never gotten around to it, primarily because it's going to be a pain in the butt. Possibly the front end in an Aleph would be less picky (I'm using the differential in my amp as a phase splitter and need the 'infinite resistance' that a current source provides in order to make the positive & negative outputs balance). If I were to tinker with Nelson's circuit (other than what I've already done), I'd probably start by regulating the rail going into the front end. I set up my circuit boards so that it would be trivial to run the front end board off of a separate power supply, whether regulated or possibly a higher voltage. That will have to wait until I clear out some other projects, though.

Current Sources Revisited


I found your thread on current sourcing the phase splitter of your tube amp. and I think I might know the cause of your problem.
Current sourcing the phase splitter(or differential) is going to limit the ability(due to current starvation) to drive,I assume in your case,the output tubes. Increasing the current through the phase splitter should somewhat correct this. Intrestingly enough a current mirror(yuk!)could be helpful in this situation(equal pull-up and pull-down).
I came across this problem in the original application notes(Hitichi) for the drive circuit, for their power mosfets. The circuit consisted of an input differential pair followed by a gain stage differential, with a Vbe multipli(bias), some varaitions just used a variable resistor, attached to one of the legs of the second differential. The drive for the outputs was taken off this point.
The first differential was current sourced, but when you tried to source the second differential distortion increased dramitaically(slewing). The problem turned out to be current starvation and a resistor worked better. Replacing the collector resistors with a current mirror reduced the distortion even more,for the above mentioned reasons.
The only other work around was to increase the current thru the current source and change the value of the collector resistors. The downside was that heat dissipation was up and the current mirror was the only elegant solution,without resorting to drivers.
The problem you will face is there is no such thing as a P-channel tube, so the current mirror will have to be built with mosfets or bipolars besides current mirrors tend to cancel second harmonic products which might be not so good sonically.
I hope I have not meandered too much off the point of this thread.


[Edited by jam on 05-29-2001 at 02:35 PM]
Increasing the drive is a good idea, but I don't think it's the cure to this problem as the sonics were pretty much the same even at low volume, hence not drawing grid current (AB bias).
Mind you, the sound of the amp would probably have satisfied a lot of people, even with the current source(s). It's just that I was determined to do better, sonically. Resistors happened to be the solution. One of these days, I'll get around to trying a resistor (& negative rail) under the front differential...when I get time...

Now, that's an idea that I can live with!
(Clearly, you aren't going to be happy until you've got me working 48 hours a day...*sigh*...actually, I'm smacking myself in the forehead for not thinking of it myself.)
I'll have to drag out the schematics and see what the raw AC going into the bias regulator is, but it should work just fine.