The MONGREL (supersym II)

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Ahhh - At it again , to refine what I know already works. The original supersym(s) are now a year old , they have not gone to the fiery pit :flame::flame: ... they even are very pleasing to listen to :D .

I just had to try for more stability , lower THD and the "MONGREL" is what I came up with. It is STILL the same core circuit-a single ended LPT with the Otala/Hitachi VAS but with a few improvements.

1. - 2 current sources and a cascode on the differential - much better linearity and PSRR in the "front end" , also.. ABSOLUTE signal symmetry at the input pair.

2. - lower differential current,slightly higher VAS current - better for running standard type 2 or 3 EF's with difficult loads and better linearity .

3. - Standard double EF but with type 3 current sources - carrier "suckout" ,less Xover distortion.

4. - better Vbe (complimentary)


Picture 1 is the circuit.

Picture 2 is the BEST :) FFT I have ever produced from a class B. The distortion is damn low , too. (pix 3)

On this set of simulations , I nearly was able to go to sub 10pF for Cdom and the simulation did not "choke". Soon I will tweak the open loop gain and global compensation but I suspect it is already close by the impressive stability I have already observed.
It seems Mr. Carlos (DX) saw the first one (supersym) , because now his amps are "locked and loaded" full of CCS's , this is how you really use them .... each of the 4 (CCS's) has a very valuable electrical as well as a noticeable sonic improvement. We don't waste valuable part around here !!

OS

Any comments or suggestions would be highly appreciated.
 

Attachments

  • mongrel.gif
    mongrel.gif
    94 KB · Views: 13,633
  • FFT 2_3 power.gif
    FFT 2_3 power.gif
    21.4 KB · Views: 12,062
  • full power thd.gif
    full power thd.gif
    12 KB · Views: 10,926
Last edited:
Ostripper: eventhough you still want to fiddle with the OLG and feedback, then the project is still very impressive.
I hope many people will buildt it just like Symasym.
One question: What are the ethomylogical content of the names of your amps? How did you come up with these names "Symasym", "Mongrel" ?
They sound very cool and unique.
 
Thank you , Hugh.

I did place the DC feedback components on the wrong side of the main feedback resistor. This is corrected in (attachment 1) . This brought max OLG to 65db and unity to a respectable 1.2mhz (attachment 2)... This amp would make a VERY good high output video amplifier. :):) I am happy. I suspect even with the changes 80v/uS slew would be a walk in the park for this circuit.

Do you feel the EFII with CCS is better sounding and lower distortion?

It is better sounding as well as more stable. The CCS's lower the high order harmonics of the Xover distortion by 6-10DB ! As a added bonus the OPS is more stable. This amp can almost work without Cdom .. WOW :eek: at 12pf it breaks out at 1.2mhz (unity gain point ?-attachment 3) and will run happy at 22p. Even while oscillating , it still stays electrically stable (it still amplifies) , this is unlike all my other simulations.

(Attachment 4-5) is the best symmetry I have seen from a single ended input stage.

By Nrik -One question: What are the ethomylogical content of the names of your amps? How did you come up with these names "Symasym", "Mongrel" ?

The symasym was mike brittner's creation , I upped the power and used newer parts to make the "supersym" . The "MONGREL" is :snoopy: ... part symasym ,part doug self type 3 OPS , inproved pioneer input stage (CCS's , not a stupid resistor), and economical output devices... A REAL MUTT.

So the "boys" can play I have attached "mongrel.ZIP" , LT simulation. Unzip "mongrel.asc" and "mongrel.txt" into same folder or desktop. NO models to hunt down , all you need is in the text. The simulation reads from the text. :)
OS
 

Attachments

  • osc_7pf.gif
    osc_7pf.gif
    27.4 KB · Views: 1,361
  • correct_OLG.gif
    correct_OLG.gif
    36.6 KB · Views: 2,791
  • correct_schema.gif
    correct_schema.gif
    62.5 KB · Views: 10,561
  • inputperfect.gif
    inputperfect.gif
    36.9 KB · Views: 1,325
  • vas_OPS.gif
    vas_OPS.gif
    36.8 KB · Views: 1,051
  • mongrel.zip
    6.7 KB · Views: 701
This looks promising as a high power version of Symasym, which is a great sounding amplifier (I'm using 2 stereo ones!) :)

I look forward to seeing a PCB and mabye European equivalent parts. Will this amplifier be compatible with Symasym in terms of phase response, to be used in a bi/tri amplified active setup with the original?
 
I look forward to seeing a PCB and mabye European equivalent parts.

I already have most of this amp and some of the improvements up and running. (attachment 1) I added the current sources with 1 trimmer each to adjust the standing current between 25-60ma. at 40+ma this amp sounds nearly identical to the 200 watt class A $7000 genesis amp which I a/b 'ed it to.

phase wise this amp is nearly identical to the original "symasym" the cascode has 0 effect on phase and the CCS's at the OPS have no effect either.

As far as parts...
Input - ksa992/ksc1845 or 2n5551/5401 or any Vceo 160-< / Hfe 60 - 200

Vas - mje340/350 mediocre - 2sa1381 / 2sc3503 superior (video amp - Hfe 100 +)

drivers - mje15033/32 or 2sa1837 / 2sc4793

Any outputs MJ or MJW or NJW (on semi) Hfe 25 - 50 (250v+ Vceo)

PS voltage +-55 to +- 85VDC . on shutdown amp below still works down to +- 10v !! No turn on "thump" as well.

In both the real amp and the sim I have used all these with only minor adjustments. Very wide parts margin for this circuit.
OS
 

Attachments

  • supersym.jpg
    supersym.jpg
    181.9 KB · Views: 3,067
I got it . No cutoff like the krill or super class A. A Little positive feedback into my type 3 CCS's and :) good to go.. !

Look at the plots (1,2) . One issue , at high power , OPS become more typical Class B. At 10-20w neither output cuts off and the FFT shows .0008 or less and almost no 3rd or 5th harmonic.

Circuit is the last attachment.
OS
 

Attachments

  • mongrelnocutoff.gif
    mongrelnocutoff.gif
    105.8 KB · Views: 2,123
  • nocutoff.gif
    nocutoff.gif
    12.4 KB · Views: 714
  • nocutoffoverview.gif
    nocutoffoverview.gif
    23.9 KB · Views: 2,574
I did a .step command with input as the variable and this shows that it is normal for the type 3 to "resist" cutoff. Hence the low distortion components. I only augmented this effect with a little positive feedback.

This is shown in attachment 1 at various output levels.

OS
 

Attachments

  • steppedfft.gif
    steppedfft.gif
    26.9 KB · Views: 2,088
You could bypass R47 and R48 with small (below 100pF) caps to speed up the output stage, if you're adventurous.

I revamped the simulation commands and modified the circuit a little to decrease simulation noise floor. The high noise floor was exaggerating THD figures. The spectrum also changed, I don't know what the cause of this is, however I have noticed that high simulation noise floors in LTSpice can create a phantom dominant 2nd harmonic.

I replaced the signal caps with voltage sources, since caps in the signal path create a slanting FFT noise floor due to imperfect operating point solution.

- keantoken
 

Attachments

  • Mongrel_Ksim.zip
    6.8 KB · Views: 562
Thank you , kean. I now know more. :) what a nice low noise floor !

Wow , this amp has ridiculous low 5th and 7th :)devilr: the evil harmonics :devilr:) Too bad everyone is too stuckup to build this amp. They would rather build schoolboy 30 year old designs and discuss blackgate caps and other esoteric B_S. :mad:

A shame , it is still simple , 25 components and a cheap $9 OPS (njw0281/0302) which can overwhelm any speaker out there.

Edit : 20k is impressive , too ... for a simple amp.

OS
 

Attachments

  • low.gif
    low.gif
    41.9 KB · Views: 2,788
  • low20k.gif
    low20k.gif
    38.6 KB · Views: 2,747
Last edited:
it is still simple , 25 components

It could be said that simplicity is in the eye of the beholder. Personally I try not to shy away from increased part count. In one way DIY is not a good way to "push the edge" of audio electronics; not enough money and not enough time.

I notice your amp has 63db of open loop gain with a corner frequency of 600Hz; ouch. If you play with the compensation a bit, I believe you will see improvement in the FFT.

If you're adventurous, I'll suggest what I would do. The supremely enlightened Michael Bittner granted us his favor and revealed unto us Symasym some time ago. I wonder if his unconventional use of compensation didn't go unnoticed. Most people just add a cap and then are done with it. The high-value caps I use in my headphone amp would send it into convulsions if I didn't decrease the Q factor with series resistance; but when I do, look at the result. I think there is much to be gained by experimenting with compensation.

I believe this type of compensation is also more stable into capacitive loads; it reacts to current draw instead of voltage output. The drawback of the conventional methods is that the compensation doesn't react if the output voltage doesn't change - hence indomitable instability into capacitive loads. Voltage-type compensation is counter-intuitive for voltage drive amps isn't it? Voltage drive amps are immune to inductive loading, little need to protect against it.

Notice the dip in phase required to produce the steep rolloff. Savor the economy.

The spectrum benefits much as well. 11th harmonic has been reduced by 22db! Before and after OLG and FFT plots.

- keantoken
 

Attachments

  • Mongrel_OLG_Altercomp.PNG
    Mongrel_OLG_Altercomp.PNG
    15.6 KB · Views: 2,553
  • Mongrel_FFT.PNG
    Mongrel_FFT.PNG
    30.3 KB · Views: 2,615
  • Mongrel_OLG.PNG
    Mongrel_OLG.PNG
    15 KB · Views: 2,661
  • Mongrel_FFT_Altercomp.PNG
    Mongrel_FFT_Altercomp.PNG
    30.5 KB · Views: 392
Here is the simulation file.

If you look at the placement of C4 in post 9, this cap is injecting switching trash from the output stage directly into the VAS input, which the LTP has to shunt away. This is another reason to avoid that type of compensation.

- keantoken
 

Attachments

  • Mongrel_Ksim.zip
    10.4 KB · Views: 291
I just noticed, in your posts the higher order harmonics are much lower than mine. When I first opened your amp in LTSpice it was anything but stable. I had to increase compensation drastically to get phase margin within bounds. I'm not sure if this is normal, but it explains your better FFT results.

Also, don't forget the 5pF cap in the feedback network! The transistor parasitics here can be seen as a capacitor from Q4's base to ground. This increases VHF response and instability, causing a blip in the VHF if you look at the OLG response (set your AC analysis farther than 10MHz!). The 5p cap swamps the parasitics out and increases stability.

- keantoken
 
Last edited:
I just noticed, in your posts the higher order harmonics are much lower than mine.

I destabilized this amp on purpose with the sub 22pF Cdoms. You mean a 5Pf in parallel with the main GFB resistor ? They call that lead compensation , I am not using that in the working model and have no oscillations. Still , to be on the safe side , why not ?

I had to increase compensation drastically to get phase margin within bounds.

You mean the 47Pf Cdom ? 56pF takes the overshoot out of the squarewave plots and set unity gain at 800khz.
Also , I got -102 degree margin at unity. would low "90 ish" be better ?


OS
 
Last edited:
If you're using small Cdoms, I'm sure the 5p cap couldn't hurt. I just cringe when I see that blip. It gets in the way when I'm trying to adjust compensation and stuff, just easier to have it out of the way.

There seem to be a lot of amps that don't adhere to the -100 degree phase margin standard, so it can't be too bad.

- keantoken
 
Last edited:
Keen , submit your "version" (another ksim) of the mongrel with the changes and I will check it out.
This time , don't forget the output inductor ;) .

On your second plot above "the dip" was a cause for concern to Andy C , he called it a "2-pole response" and said it might
cause problems ?????


ThanX , OS
 
Last edited:
I think Andy is right about it being two-pole response, but it's not done in the same way it's done in this article:

http://www.analogzone.com/col_0628.pdf

I haven't had problems with my headphone amp, though admittedly power amps are different.

The output inductor is present in the simulation, I just use a different symbol than you do. I take the output FFT from the before the inductor, not after.

- keantoken
 
Wow , this amp has ridiculous low 5th and 7th :)devilr: the evil harmonics :devilr:) Too bad everyone is too stuckup to build this amp. They would rather build schoolboy 30 year old designs and discuss blackgate caps and other esoteric B_S. :mad:

This is actually "my sort of amp"! Having built the Symasym (twice) I've been very impressed with it's neutral, clean performance. This amp seems like a similar design, excellent technical specs but more juice! Only thing is, I don't need one right now :eek:

I asked about compatibility with Symasym as I'm running a 3-way tri-amplified setup and in the back of my mind had sort of planned to, one day, integrate the 2 Symasym and 1 P101 into a 6-channel unit, for tidyness. However, it seems as though this design might be more compatible with the Syms while providing that extra current for the bass section, OTOH the P101 also seems quite excellent in this regard so I don't really know! I measured it as having the same phase as the Syms at least.

I think it's great work anyhow, I certainly don't mind a more complex design if the parts go toward creating better performance :)
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.