Low watt Aleph J

This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
I have built the Aleph J from the First Watt schematic and it sounds amazing. The problem with that amp is it runs a little hot. In a cold room in the winter a couple inches off the floor, (around 55 to 60 degrees F) after an hour the heat sinks get hot enough that I can only keep my hand on them for 4 seconds. So in warmer temps it would be way too hot. I was thinking about larger heat sinks when I saw Gray's recipe for Aleph J X . He says to pick the output watts and select a power transformer based on that wattage. This would also affect the current used. So if I lowered the rail voltage I could make the amp run cooler.

I was wondering what the lower limit is for choosing the output watts. I am using high efficiency full range speakers and I do not need a lot of power. With the amp the way it is now I have the volume barely turned up to get loud volume. I would like to have only 5 watts per channel output. That would be plenty.

Would it be possible to bias the amp for 5 watts output with + -11V rails?

The figures I get from Gray's recipe for 5 watts are;
current and voltage
target rail voltage
transformer voltage
transformer VA x2
double for stereo

So a 8-0-8V 80VA transformer would be a little higher voltage and have plenty of current, if this would be even possible.

The one and only
Joined 2001
Paid Member
You can pick your rails for the peak wattage you need (which
will be twice the rms figure), but you should allow an additional
5 volt margin. So for 5 watts, you need a 10 watt peak, which
means a +/-9 volt peak, so you should have at least +/-
14 volts, and a bias figure of at least 0.75 A - more if you
can get away with it.

An easy way to reduce the dissipation if you only need a few watts is to pull a pair of output devices. If you lift one leg of the Source resistors in one of the current source devices and one on the gain side along with a pair of the output current sense resistors. Then you could reinstall easily, should you want more power next winter.

Another option is to increase the power resistors (Source and current sense) from 0R47 to 0R68, which would reduce your bias ~20%.
If you have a thick aluminum face plate on your chassis you might use it as an extra heatsink and put a couple of mosfet cap. multipliers there. That would lose you about 4v and then just turn the bias down a little. Sure would be cheaper than another transformer.
BobEllis, I tried lifting those resistors but when I powered the amp, all I got was static/hiss and the speakers pushed out. Put them back everything still works. So just to make sure I lifted the right resistors, they are R17,R19,R20,and R21?

Bill Fuss, to adjust the bias on that side requires adjusting only R27? You would not have to adjust R26 or R25?
Last edited:
You obviously lifted the wrong resistors. If you lift R16 instead of R17 the whole thing will go to hell. Never do anything like this with spkrs hooked up. Always look for output offset after any adjustment or modification with a meter. You are lucky your voice coils are still intact, but they could still be damaged severely.

R27 is the bias trim resistor. Without it the amp will bias to around .63V across R16, which is around 1.3A per fet. The lower it is the less bias current you will have.
Those are the ones to pull on the schematic I have. I guess you should pull the devices instead, or as Bill points out you may have pulled R16 instead of R17.

You can try tacking something smaller across R27 - Another 68K or something less should take a sizeable chunk out of the bias. Based on the Aleph 30 Schematic with voltages that someone posted, bringing R27 down to 26K ought to get you around .4A per output device, right around your initial target. You won't need to drop the rail voltage unless you really want to.
Before I lifted the resistors I double checked the schematic, layout for the pcb and followed the traces on the pcb to make sure they were numbered correctly. So I know I lifted the correct resistors.

Could I use for testing a 1000uF 35V electrolytic capacitor between the positive output of the amp and the speaker that could be removed after checking for DC offset? Or would there be an advantage to just getting some power resistors. I saw some 220W 6.8Ohm on bay.

I'm going to look into adjusting the bias and see how that sounds.
Sorry this is my first solid state amp so this is a learning experience. And modifying circuits helps me to understand what is going on. The PCB is by Peter Daniel.

With the amp hooked up as originally wired the DC offset measured is .025V and .03V. Should I try to adjust that lower?

My new power transformer arrived. A 10-0-10, 120VA from Triad. I also installed a 47k resistor parallel with the bias trim R27. Tried it out and it ran way cooler. The volume was way less sensitive, I could turn it up a lot more but it is still loud. After a few hours I could keep my hand on the heat sinks as long as I liked. But when I checked the transformer, it was hot. I could only keep my hand on it for 10 seconds. Checking the current being used I found each mosfet was pulling .978A and the amp total 3.914A. The transformer is rated at 6A. Looks like it could handle it but I should have a bigger transformer.

It also seemed like it was not as easy to listen too. When I first built the amp and listened to it I was amazed by the smooth non-fatiguing sound it produced. And after it warmed up the sound became more textured. Now the amp still had loads of detail but not as smooth and textured.

So I decided to try again at lifting the resistors to take a pair of mosfets out of each channel. I checked the outputs and they were -10.2V and -10.6V offset. I checked with the meter around the circuit until I found the problem. On the schematic R13 from the short protection circuit connects to R18. On the board it connects to R19. So I was in fact lifting the wrong resistor. Doh! :eek: That caused Q3 to open and send the negative rail to the output.

I reconnect R19 and lifted R18 on both channels and checked the outputs. DC offset .025V and .03V. I hooked up the speakers and listened. Much better volume. I could turn it all the way up and still stand in the room. Still gets loud just not as loud. But the sound was the same as before. Did not sound like when I first built it.

I checked the current. 1.87A for the amp. While reading other posts here I see that the more bias means better sound. So I tried removing the 47k resistor from the bias. Checking the current. 2.35A. The sound is like it was before. So now I wanted to know what it would sound like with more bias. I had read you can remove the bias trim resistor to make an Aleph sound its best. Thats what I did. I lifted R27. Check current 2.55A. Not too much of an increase. The sound? More texture to everything especially acoustic stringed instruments. Seems like the low end got a little more pronounced. Like the amp does not have to try as hard now. Very nice.

The heat sinks do not get hot and nether does the transformer. :D

Now for some questions.

BobEllis, you said I could increases the power resistors from .47R to .68R to drop the bias. Could I decrease the resistors to .33R to increases the bias? Could those resistors be lowered more than that? And would I have to adjust the short protection circuit too?

Also I do not understand the function of the power resistors R20-R23. The only thing I see they could affect would be what I think is the AC gain(C3,R24), which I am also not sure on how it functions.

Thanks for the help
Everything you are saying makes perfect sense.

R20-23 ARE part of the AC current gain mechanism, which is a ratio between R15,R24, and R20-23 and R16-19. If you maintain the ratio between your sense resistors and your source resistors all will be fine. If you eliminate fets on the output you should also eliminate the same number of sense resistors. Likewise, if you change your source resistors to .33 ohms you should also change your sense resistors to .33 ohms.

.33 ohms may be too much current for one fet to handle, depending on the intimacy of contact with the heatsink and heat transfer. That would be approaching 50 watts per fet. I think I would stop at .39 ohms, which is around 40 watts.

I hope I was coherent enough to understand.

While waiting for the transformer I installed some more capacitors in the power supply. I had 30mF on each rail and I increased that to 75mF. That gave the amp a big boost in the low end. Very nice. The next thing I would like to try installing some chokes for a CLC filter.

Then my 200VA 20VCT transformer rated at 10A arrived. Time to change the current sense and source resistors to .33ohms which should raise the bias from 2.55A to 3.6A. I was not sure if that will do anything much to the sound since the amp sounds so good already.

I switched the resistors out, turned it on and gave it a listen. With no warm up it sounded good. As it warmed up I started to notice the ambiance in/of the recordings that I never noticed before. Reverbs, recording room reflections, and other things that give sounds/recordings depth and space started to reveal themselves. Delicate sounds sounded amazingly delicate. Everything seemed very precise and continued to get more focused until I started to experience holographic imaging. I could not believe the improvement in sound quality. I could hear instruments and voices so clearly and placed pinpoint on the sound stage. Sitting there with my eyes closed I could hear everything! I had never experienced this before. As I listened to different types of music I found myself listening to songs that I had not cared to listen to before. Now they sounded so amazing.

Now I am thinking, is this now the best it can sound or if I raise the bias more will it sound even better? What is the limit where the amp won't sound more refined from increasing the bias? After a couple of hours the heat sinks are hot but I can keep my hand on them for 15 seconds before it becomes uncomfortable. The transformer is just warm. I measured the bias and the amp is at 3.76A. 1.83A per channel at +/-12V I think that the MOSFETs are at 44.5 watts. I would have to lower the rails more to increases the bias anymore. Keeping less than 50 watts.

Why is it best to keep the dissipation less then 50 watts?

Official Court Jester
Joined 2003
Paid Member
50W - magic number for keeping white poof inside power mosfet

also - 50C - magic number for keeping them well off flue


  • TEMP.jpg
    39.5 KB · Views: 780
  • Like
Reactions: 1 user
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.