Aleph3 problem

rtirion

Member
2000-11-24 12:02 am
I just finished a prototype Aleph3 and fired it up. It seemed to work OK but the voltage across R120, R121 was only 0.28 volt. Sound was distorded at higher volume as well.

I also noticed Vbe for Q105 was only 0,5 Volt so Q105 was not completly conducting. btw Q105 is a BC550C instead of MPSA18.

So I lowered the value for the base resistor R115 from 1K to 500 Ohm. Voltages across R120, R121, R122 and R123 is now 0,51 Volt each and Vbe for Q105 is 0,67 Volt. Sound is clean.

My questions are: Is this the proper solution? Anyone experienced a similar problem?.

I have used the Aleph 3 Current source in a ZEN and this amplifier runs flawless. So this problem puzzles me.

Thanks in advance
 
In another thread I complained about sound of Aleph.
I have just realized similar problem I solved.
Your problem should be due to low gain of the BC. Bias current you can adjust by Base-Collector resistor. I use the trimpot here.Bias is set to 1 per rail (I have Aleph 30, it means 0.3A per IPFP). Is it enough or not?
 

rtirion

Member
2000-11-24 12:02 am
Thank you for your reply Mr. Pass. In a first inspection I did not find any faults around the current source. I have just finished a thorough examination, checking every individual net, no luck or fault for that matter. So I have decided to build a second channel from scratch to see if I can reproduce the same situation.

Koy, I have looked into the DC Gain (Hfe) issue you are suggesting and found that a fairchild MPSA18 has min_Hfe of 500, max_Hfe of 1500 (no typical rating specified) compared to a philps BC550C of min_Hfe 420, max_Hfe 800, typical_Hfe 520.
I have measured the Hfe for the BC550C before I put it in the circuit and it was 478.
So your suggestion is defintily worth some furhter investigation.

But...
As I metioned in my previous post, I build a ZEN with a Aleph 3 current source with some salvaged parts and I have used a 2n2222 (measured Hfe = 150) and this thing runs flawless. I rechecked again measuring all the relevant voltages and everything is OK. The ZEN that is.

So I am still puzzled

More to follow soon
 

rtirion

Member
2000-11-24 12:02 am
In addition to my previous post, I remembered that the ALEPH 30 uses a Zetex ZTX 450. From their datasheet I found that min_Hfe is 100, max_hfe is 300. This is well below the BC550C. Even the one I have used.

So, to me this pretty much rules out the BC550C as a suspected candidate.

Also if I slightly understand the function of Q105 than Hfe is not the primary issue for this component in this circuit. Minimum Hfe of around 100 should be just fine.
As evidence for this I want to use my still going strong ZEN with Aleph3 current source and 2n2222 for Q105

And this story will continue...
 

rtirion

Member
2000-11-24 12:02 am
So the story continues...

I swapped R113, 47K for a 100K and removed "my solution" (paralleled 1k over R115).
Powered up and tosed a CD in the player. Dissapointment all over. The sound is distorted at a lower level than before. btw R120, R121, R122, R123 showed the correct voltage accross each, 0,5 Volt.

So I shut everything down and reinstalled the 47K and "my solution". Sound is clean again and after listening to it for an hour or so I would say a little forward in the mid and high section. I am not able to say what the sound stage and or stereo image is, since I only have the first channel finished. As soon as I have finished the second channel I will try to comment on this.

Koy, as for your problem:
What type of cap did you use for C11?
I hope this is not one of those "famous orange drop" things or a tantalum. In the unfortunate event that it is one of those, I would suggest changing it for a normal good quality electrolytic cap, say philips, trobo, or elna-cerafin.
This cap can have a profound influence on the overall sound of an amplifier. I regret to say I had to find out the hard way when I build the John Linsley Hood amp about a year ago. I used the orange drop buggers and the JLH sounded dreadfull. Took me some time to find this problem then. Difference between philips, trobo or elna-cerafin are not noticable. At least in my own experience in this type of circuitry. C11 is part of the feedback circuit (R7, R6, R3, C11) and any influence from C11 will more or less be amplified again, and could cause your problem.
Even if you don't have an orange drop in your circuit, I think it is worth at least a try. (Not to much work, dirt cheap, and if it will not solve, it will surely not harm, I think)
While you are at changing caps you could consider changing C9, C10 as well in case of the orange dr&*(ps.
I suspect these are not as critical as C11, but if changing only C11 improves things, than you'd better change C9, C10 as well.

I took the assumption that your are driving your ALEPH30 unbalanced, so -IN is connected to ground.


[Edited by rtirion on 11-14-2001 at 07:10 PM]
 

rtirion

Member
2000-11-24 12:02 am
Koy, checking the schematic for the Aleph30 regarding your question:

.......is set to 1 per rail (I have Aleph 30, it means 0.3A per IPFP). Is it enough or not?

I would say the answer is no.

As you can see R34, R35, R36 are dead parallel. Doing some basic math this works out to a resistance of 0.1566 Ohm. (3 parallel R's of 0.47 Ohm)
Across this parallel bunch there should be a voltage of 0.25 Volt. So the current through R34, R35 an R36 works out to be 0.25Volt/0.1566 Ohm = 1.596A.
Each resistor will handle 1/3 of this current. So will each of the IRF244's.
Same story for R37,R38,R39 and Q9, Q10, Q11.
So 1.596, let's say 1.6 Amps per rail which is 60% more than the 1 Amp you currently use. I would consider this to be significant.

Please provide for adequate heatsinks if you increase the quiescent current for your amp, or you could be facing trouble.

Hope this helps a little.



[Edited by rtirion on 11-15-2001 at 05:55 PM]
 
Your amp: And the bias current was changed when moving the C-B resistor value up? Strange. Only explanation I see is higher base current for the BC transitor ( having 500ohm base resistor) - ie higher gain. Zetex maybe behave differently.
Chi,chi - you helped me a lot. The more apparent mistake the harder to find it. I checked it yesterday. On 0.47 ohm I have about 0.370mV - ie about 0.8A per IRFP. Together 2.4A per channel. Pass states cca 1A per trans - ie 2A together I belivede it to be better. After you information I noticed small figures in the schematics - not one but 3x 0.47 resistor!. I think it could be the problem. Not 0.8A but 1.5A per transistor. As to cap - I use ordinary electrolyts ( some Asia - Hitano or so), I try to add one polypropylen 0.1uF paralelly. Thanks!
 

rtirion

Member
2000-11-24 12:02 am
I am happy to say I found the solution for my problem!
After studying the ALEPH3 schematic again, I concluded that a possible solution was to be found around R111/R112. So I removed the old "solution" (parallel 1K at R115) and raized R111 to 3K. Fired up and the change was huge. Almost no distortion. Voltage across R120..R123 was up from 0.28 to 0.42V each. So shutdown again and I decided to add an extra 1.5K to R111 for a total of 4.5K. And then it struck me: I did see this before in the schematics of the other ALEPH's namely the A2, A4 and A5. So I studied all of them and decided to implement the values of the A5. So R111 is 4.75K (was 1.5K), R112 is and was 1.5K and R113 is 221K (was 47K).
After this change, no more noticable distortion, voltage across R120..R123 is 0.49V each.
So now I am the proud owner of an ALEPH3 with a touch of ALEPH5.
I will move forward in finishing the second channel and suitable housing for this unit. After that sit back and enjoy.

Cheers
 

rtirion

Member
2000-11-24 12:02 am
Yesterday I finished the second channel of the ALPEH3 with the ALEPH5 mod for the current source. Fired it up and it was immediatly perfect.
I had a short tryout with the pair, and I am very happy.
Now I will start the housing and defintive supply. I will try to get some picturess posted in a few weeks so you all can see the fruits of my labor.

Koy,
I like to hear which simulation software you used. I have tried a few, but either I make a lot of mistakes using these packages or the results are not always real world reflections. In short I hate them so far. Of course this can change over time.
For now, I will stick to my method of building, trying, and optimizing a test circuit first, and second start a simulation to confirm (or not) the real world results.
Used this method on power supplys and was very succesfull. Tried simulations on even the simplest amps and had mixed results.(most where hopeless)
Maybe we should start a new thread on this subject?
 
rtirion, I owe you a huge thanks!

I was suffering distortion at low power levels with the Aleph30 I'm building. I read this thread, and noticed that the Aleph30 also uses 1.5K and 47K in the active current source. I changed these values to 4.75K and 221K and the problem disappeared. THANKS!

Of course, this means that my Aleph 30 has 3.8A of bias current. I'm driving a MTM setup with Focal midranges and Accuton tweeter - the impedance dips to about 4 Ohms, so I figure the extra bias current helps with the lower impedance, too.

Sud
 
rtirion,

> What type of output device did you use?

I used IRF244s for the output devices, in 3 pairs.

>Did you use MPSA18 or something else for the BJT's?

I originally used MPSA18, then tried ZTX450. There was no difference in bias current. I couldn't comment on whether there was a difference in sound, because I've only applied your resistor changes with the ZTX450 installed.

Cheers,
Sud
 

rtirion

Member
2000-11-24 12:02 am
Koy,

Changed the 1.5k to 4.7k because getting the value for 47k up did not solve everything! You can read this in my earlier replys to this thread.
I do think the most important resistor to set the quiescent current is not the R111 1.5K/4.7K or the R113 47K/221K. I think 0.47 Ohm source resistors are the most important ones. I regard the resistors R111,R112,R113 as the construction to get a Vgs that is needed for proper/sufficient conduction of the Q106.

I did not (software) simulate this in any way but actually build different setups and took measurments for each setup.
Painstakingly, Time consuming? Yes but at least I got some real world values. And I like the end result.

Sud,
Happy listening and thanks for your reply

Regards