Waveform distortion on Aleph 30!

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Dear all,

Its been 2 years now that I am very happy with my aleph 30 clone based at exactly the published schematic. Only difference is that my power supply is at +/-28VDC biased at 1,6A per channel, also current gain is exactly 50%. Although i am satisfied with the overall sound i decided to take some measurements and i did.
My surprise was very big though to see that the amplifier is not behaving normally(at least to my understanding). To explain that I have to say that I found 18Vrms on 7,5 Ohm at 1Khz with excellent sinus and square. But at 10Khz I only have 13Vrms and very bad sinus as well as square, so bad that both look identical! At 20Khz the waveforms are funny, to say the least, and there is only 9Vrms available with tha same input (1,8Vrms). What I wanted to know is if this behaviour is typical for this topology or my amp got somekind of problem. Forgot to say that after the measurements I added 15pf extra at the feedback loop (25p total) and things got slightly better but certainly not good at all! All input and output mosfets are matched to within 10mV as well as all 0,47 Ohm resistors, offset is 15mV at start and 3mV after 10minutes, heatsink temp is 60 degrees celsius with 25 abient. Any answer will be greately appreciasted as am in the proccess of building my aleph X, but if these are typical I will quit this project as my gainclones reproduce all aveforms excellent at least to 30Khz and alot cooler and also I am not always able to say which is which in a blind testing! (pictures of the waveforms will follow asap). Terribly sorry for the long post, but I really had to tell about it!

Sincerely Panos :dead:
 
I also noticed a bump on the waveform at frequencies > 10 kHz, the shape was very similar to the one shown by Algar Emi in Aleph30 thread.

See
http://www.diyaudio.com/forums/show...perpage=15&highlight=algar aleph&pagenumber=9


Raising the compensation caps to 33pF Silver-Mica things got much better, altough gain at 20 kHz is slightly lower than at 1 kHz, but this is a minor concern.

I also read that current balance in the diff pair might be related to such behaviour.

Cheers

Andrea
 
Did you manage to solve it?

Hi Panos,

How is this going? I have bought the mosfet for my A30, but your problem fears me! I wouldn't know how to fix your problem, so maybe I wait for this solution before try to machining something.
Besides, gainclone is so cute and cheap... is tempting me... you devil... oh,... it's sooo cheap....hmmmm... come to me .... I'm falling... embrace me, oh chip.....
 
Hi Panoz,

is this behaviour only shortly before the amp clips or also when you measure at 2 or 4 volts rms?

I measured my Aleph5 up to 200 kHz and found a -3dB point of around 100kHz and no dependency on frequency. Square wave was also ok at 10 and 20kHz.

Did you also have a look at the input? Often the signal generators have a very high output impedance and do funny things at higher frequencies into a low impedance/capacitive input.

william
 
Disabled Account
Joined 2002
panos29 said:

Only difference is that my power supply is at +/-28VDC biased at 1,6A per channel...

I believe that you have read this:
 

Attachments

  • np_comment.jpg
    np_comment.jpg
    45.3 KB · Views: 873
Some light at the end of the tunnel!

O.k
After alot of search on the amp and many hours :hphones: , I found out that the current source should be at exactly 50% setting otherwise distorts all signals above 10Khz very much. Moreover I found that feedback cup should be more than 10p(to my amplifer) towards 15-18p to remove the square wave ringing and distortion. Additionaly the 1n cap between the gate and and drain of negative side mosfets is very much needed as installed in the aleph3 and aleph5. Biasing the circuit to 3 amp without the active current source in the game(R21 removed) gave the best square and sine wave responce by far. Obviously the active current source is more of a marketing trick to get some more power in 4 ohms(flames welcomed :flame: ), or on the other hand maybe there is something I do not understand at all, maybe there is not a direct relation between sound quality and square wave responce. O.k I know this is something overtalked in here but I wish somebody to remind the exact procedure to measure the current source with and without R21(aleph30)? Measurements were taken with a tektronix 475 and a Fluke 8840A (true rms option installed). I will post the exact measurements hopefully tonight. thanks to everybody for the interest!
 
Some new measurements were made right now. Vcc=+-28, bias 1,6A, at 11Vrms on 7,5Ohm square excellent at 1khz to 5khz, good to 40Khz, sine wave perfect to 60Khz. At 15vrms square excelent at 1khz good to 5khz acceptable to 20khz, sine excellent to 50khz. At 4ohm, excellent square 1-5khz good to 40khz, sine perfect to 60khz. Frequency respose at 7,5ohm from 100hz 15Vrms to 30khz 14.7Vrms. Notes: in all measurements distortion is evident in increasing frequency or power output. Current source behaviour is : increasing the gain corrects the sine wave but distorts the square wave so I choose a middle solution. Can someone guide me to measure the current gain please?
 
Current gain measurements

Forgot to mention the AC current measurements on the 0,47ohm power resistors measured with 9Vrms ouput power on 8Ohm load. So, on the 6 paralleled output resistors I got 83,3mV AC, on the negative side resistors 87,3mV AC on each one and on the current source resistors 89.5mV AC on each one. R21 is a 2k trimmer. What is the current gain and how can I calculate it?
 
Hi Panos,


I think the current gain is defined as the percentage of the total output current that is delivered by the active current source.

So at 1A bias and 2A peak output current the ac-current-gain is 50%. (1A extra)

At 1A bias and 3A peak output current ac-current gain would be 66.6%

The simplest way to measure it is to measure the bias and the peak output current (not directly but with the voltages over source and output resistors/or load resistor)

If total bias is b and peak output current is c the ac-current-gain would be ( 1-b/c) x 100%

You can also look at the ac-currents through the source resistors. At 50% these should be the same for the positive and negative side (source and output) ((with a load on the amp of course))

So when you have 89.5mv and 87.3mV your ac-current gain is a tad over 50% (50.6%).


william
 
Re: Current gain measurements

panos29 said:
Forgot to mention the AC current measurements on the 0,47ohm power resistors measured with 9Vrms ouput power on 8Ohm load. So, on the 6 paralleled output resistors I got 83,3mV AC, on the negative side resistors 87,3mV AC on each one and on the current source resistors 89.5mV AC on each one. R21 is a 2k trimmer. What is the current gain and how can I calculate it?


According to your measurements you have: 1,06 A output current, 557 mA on the negative side and 571 mA on the current source. The output current should be the sum of the other currents (well, minus the current flowing in the feedback resistor..), but taking in account wirewound resistors value spread (often 10%) your measurements are OK.

Therefore your AC current gain is slightly higher than 50% (calculated values can be between 50.6% and 54%, depending on which currents you assume as "good"), so I think everything's OK.

To make a more precise measurement there is another thread where such a method is described.

Cheers

Andrea
 
Measurements

I finally found some time to take some pics of the measurements I described earlier in this thread. After 1 hour warming at 60 degrees celsius, +-28v, biased at 1,6A. At top of scope is an RMS AC voltmeter and coresponding readings. No1 and 2 pics are 1Khz sine and then 1khz square just before clipping using the same setting on my signal generator, just pushing the square button after sine.
No3 and 4 the same at 10khz.
No5 and 6 the same at 20khz
No7 clipping sine at 1khz
No8 testbed

And my final question is: are these measurements typical of aleph topology or something is wrong? Soundwise the amp got lots of power and good control to my 15" Tannoys, but compared to my gainclone is a bit tiring to the ear although a bit more "alive" but less gentle and analytical compared to the simple gainclone using the same powersupply.
 

Attachments

  • 1.jpg
    1.jpg
    34.3 KB · Views: 641
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.