DC offset / Bias setting for Aleph 2

Hi all,

After listening to my Aleph 2 for over a couple weeks now, I've noticed that the heatsinks on the left channel is warmer than the right channel. So, I took some measurement. The DC reading on the speaker output terminal shows 30mv on the left channel and 10mv on the right channel. I suspect this is the main cause for the warmer heatsink on the left channel. (I assumed this is what you call DC offset reading).

The following link is what I came up after doing a search on "biasing" and I have spend the morning reading about it.

To cut the long story short (& after my couple hours of reading), it basically boils down to what was describe in the service manual and that is:

- changing the source resistors R40-45 would change the bias current draw (so, to increase bias, decrease resistor value)
- R19 trims the DC current value
- R21 & C10 adjust the current against output current as sensed by the voltage across R22-27.

My questions:
1. In my case, to lower the DC offset, am I correct to say that I would have to increase (or decrease?) the value of R21 or ...
2. Increase the total parrallel resistance value of R22-27? (by adding another resistor of same value in parrallel .... to do it the easy way?)
3. If my line of thinking is correct, how does one calculate resistance value of R21 or R22-27 so that I can decrease the DC offset from 30mv to 10mv or it could be the other way round, increase from 10mv to 30mv. I don't mind running the amp hotter.
4. If my line of thinking is totally off, may be I need to adjust the bias current draw first by changing resistance value of R40-45 to lower the bias current and then the DC offset will correct by itself?
5. An observation: After I turn off the left channel amp, the blue led light will still show some light even after 10 minutes while the other channel blue led will be completely off after 3 minutes. Looking at the schematics, this led light is connected to the source bias resistor R40-45. Would this observation/hint provide any help in figuring out a solution?

A footnote:
Both channels are working perfectly. I won't say one channel sounds better/worse than the other.

I'm aware that today is Sunday and probably everybody is on the beach enjoying the sun .... but if anybody is home on their computer, please take a look at my question. Thanks in advance.
The 30mv and 10 mv reading you took is the DC offset. This is a result of the small mismatch in the differential transistors. To reduce it closer matching is required. It is within tolerance on both channels and is probably not the cause of the difference in temperatures. The service manual states DC offset is less than 100mv for the commercial Aleph 2.
The temperature difference is due to one channel drawing more or less current. Measure carefully the voltage across the source resistors of each channel and use Ohms law to calculate current draw. One channel will probably be lower. You can adjust the value of R19 to raise or lower the current draw to some extent. If this is not enough ( it should be ) you would then have to lower the source resistance of the cooler channel.

Thanks for the quick response!

I was going to post to let you guys know that I do have matched output transistors to within 10 to 20mv and I also have matched input transistors, Q1 and Q2. So, you're saying DC offset is not the cause of the temperature diference ... that's good. I did noticed that the service manual says <100mv for DC offset is acceptable.

o.k., so, the problem is ... the DC bias WAS too high for one channel in my case.

I'll measure the voltage across the source resitor later and do the math of I=dV/R and add up all the current value of each source resistor to get the total current draw.

Assuming the DC bias was too high (measure later as mentioned above), I'll have to INCREASE the resistance value of R19 to trim the current to the output stage. Correct? I was hoping that there is a formula that I could use to calculate the value of R19 for a certain value of DC bias value so that I can get a closer number. Just from the top of your head, what value would you suggest for me to try for R19 (to start with) to LOWER the DC bias? Let's take an example, assume the current draw is 3.5 amp now.

Just a side comment. I read one time about using match resistors. Have not read into how to do that yet. But in this case, if I did use matched source resistors, I would NOT have this difference of DC bias / temperature? True?


2000-12-31 9:31 am
the 10 vs. 30mV dc offset its just fine. What you need to do is to measure a bunch of voltages or currents and make sure that they are the same. Did you measure all the components before doing up the PCB??
To figure out where is the difference coming from you should first make sure that the power mosfet are fed the same bias voltage then, if it is so, you can measure the voltage with respect to ground before and after ALL the resistors in series with the power mosfets. From these measurements you can calculate the voltage drop across the resistor. I would not assume that the resistance is really what it says it is even if you measured it before soldering them in. The only way to know for sure is to take it out and I think you want to avoid doing that. In principle the average voltage drop of the 12 resistors should be the same for both channels. If it isn't that is your problem. Maybe one or two of the 12 mosfet unit (by unit I mean the mosfet and its resistor) is misbehaving you can easily identify that by this analysis and replace the bad boys.

Potentiometer ... this is getting very interesting!!

I could have a potentiometer permanently installed on the back panel and have the option of increasing the bias for serious listening (to me ... that means loud!) and lowering the bias for not so serious listening. And of course, put a little fan next to the amp on higher bias .... just in case.

To go a step further, it would be nice to permanently install a current meter to show the amount of current draw ... just like the look of the commercial X series Pass Amps.
Let`s try that even simplier. A finished my aleph 2 monoblocks to and this are concret numbers you need to check. I supouse you are runing a2 version of 12 pieces IRFP 240 (or similar) - 6 for + and 6 for - chanal of the amp. Your source resistors are 1ohm rated. Go and mesure voltage acros each of them. You shoud get the reading of aproks. 0,45-0,55V per resistor. So the total should be about 3amps per positiv chanal(6x) and the same on negativ - output chanal. Those 0,5 values shoud be prity close together. If that is the fact and your bias for 6 devices is to low or to high, you can try chancge the bias for coplet package of 6 devices as the guys advised before. If the values are very diferent you shoud check out your source resistors, insulation and fets (matching).

rgds Marijan

raising the value of R19 INCREASES the bias!
When you set a different bias value you must recheck you ac current gain setting and maybe adjust R21.

Normally the left and right channel bias is the same when using matched parts. It´s possible that the two fets that set the bias for the left and right channel are a bit different (the ones connected to the sence input of the current source). I would recheck the Vgs for those two.


more current through R19 means less current through the 1k resistor and less voltage drop over the source resistor and less current going through the fet.........

Voltage over the source resistor = Vbe (0,66v) + V over the 1k resistor (don't have a scheme at hand so I don't know the number) This depend on the current going through the transistor and this again depends on the hfe of the latter.


Update on the DC Bias setting ..... in case anybody wants to know.

I have changed R19 on the "cooler" heatsink channel from 56.2K to 81K and now the DC Bias on both channel are about the same. The "cooler" heatsink channel used to have a reading of about 2.3A and now it's 2.6A. The other channel is about 2.7A. I do realize that they are slightly higher than 2.5A but I think I prefer it to run a little hotter. I run out of resistors to try to match the two channels bias exactly.

A side note:
Now both channel has about the same DC offset reading too. I guess this is call killing two birds with one stone!
More information on Bias adjustment

While surfing the web this morning, I saw this thread http://www.diyaudio.com/forums/showthread.php?threadid=3589 by William that goes into great details on his tweaking of his Aleph 5. It talks about Bias adjustment, gain adjustment, power output manipulation, etc . My previous "search" did not get me to his thread and thus I started this thread. I thought I mention it here so that at least I know where to look next time.