My joyful projects

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

These are my future & ongoing audio amp's projects .
Because it is fun to design them and hear .
I prefer the sziklai/cfp sound overall and prefer lot of degeneration of the emitters in the voltage amplifiers because I believe that the solid state sound is less predominant .

The first is one that has harsh sound sometimes but full of dynamics .(In this topology I've heard a lot of dynamics,the drums and cymbals are great) .
A lot of amp's has the idle current dependent with the ambient temp of the vas and also the temp of the power heatsink .
The focus here was to reach low temperature drift over the idle current versus the components temperature ,also the vas current is temp-compensated (Q4 and J1 are sticked together ) .Also the multiplied diode Q14 share the same heatsink only with Q7,13,9 and 10 .The power bjt's are alone on other heatsink .
With this vas load (current source), the vas has the same current in it with the temp, so the idle current of the power bjt's don't depend on the vas current source temperature .
In simulators and also tests with audio souncard,scope, the thd is very low at 1khz and also at 20khz ,also at any power the thd is almost the same 0.002% at 20khz.
At 1khz the thd is less than 10ppm .The low thd and imd is achieved only if the power bjt's are biased with 10 to 15mA.
Because the thd is smallest at 10 to 15 mA idle current for power bjt's it was a little tricky to design it to keep the idle current stable with the temp.
The slew rate is also high and if one of the + or - supplies is not working , the dc voltage is not present at the output compaired to the other amp's which share almost the same topology.
This has also the smallest imd but :
The sound is very accurate and not so musical .

So let's get further
 

Attachments

  • joy.JPG
    joy.JPG
    156.9 KB · Views: 644
Last edited:
The second is a my_ref modded a little ,with Lm4562/lme49720 and all dc connected .No caps through the signal path .
If I have dc offset from the source(dac,etc) ,depending on the source I have an switch whitch inserts a capacitor in the input .

The sound is full and rich within all the octaves but has less power and to my speakers seems that is not enough .
The dynamics are less compaired to the first schematic, good enough but not outstanding.
 

Attachments

  • my_ref.JPG
    my_ref.JPG
    96 KB · Views: 602
Last edited:
My dynaudio speakers eats a lot of current and a my ref is not enough .
The third one which is not done yet, I hope that will sound great .
It has the highest bandwith from all , almost 20 Mhz,no global negative feedback and no phase turning in open loop .The thd is 0.02 at 20khz at 3 watts .I will study the type of compensation .
 

Attachments

  • casc_vas.JPG
    casc_vas.JPG
    129.5 KB · Views: 571
  • fft_sim_casc.JPG
    fft_sim_casc.JPG
    99.7 KB · Views: 535
Last edited:
And the 4'th which I hope to finish it this year is a simple current feedback ,the idea being that less components =better sound .
This one has high slew rate ,small thd at low power and high to medium bandwidth .
We will see at the end what the ears will choose :)
 

Attachments

  • simplecfb.jpg
    simplecfb.jpg
    50.6 KB · Views: 515
Hi Guys

The first project is just the Blameless amp with a different current source for the diff amp and a buffer added between the VAS and CFP. The latter is beneficial where the former is not.

I would suggest that you found its sound to be dull, lifeless and boring because in fact you are hearing a transparent amp for the first time. Many have this same experience. Once you train your ear to recognise truly low distortion sound, you realise how much distortion is present in other things you hear.

Having a transparent PA will highlight deficiencies in the rest of the system. Or, if your speakers are also quite flat response with low-THD, you are hearing them play true sound for the first time.

You get used to certain distortions or frequency emphasis and likely even prefer some emphasis. When that is removed, of course the sound will seem to be lacking. The hard part is in assessing what it is we really like to hear and accepting that this may be different from technically accurate reproduction.

Otherwise, it is fun to play with lots of different circuit types.

Have fun
Kevin O'Connor
 
Hello .
The sound is very accurate and not so musical .

I like the design you have and wonder why you didn't find the sound musical. Something to learn, something to fix perhaps. It looks like a good design, LTP with lots of emitter degeneration and current mirror to maintain good balance and improve the gain (which was reduced with degeneration). I'm not sure about the current source - it plays such an important role as it is used for the LTP and VAS, I wonder if it deserves a decoupling capacitor in there somewhere, perhaps on base of Q4 to ground. Some people find that a bootstrapped VAS load sounds better. I'm also not so sure about the cap multiplier (Q5) and it's possible impact on the sound.

The feedback scheme is the most interesting part. The benefit of a Singleton input is the 2nd harmonics giving the sound more richness. The current mirror you have may spoil this and prevent it - I would have to see some simulation.

The fourth design looks like something I have been considering - a marriage of Gaborbela's "20 year" amplifier and Rod's P3A output stage. I worry a bit about the bias stability though - the CFP output needs very accurate bias and the symmetric design is a poor topology in basic form for bias stability. I think this needs more work.
 
Last edited:
My dynaudio speakers eats a lot of current and a my ref is not enough .
The third one which is not done yet, I hope that will sound great .
It has the highest bandwith from all , almost 20 Mhz,no global negative feedback and no phase turning in open loop .The thd is 0.02 at 20khz at 3 watts .I will study the type of compensation .

It took me a minute to realize this is a double differential input stage, two LTPs. I've never used this topology but my Bryston B60 amplifier uses it and it provides for a very clean undistorted sound with strong bass. If I understand properly, the LTP drives a folded cascade VAS ?
 
Does anyone ever consider that its not always the amp that is responsible for poor sound. Its just one link in a system, the others could be responsible for poor sound. An ideal amp is a wire with gain so its not supposed to sound like anything, its suposed to sound just what your input into it sounds like. If you dont like it, instead try changing your input source which is likely the real culprit. Then there is the speakers used that also could have a bad effect. I find high performance amps such as blameless amp to be excellent, they reproduce the incoming signal quite accurate, the problem usually is inferior source and speakers but I often see the amp is blamed whenever the sound is not to ones liking. Its not a graphic equiliser or sound enhancer, if that is what you need to get better sound then introduce this in your system or simply change the other links, that is if you have a good amp with good specs, usually that is a accurate amp.
 
Hi Guys

The first project is just the Blameless amp with a different current source for the diff amp and a buffer added between the VAS and CFP. The latter is beneficial where the former is not.
Otherwise, it is fun to play with lots of different circuit types.

Have fun
Kevin O'Connor

Hello Struth ,

Well it seems to be but may be not ....
How is the gm in contrast with the blameless regarding diff amp ?What value of slew rate you suppose here ? :)

The second important thing is the "unusual" current source for the vas & diff amp .Why do you think is done in that way ?
The response is the thermal drift ,In every amp the vas current has a high thermal drift, in this one no drift exists .The second thing related to the current source is that if one of the rails are missing, there will be no voltage at the output which ,at blamelless this is present .

Also have a look at the gm of the vas ...
These topologies has a big flaw regarding the psrr of the negative rail ,in this version this is eliminated .And if we take a closer look we can see a "current" feedback also :)

I don't know the thd20 for blameless can you provide it ?

So what it seems to be may be it isn't :)

PS :When I had tested/hear the amp(2 years) that version missed the psrr benefit also the temperature drift for the vas and also another gm for vas&diff amp .But a my_ref amplifier modded I like more than that version .

Thank you and greetings ,
 
Last edited:
I like the design you have and wonder why you didn't find the sound musical. Something to learn, something to fix perhaps. It looks like a good design, LTP with lots of emitter degeneration and current mirror to maintain good balance and improve the gain (which was reduced with degeneration). I'm not sure about the current source - it plays such an important role as it is used for the LTP and VAS, I wonder if it deserves a decoupling capacitor in there somewhere, perhaps on base of Q4 to ground. Some people find that a bootstrapped VAS load sounds better. I'm also not so sure about the cap multiplier (Q5) and it's possible impact on the sound.

The feedback scheme is the most interesting part. The benefit of a Singleton input is the 2nd harmonics giving the sound more richness. The current mirror you have may spoil this and prevent it - I would have to see some simulation.

The fourth design looks like something I have been considering - a marriage of Gaborbela's "20 year" amplifier and Rod's P3A output stage. I worry a bit about the bias stability though - the CFP output needs very accurate bias and the symmetric design is a poor topology in basic form for bias stability. I think this needs more work.
3-th
It took me a minute to realize this is a double differential input stage, two LTPs. I've never used this topology but my Bryston B60 amplifier uses it and it provides for a very clean undistorted sound with strong bass. If I understand properly, the LTP drives a folded cascade VAS ?
Hello Bigun ,for the 4-th version I was thinking from begining to insert a opamp only in dc to make dc servo(I will have the option to mount it or not ,I will see at the tests) ,cheaper and simple !
Anyway even with big variation of beta the dc output is stable ,under +/- 50mv .For the forth version I wanted a low number of active devices in therms of voltage gain .Here we have 3 .
And I wanted only one phase delay through the feedback .
Also an super high bandwidth in closed loop through negative reaction ~30Mhz .This should improve the harmonical response .

And yes ,for the 3-th version LTP drives a cascode VAS .Doing this I have less phase turning in open loop .Also by not decopling the cascode I have a smaller gain but a great benefit ,small compensation caps .


@Manso ,why a man climb on a mountain ? For the same goal like here ...for pleasure ! :)
 
Last edited:
Hi Guys

Dadod, I find that the schematics drawn in simulators are of the worst quality, as the ones above are. Some guys draw poor schemos even in Eagle or other dedicated schematic and PCB software.

It would be nice if lines were not drawn through through the part names and geographic designations.
It would be nice if lines were drawn across other components.
It would be nice of part values of one component were not overlayed by other part values for other parts, or by other parts.

I think clutter is universally disliked.

Apart from that, there are as many variations of the Lin/Blameless as there are builders.

Have fun
Kevin O'Connor
 
The 4th topology has splitted in two varations(no dc caps) 4a and 4b(cascode vas) .
Since I don't have time to try them I can only post them :D
I think I will make a common pcb for output stage and I will change only the gain pcb's through connectors to make diference on them .
And I will bias the output stage at 250-350 mA .Enough for 2 watts class A .
 

Attachments

  • joy_4a.JPG
    joy_4a.JPG
    60.9 KB · Views: 116
  • joy_4b.JPG
    joy_4b.JPG
    59.8 KB · Views: 108
Last edited:
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.